Solvent Extraction Principles and Practice PDF

Solvent Extraction
and Praciice
Second Edition, Revised and Expanded
edited by
Jan Rydberg
ChaZmers University of TechnoZogy
Goteborg, Sweden
Michael Cox
University of Hertfordshire
Hatjield, Hertfordshire, United Kingdom
Claude Musikas
Commissariat a I’ Energie Atomique
Paris, France
Gregory R. Choppin
Florida State University
Tallahassee, Florida, U.S.A.
MARCEL
MARCELDEKKER,
INC. NEWYORK * BASEL
DEKKER
Previous edition published as Principles and Practices of Solvent Extraction
(Rydberg, J. et al., Eds.), Marcel Dekker, 1992.
Although great care has been taken to provide accurate and current information,
neither the author(s) nor the publisher, nor anyone else associated with this pub-
lication, shall be liable for any loss, damage, or liability directly or indirectly caused
or alleged to be caused by this book. The material contained herein is not intended to
provide specific advice or recommendations for any specific situation.
Trademark notice: Product or corporate names may be trademarks or registered
trademarks and are used only for identification and explanation without intent to
infringe.
Library of Congress Cataloging-in-Publication Data
A catalog record for this book is available from the Library of Congress.
ISBN: 0-8247-5063-2
This book is printed on acid-free paper.
Headquarters
Marcel Dekker, Inc., 270 Madison Avenue, New York, NY 10016, U.S.A.
tel: 212-696-9000; fax: 212-685-4540
Distribution and Customer Service
Marcel Dekker, Inc., Cimarron Road, Monticello, New York 12701, U.S.A.
tel: 800-228-1160; fax: 845-796-1772
Eastern Hemisphere Distribution
Marcel Dekker AG, Hutgasse 4, Postfach 812, CH-4001 Basel, Switzerland
tel: 41-61-260-6300; fax: 41-61-260-6333
World Wide Web
http://www.dekker.com
The publisher offers discounts on this book when ordered in bulk quantities. For
more information, write to Special Sales/Professional Marketing at the headquarters
address above.
Copyright # 2004 by Marcel Dekker, Inc. All Rights Reserved.
Neither this book nor any part may be reproduced or transmitted in any form or by
any means, electronic or mechanical, including photocopying, microfilming, and
recording, or by any information storage and retrieval system, without permission in
writing from the publisher.
Current printing (last digit):
10 9 8 7 6 5 4 3 2 1
PRINTED IN THE UNITED STATES OF AMERICA
Preface
The partition of a solute between two immiscible solvents is a major tech-

nique in separation, both in the laboratory and in industry. This technique has
been designated liquid–liquid distribution by the International Union of Pure
and Applied Chemistry (IUPAC), but it is more commonly called solvent
extraction, the term we use. Solvent extraction has been developed in a broad
field of applications: (1) for studying inorganic and organic complex equilibria,
either for fundamental understanding of solution equilibria and kinetics or
for developing selective separation schemes; (2) for separations in analytical
chemistry; (3) for large-scale industrial separation processes in the inorganic,
organic, pharmaceutical, and biochemical industries; and (4) for industrial
waste treatment. It also serves as a base for other analytical techniques, such
as liquid partition chromatography and ion-selective electrodes.
The many uses of solvent extraction make the subject important for
university students of chemistry and chemical technology. Some universities
offer special courses in solvent extraction, whereas others include it as a
minor topic in more comprehensive courses. Pilot-scale experiments on
solvent extraction are common in chemical engineering curricula. Because of
the breadth of the subject, the treatment in such courses is often scanty, and
a satisfactory text is difficult to find in a form suitable for use directly with
students.
To meet this demand, we felt it desirable to develop a simple text
suitable not only for students but also for chemists in the field. However, no
single scientist or engineer can be an expert in all parts of the field. There-
fore, it seemed best to develop the book as a joint project among many
iii
iv Preface
expert authors, each of whom has decades of experience in research,

teaching, or industrial development. The product is a truly international
book at a high scientific and technical level. The task of making it homo-
geneous, consistent, and pedagogical has been undertaken by the editors.
The book is directed to third- to fourth-year undergraduate and post-
graduate chemistry and chemical engineering students, as well as to
researchers and developers in the chemical industry. The book is also intended
for chemical engineers in industry who either have not kept up with modern
developments or are considering use of this technique, as well as engineers
who already are using this technique but desire to understand it better.
Furthermore, the book should be useful to researchers in solvent extraction
who wish to learn about its applications in areas other than their own.
The first edition of this book was published more than 10 years ago.
Since then four large international conferences have summarized the latest
developments in the field of solvent extraction. We have tried to incorporate
the latest achievements of a fundamental type in this text, for instance, new
types of solvent matrices and industrial applications, without burdening the
text with new organic extractants or solvent combinations, which appear
almost daily in the specialist literature. As far as possible, we have also
rationalized the earlier text, concentrating on or removing outdated infor-
mation. In doing so we have added new contributors. We therefore hope
that this text will be met with the same enthusiasm as the first edition.
After an introduction (Chapter 1), the following five chapters (Chapters
2–6) present the physical principles and formal expressions used in solvent
extraction. They are followed by eight chapters (Chapters 7–14) of various
industrial applications and two concluding chapters (Chapters 15 and 16)
indicating the research frontiers and future developments in technology.
We, the editors, want to express our gratitude to the contributors, who
made this book possible through their helpful suggestions and extensive
efforts. Prof. Manuel Aguilar, Prof. Eckhart Blass, Prof. José Luis Cortina,
Dr. Pier Danesi, Prof. Ingmar Grenthe, Prof. Jan-Olov Liljenzin, Dr. Marian
Czerwiński, Prof. Philip Lloyd, Prof. Yizhak Marcus, Prof. Jerzy Narbutt,
Dr. Susana Pérez de Ortiz, Dr. Hans Reinhardt, Dr. Gordon Ritcey,
Prof. Ana Marı́a Sastre, Dr. Wallace Schulz, the late Prof. Tatsuya Sekine,
Prof. David Stuckey, Dr. Hans Wanner, and Prof. Ronald Wennersten
have worked hard and successfully on their own chapters and have given
us much valuable guidance.
Jan Rydberg
Michael Cox
Claude Musikas
Gregory R. Choppin
Contents
Preface iii
Contributors vii
1. Introduction to Solvent Extraction 1

Michael Cox and Jan Rydberg
2. Principles of Solubility and Solutions 27

Yizhak Marcus
3. Complexation of Metal Ions 81

Gregory R. Choppin
4. Solvent Extraction Equilibria 109

Jan Rydberg, Gregory R. Choppin, Claude Musikas,
and Tatsuya Sekine
5. Solvent Extraction Kinetics 203

Pier Roberto Danesi
6. Ionic Strength Corrections 253

Ingmar Grenthe and Hans Wanner
7. Development of Industrial Solvent Extraction Processes 277

Gordon M. Ritcey
v
vi Contents
8. Principles of Industrial Solvent Extraction 339

Philip J. D. Lloyd
9. Engineering Design and Calculation of Extractors for

Liquid–Liquid Systems 367
Eckhart F. Blass
10. Extraction of Organic Compounds 415

Ronald Wennersten
11. Solvent Extraction in Hydrometallurgy 455

Michael Cox
12. Solvent Extraction in Nuclear Science and Technology 507

Claude Musikas, Wallace W. Schulz,
and Jan-Olov Liljenzin
13. Analytical Applications of Solvent Extraction 559

Manuel Aguilar, Jose´ Luis Cortina,
and Ana Marıá Sastre
14. The Use of Solvent Extraction in the Recovery of Waste 609

Michael Cox and Hans Reinhardt
15. Recent Advances in Solvent Extraction Processes 651

Susana Pe´rez de Ortiz and David Stuckey
16. Computational Chemistry in Modeling Solvent

Extraction of Metal Ions 679
Jerzy Narbutt and Marian Czerwiński
Appendix 715
Index of Compounds 725
Subject Index 741
Contributors
Manuel Aguilar Department of Chemical Engineering, Universitat Poli-

tècnica de Catalunya, Barcelona, Spain
Eckhart F. Blass* Professor Emeritus, Lehrstuhl A für Verfahrenstechnik,

Technische Universität Munich, Munich, Germany
Gregory R. Choppin* Department of Chemistry, Florida State University,

Tallahassee, Florida, U.S.A.
José Luis Cortina Department of Chemical Engineering, Universitat

Politècnica de Catalunya, Barcelona, Spain
Michael Cox* Department of Chemical Sciences, University of Hertford-

shire, Hatfield, Hertfordshire, United Kingdom
Marian Czerwiński Zawiercie University of Administration and Manage-

ment, Zawiercie, Poland
Pier Roberto Danesi Seibersdorf Laboratories, International Atomic

Energy Agency, Vienna, Austria
*Retired.
vii
viii Contributors
Ingmar Grenthe* Professor Emeritus, Department of Inorganic Chemistry,

Royal Institute of Technology, Stockholm, Sweden
Jan-Olov Liljenzin* Professor Emeritus, Department of Nuclear Chemistry,

Chalmers University of Technology, Göteborg, Sweden
Philip J. D. Lloyd Energy Research Institute, University of Cape Town,

Rondebosch, South Africa
Yizhak Marcus* Professor Emeritus, Department of Inorganic and Ana-

lytical Chemistry, The Hebrew University of Jerusalem, Jerusalem, Israel
Claude Musikas* Département des Procedes de Retraitement, Commis-

sariat à l’Energie Atomique, Paris, France
Jerzy Narbutt Institute of Nuclear Chemistry and Technology, Warsaw,

Poland
Susana Pérez de Ortiz Department of Chemical Engineering, Imperial

College, London, United Kingdom
Hans Reinhardt MEAB Metallextraktion AB, Göteborg, Sweden
Gordon M. Ritcey Gordon M. Ritcey & Associates, Inc., Nepean, Ontario,

Canada
Jan Rydberg* Department of Nuclear Chemistry, Chalmers University of

Technology, Göteborg, Sweden
Ana Marı́a Sastre Department of Chemical Engineering, Universitat

Politècnica de Catalunya, Barcelona, Spain
Wallace W. Schulz Consultant, Albuquerque, New Mexico, U.S.A.
Tatsuya Sekine{ Department of Chemistry, Science University of Tokyo,

Tokyo, Japan
David Stuckey Department of Chemical Engineering, Imperial College,

London, United Kingdom
*Retired.
{
Deceased.
Contributors ix
Hans Wanner Swiss Federal Nuclear Safety Inspectorate, Villingen-HSK,

Switzerland
Ronald Wennersten Department of Chemical Engineering, Royal Institute

of Technology, Stockholm, Sweden
1
Introduction to Solvent Extraction
MICHAEL COX* University of Hertfordshire, Hatfield, Hertfordshire,
United Kingdom
JAN RYDBERG* Chalmers University of Technology, Göteborg, Sweden
1.1 WHAT IS SOLVENT EXTRACTION?

The term solvent extraction† refers to the distribution of a solute between two
immiscible liquid phases in contact with each other, i.e., a two-phase distribu-
tion of a solute. It can be described as a technique, resting on a strong scientific
foundation. Scientists and engineers are concerned with the extent and dynamics
of the distribution of different solutes—organic or inorganic—and its use scien-
tifically and industrially for separation of solute mixtures.
The principle of solvent extraction is illustrated in Fig. 1.1. The vessel (a
separatory funnel) contains two layers of liquids, one that is generally water
(Saq) and the other generally an organic solvent (Sorg). In the example shown, the
organic solvent is lighter (i.e., has a lower density) than water, but the opposite
situation is also possible. The solute A, which initially is dissolved in only one
of the two liquids, eventually distributes between the two phases. When this
distribution reaches equilibrium, the solute is at concentration [A]aq in the aque-
ous layer and at concentration [A]org in the organic layer. The distribution ratio
of the solute
D = [A]org /[A]aq (1.1)
*Retired.
†The International Union of Pure and Applied Chemistry (IUPAC) recommends the use of the
term liquid-liquid distribution. However, more traditionally the term solvent extraction (sometimes
abbreviated SX) is used in this book.
1
2 Cox and Rydberg
Fig. 1.1 A schematic representation of solvent extraction (liquid-liquid distribution).

A solute A is distributed between the upper layer, for example an organic solvent, and
the lower layer, an aqueous phase.
Introduction to Solvent Extraction 3
is defined as the ratio of “the total analytical concentration of the substance in

the organic phase to its total analytical concentration in the aqueous phase,
usually measured at equilibrium” [1], irrespective of whether the organic phase
is the lighter or heavier one. If a second solute B is present, the distribution
ratio for the various solutes are indicated by DA, DB, etc. If DB is different from
DA, A and B can be separated from each other by (single or multistage) solvent
extraction. D is also called the distribution coefficient or distribution factor; we
here prefer the expression distribution ratio.
For practical purposes, as in industrial applications, it is often more popu-
lar to use the percentage extraction %E (sometimes named the extraction fac-
tor), which is given by
% E = 100 D /(1 + D) (1.2)
where D is the distribution ratio of the solute (or desired component). For D =
1, the solute is evenly distributed between the two phases. A requirement for
practical use of solvent extraction is that a reasonable fraction (percentage) of
the desired component is extracted in a single operation (or stage).
Solvent extraction is used in numerous chemical industries to produce
pure chemical compounds ranging from pharmaceuticals and biomedicals to
heavy organics and metals, in analytical chemistry and in environmental waste
purification. The scientific explanation of the distribution ratios observed is
based on the fundamental physical chemistry of solute–solvent interaction,
activity factors of the solutes in the pure phases, aqueous complexation, and
complex-adduct interactions. Most university training provides only elemen-
tary knowledge about these fields, which is unsatisfactory from a funda-
mental chemical standpoint, as well as for industrial development and for pro-
tection of environmental systems. Solvent extraction uses are important in
organic, inorganic, and physical chemistry, and in chemical engineering, theoret-
ical as well as practical; in this book we try to cover most of these important
fields.
None of the authors of this book is an expert in all the aspects of solvent
extraction, nor do we believe that any of our readers will try to become one.
This book is, therefore, written by authors from various disciplines of chemistry
and by chemical engineers. The “scientific level” of the text only requires basic
chemistry training, but not on a Ph.D. level, though the text may be quite useful
for extra reading even at that level. The text is divided in two parts. The first
part covers the fundamental chemistry of the solvent extraction process and the
second part the techniques for its use in industry with a large number of applica-
tions. In this introductory chapter we try to put solvent extraction in its chemical
context, historical as well as modern. The last two chapters describe the most
recent applications and theoretical developments.
4 Cox and Rydberg
1.2 SOLVENT EXTRACTION IS A FUNDAMENTAL

SEPARATION PROCESS
Under normal conditions, matter can appear in three forms of aggregation: solid,
liquid, and gas. These forms or physical states are consequences of various
interactions between the atomic or molecular species. The interactions are gov-
erned by internal chemical properties (various types of bonding) and external
physical properties (temperature and pressure). Most small molecules can be
transformed between these states (e.g., H2O into ice, water, and steam) by a
moderate change of temperature and/or pressure. Between these physical states—
or phases—there is a sharp boundary (phase boundary), which makes it possi-
ble to separate the phases—for example, ice may be removed from water by
filtration. The most fundamental of chemical properties is the ability to undergo
such phase transformations, the use of which allows the simplest method for
isolation of pure compounds from natural materials.
In a gas mixture such as the earth’s atmosphere, the ratio of oxygen to
nitrogen decreases slightly with atmospheric height because of the greater gravi-
tational attraction of oxygen. However, the gravitational field of the earth is not
enough for efficient separation of these gases, which, however, can be separated
by ultracentrifugation and by diffusion techniques. In crushed iron ore it is pos-
sible to separate the magnetite crystals Fe3O4 from the silicate gangue material
by physical selection under a microscope or by a magnetic field. In chemical
engineering such separation techniques are referred to as nonequilibrium pro-
cesses. Other common nonequilibrium processes are electrolysis, electrophore-
sis, and filtration.
In contrast to these we have the equilibrium processes of sublimation,
absorption, dissolution, precipitation, evaporation, and condensation, through
which the physical states of solid, liquid, and gas are connected. For example,
the common crystallization of salts from sea water involves all three phases.
Distillation, which is essential for producing organic solvents, is a two-step
evaporation (liquid ⇒ gas) condensation (gas ⇒ liquid) process.
In Fig. 1.2, phase transformations are put into their context of physical
processes used for separation of mixtures of chemical compounds. However,
the figure has been drawn asymmetrically in that two liquids (I and II) are
indicated. Most people are familiar with several organic liquids, like kerosene,
ether, benzene, etc., that are only partially miscible with water. This lack of
miscibility allows an equilibrium between two liquids that are separated from
each other by a common phase boundary. Thus the conventional physical system
of three phases (gas, liquid, and solid, counting all solid phases as one), which
ordinarily are available to all chemists, is expanded to four phases when two
immiscible liquids are involved. This can be of great advantage, as will be seen
when reading this book.
Fig. 1.2 Model of a four-phase system consisting of two liquid phases (e.g., an aque-
ous and an organic phase) in equilibrium with a gas phase and a solid phase.
Solutes have differing solubilities in different liquids due to variations in

the strength of the interaction of solute molecules with those of the solvent.
Thus, in a system of two immiscible or only partially miscible solvents, different
solutes become unevenly distributed between the two solvent phases, and as
noted earlier, this is the basis for the solvent extraction technique. In this con-
text, “solvent” almost invariably means “organic solvent.” This uneven distribu-
tion is illustrated in Fig. 1.3, which shows the extractability into a kerosene
solution of the different metals that appear when stainless steel is dissolved in
aqueous acid chloride solution. The metals Mo, Zn, and Fe(III) are easily ex-
tracted into the organic solvent mixture at low chloride ion concentration, and
Cu, Co, Fe(II), and Mn at intermediate concentration, while even at the highest
chloride concentration in the system, Ni and Cr are poorly extracted. This is
used industrially for separating the metals in super-alloy scrap in order to re-
cover the most valuable ones.
The three main separation processes between solid, gas, and liquid have
long been known, while solvent extraction is a relatively new separation tech-
nique, as is described in the brief historical review in next two sections. Never-
theless, because all solutes (organic as well as inorganic) can be made more or
less soluble in aqueous and organic phases, the number of applications of sol-
vent extraction is almost limitless. Since large-scale industrial solvent extraction
is a continuous process (in contrast to laboratory batch processes) and can be
6 Cox and Rydberg
Fig. 1.3 Percentage of extraction of various metals from a solution of dissolved stain-
less steel scrap, at 40oC. The organic phase is 25% tertiary amine (Alamine 336), 15%
dodecanol (Loral C12) and 60% kerosene (Nysolvin 75A). The aqueous phase is a CaCl2
solution at pH 2.
made more selective than the conventional gas–liquid–solid separation tech-

niques, it offers numerous industrial possibilities to achieve desired separation
efficiently and economically.
1.3 EARLY STEPS TOWARD THE USE

OF SOLVENT EXTRACTION
Around 500 B.C.E. the Greek philosophers recognized four elements: earth, wa-
ter, phlogiston (⬃air), and fire. This view harmonizes with the present concept
of three physical states of aggregation (solid, liquid, gas) and heat. Aristotle
(⬃350 B.C.E.) emphasized that these “elements” were not eternal, but could be
changed into each other. Five thousand years earlier, scientists already had
found that when certain green minerals were heated in a coal fire, metallic
copper was obtained. In Aristotle’s time it was known how to produce metals
such as copper, gold, tin, lead, silver, iron, mercury, and arsenic. Even earlier,
by transmuting certain “earths” with fire, ceramics and glasses had been pro-
duced. Many of these arts were probably developed by the Egyptians, the first
true chemists. The word “alchemy” is derived from Arabic and Greek and is
supposed to mean “art of transmutations as practiced by the Egyptians.” Fruit

juices were fermented, oils and fats were squeezed out of vegetables and animal
parts, and purified by digestion with earths, bones, etc. Crucibles, retorts, and
even distillation equipment seem to have been in use; see Fig. 1.4. We must
think of these early alchemists as endlessly mixing, heating, boiling, digesting,
cooling, etc., everything they could collect from nature. The purpose of these
transmutations varied: for lamps, weapons, pigments, perfumes, and poisons;
for cosmetics and medicines to prevent aging and to prolong life (elixir vitae);
for tanning chemicals, soap, anesthetics, and also for making gold. In fact, they
did succeed in producing gold-like metals (e.g., brass). For example, it is known
that they heated odorous leaves in alkaline water with fats and oils, so that
ointments and perfumes could be enriched in the cooled solidified fat, and that
these products were extensively used in the ancient courts, and perhaps even
among the general population. If this is considered to be solvent extraction, it
is truly one of the oldest chemical techniques. It is also likely that the Egyptians
knew how to distill alcohol, long before it is described by the Arab Kautilya
and the Greek Aristotle about 300 B.C.E. (See also Ref. [2].)
This experimentation more or less came to a halt during the Greek civili-
zation. The Greeks were philosophers, and not so much experimentalists; Aris-
totle was a philosopher and a systematizer (systems technician, in modern lan-
guage), not an experimentalist. The Greeks were followed by the Romans who
were administrators, and by the Christians who considered alchemy to be un-
godly. Although alchemy was practiced during subsequent centuries, particu-
larly in the Arabian world, it became suspect and was banned by many rulers
(though encouraged in secrecy by others). About 500 years ago, alchemy was
rather openly revived in Europe, particularly at local courts, and progressed
within a few centuries into modern science.
Digestion of various earths (or digested earths) with alcohol produces
many organic solvents (ether, acetone, etc.). These solvents could be obtained
in pure form through distillation. Such organic solvents could have been pro-
duced many thousands years ago, because of the obvious knowledge of distilla-
tion [2]. However, 200 years ago only a few pure solvents seem to be known:
besides the natural water and oils (and kerosene) only alcohol, ether, and “ethe-
ric oils” were acknowledged. It is difficult to trace organic solvents far back in
history. The reason may simply be that organic compounds obtained by distilla-
tion of mixtures of natural products were found to be rather uninteresting (ex-
cept for alcohol), because at that time they seemed to have very little practical
value, and they certainly could not be used to produce gold. Because solvent
extraction requires pure organic solvents of limited aqueous miscibility, it is
then understandable why solvent extraction historically is considered (perhaps
falsely) to be a newcomer among chemical separation methods.
Rydberg_5063-2_Ch01_R3_02-09-04 11:07:00
Fig. 1.4 (a) Equipment used by alchemists, according to an Alexandrian manuscript

(about 300 B.C.E. to A.D. 300). (b) Apparatus for fractionated distillation. Front page of
Philosophi ac Alchimistae Maximi by Johannes Greininger, Strasbourg, 1531. The origi-
nal work is ascribed to the great eighth-century Arab alchemist, Abu Musa Jabir.
8
1.4 SOLVENT EXTRACTION IN THE DEVELOPMENT

OF MODERN CHEMISTRY
The early nineteenth century saw an enormous development in chemistry. Not
only were numerous inorganic and organic compounds synthesized, they were
also characterized through skillful analyses. Pure compounds, absolutely essen-
tial for the analyst, were identified from measurements of weights and volumes,
melting and boiling points, densities, refraction, and crystal structure. It was
then found that many inorganic salts, such as chlorides of iron, mercury, and
gold, dissolved in and could be recrystallized from alcohols and ethers.
1.4.1 The Distribution Ratio

and the Extraction Constant
Inorganic extractions seem to have come into practical use without any great
notice. Although Peligot in 1842 reported that uranyl nitrate could be recrystal-
lized from ether, he never mentioned extraction of this salt from aqueous solu-
tions. In textbooks after 1870, however, it is stated briefly that “ether can even
withdraw sublimate (HgCl2) from aqueous solution.” It was also reported, for
example, that cobalt thiocyanate is weakly extracted by ether, better by amyl
alcohol, and even better by a mixture of both.
The practical use of solvent extraction for separation and purification of
different substances led Berthelot and Jungfleisch to investigate the distribution
of a large number of organic and inorganic compounds between ether or carbon
disulfide and water. In 1872, they introduced the term distribution factor (“coef-
ficient de partage”) [Eq. (1.1)] to describe how the distribution of a solute de-
pended on its concentration in the organic and aqueous phases. However, the
regularities observed were not general.
About 20 years later, Nernst realized that it was necessary to take into
account the different reactions of the solute in each phase, such as dimerization
in the organic phase and dissociation in the aqueous phase. Thus the distribution
of benzoic acid (HBz) between the organic (benzene) phase and water could be
written
Organic (benzene) phase: ← H Bz
2 HBz → 2 2
↓↑
← +
Aqueous phase: HBz → H + Bz −
In 1891, Nernst realized that only if the solute has the same molecular weight
in the organic phase as in the aqueous phase, the distribution ratio would be
independent of the concentration of the solute, or distribuend. He proposed the
simple relation:
10 Cox and Rydberg
K D = [HBz]org /[HBz]aq (1.3)

and demonstrated the validity of this principle for the distribution of a large
number of organic and inorganic compounds between organic solvents and wa-
ter (the Nernst distribution law). KD is referred to as the distribution constant.
[Note that KD is sometimes termed partition coefficient (or constant), and then
abbreviated P.]
The primary parameter in solvent extraction is the measured distribution
ratio, where it is up to the writer to define what is being measured, indicating
this by an appropriate index. In the Nernst distribution experiment described
earlier, the analytically measured concentration of benzoic acid is in the aqueous
phase [Bz]aq,tot = [HBz]aq + [Bz−]aq, and in the organic phase [Bz]org,tot = [HBz]org +
[H2Bz2]org. Thus the measured distribution ratio, abbreviated DBz, becomes
[Bz]org,tot [HBz]org + 2[H 2 Bz 2 ]org
DBz = = (1.4)
[Bz]aq,tot [HBz]aq + [Bz − ]aq
From this it follows that the analytically measured distribution ratio is a constant
only in systems that contain a single molecular species of the solute (or desired
component).
Such considerations were extended to metal complexes in 1902 by Morse,
who studied the distribution of divalent mercury between toluene and water at
various Hg2+ and Cl− concentrations. By taking complex formation in the aque-
ous phase into consideration Morse could determine the formation constants of
HgCl+ and HgCl2 from distribution measurements, as well as the distribution
constant of the neutral complex HgCl2. The overall extraction reaction can be
written
← HgCl (org)
Hg 2 + (aq) + 2 Cl − (aq) → (1.5a )
2
for which one defines an extraction constant, Kex

[HgCl2 ]org
K ex = (1.5b)
[Hg 2+ ]aq [Cl− ]aq
2
Although Kex neither describes the intermediate reaction steps (see Chapters 3
and 4) or kinetics of the reaction (Chapter 5), nor explains why or to what
extent HgCl2 dissolves in the organic solvent (Chapter 2), the extraction reaction
is a useful concept in applied solvent extraction, and Kex values are commonly
tabulated in reference works [3–4].
1.4.2 Extractants
During the years 1900 to 1940, solvent extraction was mainly used by the or-
ganic chemists for separating organic substances. Since in these systems, the
solute (or desired component) often exists in only one single molecular form,
such systems are referred to as nonreactive extraction systems; here the distribu-
tion ratio equals the distribution constant.
However, it was also discovered that many organic substances, mainly
weak acids, could complex metals in the aqueous phase to form a complex
soluble in organic solvents. A typical reaction can be written
← MA (org) + z H + (aq)
M z + (aq) + z HA(aq or org) → (1.6)
z
which indicates that the organic acid HA may be taken from the aqueous or the
organic phase. This is an example of reactive extraction. It became a tool for
the analytical chemist, when the extracted metal complex showed a specific
color that could be identified spectrometrically. The reagent responsible for
forming the extractable complex is termed the extractant.
The industrial use of solvent extraction of inorganic compounds grew out
of the analytical work. As both areas, analytical as well as industrial, needed
both better extractants and an understanding of the reaction steps in the solutions
in order to optimize the applications, theoretical interpretations of the molecu-
lar reactions in the solutions became a necessity, as will be described in later
chapters.
The increased use of computer graphics for modeling molecular structures
and chemical reactions has opened a path for the synthesis of tailor-made ex-
tractants. Thus the future promises new varieties of extractants with highly se-
lective properties for the desired process.
1.4.3 Industrial Use of Solvent Extraction

A surge in interest in solvent extraction occurred in the decades of the 1940s and
1950s initiated by its application for uranium production and for reprocessing of
irradiated nuclear materials in the U.S. Manhattan Project. The first large-scale
industrial solvent extraction plant for metals purification was built in 1942 by
Mallinckrodt Chemical Co., St. Louis, for the production of ton amounts of
uranium by selective extraction of uranyl nitrate by ether from aqueous solu-
tions. The high degree of purity (>99.9%) required for use of uranium in nuclear
reactors was achieved. An explosion led to the replacement of the ether by other
solvents (dibutylmethanol and methylisobutylketone). At the same time new
types of more efficient metal extractants were introduced, e.g., tri-n-butylphos-
phate in 1945 and trioctylamine in 1948. This activity became a great stimulus
to the non-nuclear industry, and solvent extraction was introduced as a separa-
tion and purification process in a large number of chemical and metallurgical
industries in the 1950s and early 1960s. For example, by leaching copper ore
with sulphuric acid followed by extraction of this solution with an organic hy-
droxyaryloxime dissolved in kerosene, several million tons of copper (30% of
12 Cox and Rydberg
world production) is now produced annually. This and many other processes are
described in later chapters.
For these applications, the technique of solvent extraction had to be further
developed and with this a new terminology was also developed. This can be
illustrated by considering a process where a desired component in an aqueous
solution is extracted with an organic reagent (extractant) dissolved in another
organic liquid; note here that the term “organic solvent” is not used because of
possible confusion. The term “solvent” could be used for the whole organic
phase or for the organic liquid in which the organic extractant is dissolved. Thus
the term generally given to the latter is (organic) diluent.
While in laboratory experiments the extraction vessel may be a test tube,
or more conveniently some kind of separation funnel (Fig. 1.1), this is not suited
for industrial use. Industry prefers to use continuous processes. The simplest
separation unit is then the mixer-settler, or some clever development of the same
basic principle, as described in Chapter 9. Figure 1.5 pictures a simple mixer-
settler unit, here used for the removal of iron from an acid solution also contain-
ing nickel and cobalt (the same systems as in Fig. 1.3). The mixer (or contactor)
is here simply a vessel with a revolving paddle that produces small droplets of
one of the liquid phases in the other. This physical mixture flows into and
slowly through the separation vessel, which may be a long tank; through the
influence of gravity the two phases separate, so that the upper organic kerosene–
octanol–amine phase contains the Fe(III) and the lower aqueous CaCl2 phase
contains the Co(II) and Ni(II). Numerous variations of the construction of
mixer-settlers (or MS-units, as they are abbreviated) exist (Chapter 9), often
several joined together into MS-batteries. Many such will be described later on.
A diagram of a full basic process is given in Fig. 1.6 to illustrate the
common terminology. The incoming aqueous solution is called the feed. It is
Fig. 1.5 The principle of a mixer-settler unit, e.g., for separation of iron(III) from
nickel and cobalt.
Fig. 1.6 Typical flowsheet of a solvent extraction circuit.
contacted with the (recycled) solvent phase in a mixer-settler unit. Here we do

not indicate the exact type of unit, but only its function (extraction), as com-
monly is done. After extraction and separation of the phases, the depleted phase
becomes the raffinate and the enriched solvent phase becomes the extract or
loaded (or pregnant) solvent. The raffinate may undergo a solvent recovery
stage to remove any entrained solvent before exiting the process. The extraction
process is rarely specific so that other solutes may be co-extracted with the main
component. These impurities may be removed with an aqueous scrub solution
in a scrub stage producing a scrub extract and a scrub raffinate containing the
impurities. The latter may return to the feed solution to maintain an overall
water balance. The scrubbed extract is now contacted with another aqueous
solution to strip or back-extract the desired component. The stripped solvent
then may undergo some regeneration process to prepare the solvent phase for
recycle. The loaded (pregnant) strip solution then is treated to remove the de-
sired product and the strip solution is recycled. One of the important aspects of
this flowsheet is that, wherever possible, liquid phases are recovered and recy-
cled. This is important from both an economic and an environmental standpoint.
14 Cox and Rydberg
Figure 1.6 shows a situation where each process—extraction, scrubbing, and

stripping—occurs in a single operation or stage. This is generally not efficient
because of the finite value of the distribution coefficient. Thus if D has a value
of 100 (i.e., E = 99%), then after one extraction, the organic extract phase will
contain approximately 99 parts and the aqueous phase 1 part. To achieve a
greater extraction, the aqueous raffinate should be contacted with another por-
tion of the solvent after which the new organic phase contains 0.99 parts, and
the aqueous raffinate now only 0.01 part. Therefore, two extraction stages will
provide 99.99% extraction (with “2 volumes” of the organic phase, but only “1
volume” of the aqueous phase).
Three different ways of connecting such stages are possible: namely, co-
current, cross-current, and counter-current (see Fig. 1.7). In co-current extrac-
tion, the two phases flow in the same direction between the various contactors.
A simple inspection of the diagram will show that with this configuration no
advantage is gained over a single contact because, providing equilibrium is
reached in the first contactor, the separated flows are in equilibrium when enter-
ing the second contactor so no change in relative concentrations will occur. In
the second configuration (b), cross-current, the raffinate is contacted with a
sample of fresh solvent. This is the classical way of extracting a product in the
laboratory when using a separatory funnel and will give an enhanced recovery
of the solute. However, on an industrial scale, this is seldom used because it
Fig. 1.7 Arrangements of solvent extraction stages: (a) co-current; (b) cross-current;
and (c) counter-current.
results in the production of a multitude of product phases containing a reducing

concentration of the desired solute. These have to be combined before stripping
resulting in a much larger volume of loaded solvent to be treated with conse-
quences for plant size and economics. The third configuration (c), counter-cur-
rent, is the one generally chosen by industry. The phase volumes remain con-
stant and by feeding the two phases, feed and solvent, at opposite ends of the
bank of contactors, the driving force for extraction, i.e., the solute concentration
difference between the two phases, is maximized. Chapters in the second part
of the book will extend this discussion.
The need to use multiple extraction to achieve efficient extraction required
the development of new types of continuously working extractors, especially
mixer-settlers and pulsed columns, which were suitable for remotely controlled
operations. These new extractors could be built for continuous flow and in mul-
tiple stages, allowing very efficient isolation of substances in high yield. A good
example is the production of rare earth elements in >99.999% purity in ton
amounts by mixer-settler batteries containing hundreds of stages. These topics
will be further developed in Chapters 6 and 7.
In the early analytical applications of solvent extraction, optimal extrac-
tion or separation conditions were obtained empirically. This was unsatisfactory
and general mathematical descriptions were developed by a number of research-
ers in many countries. This was especially important for large-scale industrial
use and is an activity that continues today almost entirely with computers.
1.5 MODERN TRENDS IN APPLICATIONS

OF SOLVENT EXTRACTION
What is the future of solvent extraction? We will try to indicate some areas in
which we expect important advances in the future. Some are more theoretical,
others are applied, but all are related to other areas of chemistry or to other
fields of science and technology. The simple answers given here do not require
the reading of the subsequent chapters, but do require some good fundamental
chemical knowledge.
1.5.1 Solution Theories

A 10% change in the distribution ratio, which can be measured easily and ac-
curately in the distribution range 0.01–100 (the one most commonly used), cor-
responds to an energy change of only 0.2 kJ/mole. Thus distribution ratio
measurements offer a method to investigate low energy reactions in solutions,
such as weak solute–solute and solute–solvent interactions, especially in the
organic phase. So far, this technique has been only slightly exploited for this
purpose. It is particularly noteworthy to find how few thermodynamic stud-
16 Cox and Rydberg
ies have been made of the distribution or extraction constants, as enthalpy and
entropy values give a good indication of the driving force of the extraction
and also may indicate the structure of the molecular species in the organic
solvent.
Separation of metals by solvent extraction is usually based on the various
complexing properties of the metals (Chapter 3). Separation systems may be
chosen on the basis of complexity constants obtained from the literature. How-
ever, the literature often shows different values for “same systems” causing
considerable concern for process design chemists. There is an obvious need for
an objective presentation of the uncertainty in the published equilibrium con-
stants, however conditional they may be.
Although theories of solution (Chapter 2) and formation of extractable
complexes (Chapters 3 and 4) now are well advanced, predictions of distribution
ratios are mainly done by comparison with known similar systems (Chapter
3). Solvatochromic parameters, solubility parameters, and donor numbers, as
discussed in Chapters 2–4, are so far mainly empirical factors. Continuous ef-
forts are made to predict such numbers, often resulting in good values for sys-
tems within limited ranges of conditions. It is likely that these efforts will suc-
cessively encompass greater ranges of conditions for more systems, but much
still has to be done. In the future, theory may allow the assignment of exact
numbers to the solvatochromic parameters, thus also permitting theoretical pre-
dictions of distribution constants for known as well as for hypothetical (not yet
synthesized) extractants.
Chemical quantum mechanics (Chapter 16) are now promising to contrib-
ute to prediction of distribution ratios, particularly when used in interactive com-
puter modeling of chemical structures and reactions. This will provide better
understanding of solute–solvent interactions and the theory of solubility, which
is the foundation for prediction of distribution ratios. Progress is likely to speed
up as efficient computing programs with large data bases become more easily
available on the market at more attainable prices. Such interactive computer
research will be able to explain poor extractability due to steric hindrance, hy-
dration, the synergistic effect of specific adducts, etc., and, as a consequence,
give clues to better extraction reagents and conditions.
A combination of known extractants can increase the net extraction both
in degree and kinetics through synergistic extraction. This subject also is poorly
explored, though it is of considerable importance to industrial solvent extraction
processes.
1.5.2 Kinetics
All chemical reactions occur with a certain rate. For aqueous systems, in which
no redox reactions occur (e.g., simple complex formation), equilibrium is often
attained rapidly, which is also true for adduct reactions in the organic phase, all
at normal temperatures. However, the transfer of a solute from an aqueous solu-
tion to an organic solvent occurs via a phase boundary; such reactions depend
on several parameters and may be quite slow. Therefore, the kinetics of solvent
extraction, which is of paramount importance to all industrial applications, is
largely determined by the interfacial chemistry. To speed up phase transfer, the
engineer tries to maximize the interfacial surface by, e.g., violent stirring, pro-
ducing billions of small droplets. Though the theory is well advanced (see Chap-
ters 5 and 9), the interface is very complicated and its properties are difficult to
investigate, particularly at the molecular level. Few applicable techniques are
available, and the results are often difficult to interpret. The coalescence of the
droplets (to produce two clean phases that are easy to separate) presents another
interfacial surface problem, and so-called phase reversals may further compli-
cate the separation. This is particularly important in the development of new
types of contactor. Much progress is required here and as such presents a chal-
lenge to future scientists and engineers.
The largest research effort in extraction kinetics is likely to be in the
development of solvent extraction related techniques, such as various versions
of liquid chromatography, liquid membranes, etc. These techniques require a
detailed knowledge of the kinetics of the system to predict the degree of separa-
tion.
It is a practical fact that most industrial solvent extractions are carried out
under nonequilibrium conditions, however close the approach may be; for exam-
ple, centrifugal contactor-separators (Chapter 9) rarely operate at distribution
equilibrium. An interesting possibility is to expand this into extractions further
from equilibrium, if the kinetics of the desired and nondesired products are
different. Such operations offer a real technlogical challenge.
1.5.3 Equipment and Processes

Laboratory solvent extraction studies are often carried out with test tubes, a
process that is cheaper but more laborious, time consuming, and usually yields
more scattered data, than using semicontinuously stirred baffled beakers or
Lewis-type cells. Highly precise distribution data can be obtained efficiently
either “mechanically” by continuous flow centrifugal separators, or “physically”
by using special phase separating filter devices; some of these methods are
described in Chapter 4. Although these techniques are going to be improved,
new, more advanced methods are not likely to appear in the near future.
Specific reagents with exotic structures are costly to produce, which re-
quires the experimentalist to use as small amounts as possible of the reagents.
This leads to a desire to shift from “milli-experiments” (e.g., mixing 10 mL of
each phase) to micro experiments (mixing 50 µL of each phase), which is now
18 Cox and Rydberg
taking place, to even smaller sizes (volumes and amounts)—the nano experi-
ments. Progress in this field requires a high degree of ingenuity.
Although single-stage laboratory techniques provide the first step toward
multistage industrial processes, such process development usually requires
small-scale multistage and pilot-plant scale equipment. A large number of excel-
lent designs are available, and we consider further fundamental improvements
unlikely.
The industrial application of solvent extraction is a mature technique, and
it is now possible to move from laboratory experiments on a new extraction
system to full industrial practice with little technological risk. There is a suffi-
cient variety of large-scale equipment available to cope with most problems
encountered in application, although much of the equipment remains rather mas-
sive. Attempts to miniaturize, for instance, by using centrifugal forces to mix
and separate phases, still has to be developed further.
Many industrial processes begin with a leaching step, yielding a slurry
that must be clarified before solvent extraction. The solid–liquid separation is a
costly step. The solvent extraction of unclarified liquids (“solvent-in-pulp”) has
been proposed to eliminate solid–liquid separation. The increased revenue and
reduced energy cost make this an attractive process, but many problems remain
to be solved: loss of metals and extractants to the solid phase, optimization of
equipment design, effluent disposal, etc.
An essential step in industrial solvent extraction is the regeneration of the
extractant. This can be done in many ways, e.g., by distillation, evaporation, or
stripping (back-extraction). While distillation and evaporation do not discrimi-
nate between solutes (the diluent is simply removed by heating), stripping, by
careful choice of strip solution and conditions, can be made highly selective.
Alternatively, all the solutes can be stripped and then subjected to a selective
extraction by changing the extractant; examples of both types of process will be
found in Chapter 13. The possibilities are many, and it may be worthwhile to
explore new paths.
Membrane extraction is a relatively new technique for solvent extraction,
in which a solute is transferred from one aqueous phase to another through a
membrane holding an extractant dissolved in a diluent. This ingenious scheme
has been only slightly explored, though it offers great potential for the future,
e.g., for waste water cleaning.
The step from laboratory experiments via pilot plants to industrial scale
requires serious consideration of all the points here; practical experience is in-
valuable in order to avoid mistakes and excess costs, as indicated in Chapter 7.
1.5.4 Organic Chemistry and Biochemistry

There is a continuing demand for selective extractants that can be developed,
for example, by tailor making organic molecules with pre-organized metal bind-
ing sites. The development of such extractants may be designed on computer

based models. Environmental considerations require that the extractants and dil-
uents used are either nontoxic or recovered within the process. The increasing
combination of extraction and distillation, as used in biotechnology (Chapter
10), places a whole host of new demands on the diluents employed.
In the extraction of biologically active compounds, care must be taken to
avoid the loss of activity that often occurs by contact with organic diluents.
Thus a series of systems have been developed specifically with these compounds
in mind. The first of these uses mixtures of aqueous solutions containing poly-
mers and inorganic salts that will separate into two phases that are predom-
inately water. A second system uses supercritical conditions in which the
original two-phase system is transformed into one phase under special tempera-
ture-pressure conditions. Also the active organic compound can be shielded
from the organic diluent by encapsulation within the aqueous center of a micelle
of surface active compounds. All these systems are currently an active area for
research as is discussed in Chapter 15.
1.5.5 Separation of Inorganic Compounds

Separation schemes have been developed for all the elements in the periodic
system, both in ionic and molecular forms. Some of these substances are toxic,
putting severe restraints on plant operation and the environment. For example,
many solvents are bioactive, and increasingly stringent control is needed to pre-
vent their spread in the biosphere. Though the techniques are advanced, de-
mands for improved safety will persist and require new types of equipment or
processes.
The production of electricity from nuclear fission energy is accompanied
by formation of radioactive waste, of which the larger hazard is the presence of
long-lived transuranium isotopes. The problems associated with this waste are
still debated, but if the transuranium isotopes could be removed by “exhaustive”
reprocessing and transmuted in special nuclear devices, the hazard of the waste
would be drastically reduced (Chapter 12). This may require new selective ex-
tractants and diluents as well as new process schemes. Research in this field is
very active.
The mineral industry, where ores produce pure metals, requires a number
of process stages that can be carried out either through pyrometallurgical treat-
ment or by the use of wet chemical separation processes (hydrometallurgy). The
former produces off gases and slags, which may be environmentally hazardous.
The latter introduces chemistry into the mineral industry, and is often met with
strong criticism among traditional metallurgists. It is therefore up to the hydro-
metallurgical chemist to develop schemes for metal production from ores that
are economically competitive and environmentally advantageous. Here, the field
is rather open for new developments. A specifically interesting trait is the use
20 Cox and Rydberg
of “in-situ” leaching of ore bodies, which circumvents conventional mining with

its sometimes devastating environmental effects, e.g., in open-pit mining.
In ore processing, the essential minerals are enriched through flotation of
the crushed ore, using an aqueous solution containing flotation reagents (we
may label them flotants), that selectively attach to the mineral surface and—
with the aid of air bubbles—lift them up to the surface where they can be
skimmed off, while the “gangue” material stays at the bottom. Alternatively, the
flotants are made magnetic, allowing the use of magnetic separation as com-
monly used in the mineral industry. These flotants are chemically very similar
to the chelating extractants used in solvent extraction processes. An interesting
subject here is collaboration between these two mineral processing fields to
produce new and more effective reagents and separation processes.
1.5.6 Environmental Aspects

The future will bring further increase in concern over the environmental impact
of chemical operations. The liquid effluents must not only be controlled, they
must also be rendered harmless to the environment. This requires removal of
the hazardous substances. For many of the dilute waste solutions, solvent extrac-
tion has proved to be an effective process. This is even more true for recycling
of mixed metals from various industries. Nevertheless, the increasing amounts
of wastes from human activities require much more to be done in this field.
In principle, solvent extraction is an environmentally friendly process with
no air or water pollution, provided the plant flows are properly designed. It
could, therefore, replace many of the present polluting processes. A particular
problem, however, is the solvent extraction effluents, which may contain bio-
chemically active substances posing “new” hazards to the environment. These
can be handled by various solid sorbents, which then can be incinerated, but the
advantage of the solvent extraction process may be lost. There is, therefore, a
demand for biodegradable and environmentally benign solvent phases. In the
future, additional attention is required to this field.
1.5.7 Spin-off
The principle of solvent extraction—the distribution of chemical species be-
tween two immiscible liquid phases—has been applied to many areas of chemis-
try. A typical one is liquid partition chromatography, where the principle of
solvent extraction provides the most efficient separation process available to
organic chemistry today; its huge application has become a field (and an indus-
try!) of its own. The design of ion selective electrodes is another application of
the solvent extraction principle; it also has become an independent field. Both
these applications are only briefly touched upon in the chapter of this book on
analytical applications (Chapter 14), as we consider them outside the scope of
this work. Nevertheless, fundamental research on solvent extraction will provide

further important input into these two applications. It is likely that fundamental
solvent extraction research will continue to contribute to the development of
further selective analytical techniques.
The driving force of the transport of salts, proteins, etc., through the cell
membrane from the nucleus to the body fluids, and vice versa, is a complicated
biochemical process. As far as is known, this field has not been explored by
traditional solution chemists, although a detailed analysis of these transfer pro-
cesses indicates many similarities with solvent extraction processes (equilibrium
as well as kinetics). It is possible that studies of such simpler model systems
could contribute to the understanding of the more complicated biochemical pro-
cesses.
Solvent extraction deals with the transport of chemical substances from
one phase into another one, the chemical kinetics of this process, and the final
equilibrium distribution of the substances between the two phases. Such trans-
port and distribution processes are the motors that make life in biological sys-
tems possible. Fundamental studies of such “solvent extraction” processes con-
tribute to the better understanding of all processes in nature. Here, only the lack
of imagination stands in the way of important new scientific discoveries.
1.5.8 Conclusion
Solvent extraction is a mature technique in that extensive experience has led to
a good understanding of the fundamental chemical reactions. At the same time,
compared to many other chemical separation processes like precipitation, distill-
ation, or pyrometallurgical treatment, the large-scale application of solvent ex-
traction is, nevertheless, a young technique. New reagents are continually being
developed, spurred on by computer modeling, and more efficient contacting
equipment is coming into use. Considering such factors as demands for higher
product purity, less pollution, and the need for recovering substances from more
complex matrices and more dilute resources, the efficiency and high selectivity
of solvent extraction should make it an increasingly competitive separation pro-
cess both in research and in industry.
1.6 HOW TO USE THIS BOOK

The first part of the book contains basic solution theory and thermodynamics
(Chapter 2); a survey of the effects on the distribution ratio of changes in param-
eter values, such as concentration of metals, complex formers and other reac-
tants, pH, temperature, etc., (Chapter 3); measurement techniques, data collec-
tion, evaluation, and interpretation (Chapter 4); and kinetics (Chapter 5). The
ionic strength is an essential factor in all aqueous systems, and how to cope
22 Cox and Rydberg
with it is treated in a fundamental way in Chapter 6. These chapters sometimes

go into such depth that they may mainly interest only the theoretical chemist or
professional engineer; in some chapters we offer some choice by placing the
most detailed parts in smaller type. It is not necessary to read these chapters
before the applied ones, and some can be read independently, before or after
the other more applied chapters.
The second part deals with applications of solvent extraction in industry,
and begins with a general chapter (Chapter 7) that involves both equipment,
flowsheet development, economic factors, and environmental aspects. Chapter
8 is concerned with fundamental engineering concepts for multistage extraction.
Chapter 9 describes contactor design. It is followed by the industrial extraction
of organic and biochemical compounds for purification and pharmaceutical uses
(Chapter 10), recovery of metals for industrial production (Chapter 11), applica-
tions in the nuclear fuel cycle (Chapter 12), and recycling or waste treatment
(Chapter 14). Analytical applications are briefly summarized in Chapter 13. The
last chapters, Chapters 15 and 16, describe some newer developments in which
the principle of solvent extraction has or may come into use, and theoretical
developments.
Apart from the first six chapters, each chapter has been written more or
less independently but, we hope, consistently.
1.7 SOLVENT EXTRACTION NOMENCLATURE

Because scientists and engineers in chemistry use different symbols, a single set
valid for all chapters in this book is not feasible. Moreover, it is desirable that
both groups be familiar with the symbols used within the subfields in order
to enhance communication. The symbols in this book closely follow IUPAC
recommendations (see reference list). In the Appendix of this book, lists are
given of the most commonly used symbols, except for those in Chapter 9, which
has its own list of symbols.
1.8 LITERATURE ON SOLVENT EXTRACTION

Recommendations of solvent extraction terminology have been published by the
International Union of Pure and Applied Chemistry (IUPAC) [1,3–5]. Some
excellent monographs on solvent extraction are listed in section 1.8.1; unfortu-
nately, the older ones may only be available in large science libraries. Most of
them deal either with a small sector of the field or are very comprehensive.
Finally, in section 1.8.2, we have collected edited versions of proceedings of
the International Conferences of Solvent Extraction (ISEC), which normally are
held every three years.
Because solvent extraction is used extensively as a technique in research,

publications with this technique appear in all types of chemical journals. This
is also true for its industrial applications that are often described in chemical
engineering journals.
1.8.1 General References on Solvent Extraction

1955 L. Alders, Liquid-liquid Extraction, Elsevier, New York, 206 p.
1955 E. Hecker, Verteilungsverfahren in Laboratorium, Verlag Chemie GmbH,
Weinheim, 230 p.
1957 G. H. Morrison and H. Freiser, Solvent Extraction in Analytical Chemistry,
John Wiley and Sons, New York, 260 p.
1963 R. E. Treybal, Liquid Extraction, McGraw-Hill Book Co, New York, 620 p.
1963 A. W. Francis, Liquid-liquid Equilibriums, Wiley-Interscience, New York,
288 p.
1964 J. Stary, The Solvent Extraction of Metal Chelates, Pergamon Press, New
York, 240 p.
1969 Y. Marcus and S. Kertes, Ion Exchange and Solvent Extraction of Metal
Complexes, Wiley-Interscience, New York, 1037 p.
1970 Y. A. Zolotov, Extraction of Chelate Compounds (transl. from Russian
Ed. 1968), Humprey, Ann Arbor, 290 p.
1970 A. K. De, S. M. Khopkar and R. A. Chalmers, Solvent Extraction of
Metals, Van Nostrand Reinhold, New York, 260 p.
1971 C. Hanson (Ed.), Recent Advances in Liquid-Liquid Extractions, Perga-
mon Press, New York, 584 p.
1977 T. Sekine and Y. Hasegawa, Solvent Extraction Chemistry, Marcel Dek-
ker, New York, 919 p.
1983 T. C. Lo, M. H. I. Baird and C. Hanson, Handbook of Solvent Extraction,
Wiley-Interscience, New York, 1980 p.
1984 G. Ritcey and A. W. Ashbrook, Solvent Extraction. Part I and II, Elsevier,
Amsterdam, 361+737 p.
1992 J. D. Thornton (Ed.), Science and Practice of Liquid-Liquid Extraction,
Oxford University Press, Oxford 1992, 2 volumes.
1994 J. C. Godfrey and M. J. Slater (Eds.) Liquid-Liquid Extraction Equipment,
John Wiley and Sons, Chichester 1994, UK.
1.8.2 Proceedings of International

Solvent Extraction Conferences (ISEC)
1965 Harwell—Solvent Extraction Chemistry of Metals (Eds. H. A. C. McKay,
T. V. Healy, I. L. Jenkins and A. Naylor), Macmillan 456 p., London
(1965).
24 Cox and Rydberg
1966 Gothenburg—Solvent Extraction Chemistry (Eds. D. Dyrssen, J. O. Liljen-

zin and J. Rydberg), North-Holland 680 p., Amsterdam (1967).
1968 Haifa/Jerusalem—Solvent Extraction Research (Eds. A. S. Kertes and Y.
Marcus), John Wiley and Sons, 439 p., New York (1969).
1971 The Hague—Solvent Extraction (Eds. J. G. Gregory, B. Evans and P. C.
Weston), Society of Chemical Industry, 3 vol. 1566 p., London (1971).
1974 Lyon—Proceedings of ISEC ’74, Society of Chemical Industry, 3 vol.
2899 p. London (1974).
1977 Toronto—CIM Special Volume 21 Canadian Institute of Mining and Met-
allurgy, 2 vol. 807 p., Montreal (1979).
1980 Liège—International Solvent Extraction Conference ISEC’80, Association
d’Ingeneur de l’Université de Liège, 3 vol. 1300 p., Liège (1980).
1983 Denver—Selected Papers, AIChE Symposium Series, No 238, Vol. 80,
1984, American Institute of Chemical Engineers, 177 p., New York
(1984).
1986 Munich—Preprints of ISEC ’86, DECHEMA, 3 vol. 2230 p., Frankfurt-
am Main (1986).
1988 Moscow—Proceedings of ISEC ’88, 4 vol. 1495 p., Vernadsky Institute
of Geochemistry and Analytical Chemistry of the USSR Academy of Sci-
ences, Moscow (1988).
1990 Kyoto—Solvent Extraction 1990 (Proceedings of ISEC’90) (Ed. T. Se-
kine) Elsevier, 2 vol. 1923 p., Amsterdam m.m. (1992).
1993 York—Solvent Extraction in the Process Industries, Proceedings of ISEC
’93,(Eds. D.H.Logsdail and M.J.Slater) Elsevier Applied Science, 3 vol.
1828 p., London & New York (1993).
1996 Melbourne—Value Adding Through Solvent Extraction, Proceedings of
ISEC ’96, (Eds. D. C. Shallcross, R. Paimin, L. M. Prvcic), The Univer-
sity of Melbourne, Parkville, 2 vol. 1684 p., Victoria, Australia (1996).
1999 Barcelona—Solvent Extraction for the 21st Century (Proceedings ISEC
’99) (Eds. M. Cox, M. Hidalgo, and M. Valiente), Society of Chemical
Industry, 2 vols. 1680 p., London (2001).
2002 Cape Town—Proceedings of the International Solvent Extraction Confer-
ence, ISEC ’02 (Eds. K.C. Sole, P.M. Cole, J.S. Preston and D.J. Robin-
son), Chris van Rensburg Publications (Pty) Ltd, 2 vol. p. Johannesburg,
South-Africa (2002). Also on CD.
REFERENCES
1. Freiser, H.; and Nancollas, G. H.; Compendium of Analytical Nomenclature. Defini-
tive Rules 1987. IUPAC. Blackwell Scientific Publications, Oxford (1987).
2. Blass, E.; Liebl, T.; Häberl, M.; Solvent Extraction—A Historical Review, Proc. Int.
Solv. Extr. Conf. Melbourne, 1996.
3. Högfeldt, E.; Stability Constants of Metal-Ion Complexes. Part A: Inorganic Li-

gands. IUPAC Chemical Data Series No. 22, Pergamon Press, New York (1982).
4. McNaught, A. D.; and Wilkinson, A.; IUPAC Compendium of Chemical Terminol-
ogy, Second Edition, Blackwell Science (1997).
5. IUPAC, Quantities, Units and Symbols in Physical Chemistry, Third Edition, (Ed.
Ian Mills), Royal Society of Chemistry, Cambridge 2002.
2
Principles of Solubility and Solutions
YIZHAK MARCUS* The Hebrew University of Jerusalem, Jerusalem, Israel
2.1 INTRODUCTION
Solvent extraction is another name for liquid–liquid distribution, that is, the
distribution of a solute between two liquids that must not be completely mutu-
ally miscible. Therefore, the liquid state of aggregation of matter and the essen-
tial forces that keep certain types of liquids from being completely miscible are
proper introductory subjects in a study of solvent extraction. Furthermore, the
distribution of a solute depends on its preference for one or the other liquid,
which is closely related to its solubility in each one of them. Thus, the general
subject of solubilities is highly relevant to solvent extraction.
In a solution, the solute particles (molecules, ions) interact with solvent
molecules and also, provided the concentration of the solute is sufficiently high,
with other solute particles. These interactions play the major role in the distribu-
tion of a solute between the two liquid layers in liquid–liquid distribution sys-
tems. Consequently, the understanding of the physical chemistry of liquids and
solutions is important to master the rich and varied field of solvent extraction.
Solvent extraction commonly takes place with an aqueous solution as one
liquid and an organic solvent as the other. Obviously, the extraction process is
limited to the liquid range of these substances. Since solvent extraction is gener-
ally carried out at ambient pressures, the liquid range extends from about the
freezing temperature up to about the normal boiling temperature. If, however,
high pressures are applied (as they are in some solvent extraction processes),
then the liquid range can extend up to the critical temperature of the substance.
Supercritical fluid extraction beyond the critical temperature (such as decaffei-
nation of coffee with supercritical carbon dioxide) is a growing field of applica-
*Retired.
27
28 Marcus
tion of solvent extraction. It has the advantages that the properties of the super-
critical fluid can be fine-tuned by variation of the pressure, and that this
“supercritical solvent” can be readily removed by a drastic diminution of the
pressure, but has drawbacks related to the high temperatures and pressures often
needed.
Numerous solvents are used in solvent extraction. They can be divided in
the context of solvent extraction into different classes as follows (but see Refs.
[1] and [2] for some other classification schemes):
Class 1: Liquids capable of forming three-dimensional networks of strong
hydrogen bonds, e.g., water, poly- and amino-alcohols, hydroxy-acids, etc.
Class 2: Other liquids that have both active hydrogen atoms and donor
atoms (O, N, F; see Chapter. 3), but do not form three-dimensional networks
(rather forming chainlike oligomers), e.g., primary alcohols, carboxylic acids,
primary and secondary amines, nitro compounds with α-positioned hydrogen
atoms, liquified ammonia, etc. They are generally called protic or protogenic
substances.
Class 3: Liquids composed of molecules containing donor atoms, but no
active hydrogen atoms, e.g., ethers, ketones, aldehydes, esters, tertiary amines,
nitro compounds without α-hydrogen, phosphoryl-group containing solvents,
etc. (see Table 4.3). They are generally called dipolar aprotic substances.
Class 4: Liquids composed of molecules containing active hydrogen
atoms but no donor atoms, e.g., chloroform and some other aliphatic halides.
Class 5: Liquids with no hydrogen-bond forming capability and no donor
atoms, e.g., hydrocarbons, carbon disulfide, carbon tetrachloride, supercritical
carbon dioxide, etc.
This diversity in solvent properties results in large differences in the distri-
bution ratios of extracted solutes. Some solvents, particularly those of class 3,
readily react directly (due to their strong donor properties) with inorganic com-
pounds and extract them without need for any additional extractant, while others
(classes 4 and 5) do not dissolve salts without the aid of other extractants. These
last are generally used as diluents for extractants, required for improving their
physical properties, such as density, viscosity, etc., or to bring solid extractants
into solution in a liquid phase. The class 1 type of solvents are very soluble in
water and are useless for extraction of metal species, although they may find
use in separations in biochemical systems (see Chapter 9).
2.1.1 Properties of Liquids

If the externally imposed conditions of pressure and temperature permit a sub-
stance to be in the liquid state of aggregation, it possesses certain general prop-
erties. Contrary to a substance in the solid state, a liquid is fluid; that is, it flows
under the influence of forces and is characterized by its fluidity, or the recipro-
Principles of Solubility and Solutions 29
cal, its viscosity. For water, the viscosity is only 0.89 mPa s−1 (nearly 1 centi-
poise) at 25°C [1,2], and water is considered a highly fluid liquid.
Contrary to a fluid in the gaseous state, a liquid has a surface, and is
characterized by a surface tension. For water, the surface tension is 72 mN m−1
at 25°C [1,2]. Again, contrary to the fluid in the gaseous state, the volume of a
liquid does not change appreciably under pressure; it has a low compressibility
and shares this property with matter in the solid (crystalline, glassy, or amor-
phous) state. For water, the compressibility is 0.452 (GPa)−1 at 25°C [1,2]. These
are macroscopic, or bulk, properties that single out the liquid state from other
states of aggregation of matter.
There are also some general microscopic, or molecular, properties that are
peculiar to the liquid state. Contrary to the crystalline solid, the particles of a
liquid do not possess long-range order. Although over a short range—2 to 4
molecular diameters—there is some order in the liquid, this order dissipates at
longer distances. A particle in the liquid is free to diffuse and, in time, may
occupy any position in the volume of the liquid, rather than being confined at
or near a lattice position, as in the crystalline solid. Contrary to particles in the
gaseous state, however, the particles in a liquid are in close proximity to each
other (closely packed) and exert strong forces on their neighbors [3].
The close packing of the molecules of a substance in the liquid state re-
sults in a density much higher than in the gaseous state and approaching that in
the solid state. The density, ρ, is the mass per unit volume, and can be expressed
as the ratio of the molar mass M to the molar volume V of a liquid. Table 2.1
lists the values of the properties M and V of representative liquids that are
important in the field of solution chemistry and solvent extraction. The densities
and molar volumes depend on the temperature, and the latter are given for 25°C.
(For a discussion of industrial solvents, see Chapter 12.)
Many liquids used in solvent extraction are polar. Their polarity is mani-
fested by a permanent electric dipole in their molecules, since their atoms have
differing electronegativities. Oxygen and nitrogen atoms, for instance, generally
confer such dipolarity on a molecule, acting as the negative pole relative to
carbon or hydrogen atoms bonded to them. The dipole moment µ characterizes
such polar molecules and ranges from about 1.15 D (Debye unit = 3.336 10−30
C ⴢ m) for diethyl ether or chloroform, to 4.03 D for nitrobenzene, and 5.54 D
for hexamethyl phosphoric triamide [1,2]. Water is also a polar liquid, with a
moderate dipole moment, µ = 1.83 D. A list of dipole moments of some solvents
that are important in solvent extraction is presented in Table 2.1. Substances
that do not have a permanent dipole moment (i.e., µ = 0) are called nonpolar.
Many hydrocarbons belong to this category, but not all (e.g., toluene has µ =
0.31 D).
When nonpolar liquids are placed in an electric field, only the electrons
in their atoms respond to the external electric forces, resulting in some atomic
30 Marcus
Table 2.1 Selected Properties of Somea Water-Immiscible Solvents

M Vb µc δ
Solvent g/mol mL/mol D εb (J/mL)1/2
c-Hexane 84.2 108.7 0. 2.02 16.8

n-Hexane 86.2 131.6 0.09 1.88 15.0
n-Octane 114.3 163.5 ⬃0 1.95 15.5
n-Decane 142.3 195.9 ⬃0 1.99 15.8
n-Dodecane 170.4 228.6 ⬃0 2.00 16.0
Decalin (mixed isomers) 138.3 157.4 0. 2.15 18.0
Benzene 78.1 89.9 0. 2.27 18.8
Toluene 92.1 106.9 0.31 2.38 18.8
Ethylbenzene 106.2 123.1 0.37 2.40 18.0
p-Xylene 106.2 123.9 0. 2.27 18.1
Dichloromethane 89.9 64.5 1.14 8.93 20.2
Chloroform 119.4 80.7 1.15 4.89 19.5
Carbon tetrachloride 153.8 97.1 0. 2.24 17.6
1,1-Dichloroethane 99.0 84.7 1.82 10.00 18.3
1,2-Dichloroethane 99.0 79.4 1.83 10.36 20.0
Trichloroethylene 131.4 90.7 0.80 3.42 19.0
Chlorobenzene 112.6 102.2 1.69 5.62 19.8
1,2-Dichlorobenzene 147.0 113.1 2.50 9.93 20.5
Carbon disulfide 76.1 60.6 0. 2.64 20.3
Watera 18.0 18.1 1.85 78.36 47.9d
Methanola 32.0 40.7 2.87 32.66 29.3
Ethanola 46.1 58.7 1.66 24.55 26.0
1-Propanola 60.1 75.1 3.09 20.45 24.4
2-Propanola 60.1 76.9 1.66 19.92 23.7
1-Butanol 74.1 92.0 1.75 17.51 23.3
Isoamyl alcohol 88.2 109.2 1.82 15.19 22.1
1-Hexanol 102.2 125.2 1.55 13.30 21.8
1-Octanol 130.2 158.4 1.76 10.34 20.9
2-Ethyl-1-hexanol 130.2 157.1 1.74 4.4 19.4
Diethyl ether 74.1 104.7 1.15 4.20 15.4
Diisopropyl ether 102.2 142.3 1.22 3.88 14.6
Bis(2-chloroethyl) ether 143.0 117.9 2.58 21.20 18.8
Acetonea 58.1 74.0 2.69 20.56 22.1
Methyl ethyl ketone 72.1 90.2 2.76 18.11 18.7
Methyl isobutyl ketone 100.2 125.8 2.70 13.11 17.2
Cyclohexanone 98.2 104.2 3.08 15.5 19.7
Acetylacetone 100.1 103.0 2.78 25.7 19.5
Ethyl acetate 88.1 98.5 1.78 6.02 18.2
Butyl acetate 116.2 132.5 1.84 5.01 17.6
Propylene carbonate 102.1 85.2 4.94 64.92 21.8
Nitromethane 61.0 54.0 3.56 35.87 25.7
Table 2.1 Continued

M Vb µc δ
Solvent g/mol mL/mol D εb (J/mL)1/2
Nitrobenzene 123.1 102.7 4.22 34.78 22.1

Acetonitrilea 41.1 52.9 3.92 35.94 24.1
Benzonitrile 103.1 103.1 4.18 25.2 22.7
Quinoline 129.2 118.5 2.18 8.95 22.8
Tributyl phosphate 266.3 273.8 3.07 8.91 15.3
Source: Refs. 1 and 2.

a
Water miscible.
b
At 25°C.
c
Isolated solvent molecules, i.e., in the gaseous phase or dilute solution in an inert solvent.
d
Behaves in organic-rich aqueous mixtures as if δ ⬇ 30.
polarization. This produces a relative permittivity (dielectric constant) ε, which

is approximately equal to the square of the refractive index (at the sodium D
line, nD), or about 2. Polar molecules, however, further respond to the external
electric field by reorienting themselves, which results in a considerably larger
relative permittivity. In particular, when cooperative hydrogen bonding takes
place (see later discussion), high values of ε are achieved (e.g., 78 for water at
25°C [1,2]). Table 2.1 lists ε values of some liquids that are important in solvent
extraction. The attractive forces between charges (such as those on ions) of
opposite sign are inversely proportional to the relative permittivity of the liquid
medium in which they find themselves. Therefore, the ionic dissociation of elec-
trolytes strongly depends on the relative permittivity of the solvent that is used
to dissolve them (see section 2.6).
There are properties of liquids used as solvents, such as the liquid range,
viscosity, surface tension, and vapor pressure that, although important from the
practical and technical points of view, are not listed in Table 2.1. These can be
found in Refs. [l,2]. Tables 12.2, 13.1, and 13.2 list solvents used in extraction
with some further information.
2.1.2 Cohesive Forces in Liquids

The ability of even the most inert gases to be liquified at sufficiently low tem-
peratures and high pressures is evidence for the existence of cohesive forces
between molecules. Such forces are manifested, in their simplest form, between
the atoms of noble gases, such as argon. The motions of the electrons in the
argon atom induce a temporary electric dipole in a neighboring argon atom,
which in turn strengthens the temporary dipole in the first one. The mutual
32 Marcus
interaction of these temporary, mutually induced electric dipoles produces at-

tractive forces, called London forces or dispersion forces. These forces are
rather weak, and high pressures and drastic cooling (for argon 4.9 MPa and
−122°C) are required for these forces to be able to overcome the thermal agita-
tion of the atoms that tends to keep them apart. The energy associated with the
dispersion forces decreases with the sixth power of the distance between the
interacting particles, and thus, these forces have a very short range, although
they are present in all liquids. Hydrocarbons are often used as diluents in the
organic phase of a solvent extraction system, and generally are held together as
liquids by these dispersion forces operating between neighboring segments (e.g.,
−CH2−) of different molecules. In fact, in nonpolar substances, the only cohe-
sive forces are the dispersion forces [3].
In polar liquids, when the electric dipoles are able to arrange themselves
in a “head-to-tail” configuration, that is, when their positive end is on the aver-
age more in the vicinity of the negative ends of neighboring molecules (Fig.
2.1a,c), further attractive forces result. The energy associated with these forces
Fig. 2.1 Dipole and hydrogen bond interactions. A schematic representation of (a)
“head-to-tail” dipole-dipole attractive interactions (e.g., in tri-n-octylamine); (b) “head-
to-head” dipole-dipole repulsive interactions caused by steric hindrance (e.g., in dibutyl
sulfoxide); (c) chainlike dipole-dipole interactions (e.g., in 1-octanol); (d) a cyclic, hy-
drogen-bonded dimer (e.g., in hexanoic acid).
is proportional to the product of the dipole moments of the molecules and is

inversely proportional to the sixth power of the mean distance between them
(i.e., it has the same range as the dispersion forces). However, the structure of
the molecules may be such that they prevent a head-to-tail configuration (see
Fig. 2.1b), and the resulting head-to-head configuration causes repulsion be-
tween the polar molecules [3].
Some of the liquids that are used in solvent extraction, especially water,
interact by means of hydrogen bonding. Their molecules have a hydrogen atom
attached to a very electronegative element (mainly oxygen and, less effectively,
nitrogen), and this hydrogen atom is able to be bound to the electronegative
atom (O, N, or F) of a neighboring molecule, forming a hydrogen bridge. Such
a bridge requires a rather rigid geometry: a linear configuration of the two elec-
tronegative atoms of the neighboring molecules with the hydrogen atom be-
tween them. This bond is of considerably greater strength than dispersion and
dipole–dipole interactions. If the molecules of a substance can both donate and
accept a hydrogen bond, a cyclic dimer may result (see Fig. 2.1d) that is consid-
erably less polar than the monomers of this substance. Occasionally, water again
being an example, this hydrogen bonding is a cooperative phenomenon: if such
a bond is formed between two molecules, each becomes more likely to partici-
pate in further hydrogen bonding with other neighbors [3].
The three kinds of forces described above, collectively known as the cohe-
sive forces that keep the molecules of liquids together, are responsible for vari-
ous properties of the liquids. In particular, they are responsible for the work that
has to be invested to remove molecules from the liquid, that is, to vaporize it.
The energy of vaporization of a mole of liquid equals its molar heat of vaporiza-
tion, ∆VH, minus the pressure-volume work involved, which can be approxi-
mated well by RT, where R is the gas constant [8.3143 J K−1 mol−1] and T is
the absolute temperature. The ratio of this quantity to the molar volume of the
liquid is its cohesive energy density. The square root of the cohesive energy
density is called the (Hildebrand) solubility parameter of the liquid, δ:
δ 2 = (∆ V H − RT ) / V (2.1)
where V is the molar volume. A list of solubility parameters of representative
liquids that are important in solvent extraction is given in Table 2.1.
2.2 SOLVENT MISCIBILITY

Solvent extraction takes place through the distribution of a solute or of solutes
between two practically immiscible liquids. For a separation to be carried out
by solvent extraction, the solute has to transfer from one region of space to
another such region, which is physically separated from the first (see Fig. 1.1).
In each such region, the solute is dissolved uniformly in a (homogeneous) liquid
34 Marcus
and, at all points, is under the same pressure. Such a region of space is called a
phase. At equilibrium, at a constant temperature and pressure, a liquid phase is
homogeneous and is isotropic: it has the same properties in all directions. In
solvent extraction, we are interested in two liquid phases in contact, normally
under the same temperature and pressure, which eventually reach mass transfer
equilibrium. Such an equilibrium is a dynamic process: as many molecules of
one solvent enter the other as leave it and return to their original phase as do
the molecules (ions) of the solute(s). If each of the two liquids is originally a
pure chemical substance and there is (as yet) no solute to distribute, we speak
of a two-component, two-phase system, but the net effect is that an amount of
the substance from one liquid phase is transferred to the other and vice versa.
For two practically immiscible liquids this transferred amount is small relative
to the total amount of the two phases. There may or may not be a vapor phase
present, but this is generally disregarded in the context of solvent extraction.
2.2.1 The Phase Rule

A substance that can be added to a system independently (or removed from it,
say by precipitation or vaporization) is called a component of such a system.
The phase rule, summarizing a general behavior of nature, says:
The number of degrees of freedom in a system equals the number of compo-
nents plus two minus the number of phases.
A two-component, two-phase system, therefore, has two degrees of freedom,
that is, two external conditions, such as the temperature and pressure, specify
its composition completely. By this we mean that the composition at equilibrium
of each of the phases is uniquely determined at a given temperature and pres-
sure. For example, in the diethylether–water system at atmospheric pressure and
at 10°C, the ether-rich layer contains 1.164 mass% water (4.6 mol% water) and
the water-rich layer contains 9.04 mass% ether (2.4 mol% ether). This holds
irrespective of the relative amounts of the liquid phases present, as long as there
are two phases. When the temperature is raised to 30°C, the composition of the
two phases changes to 1.409 mass% water (5.5 mol% water) and 5.34 mass%
ether (1.35 mol% ether) in the two phases, respectively. These compositions are
uniquely determined by the (atmospheric) pressure and the temperature, two
external conditions (degrees of freedom) for this two-component, two-phase
system. [If a third phase, the vapor, is present, then only one degree of freedom
remains, and, e.g., the temperature determines the (vapor) pressure attained by
the system]. For a two-phase system, each additional component (e.g., a solute)
permits a new degree of freedom: here, the concentration of the solute in one
of the phases. When this has been fixed, its concentration in the other phase is
given by the equilibrium condition and cannot be freely chosen.
For the two-component, two-phase liquid system, the question arises as to

how much of each of the pure liquid components dissolves in the other at equi-
librium. Indeed, some pairs of liquids are so soluble in each other that they
become completely miscible with each other when mixed at any proportions.
Such pairs, for example, are water and 1-propanol or benzene and carbon tetra-
chloride. Other pairs of liquids are practically insoluble in each other, as, for
example, water and carbon tetrachloride. Finally, there are pairs of liquids that
are completely miscible at certain temperatures, but not at others. For example,
water and triethylamine are miscible below 18°C, but not above. Such pairs of
liquids are said to have a critical solution temperature, Tcs For some pairs of
liquids, there is a lower Tcs (LCST), as in the water–triethylamine pair, but the
more common behavior is for pairs of liquids to have an upper Tcs (UCST),
(Fig. 2.2) and some may even have a closed mutual solubility loop [3]. Such
instances are rare in solvent extraction practice, but have been exploited in some
systems, where separations have been affected by changes in the temperature.
Fig. 2.2 Liquid immiscibility. The guaiacol (A) + glycerol (B) system happens to have
a closed miscibility loop. The (phase) coexistence curves are shown on the left-hand side
(a) for lower temperatures, at which a lower critical solution temperature (LCST), Tcs =
40°C, is seen, and on the right-hand side (b) for higher temperatures, where a UCST,
Tcs = 82°C, is seen. The compositions of the A-rich phases′′ and the B-rich phases′ are
shown at 50°C and 70°C, respectively.
36 Marcus
2.2.2 Excess Gibbs Energy of Mixing

The mutual solubility of two liquids A and B depends, in general, on how
much the molecules of each liquid tend to attract those of its own kind, relative
to their tendency to attract those of the other. This tendency is measured by
the excess Gibbs energy of mixing of the two liquids (see section 2.4), ∆MGAB E
,
which is related to the partial vapor pressures pA and pB of the two liquids A
and B in the mixture. If the composition of the system is given by nA and nB
moles of the respective components in a given phase, their mole fractions in
this phase are
x A = nA / ( nA + nB ) x B = 1 − x A = nB / ( nA + nB ) (2.2)
The excess Gibbs energy of mixing per mole of mixture is approxi-
mately
∆ M GAB
E
= RT [ x A ln ( pA / x A pA* ) + x B ln ( pB / x B pB* )] (2.3)
where p*A and p*B are the vapor pressures of the pure liquids A and B, respec-
tively, at the given temperature T. (The approximation consists of neglecting
the interactions in the vapors that is permissible at low vapor pressures.)
The system of two liquids will split into two liquid phases at a given T
and composition x if
x A x B (∂ 2 ∆ M GAB
E
/ ∂x 2 ) < − RT (2.4)
Many liquid mixtures behave approximately as regular mixtures [3],
which are characterized by
∆ M GAB
E
= x A x B bAB (T ) (2.5)
bAB (T ) being a (generally temperature-dependent) constant that is independent
of the composition. Such regular mixtures split into two liquid phases if the
following inequality is obeyed:
bAB (T ) > RT / (2 x A x B ) (2.6)
They generally exhibit a UCST, Tcs = bAB(T )/2R. It must be remembered that
only pairs of liquids that mix at least partly behave as regular mixtures and
that Eqs. (2.5) and (2.6) cannot be used for the general case of any two liquids,
except as a very rough approximation. The value of bAB(T ) can be estimated
from the properties of the individual liquids to be mixed, by invoking their
solubility parameters [Eq. (2.1)]. The expression is:
bAB (T ) = VA* VB* ( x A VA* + x B VB* ) −1 (δ A2 − δ B2 ) (2.7)
where V *A and V *B are the respective molar volumes of the neat liquids. [Eq.
(2.7) makes bAB(T ) composition dependent and should be used only if the mo-
lar volumes are not too disparate, since otherwise the mixture is not a regular
one.] Thus, the larger the difference between the solubility parameters of the
two liquids, the more likely is bAB(T ) to be larger than the right-hand side of
Eq. (2.6), and the larger is the tendency of the two-component system to split
into two liquid phases [3].
If the mutual miscibility is small, then approximately
ln x A ′ ≈ −VB* (δ A − δ B ) 2 / RT (2.8)
where x′A is the mole fraction of component A in the B-rich phase marked by
′. A similar equation holds for x″B, in the A-rich phase, marked ″, with V *A
replacing V *B . Thus, the composition of the two phases can also be estimated
by the application of the solubility parameters. These considerations can also
be used for an estimate of the distribution of a solute between the immiscible
liquids A and B, assuming the conditions for regular mixing to be fulfilled for
all the components (see section 2.8.2).
As an example consider the ethylene glycol (A)–benzene (B) system at
25°C, assuming it to be a regular mixture (actually it is not). The relevant
quantities are V *A = 55.9 and V *B = 89.9 cm3 mol−1, δA = 32.4 and δB = 16.0 J1/2
cm−3/2. Hence for the equimolar mixture xA = xB = 0.5, according to Eq. (2.7),
bAB = 54.7 kJ mol−1 Ⰷ RT/ 2xAxB = 5 kJ mol−1, so that this system ought to split
into two liquid phases, as it does. In the benzene-rich phase (′) the ethylene
glycol content should be 0.005 wt% according to Eq. (2.8) and the benzene
content in the ethylene glycol–rich phase (′′) should be 0.31 wt%. Most regular
mixtures, however, would form a single homogeneous liquid phase, since the
condition of Eq. (2.6) is rarely met.
2.2.3 The Mutual Solubility of Solvents

An inspection of Table 2.1 shows that, of the common liquids, water has by far
the largest solubility parameter, ⬃48 J1/2 cm−3/2 [1,2]. Many organic solvents
have appreciably smaller values of δ. Pairs of liquids that have values of δ that
are less than 4 units apart have a reasonably large mutual solubility. They may
even be completely miscible. Thus, aliphatic hydrocarbons mix with one another
or with aromatic hydrocarbons to form a single homogeneous phase, as they
also do with most halogenated hydrocarbons. Pairs of liquids having values of
δ that are much more than 4 units apart are partially or nearly completely immis-
cible. This, however, does not apply well when one of the pair is water, unless
it is assigned the value δ ⬵ 30 J1/2 cm−3/2, which is comparable with that of
methanol.
Water mixes completely with the lower alcohols (methanol, ethanol, 1-
and 2-propanol, and t-butanol), as well as with most polyhydric alcohols and
the lower ether alcohols (2-methoxy-ethanol, and so on). Of the ethers, water is
miscible with tetrahydrofuran, 1,2-dimethoxyethane, and 1,4-dioxane, but not
with the higher members. Among the ketones, water is miscible with acetone
only; among the carboxylic acids, water is miscible with those up to and includ-
ing butanoic acid, but with none of the common esters (an exception being
ethylene carbonate). Among nitrogen-containing organic liquid compounds, wa-
ter is miscible with acetonitrile, but not the higher ones (except dinitriles, such
as succinonitrile). Water is miscible with primary, secondary, and tertiary
38 Marcus
amines, up to a total of about five carbon atoms in all the chains, including
pyridine, but not with nitro compounds. Water is also miscible with many difun-
ctional liquid compounds, such as diamines, alcohol amines, amides, and others,
and with such compounds as dimethylsulfoxide [1–3].
The higher members of homologous series based on these compounds
become increasingly immiscible with water as the chain length increases or as
aromatic rings are added. Consequently, these higher members can be used as
solvents in extraction systems. Table 2.2 lists the mutual solubilities, in weight
percent (wt%), of water and representative organic solvents at 25°C, unless
noted otherwise.
2.3 SOLUTE–SOLVENT INTERACTIONS

When a solute particle is introduced into a liquid, it interacts with the solvent
particles in its environment. The totality of these interactions is called the solva-
tion of the solute in the particular solvent. When the solvent happens to be
water, the term used is hydration. The solvation process has certain conse-
quences pertaining to the energy, the volume, the fluidity, the electrical conduc-
tivity, and the spectroscopic properties of the solute–solvent system. The appar-
ent molar properties of the solute ascribe to the solute itself the entire change
in the properties of the system that occur when 1 mol of solute is added to
an infinite amount of solution of specified composition. The solvent is treated
in the calculation of the apparent molar quantities of the solute as if it had the
properties of the pure solvent, present at its nominal amount in the solution. The
magnitudes of quantities, such as the apparent molar volume or heat content,
do convey some information on the system. However, it must be realized that
both the solute and the solvent are affected by the solvation process, and more
useful information is gained when the changes occurring in both are taken into
account.
2.3.1 Interactions at the Molecular Scale

The solvation process can be envisioned as occurring in several stages, although
only the sum of the stage contributions to the overall process is measurable.
First, a cavity must be created in the solvent to accommodate the solute. Then
the solute is placed in the cavity and permitted to interact with its nearest neigh-
bors, eventually forming coordinate bonds with some of them, forming a new
entity, the solvated solute. Finally, this entity may interact further with its sur-
roundings, by orienting solvent molecules, by the formation or disruption of
hydrogen bonds, or by other interactions.
If the solute is charged (i.e., if it is an ion), it will orient the dipoles of a
polar solvent by its electrical field. Donor–acceptor bonds may be formed if the
Rydberg_5063-2_Ch02_R4_02-09-04 11:07:42
Table 2.2 Mutual Solubility of Water and Some Organic Solvents at 25°C
Solvent in water Water in solvent
Solvent wt% wt%
c-Hexane 0.0055 0.010a

n-Hexane 0.00123 0.0111a
n-Octane 6.6 × 10−7 0.0095a
n-Decane 5.2 × 10−8 7.2 × 10−5
n-Dodecane 3.7 × 10−9 6.5 × 10−5
Decalin (mixed isomers) <0.02 0.0063a
Benzene 0,179 0.0635
Toluene 0.0515 0.0334
Ethylbenzene 0.0152 0.043
p-Xylene 0.0156 0.0456
Dichloromethane 1.30 0.198
Chloroform 0.815a 0.093
Carbon tetrachloride 0.077 0.0135b
1,1-Dichloroethane 5.03a 0.096
1,2-Dichloroethane 0.81a 0.187
Trichloroethylene 0.137 0.32
Chlorobenzene 0.0488b 0.0327
1,2-Dichlorobenzene 0.0156 0.309
Carbon disulfide 0.210a 0.0142
1-Butanol 7.45 20.5
Isobutyl alcohol 10 16.9
1-Pentanol 2.19 7.46
Isoamyl alcohol 2.97 9.19
1-Hexanol 0.7061 7.42
1-Octanol 0.0538
2-Ethyl-l-hexanol 0.07a 2.6a
Phenol 8.66 28.72
m-Cresol 2.51c
Diethyl ether 6.04 1.468
Diisopropyl ether 1.2a 0.57a
Bis(2-chloroethyl) ether 1.02a 0.1a
Methyl ethyl ketone 24.0a 10.0a
Methyl isobutyl ketone 1.7 1.9
Cyclohexanone 2.3a 8.0a
Acetylacetone 16.6a 4.5a
Ethyl acetate 8.08 2.94
Butyl acetate 0.68a 1.2a
Propylene carbonate 17.5 8.3
Nitromethane 11.1 2.09
Nitrobenzene 0.19a 0.24a
Benzonitrile 0.2 1b
Quinoline 0.609a
Tri-n-butyl phosphate 0.039 4.67
Note. See Table 2.1 for some miscible solvents, marked by a.

a
At 20°C.
b
At 30°C.
c
At 40°C.
40 Marcus
solute and solvent have suitable electron pair donation and acceptance proper-
ties. Hydrogen bonds between the solute and neighboring solvent molecules
may form if one or the other or both are protic, i.e., have hydrogen atoms
bonded to electronegative atoms. The solute particle may also undergo changes
in the solvation process, its internal structure, if it has one, being affected by
the strengthening or weakening of certain bonds, by a redistribution of the par-
tial electrical charges on its atoms, or by the favoring of a certain conformation
of a flexible solute molecule.
Spectroscopic measurements may, in certain cases, yield direct informa-
tion on these interactions. On the other hand, thermodynamic values, obtained
by measuring certain bulk properties of the system, require the aid of statistical
mechanical methods to be related to specific interactions between the solute and
the solvent. However, the thermodynamic aspects of the solute–solvent interac-
tions reflect the preference of the solute for one solvent over another and,
thereby, determine distribution of the solute in a solvent extraction system.
2.3.2 Thermodynamics of the Interactions

What quantity describes best the totality of these solute–solvent interactions and
how can the various contributions to them be estimated? Let the pure solute B
be vaporized in an imaginary process to a gas, and let this gas be very dilute,
so that it obeys the ideal gas laws. In this condition each particle of the solute
(molecule or ion) is very remote from any neighbor and has no environment
with which to interact. If B is polyatomic, it does have its internal degrees of
freedom, such as bond vibrations and rotation of the particle.
If this ideal gaseous particle B is introduced into a liquid A at a given
temperature and pressure, all of the solute–solvent interactions become
“switched on,” the solvation process takes place, and the Gibbs energy of solva-
tion, ∆solvGB, is released. In many cases the process of dissolution of a gaseous
solute in a liquid solvent can, indeed, be carried out experimentally—for in-
stance, when propane or carbon dioxide is dissolved in water to give a solution
at a given gas pressure.
The Gibbs energy change for the process of dissolution of the gaseous
solute B, ∆solnGB, is the driving force for the material transfer. When equilibrium
is reached, ∆solnGB becomes zero (since, at equilibrium, no more net transfer
occurs). The following equation then holds:
∆ soln GB eq = 0 = ∆ soln GB° + RT ln [cB (l) / cB (g)]eq (2.9)
where ∆solnG°B is the standard molar Gibbs energy of solvation of the solute B,
defined by Eq. (2.9) and cB is its molar concentration (moles per unit volume) in
the designated phases, (l) for liquid and (g) for gaseous. Equilibrium is generally
assumed for the processes discussed elsewhere in this chapter, and is only em-
phasized here for Eq. (2.9) by the subscript eq for ∆solnGB and the ratio of con-
centrations, cB(l)/cB(g). The concentration in the gas phase is, of course, given
by the pressure according to the ideal gas law cB(g) = nB/V = P/RT. The ratio of
the concentrations is known as the Bunsen coefficient, KB(B,A), for the solubility
of the gaseous solute B in liquid A. At low pressures and concentrations KB(B,A)
depends only on the temperature and is independent of the pressure and the
concentration.
Whether obtained from an actual experimentally feasible process or from
a thought process, ∆solnG°,B which is obtained from Eq. (2.9) by re-arrangement,
pertains to the solvation of the solute and expresses the totality of the solute–
solvent interactions. It is a thermodynamic function of state, and so are its deriv-
atives with respect to the temperature (the standard molar entropy of solvation)
or pressure. This means that it is immaterial how the process is carried out, and
only the initial state (the ideal gaseous solute B and the pure liquid solvent) and
the final state (the dilute solution of B in the liquid) must be specified.
Because it is a function of state, ∆solnG°B may be considered to be made up
additively of the contributions from the various stages in which the transfer of
the solute particle from the gaseous state into the liquid solvent has been envis-
aged by the foregoing to take place.
The first stage is the creation of the cavity in the liquid to accommodate
the solute. Obviously, work must be done against the cohesive forces of the
liquid that hold its molecules together. This work should be proportional to the
required size of the cavity, and increase as the volume of the solute increases.
An expression for this work would be
∆ cav G = Acav VBδ A2 (2.10)
where Acav is a proportionality coefficient, VB is the molar volume of the solute
B, and δA2 is the cohesive energy density of the solvent A. Thus, in a series
of solvents for a given solute, the positive contribution of cavity formation to
∆solnG°B increases with the squares of the solubility parameters of the solvents.
For a series of solutes and a given solvent, it increases with the molar volumes
of the solutes.
It is more difficult to estimate the contribution from the dispersion forces
to the solute–solvent interactions. Their energy increases with the product of
the polarizabilities of the partners, but decreases strongly with the distance
between them (being proportional to the inverse sixth power of the distance).
The polarizability is related to the molar volume, hence, to the third power of
the linear dimension of the solute or the solvent. Hence, the product of the
polarizabilities depends on the sixth power of the distance between the centers
of the interacting molecules. Consequently, these tendencies balance each
other. For large molecules (e.g., metal chelates, liquid hydrocarbons), it is bet-
42 Marcus
ter to consider the interactions between adjacent segments of neighboring mol-

ecules, which are at a constant mean distance from one another in the liquid
solution. The contribution from the dispersion forces (negative, because they
are attractive) is proportional to the surface areas of the interacting molecules,
or to the number N of segments present, and depends on their chemical natures.
If A represents again the liquid solvent, then [3]
∆ disp G = ∑ AiBA N iB N iA (2.11)
where AiBA is the (negative) interaction Gibbs energy of a pair of segments of
kind i, and the summation extends over all the different kinds of segments. A
methylene group, a halogen atom, a −CH=CH− group of an aromatic ring, or
some other functional group, generally serves as a segment.
There may be additional, specific interactions between the solute and
the solvent. Hydrogen bonds may be formed between them, particularly in
protic solvents, i.e., solvents that contain hydrogen atoms bonded to oxygen
(more rarely, nitrogen) atoms, such as water, alcohols, carboxylic acids, or
acidic phosphoric esters. Hydrogen bonds are formed with solute anions, with
hydrated ions in general (having an outer surface of water molecules), and with
neutral solutes that have a very basic atom with a lone pair of electrons that
can accept a hydrogen bond. Also, donor–acceptor bonds can be formed if the
solvent has a very basic atom in its structure donating a pair of unshared elec-
trons, if it is suitably exposed, and the solute is a cation or some acidic neutral
molecule accepting this pair.
A generalized equation for ∆solvG°B is [4]
∆ solv GB° = A0 + Aπ π * + Aα α + Aββ + Aδ δ 2 (2.12)

which describes the value of ∆solvG°B for a given solute (characterized by A0, Aπ,
Aα, Aβ, and Aδ) with a series of solvents. The solvents are characterized by their
Taft–Kamlet solvatochromic parameters: π* for polarity–polarizability, α for
hydrogen bond donation acidity, and β for hydrogen bond acceptance basicity.
Values of these solvatochromic parameters have been tabulated for many sol-
vents. (Table 2.3 gives the values for selected solvents.) As before, δ is the
solubility parameter. The first two terms on the right-hand side of Eq. (2.12)
express ∆dispG, the next two the hydrogen-bonding interactions, and the last one
∆cavG.
As an example, consider phenol as the solute and water and toluene as
two solvents. The parameters for phenol are Aπ = 5.7, Aα = −12.9, Aβ = −18.3,
and Aδ = 0.0091, whereas A0 is unspecified, but a negative quantity. With the
solvent parameters from Tables 2.1 and 2.3, the standard Gibbs energy of sol-
vation of phenol in water becomes A0 + 3.39, and in toluene A0 + 4.11 kJ mol−1.
It is seen that ∆solvG°B is lower in water than in toluene, so that the transfer of
phenol from water to toluene entails an increase in ∆solvG°. B The consequence
of this is that phenol prefers water over toluene, since work would be required
to make this transfer. It should be remembered that the standard Gibbs energies
of solvation refer to the state of infinite dilution of the solute (solute–solute
Table 2.3 Solvatochromic Parameters for Some Solvents and (Monomeric) Solutes (in Parentheses)
Substance π* α β Substance π* α β
c-Hexane 0 0 0 2-Butanone 0.60 0.06 0.48

n-Hexane −0.11 0 0 Cyclohexanone 0.68 0 0.53
Benzene 0.55 0 0.10 Acetophenone 0.81 0.04 0.49
Toluene 0.49 0 0.11 Acetic acid 0.64 1.12 0.45
p-Xylene 0.45 0 0.12 Hexanoic acid 0.52 1.22 0.45
Dichloromethane 0.82 0.13 0.10 Benzoic acid 0.80 (0.87) (0.40)
Chloroform 0.58 0.20 0.10 Ethyl acetate 0.45 0 0.45
Carbon tetrachloride 0.21 0 0.10 Butyl acetate 0.46 0 0.45
1,2-Dichloroethane 0.73 0 0.10 Propylene carbonate 0.83 0 0.40
Chlorobenzene 0.68 0 0.07 Formamide 0.97 0.71 0.48
Water 1.09 1.17 0.47 Dimethylformamide 0.88 0 0.69
Water monomer (0.39) (0.32) (0.15) Tetramethylurea 0.79 0 0.80
Principles of Solubility and Solutions
Methanol 0.60 0.98 0.66 Hexamethyl phosphoramide 0.87 0 1.00

Methanol monomer (0.39) (0.38) (0.41) Acetonitrile 0.66 0.19 0.40
Ethanol 0.54 0.86 0.75 Benzonitrile 0.88 0 0.37
Ethanol monomer (0.39) (0.36) (0.47) Nitromethane 0.75 0.22 0.06
1-Propanol 0.52 0.84 0.90 Nitrobenzene 0.86 0 0.30
2-Propanol 0.48 0.76 0.84 Triethylamine 0.09 00.71
1-Hexanol 0.40 0.80 0.84 Tri-n-butylamine 0.06 0 0.62
Trifluoroethanol 0.73 1.51 0 Dimethyl sulfoxide 1.00 0 0.76
Trifluoroethanol monomer (0.59) (0.10) Diphenyl sulfoxide 0 0.70
Phenol 0.72 1.65 0.30 Sulfolane 0.90 0 0.39
Diethyl ether 0.24 0 0.47 Triethyl phosphate 0.69 0 0.77
Diisopropyl ether 0.19 0 0.49 Tributyl phosphate 0.63 0 0.80
Tetrahydrofuran 0.55 0 0.55 Triethyl phosphine oxide 0 1.05
Dioxane 0.49 0 0.37 Triphenyl phosphine oxide 0 0.94
Anisole 0.70 0 0.32 Pyridine 0.87 0 0.64
1,2-Dimethoxyethane 0.53 0 0.41 Quinoline 0.93 0 0.64
Acetone 0.62 0.08 0.48
43
Source: Ref. 2.
44 Marcus
interactions being, therefore, absent), and that a reaction such as ionic dissocia-
tion of the phenol in water is ignored in this example.
Extensive tables of solute parameters are beyond the scope of this book.
Equations (2.10) to (2.12) are meant to show the nature of the dependencies of
the additive terms on various quantities. They enable the prediction of tenden-
cies of solute–solvent interactions for a given solute with a series of solvents or
for a series of solutes with a given solvent.
2.3.3 Solvation of Electrolytes

If the solute B is an electrolyte that dissociates into ions in the solvent in ques-
tion, the solvation of each ion separately may be considered in the hypothetical
transfer process, although experimentally only quantities pertaining to the entire
electrolyte can be measured [5]. The ions of a crystalline salt are brought from
their standard solid state into the ideal gas state by the investment of the lattice
Gibbs energy (approximately the lattice energy for a salt dissociating into one
cation and one anion). The transfer of an ion from the ideal gas to the liquid
is accompanied by reorientation of the dipoles of the solvent molecules. The
contribution to ∆solvG°B of this reorientation is given by the Born equation:
∆ elG = − Ael z 2 r −1 (1 − 1 / ε) (2.13)
where Ael = NAe2/8πε0 = 69.5 kJ mol−1 (with r in nm), z is the charge on the ion
in proton charge units, r is a distance related to the radius of the ion in nm
(Table 2.4), and ε is the relative permittivity (dielectric constant) of the solvent.
The Born equation should be applied only from a distance r from the center of
the ion that is larger than the radius of the bare ion by ∆r, which depends on
the sizes of the ion B and the solvent molecules, A. Closer to the ion, the
phenomenon of dielectric saturation sets in (Fig. 2.3), and the solvent cannot
reorient itself freely because the binding to the ion becomes too strong. The
contribution from this inner shell of solvation to ∆solvG°B must be estimated by
other means (see, e.g., Ref. [5]), but for polar solvents can be taken to be only
slightly dependent on the nature of the solvent (except that it may depend on
the thickness of this inner shell). In any event, Eq. (2.13) shows that the contri-
bution of the reorientation of the solvent by the ion depends on the square of
the charge on the ion and, reciprocally, on both the size of the ion and the
relative permittivity of the solvent.
Terms for the electrostatic interactions [Eq. (2.13)] for the region out-
side the first solvation shell, and an appropriate one for the inner region, must
be added to Eq. (2.12) for each ion of an electrolyte B, for the evaluation of
∆solvG°.
B Since cations do not accept hydrogen bonds and anions do not donate
them, except when protonated, like HSO−4, the term in α of the solvent becomes
unimportant for cations and that in β of the solvent for anions.
Table 2.4 Crystal Ionic Radii and Standard Molar Gibbs Free Energies
of Hydration of Ions
−∆hydG° −∆hydG° −∆hydG°
Ion r (nm) (kJ/mol) Ion r (nm) (kJ/mol) Ion r (nm) (kJ/mol)
H+ 1056 F− 0.133 472

Li+ 0.069 481 A13+
0.053 4531 Cl− 0.181 347
Na+ 0.102 375 Sc3+ 0.075 3801 Br− 0.196 321
K+ 0.138 304 Y3+ 0.090 3457 I− 0.220 283
Rb+ 0.149 281 La3+ 0.105 3155 OH− 0.133 439
Cs+ 0.170 258 Ce3+ 0.101 3209 SH− 0.207 303
NH+4 0.148 292 Nd3+ 0.099 3287 CN − 0.191 305
Me4N+ 0.280 412 Gd3+ 0.094 3385 SCN − 0.213 287
Ag+ 0.115 440 Ho3+ 0.090 3480 N −3 0.195 287
Ph4As+ 0.425 −32 Lu3+ 0.086 3522 BF −4 0.230 200
Mg2+ 0.072 1838 Pu3+ 0.101 3245 ClO−3 0.200 287
Ca2+ 0.100 1515 Am3+ 0.100 3297 BrO−3 0.191 340
Sr2+ 0.113 1386 Cr3+ 0.062 4010 IO−3 0.181 408
Ba2+ 0.136 1258 Fe3+ 0.065 4271 ClO−4 0.240 214
Mn2+ 0.083 1770 Ga3+ 0.062 4521 MnO−4 0.240 245
Fe2+ 0.078 1848 In3+ 0.079 3989 NO−2 0.192 339
Co2+ 0.075 1922 TI3+ 0.088 3976 NO−3 0.179 306
Ni2+ 0.069 1992 Bi3+ 0.102 3486 CH2CO−2 0.232 373
Cu2+ 0.073 2016 Ce4+ 0.080 6129 BPh−4 0.421 −42
Zn2+ 0.075 1963 Th4+ 0.100 5823 CO23− 0.178 1479
Cd2+ 0.095 1763 U4+ 0.097 6368 C2O24− 0.21 1200
Hg2+ 0.102 1766 Zr4+ 0.072 6799 SO23− 0.20 1303
Pb2+ 0.118 1434 Hf4+ 0.071 6975 SO24− 0.230 1090
UO22+ 0.28 1229 CrO24− 0.240 958
PO34− 0.238 2773
Source: Ref. 6.
These considerations are important, since solvent extraction often involves

aqueous ions that must be transferred from an aqueous solution to some organic
solvent. (The latter may contain complexing reagents that interact with the ions.)
This transfer can be envisaged as taking place by steps in which the first one
involves the freeing of an ion from its hydration before it is permitted to react
with the solvent or reagent into which it is transferred. This freeing requires the
investment of work that is the negative of the standard Gibbs energy of hydra-
tion of the ion [5,6]. The specification of this quantity requires splitting the
value of an appropriate electrolyte into individual values for the cation and the
anion. This cannot be accomplished by purely thermodynamic means, and an
extrathermodynamic assumption or convention is needed. Several such assump-
46 Marcus
Fig. 2.3 A schematic representation of the hydration layer near a small ion (left) and
a large ion (right), showing the region where the water is dielectrically saturated (with a
low relative permittivity ε′), hence electrostricted (squeezed) and immobilized. The
thickness of this layer, ∆r, depends reciprocally on the size of the ion.
tions that lead to consistent results have been proposed. One of them involves
a reference electrolyte, of which the cation and anion are large, spherical, univa-
lent, and similar in all respects except the sign of the charge. Such a reference
electrolyte is tetraphenylarsonium tetraphenylborate (TATB); the standard molar
Gibbs energy of hydration of each of its ions is ∆hydG° = −38 ± 6 kJ mol−1. On
this basis, values of ∆hydG° of other ions can be obtained, consistent with certain
other extrathermodynamic assumptions, and are listed in Table 2.4 [6].
The differences in the solvation abilities of ions by various solvents are
seen, in principle, when the corresponding values of ∆solvG° of the ions are
compared. However, such differences are brought out better by a consideration
of the standard molar Gibbs energies of transfer, ∆tG° of the ions from a refer-
ence solvent into the solvents in question (see further section 2.6.1). In view of
the extensive information shown in Table 2.4, it is natural that water is selected
as the reference solvent. The TATB reference electrolyte is again employed to
split experimental values of ∆tG° of electrolytes into the values for individual
ions. Tables of such values have been published [5–7], but are outside the scope
of this text. The notion of the standard molar Gibbs energy of transfer is not
limited to electrolytes or ions and can be applied to other kinds of solutes as
well. This is further discussed in connection with solubilities in section 2.7.
2.4 THERMODYNAMICS OF SOLUTIONS

Thermodynamics is the branch of science dealing with the energetics of sub-
stances and processes. It describes the tendency of processes to take place spon-
taneously, the effects of external conditions, and the effects of the composition
of mixtures on such processes. Thermodynamics is generally capable of correlat-
ing a variety of data pertaining to widely changing conditions by relatively

simple formulae. One approach to such a correlation involves the definition of
a hypothetical ideal system and the subsequent consideration of deviations of
real systems from the ideal one. In many cases, indeed, such deviations are
relatively small and can be ignored in a first approximation. Such examples
are, for instance, a gas under low pressure or a dilute solution of a solute in
some solvent. In many other instances (unfortunately in many that pertain to
practical solvent extraction), such an approximation is far from being valid, and
quite incorrect estimates of properties of the real systems can result from ignor-
ing the deviations from the ideal.
2.4.1 Ideal Mixtures and Solutions

2.4.1.1 One Liquid Phase
Consider two liquid substances that are rather similar, such as benzene and
toluene or water and ethylene glycol. When nA moles of the one are mixed with
nB moles of the other, the composition of the liquid mixture is given by specifi-
cation of the mole fraction of one of them [e.g., xA, according to Eq. (2.2)]. The
energy or heat of the mutual interactions between the molecules of the compo-
nents is similar to that of their self interactions, because of the similarity of the
two liquids, and the molecules of A and B are distributed completely randomly
in the mixture. In such mixtures, the entropy of mixing, which is a measure of
the change in the molecular disorder of the system caused by the process of
mixing the specified quantities of A and B, attains its maximal value:
∆ M SAB = −R[ xA ln xA + xB ln xB ] (2.14)
per mole of mixture. This is the molar entropy of mixing expected for an ideal
mixture. The molar heat of mixing of such a mixture, ∆MHAB, is zero, since no
net change in the energies of interaction takes place on mixing. Therefore, the
molar Gibbs energy of mixing, in the process that produces an ideal mixture, is:
∆ MGAB = ∆ M H AB − T ∆ M SAB = RT [ xA ln xA + xB ln xB ] (2.15)
Since the mole fractions are quantities that are smaller than unity, their loga-
rithms are negative, ∆MSAB is positive, and ∆MGAB is negative, the process of
mixing being a spontaneous process under the usual conditions. Equations
(2.14) and (2.15) suffice to define an ideal (liquid) mixture.
The solute and the solvent are not distinguished normally in such ideal
mixtures, which are sometimes called symmetric ideal mixtures. There are, how-
ever, situations where such a distinction between the solute and the solvent is
reasonable, as when one component, say, B, is a gas, a liquid, or a solid of
limited solubility in the liquid component A, or if only mixtures very dilute in
B are considered (xB Ⰶ 0.5). Such cases represent ideal dilute solutions.
48 Marcus
The molar Gibbs energy G of an ideal mixture, whether symmetric or

dilute, consists of the molar Gibbs energies of the pure components A and B,
weighted by their mole fractions, plus the molar Gibbs energy of mixing, ∆MGAB.
G = xAGA* + xBGB* + ∆ MGAB (2.16)
The chemical potential, µ, of a component of the mixture is the partial
derivative of the Gibbs energy of the mixture with respect to the number of
moles of this component present, the number of moles of all the other compo-
nents being held constant, as are also the temperature and the pressure. For the
component A in a mixture containing also B, C, . . . , the chemical potential is
µA = (∂G/∂nA)P,T,nB,nC, . . ., whether the mixture is ideal or not. In the ideal mixture
the chemical potential of A is thus obtained from Eqs. (2.15) and (2.16) on
carrying out the partial differentiation, yielding:
µ A = µ°A + RT ln xA (2.17)
As the mole fraction of A in the mixture increases toward unity, the second
term in Eq. (2.17) tends toward zero, and the chemical potential of A tends
toward the standard chemical potential, µ°A = G*A , the molar Gibbs energy (or
chemical potential) of A in the realizable standard state of pure A, in the sense
that pure liquid A is a known chemical substance. A similar equation holds for
the component B.
Consider a dilute ideal solution of the solute B (which could be gaseous,
liquid, or solid at the temperature in question) in the solvent A. Suppose that
more concentrated solutions do not behave ideally and, in particular, the state
of pure liquid B cannot be attained by going to more and more concentrated
solutions (e.g., by removing A by volatilization). It is possible to define a stan-
dard chemical potential pertaining to a hypothetical standard state of the ideal
infinitely dilute solution as the limit:
µ ∞B = lim (µ B − RT ln xB ) (2.18)
xB → 0
Although µB tends to − ∞ as xB tends to 0 (and ln xB also tends to − ∞ ), the

difference on the right-hand side of Eq. (2.18) tends to the finite quantity µ∞B,
the standard chemical potential of B. At infinite dilution (practically, at high
dilution) of B in the solvent A, particles (molecules, ions) of B have in their
surroundings only molecules of A, but not other particles of B, with which to
interact. Their surroundings are thus a constant environment of A, independent
of the actual concentration of B or of the eventual presence of other solutes, C,
D, all at high dilution. The standard chemical potential of the solute in an ideal
dilute solution thus describes the solute–solvent interactions exclusively.
A solution in which the solvent A obeys the relationship of Eq. (2.17) and
the solute B obeys the expression
µ B = µ ∞B + RT ln xB (2.19)
over a certain (low) concentration range of B in A is said to obey Raoult’s law
for the solvent and Henry’s law for the solute in this concentration range (Fig.
2.4).
The vapor pressure of the solvent in such a case is given by
pA = xA pA* (2.20)
where p*A is the vapor pressure of pure liquid A at the given temperature. Con-
versely, when Eq. (2.20) is followed by the vapor pressure of the solvent,
Raoult’s law is said to be valid and Eq. (2.17) is obeyed. The vapor pressure of
the solute is also proportional to its mole fraction, if Henry’s law is obeyed.
However, the proportionality constant is not its vapor pressure in the pure liquid
state, which may not be attainable at the given temperature. Instead, the expres-
sion
Fig. 2.4 The vapor pressure diagram of a dilute solution of the solute B in the solvent
A. The region of ideal dilute solutions, where Raoult’s and Henry’s laws are obeyed by
the solvent and solute, respectively, is indicated. Deviations from the ideal at higher
concentrations of the solute are shown. (From Ref. 3.)
50 Marcus
pB = K B(A) xB (2.21)
is followed, where KB(A) is called the Henry’s law constant, and depends on the
chemical natures of both B and A, as the molecular description of the foregoing
solutions requires. These considerations apply also when there are several sol-
utes present at low concentrations, so that Raoult’s law may apply to the solvent
and Henry’s law to each of the solutes.
2.4.1.2 Two Liquid Phases
Consider now two practically immiscible solvents that form two phases, desig-
nated by ′ and ″. Let the solute B form a dilute ideal solution in each, so that
Eq. (2.19) applies in each phase. When these two liquid phases are brought into
contact, the concentrations (mole fractions) of the solute adjust by mass transfer
between the phases until equilibrium is established and the chemical potential
of the solute is the same in the two phases:
µ B′′ = µ B∞′′ + RT ln x B′′ = µ B′ = µ B∞′ + RT ln x B′ (2.22)
(It is the difference in the chemical potentials of the solute that is the driving
force for the mass transfer.) This equation can be rewritten in the form:
x B′ / x B′′ = exp[( µ B∞′′ − µ B∞′ ) / RT ] = PB (2.23)
where xB′ /xB″ = DB is the distribution ratio of the solute B (on the mole fraction
scale) and the exponential expression is a constant (at a given temperature),
called the distribution constant of B,* PB. Equation (2.23) is an expression of
Nernst’s distribution law that states:
The distribution ratio of a solute between two liquid phases at equilibrium is
a constant, provided that the solute forms a dilute ideal solution in each phase.
2.4.2 Nonideal Mixtures and Solutions

For most of the situations encountered in solvent extraction the gas phase above
the two liquid phases is mainly air and the partial (vapor) pressures of the
liquids present are low, so that the system is at atmospheric pressure. Under
such conditions, the gas phase is practically ideal, and the vapor pressures repre-
sent the activities of the corresponding substances in the gas phase (also called
their fugacities). Equilibrium between two or more phases means that there is
no net transfer of material between them, although there still is a dynamic ex-
change (cf. Chapter 3). This state is achieved when the chemical potential µ as
*There is no internationally recommended symbol for the distribution constant, but the symbol P
(for partition constant) is common usage in this context. Coordination (complex) chemists (as in
Chapter 3) prefer the symbol KD (for the German “Konstante” for “constant”); sometimes other
symbols are used. The important point is that the symbol should be properly defined in the text.
well as the activity a of each substance are the same in all the phases. It is the
inequality of the activities of a substance in two phases that causes some of the
substance to transfer from the one (where it is higher) to the other phase, until
equality is achieved. The activity of a pure liquid or solid substance is defined
as unity. In any mixture, whether ideal or not, the activity of a component A,
aA, is related to its chemical potential by [3]:
µ A = µ°A + RT ln aA (2.24)
Therefore, if component A is the solvent in a solution, and it obeys Raoult’s
law when the solution is dilute in all solutes, then aA = xA under these conditions.
Otherwise, the solution is nonideal, and the deviation from the ideal is described
by means of the activity coefficient
f A = aA / xA (2.25)
where aA is given by Eq. (2.24) or can also be expressed by the ratio of the
vapor pressures, pA/p*A (neglecting interactions in the vapor phase, otherwise the
fugacities have to be employed). As the solution is made more dilute and ap-
proaches the dilute ideal solution and the pure solvent, the activity coefficient
fA approaches unity:
lim f A = 1 (2.26)
xA →1
Equation (2.24) can now be rewritten as

µ A = µ*A + RT ln xA + µ EA = µ*A + RT ln xA + RT ln f A (2.27)
where µ is the excess chemical potential of A in the solution and is equal to
E
A
RT ln fA, expressing the deviation of the solution from ideality.
As already mentioned, in solutions that are not symmetric mixtures it is
conventional to treat the solute B (or solutes B, C, . . . ) differently from the
treatment accorded to the solvent A. The activity coefficient is still defined as
the ratio aB/xB, but the limit at which it equals unity is taken as the infinite dilute
solution of B in A:
lim f B = 1 (2.28)
xB →0
so that
µ B = µ ∞B + RT ln xB + µ EB = µ ∞B + RT ln xB + RT ln f B (2.29)
The activity coefficients fA and fB may be smaller or larger than unity;
hence, the excess chemical potentials µAE and µBE may be negative or positive.
Depending on the sign of µAE, the solution is said to exhibit positive or negative
deviations from Raoult’s law. Similarly, positive or negative deviations from
Henry’s law can be noted (see Fig. 2.4).
52 Marcus
The considerations that lead to Eq. (2.22) also apply to the case of non-
ideal solutions, except that the terms RT ln fB′ and RT ln fB″ must be added to
the left- and right-hand sides. When rearranged into the form of Eq. (2.23), the
result is
DB = x B′ / x B′′ = PB ( fB′′ / fB′ ) (2.30)
Therefore, the distribution ratio of B remains constant only if the ratio of the
activity coefficients is independent of the total concentration of B in the sys-
tem, which holds approximately in dilute solutions. Thus, although solutions
of metal chelates in water or nonpolar organic solvents may be quite nonideal,
Nernst’s law may still be practically obeyed for them if their concentrations
are very low (xchelate < 10−4). Deviations from Nernst’s law (constant DB) will in
general take place in moderately concentrated solutions, which are of particular
importance for industrial solvent extraction (see Chapter 12).
When no distinction between the solvent and the solute in a liquid mix-
ture is made, then nonideal mixtures can still be described by means of an
expression similar to Eq. (2.16), but with the addition of a term that is the
excess molar Gibbs energy of mixing, ∆GAB E
(see also section 2.2). Thus the
Gibbs energy per mole of mixture is:
G = x B GA* + x B GB* + RT [ x A ln x A + x B ln x B ] + ∆GAB
E
(2.31)
where ∆G = xAµ + xBµ = RT [xA ln fA + xB ln fB] is a function of the compo-
E
AB
E
A
E
B
sition. It is this quantity that has been used in Eq. (2.5) to define the variable
bAB(T), which describes the tendency of the mixture to separate into two liquid
phases. The excess Gibbs energy is positive (>2RT) for mixtures that tend to
separate, but is negative when the components have a strong mutual attraction
of their molecules. For some mixtures, ∆G AB E
is positive over a part of the
composition range and negative over another part.
When a partial derivative of ∆G AB E
relative to the number of moles of a
component, say, A, is taken, the result is RT times the (natural) logarithm of
the activity coefficient of this component, RT lnf′A. When the partial derivative
relative to the temperature is taken, the negative of the excess entropy of mix-
ing, −∆S AB
E
, results. For a regular solution (see section 2.2) this quantity is zero,
so that the (excess) heat (enthalpy) of mixing is ∆H AB E
= ∆G AB
E
. The qualifier
“excess” for the enthalpy of mixing is put in parentheses, since ideal mixtures
have no enthalpies of mixing, so that if heat is evolved or absorbed when two
liquids are mixed, the mixture is not ideal and the observed heat (enthalpy) of
mixing is always an excess quantity. The enthalpy of mixing, ∆H AB E
, may or
may not be temperature dependent itself, and may be positive, negative, or
may change sign within the composition range. A negative ∆H AB E
is, again,
indicative of strong mutual interactions of the molecules of the components.
An example of such behavior is met with in aqueous ethanol (EtOH) at,
say, 75°C. At xEtOH < 0.24, the heat of mixing is negative (heat is evolved on
mixing the components), but at higher ethanol contents it is positive (heat
is absorbed and the mixture cools). In the equimolar mixture ∆H AB E
= 220
−1
but ∆G AB = 960 J mol , due to a negative entropy of mixing, but since
E
2xEtOHxwaterRT = 1446 J mol−1 for this mixture, a single liquid phase is stable.
However, for aqueous 1-butanol (BuOH) at 25°C, ∆G AB E
≥ 2xBuOHxwaterRT =
1220 J mol for the near equimolar mixture, although ∆H AB = 500 J mol−1 only,
−1 E
so the mixture is on the verge of splitting into two liquid phases (it splits when
xBuOH = 0.485).
2.4.3 Scales of Concentration

The composition of a mixture need not be given in terms of the mole fractions
of its components. Other scales of concentration are frequently used, in particu-
lar, when one of the components, say, A, can be designated as the solvent and
the other (or others), B, (C, . . . ) as the solute (or solutes). When the solute is
an electrolyte capable of dissociation into ions (but not only for such cases), the
molal scale is often employed. Here, the composition is stated in terms of the
number of moles of the solute, m, per unit mass (1 kg) of the solvent. The
symbol m is used to represent the molal scale (e.g., 5 m = 5 mol solute/1 kg
solvent). The conversion between the molal and the rational scale (i.e., the mole
fraction scale, which is related to ratios of numbers of moles [see Eq. (2.2)]
proceeds according to Eqs. (2.32a) or (2.32b) (cf. Fig. 2.4):
mB = xB / MA (1 − xB ) ≈ xB / MA (2.32a )
xB = mB /(mB + 1 / MA ) ≈ mB MA (2.32b)
where MA is the molar mass of the solvent A, expressed in kilograms per
mole (kg mol−1). For aqueous solutions, for instance, MA = 0.018 kg mol−1 and
1/MA = 55.5 mol kg−1. The approximate equalities on the right-hand sides of
Eqs. (2.32a) and (2.32b) pertain to very dilute solutions. Activity coefficients
on the molal scale are designated by γ, so that Eq. (2.29) becomes
µ B = µ ∞B (m) + RT ln mB + RT ln γ B (2.33)
∞
where, again, µ = lim(mB → 0) (µB − RT ln mB) is the standard chemical potential
B(m)
of B on the molal scale, and lim(mB → 0) γB = 1.
Another widely used concentration scale is the molar scale, that describes
the number of moles of the solute, c, per unit volume of the solution [i.e., per
1 dm3 = 1 L (liter)]. The symbol M = mol/L is used to represent this scale [e.g.,
0.5 M = 0.5 mol solute/1 liter (L) of solution]. Conversion between the molar
and molal scales is made according to Eqs. (2.34a) and (2.34b) by means of the
density of the solution, ρ, in kg L−1(= g mL−1 = g cm−3; see Fig. 2.5):
cB = mB ρ / (1 + mB MB ) ≈ mB ρ A / (1 + mB MB ) (2.34a )
mB = cB / ( ρ − c B M B ) ≈ cB / ( ρ A − cB M B ) (2.34 b)
where MB is the molar mass of the solute (in kg mol−1). The approximate equali-
ties on the right-hand sides of Eqs. (2.34a) and (2.34b) pertain to very dilute
54 Marcus
Fig. 2.5 Concentration scales. The molality mB [in mol/(kg solvent)] and molarity cB
[in mol (L solution)−1] of solute B, which has a molar mass MB 0.100 kg mol−1 and a
molar volume VB = 0.050 L mol−1, in the solvent A, which has MA = 0.018 kg mol−1 and
VA = 0.018 L mol−1 (water), are shown as a function of the mole fraction xB of the solute.
Note that the molarity tends toward a maximal value (1/VB), whereas the molality tends
toward infinity as xB increases toward unity.
solutions. For aqueous solutions at 25°C, ρA = 0.997 kg L−1, and a generally

negligible error is made if ρA is replaced by 1 for very dilute aqueous solutions.
The activity coefficient on the molar scale is designated by y, so that Eq. (2.29)
becomes
µ B = µ ∞B (c) + RT ln cB + RT ln yB (2.35)
where, as before, µ∞B(c) = lim(cB→0) (µB − RT ln cB) is the standard chemical poten-
tial of B on the molar scale, lim(cB→0) yB = 1.
As an example for such concentration scale conversions, consider 20 wt%
tri-n-butyl phosphate (TBP, B) in toluene (A). The mole fraction of the TBP
is xB = (20/MB)/[(20/MB) + (80/MA)] = 0.080, its molality is mB = (0.200/MB)/
0.800 = 0.94 m, and its molarity is cB = 0.94 ρA/(1 + 0.94 MB) = 0.65 M (MB is
taken in kg mol−1).
The quantity µB is independent of the concentration scale used, being a
true property of the solution, but the three standard chemical potentials µB∞(x),
µB∞(m), and µB∞(c) are not equal. Consequently, differences between the standard
chemical potentials of a solute in the two liquid phases employed in solvent
extraction also depend on the concentration scale used. Thus, PB defined in
Eq. (2.23) is specific for the rational concentration scale, and does not equal
corresponding quantities pertaining to the other scales. Therefore, Eq. (2.30)
might be rewritten with a subscript (x), to designate the rational scale, [i.e., with
DB(x) and PB(x)]. Similar expressions would then be
DB(m) = mB′ / mB′′ = PB(m) (γB′′/γB′) (2.36a )
for the molal scale, and
DB(c ) = cB′ / cB′′ = PB(c ) ( yB′′/yB′) (2.36b)
for the molar scale. The subscripts are often omitted, if the context unequivo-
cally defines the concentration scale employed.
2.5 AQUEOUS ELECTROLYTE SOLUTIONS

In many practical solvent extraction systems, one of the two liquids between
which the solute distributes is an aqueous solution that contains one or more
electrolytes. The distributing solute itself may be an electrolyte. An electrolyte
is a substance that is capable of ionic dissociation, and does dissociate at least
partly to ions in solution. These ions are likely to be solvated by the solvent
(or, in water, to be hydrated) [5]. In addition to ion–solvent interactions, the
ions will also interact with one another: repulsively, if of the same charge sign,
attractively, if of the opposite sign. However, ion–ion interactions may be negli-
gible if the solution is extremely dilute. The electrolyte Cν+Aν−, is made up of
ν+ positive ions, or cations, Cz+, and ν−− negative ions, or anions, Az−, or of
altogether ν = ν+ + ν− ions, and is designated in the following by the symbol
CA. A general principle that is always obeyed in electrolyte solutions is that of
the electroneutrality of the solution as a whole. This means that the number of
cations weighted by their charge numbers z+ (the number of protonic charges
each has) equals that of the anions similarly weighted with z−. Per unit volume
of the solution this can be written as:
56 Marcus
∑ c+ z+ + ∑ c− z− = 0 (2.37)
(Note that z− is a negative number, so that ν+ z+ + ν− z− = 0.) The electroneutral-
ity principle is equivalent to stating that it is impossible to produce a solution
that contains, for example, only cations or an excess of positive charge (i.e.,
that ions cannot be considered as independent components of solutions). Only
entire electrolytes are components that can be added to a solution.
It is expedient to consider the mean ionic properties of the electrolytes.
An electrolyte that dissociates into ν ions has a mean molality m± = (ν+ m+ +
ν− m−)/ν and a mean molarity c± = (ν+ c+ + ν− c−)/ν.
A useful concept that is used when the activities of electrolytes are calcu-
lated is that of the ionic strength of the solution. This is defined (on the molar
scale) as:
1
I= ∑ ci zi2 (2.38)
2
where the summation extends over all the cations and anions present in the
solution. Since the charges appear to the second power in the expression for the
ionic strength, all the terms are positive. An electrolyte that is completely disso-
ciated to univalent ions is designated as a 1:1 electrolyte, examples being aque-
ous HCl and LiNO3. If only a single 1:1 electrolyte is present at a molar concen-
tration cCA, then I = cCA. According to this notation, MgCl2, and (NH4)2SO4 are
2:1 and 1:2 electrolytes, respectively, for which I = 3 cCA. For 1:1, 2:2, . . . (i.e.,
symmetric) electrolytes, m± = mCA, and I = cCA, 4 cCA, . . . The relationship be-
tween mCA and cCA is given by Eq. (2.34.)
2.5.1 Electrolyte Activity*

The chemical potential of an electrolyte may formally (but not experimentally)
be considered as being made up from contributions of the individual kinds of
ions:
µ CA = vµ ± = v+µ + + v−µ − (2.39)
Mean ionic activities, molalities (or molarities), and activity coefficients, desig-
nated with subscript ±, are used in equations corresponding to Eqs. (2.33) or
(2.35) for electrolyte solutions:
µ CA = v µ ∞± + v RT ln m± + v RT ln γ ± (2.40)
Electrolyte solutions are typical nonsymmetrical solutions, in that the sol-
vent is treated according to Eq. (2.27) and the solute according to Eq. (2.40).
*See also Chapter 6.

The mole fraction concentration scale is generally used for the solvent water,
designated by subscript w and having molar mass Mw.
The activity of water is related formally to the molality of the electrolyte
by means of the osmotic coefficient, ϕ, of the solution:
ln aW = ln ( pW / pW* ) = − v m± M Wϕ (2.41)
As the solution becomes more dilute in the electrolyte (or, in all electrolytes
present, where ν m± is replaced by Σ νi m±i), ln aw approaches zero (aw ap-
proaches unity), and ϕ approaches unity.
A definite relationship exists between γ± and ϕ (the Gibbs–Duhem rela-
tionship), which may be expressed by the excess Gibbs energy of the solution
of an electrolyte:
∆G E = v m± RT [1 − ϕ + ln γ ± ] (2.42)
This expression is analogous to Eq. (2.3), in that (1 − ϕ) expresses the contri-
bution of the solvent and ln γ± that of the electrolyte to the excess Gibbs energy
of the solution. The calculation of the mean ionic activity coefficient of an
electrolyte in solution is required for its activity and the effects of the latter in
solvent extraction systems to be estimated. The osmotic coefficient or the activ-
ity of the water is also an important quantity related to the ability of the solu-
tion to dissolve other electrolytes and nonelectrolytes.
Electrostatic and statistical mechanics theories were used by Debye and
Hückel to deduce an expression for the mean ionic activity (and osmotic) coeffi-
cient of a dilute electrolyte solution. Empirical extensions have subsequently
been applied to the Debye–Hückel approximation so that the expression remains
approximately valid up to molal concentrations of 0.5 m (actually, to ionic
strengths of about 0.5 mol L−1). The expression that is often used for a solution
of a single aqueous 1:1, 2:1, or 1:2 electrolyte is
log γ ± = − Az+ z−I 1 / 2 /[1 + 1.5 I 1 / 2 ] + bI (2.43)
A is a constant that depends on the temperature, and at 25°C A = 0.509, and b
has the value 0.2. (Note that although I is given on the molar scale, the γ±
obtained from the semi-empirical Eq. (2.43) is on the molal scale.) However,
Eq. (2.43) does not seem to be valid for 2:2, 3:1, or higher electrolytes (i.e.,
those with highly charged ions) at practical concentrations. A much more com-
plicated expression results from the Debye–Hückel theory for the osmotic coef-
ficient of the solution, but in the range of validity of Eq. (2.43), ϕ differs from
unity by no more that about −5%. Other modifications of the theory have been
suggested to extend its validity beyond the I = 0.5 mol L−1 of Eq. (2.43), but
their consideration [3,8] is outside the scope of this book. If I for the solution
is >0.5 mol L−1, and for higher-charge-type electrolytes (2:2, 3:1, etc.), experi-
mental values of γ± and ϕ can be obtained from suitable compilations [8].
58 Marcus
2.5.2 Mixtures Containing Electrolytes

It is important to be able to estimate the values of γ± for aqueous electrolytes in
their mixtures, say B and C. An expression that has been found to be valid
under wide conditions is Harned’s rule:
log γ ± B = log γ 0± B ( I m ) − α B(C) I mC (2.44a )
and
log γ ± C = log γ 0± C ( I m ) − α C(B) I mB (2.44b)
where in Eq. (2.44a) the activity coefficient of electrolyte B in its mixture with
C is γ±B, the latter electrolyte contributing ImC to the total ionic strength on the
molal scale, Im = 1/2Σ mi z2i, and γ±B°(Im) is the activity coefficient of B when
present alone at molal ionic strength Im. A similar expression, Eq. (2.44b), holds
for log γ±C of electrolyte C in the mixture.
Consider, for example an aqueous solution that is 1 m in HCl (electrolyte B)
and 2 m in LiCl (electrolyte C), at a total ionic strength on the molal scale of
Im = 3 m. The activity coefficient of HCl in 3 m HCl is 1.316 [its logarithm is
log γ±HCl°(Im) = 0.119) and αHCl(LiCl) = 0.004 at this total molality. Hence, log
γ±HCl = 0.119 − 2 × 0.004 = 0.111 or γ±HCl = 1.291 in this mixture. Similarly,
γ±LiCl°(Im) = 1.174 (log γ±LiCl°(Im) = 0.070) for 3 m LiCl and αLiCl(HCl) = −0.013,
so that log γ±LiCl = 0.070 − 1 × (−0.013) = 0.083 or γ±LiCl = 1.211 in this mixture.
Thus, the activity coefficient of HCl is lower and that of LiCl is higher in the
mixture than in the single electrolyte solutions at the same total molality. If the
HCl is present at trace concentration in 3 m LiCl, its activity coefficient would
be somewhat reduced to γ±HCl,trace = 1.279, whereas that of trace LiCl in 3 m
HCl would be somewhat increased to γ±LiCl,trace = 1.285.
The values of αB(C) and −αC(B) are interrelated. For very simple systems
they are equal, but generally they are not and must be obtained from the litera-
ture [9]. They also depend on the total ionic strength, I, but not on the composi-
tion relative to the two electrolytes. It must also be noted that Eq. (2.44) is a
semi-empirical expression, and many systems do not obey it, requiring added
terms in ImC2 or ImB2.
When one electrolyte, say, B, is present at very low concentrations in the
presence of the other, say, C, then ImC ⬇ Im, and if this ionic strength is fixed,
then according to Eq. (2.44) the activity coefficient γ±B of the trace electrolyte
B becomes independent of its own concentration in the mixture. That is, log
γ±B = log γ±B°(Im) − αB(C)Im ≠ f(mB). The same is true if there are several electro-
lytes present, all at concentrations much lower than that of C, which is fixed.
This is the basis of the ionic medium method, where an electrolyte C, such
as sodium perchlorate, is kept in the solution at a fixed and high concentration
(e.g., 3 mol L−1). This practice permits the concentrations of reactive electro-
lytes, that is, those that provide ions that participate in reactions (contrary to the
“inert” ions of the medium electrolyte) to be varied below certain limits at will,
without changes in their activity coefficients. These are determined entirely by

the natures of the electrolytes and the fixed concentration of C. Being constant
quantities under these conditions, the activity coefficients can be incorporated into
conditional equilibrium constants for the reactions where the ions participate. (It
becomes also immaterial for the use of the constant ionic medium whether the
ionic strength is specified in terms of a constant molality or molarity, although
the numerical values of the equilibrium constants depend on this specification.)
The nonideality of electrolyte solutions, caused ultimately by the electrical
fields of the ions present, extends also to any nonelectrolyte that may be present
in the aqueous solution. The nonelectrolyte may be a co-solvent that may be
added to affect the properties of the solution (e.g., lower the relative permittiv-
ity, ε, or increase the solubility of other nonelectrolytes). For example, ethanol
may be added to the aqueous solution to increase the solubility of 8-hydroxyqui-
noline in it. The nonelectrolyte considered may also be a reagent that does not
dissociate into ions, or one where the dissociation is suppressed by the presence
of hydrogen ions at a sufficient concentration (low pH; cf. Chapter 3), such as
the chelating agent 8-hydroxyquinoline.
In any event, the activity coefficient of the nonelectrolyte, designated by
subscript N, generally follows the Setchenov equation:
log y N = log yNo + kN,CA cCA (2.45)
where yN° is the activity coefficient of the nonelectrolyte on the molar scale in
the absence of the electrolyte CA, and kN,CA is a proportionality constant that
depends on the natures of N and the electrolyte CA, but is independent of their
concentrations. The larger the molar volume and the lower the polarity of the
nonelectrolyte N, generally the larger kN,CA is. The better hydrated the ions of
the electrolyte CA are, the larger, again, kN,CA is. Since the values of kN,CA are
generally positive, the activity coefficient of the nonelectrolyte in the solution
increases in the presence of the electrolyte, and for a given activity of the former
(e.g., determined by equilibrium with excess insoluble nonelectrolyte), its con-
centration must decrease. The nonelectrolyte is then said to be salted-out by the
electrolyte. As an illustration, consider the salting-out of benzene by CsCl, LiCl,
and BaCl2: the corresponding values of kN,CA are 0.088, 0.141, and 0.334. These
values are ordered according to the strength of the hydration of these salts (see
Table 2.4). A poorly hydrated salt, such as tetramethylammonium bromide even
salts-in benzene, with kN,CA = −0.24. Equation (2.45) may be obeyed up to elec-
trolyte concentrations of several moles per liter. The contributions of several
electrolytes are, within limits, additive.
2.6 ORGANIC SOLUTIONS

Solutions in organic solvents or in mixed aqueous–organic solvents, on the
whole, behave not very differently from purely aqueous solutions. In particular,
60 Marcus
if the relative permittivity of the solvent (see Table 2.1 for values) is roughly
ε > 40, electrolytes are, more or less, completely dissociated into ions. On the
other hand, if ε < 10, only slight or practically no ionic dissociation takes place.
Furthermore, solutions in organic or mixed aqueous–organic solvents lack the
three-dimensional cooperative hydrogen bond network that characterizes aque-
ous solutions. Many “anomalous” properties of aqueous solutions that depend
on this structured nature of water become normal properties in organic solvents.
2.6.1 Electrolyte Transfer

The presence of an organic solvent in an aqueous mixture affects the activity
coefficient of an electrolyte relative to its value in water, even when the electro-
lyte is present at such low concentrations that, in the absence of the co-solvent,
effectively y± would have been unity. Accordingly, in mixed aqueous–organic
solvents there is a primary medium effect on the activity coefficient, reflecting
the interactions of the ions with their mixed-solvent surroundings, compared
with their interactions with a purely aqueous environment. If the content of
water in the mixture is continuously reduced until the pure organic component
is reached, this effect is expected to increase and reach a value characteristic of
the pure solvent. This quantity can be expressed in terms of a standard molar
Gibbs energy of transfer of the electrolyte CA from water, aq, to a new medium,
org (which may also be a nonaqueous solvent that is immiscible with water):
∆tG°(CA,aq → org) (see section 2.3); here, we use the symbols aq and org to
indicate the pure liquids. Thus, when an infinitely dilute solution of the electro-
lyte CA in the new medium is compared with a corresponding solution in water,
we have
∆ t G°(CA,aq → org) = µ ∞CA (org) − µ ∞CA (aq) (2.46)
Note that the µ∞CA are the standard molar Gibbs energies of solvation of the
(combined ions of the) electrolyte [5]. Alternatively, a transfer activity coeffi-
cient can be defined as
ln γ 0t (CA,aq → org) = ∆ t G°(CA,aq → org) / RT (2.47)
The superscript zero designates the standard state of infinite dilution, where
only solute–solvent but no solute–solute interactions take place. As mentioned
in section 2.3, tables of ∆tG°(ion,aq → org) [5–7] may be used to obtain the
values for ions, which can then be combined to give the value of ∆tG°(CA,aq
→ org), for the desired electrolyte. For example, for the solvent org = nitro-
benzene (PhNO2), ∆tG°(Na+,H2O → PhNO2) = 36 kJ mol−1, ∆tG°(Cs+, H2O →
PhNO2) = 18 kJ mol−1, and ∆tG°(I3−, H2O → PhNO2) = −23 kJ mol−1. Therefore,
∆tG° = 13 kJ mol−1 for NaI3 and −5 kJ mol−1 for CsI3 transferring from water to
nitrobenzene, signifying the preference of the former salt for water (positive
∆tG° and of the latter salt for nitrobenzene (negative ∆tG°).
When nonnegligible concentrations of the electrolyte are present in the

organic solvent, ion–ion interactions superimpose on the ion–solvent ones, or
the secondary medium effect. Although an equation similar to Eq. (2.43) may
be used for determining the activity coefficient in the new medium, it is neces-
sary to employ the appropriate value of A in this equation that depends on the
relative permittivity of the medium: A(org) = A(aq)(εaq/εorg)3/2. Unless very wa-
ter-rich mixed solvents are used, different numerical values of the parameters
in the denominator and the second term on the right-hand side of Eq. (2.43)
have to be employed.
2.6.2 Ion Pair Formation

A further complication that sets in when organic or mixed aqueous–organic sol-
vents are used, which is aggravated when the relative permittivity of the medium,
ε, falls below 40, is ion pairing. This phenomenon does occur in purely aqueous
solutions, mainly with higher-valence-type electrolytes: 2:2 and higher, and with
2:1 or 1:2 electrolytes only at high concentrations. Ion pairs may also form in
aqueous solutions of some 1:1 electrolytes, provided the ions are poorly hydrated
and can approach each other to within <0.35 nm. Such ion pairs are of major
importance in solvents that are relatively poor in water or that are nonaqueous.
Ion pairing is due to electrostatic forces between ions of opposite charges
in a medium of moderate to low relative permittivities. It should be distin-
guished from complex formation between metal cations and anionic ligands,
in which coordinative bonds (donation of an electron pair) takes place. One
distinguishing feature is that, contrary to complex formation, the association is
nondirectional in space. The association of a cation and an anion to form an ion
pair can, however, be represented as an equilibrium reaction by analogy to com-
plex formation with an equilibrium constant Kass [3,5]. If α is the fraction of the
electrolyte that is dissociating into ions and therefore (1 − α) is the fraction that
is associated, then
K ass = (1 − α) yN / cCA α 2 y±2 (2.48)
where yN is the activity coefficient of the undissociated part [treated according
to Eq. (2.45) or set, with good approximation, equal to unity] and y± is the mean
ionic activity coefficient of the dissociated part of the electrolyte. This quantity
can be calculated from the extended Debye–Hückel Eq. (2.43), provided that
the value of A appropriate for the medium having the relative permittivity ε is
used and that the ionic strength I takes into account the incomplete dissociation:
αc replaces c in Eq. (2.38).
According to Bjerrum’s theory, the association constant is proportional to a ⴢ
b times a certain function of b, Q(b). The quantity a is the ionic distance
parameter (i.e., the distance between the centers of the cation and the anion).
The quantity b is proportional to the absolute value of the product of the
62 Marcus
charges of the cation and the anion and inversely proportional to the absolute
temperature, T, the relative permittivity (dielectric constant) of the solvent, ε,
and to the distance a. At 25°C
log b = log z + z − + 1.746 − log ε − log a(in nm) (2.49)
b −4 t
The function Q(b) = 2 ∫ t e dt (t is an auxiliary variable) mentioned earlier can
be approximated by eb/b4 at high values of b, or be read from a figure [3]. The
value of Kass at 25°C is given by
log K ass (in L mol -1 ) = 3 log z + z − + 6.120 − 3 log ε + log Q(b) (2.50)
Finally, values of α can be obtained from Kass by means of Eq. (2.48), provided
that y± is calculated iteratively [use Eqs. (2.34), (2.38), and (2.43)].
Solvent-separated ion pairs, in which the first solvation shells of both ions
remain intact on pairing may be distinguished from solvent-shared ion pairs,
where only one solvent molecule separates the cation and the anion, and contact
ion pairs, where no solvent separates them (Fig. 2.6). The parameter a reflects
the minimum distance by which the oppositely charged ions can approach each
other. This equals the sum of the radii of the bare cation and anion plus 2, 1,
and 0 diameters of the solvent, respectively, for the three categories of ion pairs.
Since a appears in Eq. (2.49), and hence, also in Q(b), it affects the value of
the equilibrium constant, Kass. The other important variable that affects Kass is
the product Tε and, at a given temperature, the value of the relative permittivity,
ε. The lower it is, the larger b is and, hence, also Kass.
When the relative permittivity of the organic solvent or solvent mixture is
ε < 10, then ionic dissociation can generally be entirely neglected, and potential
electrolytes behave as if they were nonelectrolytes. This is most clearly demon-
strated experimentally by the negligible electrical conductivity of the solution,
which is about as small as that of the pure organic solvent. The interactions
between solute and solvent in such solutions have been discussed in section 2.3,
and the concern here is with solute–solute interactions only. These take place
mainly by dipole–dipole interactions, hydrogen bonding, or adduct formation.
2.6.3 Other Kinds of Interactions

Dipole-dipole interactions between polar solute particles in organic solvents
may be either repulsive (“head-head” interactions) or attractive (“head-tail” in-
teractions). The former are rare and are due to steric hindrance around one of
the poles of the dipole, which causes the ends of neighboring molecules that
carry partial charges of the same sign to approach it (see Fig. 2.1b). The attrac-
tive dipole interactions lead to solute aggregation, either to dimers, where the
“head” of each of the two partners is near the “tail” of the other (see Fig. 2.1a),
or to larger, chainlike or cyclic aggregates (oligomers), for which each member
of the chain or cycle interacts with a neighbor at both ends (see Fig. 2.1c). The
Fig. 2.6 Ion pairs. A two-dimensional representation of (a) a solvent-separated pair of

ions, each still retaining its intact shell of solvating solvent molecules; (b) a solvent-
sharing ion pair, which has lost some of the solvent between the partners, so that one
layer of solvent shared between them separates them; (c) a contact ion pair, the cation
and anion being contiguous.
dipole moment, µ, is proportional to the spatial separation of the (partial)

charges that cause the particle to be dipolar and is generally considered to “re-
side” midway between these charges. The energy of dipole–dipole interactions
depends on the product of the squares of the dipole moments, µ, of the partners
(i.e., on µ2, if there is a single kind of interacting species) and inversely on the
sixth power of the distance between the centers of the dipoles. Such an energy
is not very large, ranging between 1 and 10 kJ mol−1 (i.e., 0.4 to 4 times the
mean thermal energy, RT, that tends to disperse the aggregates). Tertiary amine
salts in hydrocarbon solvents are typical examples of such aggregated solutes.
Hydrogen bonding occurring between solute particles also leads to aggre-
gation, when the molecule of the solute contains both a hydrogen atom bonded
to an electronegative atom and another electronegative atom that can accept a
hydrogen bond. Typical of such solutes are carboxylic acids and acidic phos-
Rydberg_5063-2_Ch02_R4_02-09-04 11:07:43
64 Marcus
phate esters. Cyclic dimers are formed (see Fig. 2.1d), bonded by two hydrogen
bridges, the energy of each being about 20 kJ mol−1. Such cyclic dimers are
quite stable. If the numbers of ionizable hydrogen atoms and electronegative
accepting sites do not match, then noncyclic aggregates, possibly larger than
dimers, are formed by hydrogen bonding. One aspect of such an aggregation is
a drastic increase in the viscosity of the solution with increasing concentra-
tions.
The mean aggregation number of a solute at molality m, forming various
oligomers (i-mers, i = 1, 2, 3, . . . ), is
n% = m / ∑i mi = ∑i imi / ∑i mi (2.51)
the molality of each oligomeric species being weighted by the number of mono-
mers it contains. The equilibrium constants for the aggregation reactions can be
obtained from the dependence of ñ on m by the methods discussed elsewhere
in this book (see Chapter 3).
The third kind of solute–solute interaction, donor–acceptor adduct forma-
tion, tends to form 1:1 adducts between the molecules of two different kinds of
solute, rather than to lead to self-aggregation. Adduct formation results when
one partner has a donor atom (i.e., an atom with an exposed pair of unbonded
electrons), such as the nitrogen atom in trioctylamine, and the other has an
acceptor atom (i.e., one with an orbital that can take up such a pair of electrons),
such as antimony in antimony pentachloride or even the hydroxyl hydrogen
atom in octanol. Although adduct formation formally can be said to take place
in dipole–dipole interactions between two different kinds of molecules, this is
rare, since the strong dipoles required naturally participate also in self-aggrega-
tion. But if one kind of molecule is a donor only and the other is an acceptor
only, as occurs when octanol (the oxygen atom of which has two exposed non-
bonded pairs of electrons) in the foregoing example is replaced by antimony
pentachloride, then self-aggregation is precluded, and only mutual adduct for-
mation can take place (see following).
Hydrogen bond formation between dissimilar molecules is an example
of adduct formation, since the hydrogen atom that is bonded to an electronega-
tive atom, such as oxygen or nitrogen, is a typical acceptor atom. The ability
of molecules to donate a hydrogen bond is measured by their Taft–Kamlet
solvatochromic parameter, α, (or αm. for the monomer of self-associating sol-
utes) (see Table 2.3). This is also a measure of their acidity (in the Lewis
sense, see later, or the Brønsted sense, if protic). Acetic acid, for instance,
has α = 1.12, compared with 0.61 for phenol. However, this parameter is not
necessarily correlated with the acid dissociation constant in aqueous solutions.
The ability of molecules to accept a hydrogen bond is measured by the Taft–
Kamlet solvatochromic parameter, β, (or βm for the monomer of self-associat-
ing solutes) (see Table 2.3). This, too, is a measure of their basicity (in the
Lewis sense), also measured by the Gutmann donor number DN (discussed
later). Thus, pyridine has β = 0.64, compared with 0.40 for acetonitrile, but
Rydberg_5063-2_Ch02_R4_02-09-04 11:07:43
again, this measure is not directly correlated to the base dissociation constant
(protonation constant) in aqueous solutions. The tendency for a hydrogen bond
to be formed between two dissimilar solute molecules A and B increases with
the sum of the products of their donating and accepting parameters: αAβB + αBβA.
Adduct formation does not require formation of a hydrogen bond, and
other acceptor atoms (or molecules) are known (e.g., transition metal cations in
general, SbCl5, I2, and so on). They are collectively called Lewis acids, and they
react with electron pair donors, that are collectively called Lewis bases.
The enthalpy change during donor–acceptor adduct formation has been re-
lated by Drago to the sum of two terms: (1) the product of the electrostatic proper-
ties of the acid and the base, EA and EB; and (2) the product of their tendency
toward covalent bonding, CA and CB [10]. For the particular case, where the ac-
ceptor is specified to be SbCl5 (and the inert solvent is 1,2-dichloroethane), the
negative of this enthalpy change (in kcal mol−1, 1 cal = 4.184 J) is the Gutmann
donor number, DN [2,11]. These concepts are further discussed in Chapter 3.
2.7 SOLUBILITIES IN BINARY SYSTEMS

Binary systems are systems that involve two components, that is, two substances
that can be added individually. (Ions are not components; they have to be added
in combinations as electroneutral electrolytes; see section 2.5.) The phase rule
(see section 2.2) states that at a given temperature and pressure, when only a
single liquid phase is present, there will be one degree of freedom, and we can
choose the composition of this phase, made up from the two components, at
will. This does not preclude, however, the phenomenon of saturation, where
beyond a certain amount of solute in the liquid mixture a new phase appears.
When two phases are present at the given temperature and pressure in the binary
system, there are no longer any degrees of freedom, and the compositions of the
phases are fixed. The new phase may be a solid or a second liquid. Generally, the
solid phase is the pure solid solute, but in rare circumstances, it is a solid solution
of the two components in each other; occasionally, it is a pure solid solvate of the
original solute by the solvent. When a second liquid phase separates out, it is
generally a saturated solution itself, rather than a pure liquid. We then have a
solvent-rich dilute solution of the solute and a solute-rich concentrated solution of
it. As an example, consider a solution of phenol in water at 25°C. At this tempera-
ture, pure phenol is a solid (its melting point is 40.9°C), but when equilibrated
with water at 25°C, the saturated aqueous layer contains 8.66% by mass of phenol,
and the phenol-rich layer contains 28.72% by mass of water. However, on a mole
fraction basis, both layers appear to be water-rich phases, the one having xwater =
0.982 and the other (the “phenol” phase) having xwater = 0.678.
The solubility of a solute in a solvent is given by the composition of the
saturated solution at a given temperature and pressure. The solubility may be
expressed on any of the concentration scales: the molar (mol L−1; i.e., per liter
66 Marcus
of the saturated solution), the molal (mol kg−1; i.e., per kilogram of the pure
solvent), the mole fraction, the mass fraction (wt%; see Table 2.2), or the vol-
ume fraction scales. Only if the solute is a gas is the pressure of any signifi-
cance. For liquid and solid solutes, however, the temperature is the only variable
that ordinarily needs to be specified.
2.7.1 Gas Solubility

The solubility sB(A) of a gaseous solute B in a liquid A is given by its Bunsen
coefficient KB(B) = [cB(A)/cB(g)]eq [see Eq. (2.9)]. For a binary system, ignoring
the vapor pressure of the solvent, the partial pressure pB of the solute gas equals
the total pressure p. For sufficiently low pressures (< 1 MPa) the gas behaves
as an ideal gas, obeying the ideal gas law pV = nBRT; hence, cB(g) = nB/V = p/
RT is proportional to its pressure. According to Eq. (2.9) the solubility of the
gas in the liquid phase (A) is
sB(A) = cB(A),eq = pB (RT ) −1 exp( − ∆ solv G° B / RT ) (2.52)
The solubility of the gas is thus proportional to its pressure. Gas solubilities
are generally very low, except when the gas interacts chemically with the sol-
vent, e.g., ammonia in water. Combination of Eqs. (2.32b) and (2.34b) with
this fact leads to the mole fraction of the saturated gas solution: xB(A),eq = (MA/
ρA)sB(A) = VAsB(A), where MA, ρA, and VA are the molar mass, the density, and
the molar volume of the liquid A. For a given value of the solute–solvent
interaction, measured by ∆solvG°B, the solubility of a gaseous solute, measured
by the number of moles of solute accommodated by a given number of moles
of solvent, is higher the larger the molar volume of the solvent.
2.7.2 Solubility of Nonelectrolytes

Solvent extraction rarely involves gases, so that other cases should now be con-
sidered. Most liquid organic solutes are completely miscible with, or at least
highly soluble in, most organic solvents. The case of a liquid solute that forms
a solute-rich liquid phase that contains an appreciable concentration of the sol-
vent is related to the mutual solubility of two solvents, and has been discussed
in section 2.2. This leaves solid solutes that are in equilibrium with their satu-
rated solution. It is expedient to discuss organic, nonelectrolytic solutes sepa-
rately from salts or other ionic solutes.
Solid organic solutes often have a limited solubility, not only in water,
but also in organic solvents. Their solubility can be handled by assuming that
work must first be invested to convert the solute into a supercooled liquid at the
temperature of interest (say, 25°C, T = 298.15 K). As a good approximation,
the product of the heat capacity change on fusion, ∆FCp, and the difference
between the temperature of fusion, T F, and T is a measure of this work (the
Gibbs energy of fusion at T). This virtual supercooled liquid would have the
same solubility as an ordinary liquid of similar structure.
The particular case of the solubilities of organic solutes in water can be

dealt with by rather simple equations, based on a general equation for solvent-
dependent properties, applied to solubilities, distribution ratios, rate constants,
chromatographic retention indices, spectroscopic quantities, or heats of associa-
tion [4] [see Eq. (2.12) for an example of its application]. For the molar solu-
bilities of (liquid) aliphatic solutes B in water at 25°C the equation
log sB(water) = log[cB (g) / K B (B)] = 0.69 − 3.44(VB / 100) +
0.43 π *B + 5.15βmB (2.53)
has been found to hold well for 115 solutes (the first equality holding for
gaseous solutes). Here, VB is the molar volume of the solute (in cm3 mol−1), π*B
is its polarity–polarizability index, and βB is its hydrogen bond acceptance
ability (the subscript m denoting the β value for the monomer, if the solute B
is capable of self-association). Tables 2.1 and 2.3 present values of these pa-
rameters for some representative substances. For example, the solubility of
butyl acetate in water is 0.0437 mol L−1 (i.e., log sB = −1.36); Eq. (2.53) gives
0.69 − 3.44 × 1.325 + 0.43 × 0.46 + 5.15 × 0.45 = −1.35 or 0.0444 mol L−1 for
the solubility. The first equality on the right-hand side of the equation holds
for volatile solutes, for which cB(g) = pB/2450, pB being the saturated vapor
pressure of the neat solute in kPa. The qualitative meaning of Eq. (2.53) is that
the solubility decreases with increasing solute molar volumes (the coefficient
of VB is negative) but increases somewhat with increasing polarity-polarizabil-
ity (π*) and in particular with basicity (ability to accept a hydrogen bond, βm).
For the solubility of aromatic solutes in water at 25°C, a different equa-
tion holds:
log sB = 0.57 − 5.58(VB,vdW / 100) + 3.85β B − 0.0100(T F − 298.15) (2.54)
has been found to hold for 70 solid and liquid solutes. The last term in Eq.
(2.54) is to be used only for solutes that are solid at 298.15 K. Note that Eq.
(2.54) involves the van der Waals volume of the solute, VB,vdW, rather than its
molar volume, VB, since the former is more readily estimated than the latter
for solid solutes. Note also that the term in πB* is missing from Eq. (2.54),
which is fortunate, since it is difficult to estimate such values for solids. Again,
using the βB parameter from Table 2.3 and the van der Waals volume of p-
xylene (67.1 cm3 mol−1), Eq. (2.54) predicts log sB = 0.57 − 5.58 × 0.671 +
3.85 × 0.12 = −2.71, or a solubility of 0.00194 mol L−1, compared with the
experimental solubility of 0.00186 mol L−1. For the solid solute p-dinitroben-
zene the parameters are VB,vdW = 77.1 cm3 mol−1, βB = 0.55, TF = 455 K; hence,
the calculated value of log sB = −3.34, compared with the experimental value
−3.33.
When polychlorinated biphenyls and polycyclic aromatic hydrocarbons
are included with other aromatic solutes, the slightly modified equation
log sB = 0.27 − 5.29(VB,vdW / 100) + 3.93β B − 0.0099(T F − 298.15) (2.55)
holds for 139 aromatic solutes. The differences noted between Eq. (2.54) [or
Eq. (2.55)] for aromatic solutes and Eq. (2.53) for aliphatic ones are signifi-
68 Marcus
cant, but have not as yet been explained. However, the qualitative trends are
the same.
2.7.3 Solubility of Salts

As is well known, water is the best medium for the solubility of salts. This is
due to several factors. The high relative permittivity permits complete ionic
dissociation when the ionic charges are not too high. The resulting ions are
effectively solvated, the cations due to the fairly high electron pair donation
ability of water molecules, and the anions due to the very high hydrogen bond
donation ability of water molecules. Furthermore, due to the small size of these
molecules a relatively large number of these can thus solvate the ions. As a
consequence, the attractive electrostatic forces between the ions in the crystal
lattice of a salt are compensated by the solvation forces of the dissociated ions,
and this permits reasonable solubilities. Nonaqueous solvents generally have
many of these properties of water to a lesser degree (have lower relative permit-
tivities and anion solvating abilities and are larger, hence fewer, around the
ions), and, therefore, commonly show lower salt solubilities.
The solubility of an ionic solute, sCA, may be expressed in terms of its
solubility product, Ksp. The equilibrium between a pure solid salt, Cν+Aν− and its
saturated solution in a solvent where it is completely dissociated to ions (gener-
ally having ε > 40; see section 2.6) is governed by its standard molar Gibbs
energy of dissolution
−∆ soln G° = RT ln K sp = RT ln (aCν + aAν − ) (2.56)
For a sparingly soluble electrolyte CA, for which the ionic strength produced in
the saturated solution is very low (e.g., <10−4), the approximations aC ⬇ ν+sCA
and aA ⬇ ν−sCA can be made. With the notation ν = ν+ + ν−, this approximation
leads to
+ −
sCA = [ K sp /(ν ν+ ν ν− )]1 / ν (2.57a )
or
+ −
K sp = (ν ν+ ν ν− ) sCA
ν
(2.57 b)
that relate the solubility of the salt, sCA, to its solubility product, Ksp.
Equation (2.57) presumes not only that the solution is very dilute, but also
that the solute ions do not undergo side reactions, such as hydrolysis or self-
complexation in the saturated solution. A striking case for which, in spite of the
low solubility, these conditions are not obeyed is mercury(I) chloride (calomel,
Hg2Cl2) in water. Its solubility product at 25°C is Ksp = 1.43 × 10−18 mol3 L−3; its
solubility is sCA = 8.4 × 10−6 mol L−1, which is not equal to (Ksp/4)1/3. This is due
to the hydrolysis of the Hg2+2 ion and its disproportionation to form dissolved
Hg0 and Hg(II) (this is strongly complexed by the Cl− ions present). The precau-
tion of taking into account any side reactions must be followed when it is de-
sired to deduce the actual molar solubilities from solubility products; in particu-
lar, in higher-charge-type salts.
Solubility products can be derived indirectly from standard electrode po-
tentials and other thermochemical data, and directly from tabulated standard
Gibbs energies of formation, ∆fG°, of the ions in aqueous solution [12]. Thus,
the use of
∆ soln G° = ν + ∆ f G°(C z + , aq ) + ν − ∆ f G°(A z − , aq ) − ∆ f G°(Cν + A ν - ) (2.58)
and Eq. (2.56) gives the corresponding Ksp value in aqueous solutions.
When solubility products in nonaqueous solvents are desired, tables [5–7]
of the Gibbs energies of transfer of the ions from water to the desired solvent,
org, must be consulted. For any ion
∆ f G°(ion, org) = ∆ f G°(ion, aq ) + ∆ t G°(ion, aq → org) (2.59)
In other words, values of ∆tG°(ion,aq → org), weighted by the stoichiometric
coefficients for both cation and anion, must be added to ∆solnG°(aq) from Eq.
(2.58) to obtain ∆solnG°(org), which in turn gives Ksp and sCA in the solvent org
from Eqs. (2.56) and (2.58) [12].
The lower the relative permittivity of the solvent org, the greater is the
deviation of the activity coefficient of the electrolyte from unity at a given low
ionic strength, and the smaller must Ksp be for the approximation [Eq. (2.57)]
to remain valid. On the other hand, the values of ∆tG°(ion,aq → org) are posi-
tive for the halide and many other anions, and are positive or negative, but
smaller absolutely than the former, for univalent cations. This means that for
1:1 electrolytes, the value of ∆solnG°(org) is more positive than that of ∆solnG°(aq)
and Ksp in the organic solvent org is smaller than that in water. This trend is
generally enhanced with decreasing values of ε of the solvents. Most alkali
metal halides are much less soluble in common organic solvents than in water,
although there are exceptions (some lithium salts and some iodides).
Divalent or higher-valent cations and, in particular, transition metal cat-
ions, are likely to be covalently solvated by solvents that are strong electron
pair donors (have large solvatochromic β values). This solvation often persists
in crystals, so that the salt that is in equilibrium with the saturated solution in
such solvents may not be the anhydrous salt (nor the salt hydrate). Equation
(2.56) omits any consideration of the solvent of crystallization and pertains to
the solventless (anhydrous) salt. For a salt hydrated by n water molecules in the
crystal, the activity of water raised to the nth power must multiply the right-
hand side of Eq. (2.56) for it to remain valid. A similar consideration applies
for salts crystallizing with other kinds of solvent molecules, the activity of the
solvent in the saturated solution replacing that of water. Such situations must be
70 Marcus
avoided when Eq. (2.59) is to be applied for the calculation of solubilities. For
instance, the solubility of lead iodide, PbI2, in water is only 1.65 × 10−3 mol L−1,
but those of the solvates PbI2 ⴢ DMF and PbI2 ⴢ 2DMSO are 0.71 and 0.99 mol
L−1 in N,N-dimethylformamide (DMF) and dimethylsulfoxide (DMSO), respec-
tively.
2.7.4 Heats of Solution

When a solute is dissolved in a solvent, heat may either be evolved (as with
sulfuric acid in water, where strong heating is observed) or absorbed (as with
ammonium nitrate in water, where strong cooling is observed). Thus the disso-
lution may be either exothermic or endothermic, and the standard molar en-
thalpy of solution, ∆solnH°, is then either negative or positive. The sign of ∆solnH°
for a solid solute depends on the balance between its lattice enthalpy (or the
cohesive enthalpy of a liquid solute) and its enthalpy of solvation or that of
the ions produced by it.
Since, for a nonionic solute B, we have −RT ln sB = ∆solnG°, it follows
that the temperature coefficient of its solubility is
dsB / dT = sB ( ∆ soln H B° / RT 2 ) (2.60)
For an ionic solute dissociating into ν ions, the temperature coefficient is 1/ν
times the right-hand side of Eq. (2.60). Again, it is assumed that the solubility
is sufficiently low for the mean ionic activity coefficient to be effectively equal
to unity and independent of the temperature. When this premise is not met,
then corrections for the heat of dilution from the value of the solubility to
infinite dilution must be added to ∆solnH°B in Eq. (2.60).
The qualitative trend predicted by this equation is that, when the heat of
solution is negative (the dissolution is exothermic, i.e., heat is evolved, the
enthalpy of solvation is more negative than the lattice enthalpy is positive), the
solubility diminishes with increasing temperatures. The opposite trend is ob-
served for endothermic dissolution. An analogue of Eq. (2.58), with H replac-
ing G, and the same tables [12] can be used to obtain the required standard
enthalpies of solution of ionic solutes. No general analogues to Eqs. (2.53)–
(2.55) are known as yet.
2.8 SOLUBILITIES IN TERNARY SYSTEMS

2.8.1 Ternary Phase Diagrams
Ternary systems of interest to solvent extraction are generally two-phase sys-
tems, but occasionally ternary systems that consist of a single phase must be
considered. According to the phase rule (see section 2.2), a single-phase three-
component system at a given temperature and pressure has two degrees of free-
dom, so that the mole fractions of two of the components, say, xA and xB, can
be selected at will, that of the third being given by difference: xC = 1 − xA − xB.
It is customary to depict such a system at a given temperature on a triangu-

lar phase diagram, where a composition is indicated by a point in an equilateral
triangle (Fig. 2.7a). The sums of the distances of such a point from the sides of
the triangle are independent of the position of the point (and equal to the height
of the triangle). These distances represent the corresponding mole fractions. An
apex of the triangle represents a pure component; a side represents a binary
mixture of the components at its two ends. The addition (removal) of a compo-
nent from a given composition (point in the triangle) is represented by move-
ment along the straight line connecting the point with the apex for this compo-
nent toward this apex (or away from it). The two degrees of freedom available
in this system can be taken as freedom to choose a composition from the entire
two-dimensional area of the triangle.
Suppose now that, as the temperature is changed, the system at certain
compositions splits into two (liquid) phases. There are then composition ranges
where the system remains a single homogeneous (liquid) phase, but also another
region where two phases coexist. A line on the triangular diagram encloses such
a region of nominal compositions of the system. When a composition within
this region is chosen, two phases result, the compositions of which are located
on the enclosing line. The one remaining degree of freedom can be taken as the
freedom to choose a point on that line, but the composition of the other phase
in equilibrium with it, also on the enclosing line, is now fixed. A straight tie-
line connects the points on the enclosing line that correspond to phases in mutual
equilibrium. The amounts of the two phases at equilibrium for a nominal com-
position that is situated on any tie-line (i.e., within the two-phase region) is at
the inverse ratio of the distances of the point for this composition from the ends
of the tie-line (the lever rule; see Fig. 2.7b).
Suppose now that the temperature is changed further. The enclosed two-
phase region may grow and change shape. It may be bounded on one or two
sides (rarely all three sides) by the sides of the triangle. If this is the case, the
corresponding binary mixtures show miscibility gaps. As long as only two
phases coexist, there is still one degree of freedom. If, however, a further phase
forms, for example, by one component crystallizing (on cooling below its freez-
ing temperature), then the system becomes invariant, and its composition is
fixed uniquely by the fact that there are now three phases in equilibrium at the
temperature of interest.
In ternary systems of relevance to solvent extraction these may be two
(partly) immiscible solvents and one (solid) solute, or two (solid) solutes and a
single solvent. The latter system may constitute a partial system of quaternary
or higher mixtures that involve two liquid phases, which are solvent extraction
systems. In principle, however, a system of two solid solutes and a liquid solvent
could split into two liquid phases, one rich in the one solute, the second in the
other. In general, when one solute crystallizes out from a ternary (two solutes
72 Marcus
Fig. 2.7 Ternary phase diagrams with components A, B, and C. (a) A homogeneous
system, the composition at point P being given by the lengths of the lines marked with
Xs; the dashed line shows the changes occurring when solvent A is added (upward) or
removed (downward). (b) A heterogeneous system, splitting into two liquid phases, with
compositions marked with r, q, p, p′, q′, and r′, along the phase boundary; r and r′ denote
the BC miscibility gap; at nominal composition P there are Pp′ moles of a phase of
composition p for every pP moles of one of composition p′, at the two ends of the tie-
line. (c) A heterogeneous system with two solid solutes, B and C, and solvent A; r and
q denote the solubilities in the binaries AC and AB; at composition r′ the solution con-
tains both B and C and is saturated with solid C; at the invariant point P the solution is
saturated with both B and C.
plus solvent) system, the composition changes along a saturation line, the satu-
rated solution becoming poorer in this solute. The other solute has its own
saturation line. The point at which the two saturation lines meet is, again, an
invariant point of the system, since now there are two solids in equilibrium with
a saturated solution (i.e., three phases in all). The composition of the solution
that is saturated relative to the two solid solutes is thus given uniquely by the
fixed temperature (see Fig. 2.7c).
General solvent extraction practice involves only systems that are unsatu-
rated relative to the solute(s). In such a ternary system, there would be two
almost immiscible liquid phases (one that is generally aqueous) and a solute at
a relatively low concentration that is distributed between them. The single de-
gree of freedom available in such instances (at a given temperature) can be
construed as the free choice of the concentration of the solute in one of the
phases, provided it is below the saturation value (i.e., its solubility in that
phase). Its concentration in the other phase is fixed by the equilibrium condition.
The question arises of whether or not its distribution between the two liquid
phases can be predicted.
Given a nonionic solute that has a relatively low solubility in each of the
two liquids, and given equations that permit estimates of its solubility in each
liquid to be made, the distribution ratio would be approximately the ratio of
these solubilities. The approximation arises from several sources. One is that,
in the ternary (solvent extraction) system, the two liquid phases are not the pure
liquid solvents where the solubilities have been measured or estimated, but
rather, their mutually saturated solutions. The lower the mutual solubility of the
two solvents, the better can the approximation be made. Even at low concentra-
tions, however, the solute may not obey Henry’s law in one or both of the
solvents (i.e., not form a dilute ideal solution with it). It may, for instance,
dimerize or form a regular solution with an appreciable value of b(T) (see sec-
tion 2.2). Such complications become negligible at very low concentrations, but
not necessarily in the saturated solutions.
2.8.2 The Fragment Hydrophobicity Model*

Several models have been suggested for the estimation of the distribution ratios
of nonionic solutes between water and (practically) immiscible organic solvents.
One model takes 1-octanol to represent, in general, lipophilic (“fat-liking”) me-
dia, which hydrophobic (“water-fearing”) solutes would prefer over water. Such
media may be oils, biological lipid membranes, and, somewhat less suitably,
hydrocarbon solvents.
*See also Hansen solubility parameters, section 4.13.3.

74 Marcus
The logarithm of the distribution constant, log P, (i.e., of the distribution ratios
D extrapolated to infinite dilution) in such 1-octanol/water systems was shown
by Hansch and Leo to be additive relative to the contributions of a number Ni
of certain fragments of the solute molecules and a number Nj of certain struc-
ture factors. Each hydrocarbon fragment contributes some hydrophobicity (its
tendency to avoid water) or lipophilicity (tendency to prefer nonpolar organic
solvents) to the solute, and adds a positive quantity fi to log P; polar fragments
provide negative fi values, and certain structural features, Fj, may change the
hydrophobicity (e.g., by the shielding of fragments at branching points from
water).
log P (1-octanol/water ) = ∑ N i f i + ∑ N j Fj (2.61)
Table 2.5 lists the additive values for representative fragments f and structural
factors F. As an example, the infinite dilution distribution ratio of C2H5
C(O)OC2H5, ethyl propionate, between 1-octanol and water is obtained as fol-
lows. The carboxylate group contributes −1.49 to log P, each ethyl group con-
tributes 0.89 + 0.66 − 0.12 = 1.43 (for the methyl and methylene groups and
Table 2.5 Selected Fragment Factors f and Structure Factors F for the Calculation
of log P(1-octanol/water)
Fragment f Fragment f
>C< 0.20 —N< −2.18

>CH— 0.43 —NH— −2.15
—CH2 — 0.66 —NH2 −1.54
—CH3 0.89 —OH −1.64
—H 0.23 —SH −0.23
—F −0.38 —C(O)H −1.10
—Cl 0.06 —C(O)OH −1.11
—Br 0.20 —C(O)N(H)— −2.71
—I 0.59 —C(O)NH2 −2.18
—NO2 −1.16
—O— −1.82
Function F
—S(O)— −3.01
>SO2 −2.67
—OP(O)O2< −2.29 double bond −0.55
—S— −0.79 triple bond −1.42
—CN 1.27 n-Membered chain −0.12(n − 1)
—C(O)N< −3.04 Member of ring −0.09 n
—C(O)— −1.90 Multiple (n) halogens:
—C(O)O— −1.49 on same carbon 0.17 n
—CH— (aromatic) 0.355 on adjacent carbons 0.28(n − 1)
—C< (aromatic ring 0.225 Fragmented attached to −1.0
fusion point) aromatic ring
the chain structural feature, respectively), and altogether the calculated log P
is −1.49 + 2 × 1.43 − 0.12 = 1.25 or the distribution ratio is 17.8. The experi-
mental value of log P is 1.21 and of the distribution ratio 16.2, and it is seen
to be estimated with sufficient accuracy.
The entries in Table 2.5 pertain to the distribution between water and 1-
octanol; for any other organic solvent other solute f and F values should be
used. However, 1-octanol is a fair representative of organic solvents in general,
and the effect of using some other solvent can be dealt with by the use of a
“solvent effect” correction, which is much smaller than the solute dependency
given by the log P values. An approximation to such solvent effect corrections
may be made by using the solubility parameters, accounting in effect for the
differing amounts of work done against the cohesive energies of the solvents.
Thus, for the distributions of a solute B (also an electrolyte CA) between water
and two solvents A1 and A2 (where 1 might be 1-octanol):
log ( DB(A1 ) / DB(A2 ) ) = VB (δ 2A1 − δ 2A2 ) / RT (2.62)
If DB(A1) can be equated with P calculated from the entries in Table 2.5, then
DB(A2) in any other solvent A2 can be estimated from Eq. (2.62). Equation (2.62)
is actually a combination of four expressions of the form of Eq. (2.8) (see
section 2.2.2), two for water and solvent Al and two for water and solvent A2,
presuming them to be immiscible pairs of liquids. It employs concentrations
on the mole fraction scale, and assumes that the systems behave as regular
solutions (which they hardly do). This eliminates the use of the solubility param-
eter δ of water, which is a troublesome quantity (see Table 2.1). Solvent Al
need not, of course, be 1-octanol for Eq. (2.62) to be employed, and it suggests
the general trends encountered if different solvents are used in solvent extrac-
tion.
As an illustration of the application of Eq. (2.62), consider the distribution of
acetylacetone, having a molar volume of 103 cm3 mol−1, between various sol-
vents and water, compared with the distribution between 1-octanol and water,
for which log P = 1.04. The values of log D calculated from Eq. (2.62) at
infinite dilution of the acetylacetone are 1.29, 0.80, 0.69, 0.51, and 0.14 for 1-
pentanol, chloroform, benzene, carbon tetrachloride, and n-hexane, respec-
tively. The experimental values of log D are 1.13, 0.77, 0.76, 0.50, and −0.05
for these solvents. The order is the same, and even the numerical values are
not very different, considering that solute–solvent interactions other than the
work required for cavity formation have not been taken into account.
2.8.3 The Distribution of Nonelectrolytes

A general expression for the distribution of nonelectrolytes takes into account
all the differences in the solute–solvent interactions in the aqueous and the
76 Marcus
organic phases, provided the solutions are sufficiently dilute for solute–solute
interactions to be negligible. In such cases, the energetics of the relevant interac-
tions consist of the work required for the creation of a cavity to accommodate
the solute, and the Gibbs energy for the dispersion force interactions and for the
donation and acceptance of hydrogen bonds. For a given solute distributing
between water (W) and a series of water-immiscible solvents:
log D = AVVsolute ∆δ 2 + Aπ π*solute ∆π* + Aα α solute ∆β + Aββsolute ∆α (2.63)
where the difference factors are ∆δ2 = δsolvent2 − δW2, ∆π* = π*solvent − π*
W, ∆α = αsol-
vent − αW, and ∆β = βsolvent − βW, i.e., the differences between the relevant proper-
ties of the organic solvent and those of water. The first term on the right-hand
side is related to Eqs. (2.10) and (2.62) and the next three terms to Eq. (2.12).
Interestingly, it was established [13] that the coefficients AV, Aπ, Aα, and Aβ are
independent of the solutes and solvents employed—i.e., are “universal”—pro-
vided that the water content of the organic solvent at equilibrium is approxi-
mately xW < 0.13. Such water-saturated solvents can be considered to be still
“dry” and to have properties the same as those of the neat solvents. The values
of the coefficients are AV = 2.14 (for δ2 in J cm−3 and Vsolute = 0.01VX, where VX
is the additive McGowan–Abraham intrinsic volume [2]), Aπ ⬇ 0, Aα = −7.67,
and Aβ = −4.62. The substantial absence of a term with the polarity/polarizability
values π*solute and π*solvent is explained by the fact that the cavity formation term
in Vsolute and ∆δ2 effectively takes care also of the polar interactions. For “wet”
solvents, i.e., those with xW > 0.13 when saturated with water, such as aliphatic
alcohols from butanol to octanol and tri-n-butyl phosphate, Eq. (2.63) is still
valid with practically the same AV, Aπ, and Aβ, but with Aα = 1.04, provided that
δ2, α, and β are taken as those measured (or calculated) for the water-saturated
solvents.
The applicability of Eq. (2.63) was tested for some 180 individual solutes (with
up to 10 carbon atoms) and 25 “dry” solvents by stepwise multivariable linear
regression and for 28 solutes and 9 dry solvents by target factor analysis; essen-
tially the same conclusions and universal coefficients were obtained by both
methods. As an example of the application of Eq. (2.63), the distribution of
succinic acid between water and chloroform (a “dry” solvent with xW = 0.0048)
and tri-n-butyl phosphate (a “wet” solvent with xW = 0.497) may be cited [13].
For the former solvent log D = −1.92 (experimental) vs. −1.98 (calculated)
and for the latter log D = 0.76 (experimental) vs. 0.88 (calculated). Obviously,
succinic acid with two carboxylic groups that strongly donate hydrogen bonds
(assigned α = 1.12 as for acetic acid) prefers the basic (in the Lewis basicity,
hydrogen-bond-accepting sense) tri-n-butyl phosphate (β = 0.82, measured for
the “wet” solvent) over water (β = 0.47) and naturally also chloroform (β =
0.10).
2.8.4 The Distribution of Electrolytes

The distribution ratios of electrolytic solutes, CA, that are practically completely
dissociated into ions in both water and the solvent org, can be approximated by
DCA = (l / ν) exp [∆ t G°(CA,aq → org) / RT ] (2.64)
This expression arises from Eq. (2.40), noting that the chemical potential µCA
must be the same in the phases aq and org for equilibrium to be maintained,
and that ∆tG°(CA,aq → org) = µ∞CA,org − µ∞CA,aq by definition. The approximation
is that the activity coefficients of CA are taken to be equal (near unity) at the
low concentrations where Eq. (2.64) is valid.
This equation would be exact if the two phases org and aq were com-
pletely immiscible.
The standard molar Gibbs energy of transfer of CA is the sum ν+∆tG°(C) +
ν−∆tG°(A), where the charges of the cation Cz+ and anion Az− and the designation
of the direction of transfer, (aq → org), have been omitted. The values for the
cation and anion may be obtained from tables [5–7], which generally deal with
solvents org that are miscible with water and not with those used in solvent
extraction. However, ∆tG°(C) depends primarily on the β solvatochromic pa-
rameter of the solvent and ∆tG°(A) on its α parameter, and these can be esti-
mated from family relationships also for the latter kind of solvents.
The general qualitative rules are that ions prefer water if they are multiply
charged and small, but otherwise they may prefer solvents that have large β
values (donor properties for cations) or α values (acceptor properties for
anions). Consider, for example, the small cation Li+ and multiply charged cation
Cu2+: these cations prefer water (β = 0.47) over such solvents as methyl ethyl
ketone (β = 0.48), propylene carbonate (β = 0.40), benzonitrile (β = 0.41), nitro-
benzene (β = 0.30), and 1,2-dichloroethane (β = 0.00), and also 1-butanol (see
later), but prefer the more highly basic (although water-miscible) solvents di-
methylsulfoxide (β = 0.76), N,N-dimethylformamide (β = 0.69), and hexamethyl
phosphoric triamide (β = 1.06) over water. [Note that 1-butanol (β = 0.88) is
exceptional, being capable of hydrogen bond donation.] However, with propyl-
ene carbonate, the preference is reversed for the large univalent cations Cs+
and tetramethylammonium, (CH3)4N+. For the tetraalkylammonium ions with n-
propyl or longer alkyl chains, the preference for the organic solvents is quite
general. For the anions, water is preferred over most solvents, an exception
being 2,2,2-trifluoroethanol, with α = 1.51, larger than that of water (1.17).
However, very large anions may show a preference for the organic solvents, as
for ClO−4 with propylene carbonate and hexamethyl phosphoric triamide and for
I−3 with most organic solvents. A numerical example is given at the end of sec-
tion 2.6.1, where the transfer of Na+, Cs+, and I−3 from water to nitrobenzene is
discussed. The negative value of ∆tG° for CsI3 is related to the ability of this
78 Marcus
salt to be extracted into nitrobenzene from water, contrary to NaI3, which cannot
(having a positive ∆tG°).
A common situation is that the electrolyte is completely dissociated in the
aqueous phase and incompletely, or hardly at all, in the organic phase of a
ternary solvent extraction system (cf. Chapter 3), since solvents that are practi-
cally immiscible with water tend to have low values for their relative permittivi-
ties ε. At low solute concentrations, at which nearly ideal mixing is to be ex-
pected for the completely dissociated ions in the aqueous phase and the
undissociated electrolyte in the organic phase (i.e., the activity coefficients in
each phase are approximately unity), the distribution constant is given by
log PCA = log cCA (org ) − ν log cCA (aq) (2.65)
The distribution ratio for the electrolyte, DCA, however, is given by
log DCA = log[cCA (org ) / cCA (aq )] = log PCA − (ν − 1) log cCA (aq) (2.66)
which is seen to be a concentration-dependent quantity. Even if the constant log
PCA can be predicted, log DCA cannot, unless some arbitrary choice is made for
the aqueous concentration.
As examples, take the extraction of nitric acid by tri-n-butyl phosphate or
of hydrochloric acid by diisopropyl ether. For the former, at low acid concentra-
tions, the acid is completely dissociated in the aqueous phase, but is associated
in the organic phase. Therefore, DHNO3 increases with the nitric acid concentra-
tion according to Eq. (2.65), up to such concentrations at which the dissociation
of this acid in the aqueous phases becomes significantly lower than complete,
and then DHNO3 flattens out. For hydrochloric acid, the situation is the opposite:
it is dissociated in the ether phase up to about 3 M aqueous acid, so that DHCl is
constant, but at higher concentrations it associates in the organic phase, and DHCl
starts to increase, again according to Eq. (2.65), when the nonidealities in both
phases are taken into account.
2.9 CONCLUSION
This chapter provides the groundwork of solution chemistry that is relevant to
solvent extraction. Some of the concepts are rather elementary, but are necessary
for the comprehension of the rather complicated relationships encountered when
the solubilities of organic solutes or electrolytes in water or in nonaqueous sol-
vents are considered. They are also relevant in the context of complex and
adduct formation in aqueous solutions, dealt with in Chapter 3 and of the distri-
bution of solutes of diverse kinds between aqueous and immiscible organic
phases dealt with in Chapter 4.
For this purpose it is necessary to become acquainted, at least in a cursory
fashion, with the physical and chemical properties of liquids (section 2.1.1)
and the forces operating between their molecules (section 2.1.2). Since solvent
extraction depends on the existence of two immiscible liquid phases, solvent
miscibility, i.e., their mutual solubility, is an important issue (section 2.2). The
solvation of a solute that is introduced into a solvent, i.e., its interactions with
the solvent, is described (section 2.3.1) and the thermodynamics thereof are
elaborated (section 2.3.2), the case of electrolytes that dissociate into ions being
given special attention (section 2.3.3).
Solution chemistry depends strongly on thermodynamic relationships that
have to be mastered in order to make full use of most other knowledge concern-
ing solvent extraction. Therefore, a comprehensive section (2.4) is devoted to
this subject, dealing with ideal (section 2.4.1) as well as nonideal (section 2.4.2)
mixtures and solutions. Then again, as solvent extraction in, e.g., hydrometal-
lurgy, deals with electrolytes and ions in aqueous solutions, the relevant thermo-
dynamics of single electrolytes (section 2.5.1) and their mixtures (section 2.5.2)
has to be understood. In the organic phase of solvent extraction systems, compli-
cations, such as ion pairing, set in that are also described (section 2.6.2).
The amount of a solute that can be introduced into a solvent depends on
its solubility, be it a gas (section 2.7.1), a solid nonelectrolyte (section 2.7.2),
or an electrolyte (section 2.7.3). Ternary systems, which are the basic form of
solvent extraction systems (a solute and two immiscible solvents), have their
own characteristic solubility relationships (section 2.8.1).
In conclusion, therefore, it should be perceived that this groundwork of
solution chemistry ultimately leads to the ability to predict at least semi-quanti-
tatively the solubility and two-phase distribution in terms of some simple prop-
erties of the solute and the solvents involved. For this purpose, Eqs. (2.12),
(2.53), (2.61), and (2.63) and the entries in Tables 2.3 and 2.5 should be particu-
larly useful. In the case of inorganic ions, the entries in Table 2.4 are a rough
guide to the relative extractabilities of the ions from aqueous solutions. The
more negative the values of ∆hydG° of the ions, the more difficult their removal
from water becomes, unless complexation (see Chapter 3) compensates for the
ion hydration.
Although theories of solution (this chapter) and formation of extractable
complexes (see Chapters 3 and 4) now are well advanced, predictions of distri-
bution ratios are mainly done by comparison with known similar systems. Sol-
vatochromic parameters, solubility parameters, and donor numbers, as discussed
in Chapters 2–4, are so far mainly empirical factors. Continuous efforts are
made to predict such numbers, often resulting in good values for systems within
limited ranges of conditions. It is likely that these efforts will successively en-
compass greater ranges of conditions for more systems, but much still has to be
done.
Because a 10% change in the distribution ratio, which can be measured
easily and accurately in the distribution range 0.01–100, corresponds to an en-
80 Marcus
ergy change of only 0.2 kJ/mole, distribution ratio measurements offer a method
to investigate low energy reactions in solutions, such as weak solute–solute and
solute–solvent interactions in the organic phase. So far, this technique has been
only slightly exploited for this purpose. It is particularly noteworthy to find how
few thermodynamic studies have been made of the distribution or extraction
constants, as enthalpy and entropy values give a good indication of the driving
force of the extraction and indicate the structure of the molecular species in the
organic solvent.
REFERENCES
1. Riddick, J. A.; Bunger, W. B.; Sasano, T. K. Organic Solvents, Fourth Edition,
John Wiley and Sons, New York, 1986.
2. Marcus, Y. The Properties of Solvents, John Wiley and Sons, Chichester, 1998.
3. Marcus, Y. Introduction to Liquid State Chemistry, John Wiley and Sons, Chiches-
ter, 1977; 204–207.
4. Kamlet, J. M.; Abboud, J.-L.; Abraham, M. H.; Taft, R. W. J. Org. Chem., 1983,
48: 2877.
5. Marcus, Y. Ion Solvation, John Wiley and Sons, Chichester, 1985.
6. Marcus, Y. Ion Properties, M. Dekker, New York, 1997.
7. Marcus, Y. Pure Appl. Chem., 1983, 55:977.
8. Robinson, R. A.; Stokes, R. H. Electrolyte Solutions, Second Revised Ed., Butter-
worth, London, 1970.
9. Harned, H. S.; Robinson, R. A. Multicomponent Electrolyte Solutions, Pergamon
Press, Oxford, 1968.
10. Drago, R. S. Struct. Bonding, 1973; 15:73.
11. Marcus, Y. J. Solution Chem., 1984; 13:599.
12. Wagman, D. D.; et al., NBS Tables of Chemical Thermodynamic Properties, J.
Phys. Chem. Ref. Data, 1982, Vol. 11, Suppl. 2.
13. Marcus, Y. J. Phys. Chem., 1991;95:8886; Marcus, Y. Solvent Extract. Ion Exch.,
1992, 10: 527; Migron, Y.; Marcus, Y.; Dodu, M. Chemomet. Intell. Lab. Syst.,
1994, 22: 191.
3
Complexation of Metal Ions
GREGORY R. CHOPPIN* Florida State University, Tallahassee, Florida,
U.S.A.
3.1 METAL ION COMPLEXATION

Chapter 2 discussed the various forms of interaction between solute and the
solvent molecules (see section 2.3), which leads to a certain solubility of the
solute in the solvent phase. It was also described how the ratio of the solubility
of the solute between two immiscible solvents could be used to estimate distri-
bution ratios (or constants) for the solute in the particular system (see section
2.4). It was also pointed out that in the case of aqueous solute electrolytes,
specific consideration had to be applied to the activity of the solute in the aque-
ous phase, a consideration that also was extended to solutes in organic solvents.
However, in many solvent extraction systems, one of the two liquids be-
tween which the solute distributes is an aqueous solution that contains one or
more electrolytes, consisting of positive and negative ions that may interact with
each other to form complexes with properties quite different from the ions from
which they are formed. Such complexation is important to the relative extractive
properties of different metals and can provide a sufficient difference in extract-
ability to allow separation of the metals.
In this chapter, the factors underlying the strength of metal–ligand interac-
tion are reviewed. An understanding of how these factors work for different
metals and different ligands in the aqueous and in the organic phases can be of
major value in choosing new extraction systems of greater promise for possible
improvement in the separation of metals. The discussion in this chapter builds
on the principles of solubility of Chapter 2 and provides additional theoretical
*Retired.
81
82 Choppin
background for the material in the chapters that deal with the distribution equi-
libria, kinetics, and practices of solvent extraction systems.
3.1.1 Stability Constants

The extent of metal ion complexation for any metal–ligand system is defined
by the equilibrium constant that is termed the stability constant (or formation
constant) metal–ligand interaction. The term “metal” in metal–ligand systems
simply refers to the central metal ion (or cation), and the term “ligand” (from
Latin ligare, to bind) to the negative ion (anion) or to a neutral electron donor
molecule, which binds to the central atom. Since most ligands bind to the metal
ion in a regular sequence, equilibria are established for the formation of 1:1, 1:2,
1:3, etc., metal-to-ligand ratios. In some systems, polynuclear (i.e., polymetal)
complexes such as 2:1, 2:2, 3:2, etc., form. In this chapter, the primary concern
is with the more common mononuclear complexes in which a single metal atom
has one or more ligands bonded to it. These relatively simple complexes can
serve to illustrate the principles and correlations of metal ion complexation that
apply also to the polynuclear species. Defining M as the metal and L as the
ligand (with charges omitted to keep the equations simpler), the successive com-
plexation reactions can be written as:
M + L = ML (3.1a )
ML + L = ML2 (3.1b)
Or, generally,
MLn − i + L = MLn (3.1c)
The equilibrium constants for these stepwise reactions are expressed generally
as Kn:
K n = [ MLn ]/[ MLn−1 ][ L] (3.2a )
Sometimes a complexation reaction is considered to occur between M and an
acidic ligand HL
M + HL = ML + H (3.1d )
in which case the protonated stepwise formation constant is
*K n = [ ML][ H ]/[ M ][ HL] (3.2b)
It is common to use the overall stability constant, βpqr, where p = number of
metal, q = number of hydrogens (for protonated species), and r = number of
ligands. The value of q is negative for hydroxo species. For the reaction to form
ML,
M + L = ML; β101 = [ ML]/[ M ][ L] = K1 (3.3)
Complexation of Metal Ions 83
However, the formation of multiligand complexes is more complex: e.g.;

M + 3 L + ML3 ; β103 = [ ML3 ] / [ M ][ L]3 = K1 K2 K3 . (3.4)
This can be generalized as:
n
M + nL = MLn ; β10 n = [ MLn ]/[ M ][ L]n = ∏ K i (3.5)
i =1
The βpqr is sometimes referred to as the complexity constant.

Some ligands retain an ionizable proton. For example, depending on the
pH of the solution, metals may complex with HSO−4 , SO42− or both. In the forma-
tion of MHSO4, the stability constant may be written as:
M + HSO 4 = MHSO 4 ; K1 = [ MHSO 4 ]/[ M ][HSO 4 ] (3.6a )
or
M + H + SO 4 = MHSO 4 ; β111 = [ MHSO 4 ]/[ M ][H][SO 4 ] (3.6b)
where
β111 = K111 / K a 2
and Ka2 is the dissociation acid constant for HSO−4. Alternately, some complexes
are hydrolyzed and have one or more hydroxo ligands. For such complexes,
usually in neutral or basic solutions, the reaction may be represented as:
M + H 2O + L = M (OH)L + H (3.7a )
and
β1−11 = [ M (OH)L]/[ M ][H]-1[ L] (3.7 b)
or
*β111 = [ M (OH) L][H] / [ M ][ L] (3.7c)
In general, we use βn to represent β10n in this text.
3.1.2 Use of Stability Constants

If the concentration [L] of “free” (uncomplexed, unprotonated, etc.) ligand at
equilibrium can be measured by spectrometry, a selective ion electrode, or some
other method, the stability constant can provide direct insight into the relative
concentrations of the species. This is achieved by rewriting Eq. (3.5) in the
form:
[ MLn ] /[ M ] = βn ⋅[ L]n (3.7d )
−2
For example, if β1 = 10 and [L] = 10 M, there are equal concentrations of M
2
and ML in the solution since β1 ⴢ[L] = 102 10−2 = 1. This relationship is indepen-
84 Choppin
dent of the amount of metal in the solution and the equality [M] = [ML] = 1 is
valid whether the total metal concentration is 10−9 M or is 1M as long as the
free ligand concentration [L] = 0.01M. For example, if the ligand is present only
as L and ML, and if the total metal concentration, [M] + [ML], is 10−9 M, the total
ligand concentration is [L] + ML = 0.01 + 0.5 × 10−9 = 0.01 M. However, if the
total metal is 1.00 M and [M] = [ML], the total ligand is 0.01 + 0.50 = 0.51 M.
Consider a system in which β1 = 102, β2 = 103 and β3 = 2 × 103. The con-
centrations of each complex can be calculated relative to that of the free (un-
complexed) metal for any free ligand concentration. If [L] = 0.1 M,
[ ML] /[ M ] = β1 ⋅[ L] = 102 × 10−1 = 10
[ ML2 ] /[ M ] = β2 ⋅ [ L]2 = 103 × 10−2 = 10
[ ML3 ] /[ M ] = β3 ⋅[ L]3 = 2 × 103 × 10−3 = 2
Thus, abbreviating the total metal concentration Σ[MLn] = [MT], and defining
the mole fraction of each metal species (i.e., of M, ML, ML2, and ML3) as
the ratio of the concentration of that species to the total metal concentration by
KML,i = [MLi]/[MT], then
K M = 1 /(1 + 10 + 10 + 2) = 0.43
K ML = 10 /(1 + 10 + 10 + 2) = 0.43
K ML = 10 /(1 + 10 + 10 + 2) = 0.43
K ML = 2 /(1 + 10 + 10 + 2) = 0.087
Such calculations can be done for a series of free ligand concentrations to
generate a family of formation curves of concentration or mole fraction of metal
complex species as a function of the concentration of the free ligand. Such curves
are shown in Figs. 3.1 and 3.2. These calculations are particularly useful for trace
level values of metal as they require only knowledge of the free ligand concentra-
tions and the βn. Values of stability constants can be found in Refs. [1,2].
3.1.3 Stability Constants and Thermodynamics

The examples in section 3.1.2 of calculations using stability constants involve
concentrations of M, L, and MLn. Rigorously, a stability constant, as any thermo-
dynamic equilibrium constant should be defined in terms of standard state con-
ditions (see section 2.4). When the system has the properties of the standard
state conditions, the concentrations of the different species are equal to their
activities. However, the standard state conditions relate to the ideal states de-
scribed in Chapter 2, which can almost never be realized experimentally for
solutions of electrolytes, particularly with water as the solvent. For any condi-
tions other than those of the standard state, the activities and concentrations are
related by the activity coefficients as described in Chapter 2, and especially
Fig. 3.1 Formation curves of the fraction of Cd(II), Xi, in the species Cd(NH3)2i +, where
i = 0 to 4, as a function of the uncomplexed concentration of NH3.
as discussed in Chapter 6. Thus, for the thermodynamic constant, K°,

1 for the
reaction
M + L = ML
one obtains
aML [ ML]y ML [ ML]y ML
K °1 = = = = β1 ⋅ y ML / y M y L = β °1 (3.8)
aM a L [ M ]y M [ L]y L [ M ][ L]y M y L
Sometimes the activities are replaced by braces, e.g., aML = {ML}; here we prefer
the former formality. The estimation of yi values by equations such as Eqs.
(2.38) and (2.43) is unreliable for ionic strengths above about 0.5 M for univa-
lent cations and anions and at even lower ionic strengths for polyvalent species.
Consequently, values of β°i are rarely calculated, except for very exact purposes
(see Chapter 6). Instead, measurements of equilibrium concentrations of the
species involved in the reaction are used in a medium of fixed ionic strength
where the ionic strength, I, is defined as I = 1⁄2 Σ ci z2i M [see Eq. (2.38)]. A
solution of fully ionized CaCl2 of 0.5 M concentration has an ionic strength I =
1⁄2 〈0.5 × 22 + 2 × 0.5 × (−1)2〉 or 1.5 M. Stability constants are reported, then,
as measured in solutions of 0.1 M, 0.5 M, 2.0 M, etc., ionic strength. In practice,

86 Choppin
Fig. 3.2 Formation curves of the fraction of Eu(III), Xi, in the species Eu(OAc)3−i
i ,
where i = 0 to 4, as a function of the uncomplexed acetate anion concentration.
the concentrations of the complexing metal and ligand may be so low (e.g., 0.01
M) that if an inert (i.e., noncomplexing) electrolyte is added at higher
concentration (e.g., 1.00 M), it can be assumed that the ionic strength remains
constant during the reaction. However, such a practice, referred to as a constant
ionic medium, is not always possible if the complexation is so weak that rela-
tively high concentrations of metal and ligand are required.
Consider the complexation of a trivalent metal by a dinegative anion. As-
sume M(ClO4)3 is 0.01 M and Na2SO4 is 0.05 M. To obtain an ionic strength of
1.00 M, if no complexation occurs, requires a concentration of NaClO4 calcu-
lated to be:
[ NaClO 4 ] = 1.00 − 1 / 2〈 0.01 × (3) 2 + 3 × 0.01( −1) 2 + 2 × 0.05 × (1) 2
+ 0.05 × (2)2 〉 = 1.00 − 0.21 = 0.79 M
If all the M3+ is complexed,
[ MSO +4 ] = 0.01 M,
[SO 24- ] = 0.05 − 0.1 = 0.4 M
[Na + ] = 2 × 0.05 = 0.10 M,
[ClO-4 ] = 3 × 0.01 = 0.3 M and
[NaClO 4 ] = 0.79
The ionic strength would be:

I = 1 / 2〈 0.01 × (1) 2 + 0.04 × ( −2) 2 + (0.10 + 0.79)(1) 2
+ (0.03 + 0.79)( −1) 2 〉 = 0.94 M
Therefore, a 6% error in the ionic strength arises from not adding additional
NaClO4 during the measurement of the stability constant. Such a change in the
ionic strength might result in significant changes in the yi values of Eq. (3.8). If
the ionic strength is maintained constant during the measurement, Eq. (2.43)
indicates that the values of yi should remain constant. Thus, Eq. (3.8) can be
written for constant ionic strength as:
β °n = β n ⋅ y ML / y M y Ln = β n ⋅ C (3.9)
Since C has a particular value for each value of ionic strength, I, the βn values
can only be used for solutions of that value of I. In fact, the problem can be
even more complicated. Equation (2.43) implies that yi is determined by the
value of I and is not dependent on the electrolyte used to obtain the ionic
strength. For example, β values have been measured in solutions with KNO3 as
the added inert electrolyte while other experimenters have used NaClO4, LiCl,
NaCl, etc. In I ≤ 0.1 M solutions, the differences in β values from these different
solutions are small but when the ionic strength increases above 0.1 M, signifi-
cant differences may arise. Therefore, both I, the ionic strength, and the salt that
is used to set the ionic strength should be specified for β values. For further
details see Chapter 6 and Ref. [3].
3.2 FACTORS IN STABILITY CONSTANTS

Many factors play a role in establishing the value of the stability constants of a
particular metal–ligand system. J. Bjerrum [4] considered such factors and their
effect on the successive stability constants. The ratio of two stepwise constants
is defined as:
log K n / K n+1 = Tn / Tn+1 (3.10)
Bjerrum divided this Tn,n+1 value into two terms: Sn,n+1, which accounts for statis-
tical effects and Ln,n+1, which accounts for all effects attributable to the nature of
the ligand, including electrostatic effects.
3.2.1 Statistical Effect

When simply hydrated, the number of water molecules bonded to the metal
(central) ion correspond to a value N, the coordination number, which is also
termed the hydration number. In complexation, ligands displace the hydrate
waters, although not necessarily on a 1:1 basis. Charge, steric, and other effects
may cause the maximum number of ligands to be less than N. For example, in
88 Choppin
aqueous solutions Co2+ is an octahedral hexahydrate, Co(H2O)62+, but complexes

with chloride to form the tetrahedral species CoCl42−. Trivalent actinides have
large hydration numbers—usually 8 or 9—but may form complexes with N <
8 with bulky anionic ligands.
In his treatment of the statistical effect, Bjerrum assumed that the coordi-
nation number was the same for hydration and complexation. If a ligand with a
single binding site approaches the metal, its probability of interacting is propor-
tional to the number of available bonding sites on the metal. On the other hand,
the probability of dissociating a ligand is proportional to the number of ligands
present in the complex. Bjerrum concluded that the stepwise stability constant,
Kn, should be proportional to the ratio of the available free sites at the complex
to the number of occupied sites in the complex. For, example, for the complex
ML, the probability of ML formation is N and that of ML dissociation is 1. Thus
K1 ∝ N / 1
Similarly, for the ML2 formation by ML + L, only N − 1 sites are available on
the M since one site is already occupied, but the probability of dissociation of
L from ML2 is proportional to 2:
K 2 ∝ ( N − 1) / 2
In general, for MLn−1 + L = MLn
K n ∝ ( N − n + 1) / n (3.11)
From these statistical arguments the ratio of any two successive constants would
be expressed by:
Kn +1 N−n n
= ⋅ (3.12)
Kn N − n +1 n +1
For a metal cation of N = 6 that complexes with a monodentate ligand (one with
a single binding site), the statistical effect gives the following relationship for
K1 through K4 and for β1 through β4
5 5
K2 = K1; β2 = (β1 ) 2
12 12
8 2 5
K 3 = K 2 = K1; β3 = (β1 )3
15 9 54
9 1 5
K 4 = K 3 = K1; β4 = (β1 ) 4
16 8 432
Thus, for a hypothetical system in which β1 = 100, we can calculate K2 ⬵ 42,

and β2 = 4200, while K3 = 22 and β3 = 9.2 × 104, etc. This can be compared to
the experimental values reported for cadmium complexation by chloride,* i.e.,

β1 = 30, β2 = 150, β3 = 100. The disagreement shows that factors other than the
statistical effect serve to further reduce K2, K3, etc., in the Cd2+ + Cl− complex-
ation system. However, the statistical effect calculations can provide an upper
limit on the Kn+1 / Kn ratio as the ligand effects would act to further lower Kn+1
in all but a few unusual situations.
For bidentate chelate ligands, if there are a maximum of 6 coordination
sites, only 3 ligands can be accommodated. In this system, K2 = 1/3 K1 and
K3 = 1/3 K1 [or K3 = 1/9 (K1)2].
3.2.2 Electrostatic Effect

In section 2.3.3, the Born equation was introduced in discussing the electrostatic
interactions when an ion is transferred from a vacuum to a solvent of dielectric
constant ε. The Born equation can be used also to estimate the electrostatic
interaction between an anion and a cation in a solution. The equation in this
case has the form:
∆Gel = − Ael z+ z− / ε r (3.13)
where Ael is the constant (section 2.3.3), z+ and z− are the cationic and anionic
charges, ε is the dielectric constant, and r is the distance between the charge
centers (i.e., the sum of the radii of the cation and anionic bonding group).
Although Eq. (3.13) has the proper form to calculate ∆Gel, in practice, empirical
values of ε and/or r are obtained by putting an experimental value of ∆Gel into
Eq. (3.13). These empirical parameters can be used, in turn, to calculate ∆Gel
for other similar systems.
Equation (3.13) indicates that for complexation systems in which the bond-
ing is strongly electrostatic (minimal covalent contribution), for the same ligand
(so the structural effects are constant) the stability constant should be related to
the charge of the cation divided by r. Since the anionic radius is constant, log
βn for a series of metal–anion complexes (e.g., lanthanide fluorides) could be
expected to correlate with zn/rn, where rn is the cationic radius. Figure 3.3 shows
the correlation of log β1 for a number of metals complexed with the fluoride
anion and the zn/rn of the metals. The good correlation confirms the strongly
electrostatic bonding in the 1:1 complexes of these metals with the very electro-
negative F− anion. In Fig. 3.3, and many of the successive figures, no “points”
are given as the spread, i.e., statistical or methodological uncertainty is fairly
large in many systems.
Another useful correlation can be derived from Eq. (3.13). The proton
*The literature may contain several differing equilibrium constants for the same reaction [1,2] as
measured in different laboratories or by different techniques.
90 Choppin
Fig. 3.3 Correlation of log β1 of fluoride complexation vs. z+/r+ of the metal.
associates with ligands by electrostatic bonding, therefore, Eq. (3.13) should be

applicable to proton association as well as to metal complexation when the latter
is strongly electrostatic. If this is correct, log β1 for metal complexation should
correlate with log Ka for proton dissociation (or K−a1 for proton association, i.e.,
pKa). This assumes no structural effects in the metal complex so such a correlation
can be expected only with series of structurally similar ligands, and particularly
for the first stability constant. Figure 3.4 shows the excellent correlation found for
log β1 for samarium (III) with pKa of a series of monocarboxylate ligands.
This correlation also exists for ligands with more than one binding site
and for successive stability constants. In Fig. 3.5 the log K1 and log K2 values
for UO22+ complexation correlate well with the sum pKa1 + pKa2 for a series of
salicylate type ligands. Salicylic acid has a hydroxy and a carboxylic acid group
ortho to each other on a benzene ring. The protons of both groups can be disso-
ciated and the UO22+ binds to both groups in the resulting complex. The related
ligands in Fig. 3.5 have other species such as Cl, NO2, SO3, etc., bonded also
to the benzene ring but these do not interact with the UO22+. Figure 3.6 shows
the structure of the UO2 complex with the 3,5-dibromosalicylate ligand.
Fig. 3.4 Relationship between the stability constant, β101, for formation of SmL2+ and
the acid constant, pKa, of HL: (1) propionic acid; (2) acetic acid; (3) iodoacetic acid; (4)
chloroacetic acid; (5) benzoic acid; (6) 4-fluorobenzoic acid; (7) 3-fluorobenzoic acid;
and (8) 3-nitrobenzoic acid.
3.2.3 Geometric Effects

Among the many other factors that are grouped into the Tn,n+1 factor in section
3.4, the geometry of the complex plays a significant role in many complexes.
The importance of the relative sizes of the cations and anions in determining
the geometrical pattern in ionic crystals has been long recognized and the same
steric constraints can be expected to be present in complexes. In Table 3.1, the
coordination number and the geometric pattern are listed for various ratios of
the radius of the metal cation to that of the anionic ligand. Co2+ and H2O have
a radius ratio of about 0.5 and, as predicted, octahedral Co(H2O)62+ is the hy-
drated species. Co2+ and Cl− have a radius ratio close to 0.3 and, again as
predicted CoCl−4 is the complex which forms in tetrahedral solutions of high
chloride concentration. Figure 3.7 shows common stereochemistries of the ele-
ments [5].
For electrostatic interaction, the radius ratio can be used as a guide to the
possible geometry of the complex. However, when the metal–ligand bond has
a significant covalent contribution, the geometry is fixed by the necessity to
have the bonding orbitals of the metal and the donor atom of the ligand to
overlap. Generally, this involves hybrid bond orbitals of the metal. The sp3
and dsp2 hybrid orbitals result in coordination number 4 with tetrahedral (sp3)
and square planar (dsp2) geometry. For coordination number 6 the d2sp3 hybrid
92 Choppin
Fig. 3.5 Correlation of log Kn for formation of UO2Ln (n = 1 or 2) with ΣpKa of a

series of salicylates.
orbitals produce an octahedral configuration. Thus NiCl42− is tetrahedral while

Ni(CN)42− is square planar and Ni(NH3)62+ is octahedral. Both NiCl64− and Ni(CN)64−
would have too much electrostatic repulsion to form the octahedral complexes.
The reason why the cyanide forms a planar complex is because CN− tends to
form bonds that are more covalent than those formed by Cl−, and that the dsp2
Fig. 3.6 Structure of the 1:1 complex uranyl dibromosalicylate.

Table 3.1 Range of Radius Ratio Values for Various Cationic

Coordination Numbers
Coordination number Geometric pattern Radius ratio (r+/r−)
2 Linear ≤0.15
3 Triangular 0.15 − 0.22
4 Tetrahedral 0.22 − 0.41
4 Planar 0.41 − 0.73
6 Octahedral 0.41 − 0.73
8 Cubic >0.73
configuration (square planar) is associated with complexes with more covalent

nature than the tetrahedral complexes of sp3 configuration. However, neutral
ligands such as H2O and NH3 can form octahedral structures Ni(H2O)62+ and
Ni(NH3)62+. Figure 3.8 shows examples of these geometric structures.
In summary, the geometry of transition metal complexes is determined by
the necessity (1) to group the ligands about the metal to minimize electrostatic
repulsions and (2) to allow overlap of the metal and ligand orbitals. The first
Fig. 3.7 Common stereochemistries of the elements. (After Ref. 4.)

94 Choppin
Fig. 3.8 Geometric structures of tetrahedral NiCl24−, planar Ni(CN)24− and octahedral
Ni(H2O)26+.
requirement favors a tetrahedral configuration for coordination number 4, as the

ligands are farther apart than in the square planar geometry. However, if overlap
of orbitals is a stronger requirement, and a d orbital can be included in the
hybridization, the dsp2 square planar geometry gives the more stable complex.
In the octahedral complexes of coordination number 6, secondary structural ef-
fects can be observed that can be attributed to differences in the electron distri-
bution among the metal d orbitals. Such effect can be deduced from Fig. 3.9 for
the divalent transition metals. The metal ion radius would be expected to de-
crease smoothly between Mn2+ and Zn2+, resulting in a regular increase in logβ
as shown by the dashed lines connecting the Mn2+ and Zn2+ values for β103(en).
However, the electrostatic field of the ligand anions (or ligand dipoles) causes
an asymmetric pattern of electron distribution in the d orbitals of the metals,
which results in increased stability of the complexes. Mn2+ with five electrons
in d orbitals has a half-filled set of 3d orbitals while Zn2+ with 10 d electrons
has a completely filled set of 3d orbitals. Neither of these configurations allows
asymmetric distribution, so no extra complex stability is present. The increased
Fig. 3.9 Variation of log β101 for complexation by ammonia (NH3) and ethylenediamine
(en) of some first-row divalent transition metal. The electron occupations of the 3d orbit-
als are listed for each metal.
complex stability between Mn and Zn due to the asymetric electron distribution

is known as ligand field stabilization and is a maximum for metals with d3 and
d8 configurations.
Notice in Fig. 3.9 that the expected maximum at the d8 configuration
(Ni ) is not present inasmuch as the d9(Cu2+) complexes are more stable. This
2+
extra stability is due to the Jahn–Teller effect by which d9 systems have excep-
tional stability when only two or four of the six coordinate sites of the metal
96 Choppin
are occupied by the ligand. These sites are in the square plane of the octahedron.
A loss of this extra stabilization is shown for Cu(en)32+ formation when the three
ethylenediamine (a bidentate ligand) molecules occupy all six coordination sites.
As expected, the ligand field stabilization (the difference between log β103 and
the dashed line between Mn2+ and Zn2+) is greater for Ni(en)32+ than for Cu(en)32+.
3.2.4 The Chelate Effect

Figure 3.9 also reflects the “chelate effect.” Ligands that can bind to a metal in
more than one site are said to form chelates. If a ligand binds in two sites, it is
termed bidentate, if three, terdentate, etc. Ethylenediamine, H2N−C2H4−NH2,
(often abbreviated en) is a bidentate ligand inasmuch as it can bind through the
two nitrogen atoms; see Fig. 3.10. Ethylenediaminetetraacetate (EDTA) is a
hexadentate ligand [(OOCⴢCH2)2 NC2H4N(CH2ⴢCOO)2]4− binding through two
nitrogens and an oxygen of each of the 4 carboxylate groups.
Chelates are commonly stronger than analogous, nonchelate complexes.
Ammonia and ethylenediamine both bind via nitrogen atoms and log β2 for
M(NH3)22+ formation can be compared with log β1 for M(en)2+ formation while
log β4 for M(NH3)42+ and log β2 for M(en)22+ can be similarly compared, to ascer-
tain the stabilizing effect of chelate formation. In Fig. 3.9, the curves for
Cu(NH3)22+ and Cu(en)2+ are roughly parallel, with Cu(en)2+ being more stable
by about 2 units of log β. The increase in stability for double chelation is seen
to be almost 6 units from the values of log β104 for Cu(NH3)42+ and log β102 for
Cu(en)22+. The extra Jahn–Teller stabilization for Cu2+ complexation is roughly
independent of whether the complex is a chelate or not.
Chelate complexes are very useful in many solvent extraction systems.
Most often, chelating ligands are organic compounds that provide the possibility
of solubility in the organic phase. When a metal ion forms an octahedral com-
plex with three bidentate organic ligands, it is surrounded by an outer “skin” of
organic structure that favors solubility in an organic solvent, and, hence, extrac-
tion from the aqueous phase.
Fig. 3.10 Structures of Cd(NH3)24+ and Cd(en)22+.

3.3 MODELS OF COMPLEX FORMATION

3.3.1 Early Models
A. Werner had clarified many of the structural aspects of complexes, including
chelates, in the early part of this century. A further major advance in developing
a theory of how and why metal ions form complexes was made by N. V. Sidg-
wick in 1927. He noted that the number of ligands that bond to a metal ion
could be explained by assuming the metal ion accepted an electron pair from
each ligand until the metal ion completed a stable electronic configuration. He
called the metal the “acceptor” and the ligand the “donor.” A few years earlier,
G. N. Lewis had introduced a generalized acid-base theory in which “the basic
substance furnishes a pair of electrons for chemical bond; the acid substance
accepts such a pair” [6]. Combining these views, the acceptor (metal) becomes
an “acid” and the donor (ligand) is a “base.” In these models, complexation was
one class of the general acid-base reactions.
In the 1950s it was recognized that metal cations could be divided into
two general classes. The first, or “class a,” cations behaved like H+ and their
stability constants in aqueous solution had the order:
F >> Cl>Br>I
O >> S>Se>Te
The second, or “class b,” cations had stability constants that followed the reverse
sequence: i.e.,
I >> Br > Cl > F
Te ~ Se ~ S >> O
Complexes of “class a” metals are more ionic, while those of the “class b”
metals are more covalent. Generally, the metals that form tetrahedral complexes
by using sp3 hybrid orbitals are “class a” types. Those forming square planar
complexes by using dsp2 hybrid orbitals are normally “class b” types.
3.3.2 Hard/Soft Acids and Bases (HSAB)

A popular model for predicting whether a metal and a ligand are likely to react
strongly or not is the hard-soft acid-base principle [7]. In general, hard acids
are cations that are in the “class a” group and the log of their stability constants
shows a correlation with the pKa of the ligand base of the complex. Soft acids
are usually in “class b” and their log β values correlate with the redox potential,
E°, or the ionization potential, IP, of the ligand base. Ligands that are hard bases
tend to have higher pKa values, while soft bases have large E° or IP values. The
important HSAB principle states that hard acids react strongly with hard bases
and soft acids react strongly with soft bases. This is only a general principle, as
98 Choppin
some hard acid/hard base pairs do not interact strongly and neither do some
of the soft acid/soft base pairs. The exceptions to the general rule have fac-
tors other than inherent acidity and/or basicity, which are more important in
the interaction. Nevertheless, the HSAB principle has proven a very useful
model for a large variety of complexation reactions, partially because of its
simplicity.
Commonly, soft acids have low-lying acceptor levels while soft bases
have high-lying donor orbitals. As a consequence, the bond in the complex
results from sharing the electron pair. Softness is thus associated with a tendency
to covalency and soft species generally have large polarizabilities. In hard
(metal) acids, the acceptor levels are high and in hard bases, the donor levels
are low. The large energy difference is so large that the cation acid cannot share
the electron pair of the base. The lack of such covalent sharing results in a
strongly electrostatic bond.
The principal features of these species can be listed as follows:
1. Hard species: difficult to oxidize (bases) or reduce (acids); low polarizabili-
ties; small radii; higher oxidation states (acids); high pKa (bases); more
positive (acids) or more negative (bases) electronegativities; high charge
densities at acceptor (acid) or donor (base) sites.
2. Soft species: easy to oxidize (bases) or reduce (acids); high polarizability;
large radii; small differences in electronegativities between the acceptor and
donor atoms; low charge densities at acceptor and donor sites; often have
low-lying empty orbitals (bases); often have a number of d electrons (acids).
Following these guidelines, cations of the same metal would be softer for lower
oxidation states, harder for higher ones. In fact, Cu+ and Tl+ are soft while Cu2+
is borderline and Tl3+ is hard. Even though Cs+ has a large radius and low
charge density, the low ionization potential is sufficient to give it hard acid
characteristics. This illustrates that the properties listed may not be possessed
by all hard (or soft) species, but can serve as guides as to what characteristics
can be used to predict the acid-base nature of species.
3.3.3 Qualitative Use of Acid-Base Model

Using these properties, a number of species have been placed in the hard, soft,
or borderline categories in Table 3.2. This table can be used to predict, at least
qualitatively, the strength of complexation as measured by the stability con-
stants. For example, Pu4+ is a hard acid, F−, a hard base, and I−, a soft base. This
leads to the prediction that log β1(PuF3+) would be larger than log β1(PuI3+); the
experimental log β1 values are 6.8 and <1.0, respectively. By contrast, since Cd2+
is a soft acid, log β1(CdF+) could be expected to be less than log β1(CdI+); the
respective values are 0.46 and 1.89. However, many metals of interest such as
Table 3.2 Partial List of Hard and Soft Acids and Bases
A. Acids:
Hard
+1 ions H+, Li to Cs
+2 ions Mg to Ba, Fe(II), Co, Mn
+3 ions Fe(III), Cr(III), Ga, In, Sc, all Ln (III), all actinides (III)
+4 ions Ti, Zr, Hf, all actinides (IV)
−y1 ions Cr (VI), VO, MoO3, AnO2, Mn(VII)
Borderline +2 ions Fe, Co, Ni, Cu, Zn, Sn, Pb
+3 ions Sb, Bi, Rh, Ir, Ru, Os, R3C+, C6H+5
Soft BH3
+1 ions Cu, Ag, Au, Hg, CH3Hg, l
+2 ions Cd, Hg, Pd, Pt
B. Bases:
Hard H2O, ROH, NH3, RNH2, N2H4, R2O, R3PO, (RO)3PO,
−1 ions OH, RO, RCO2, NO3, ClO4, F, Cl,
−2 ions O, R(CO2)2, CO3, SO4
−3 ions PO4
Borderline C6H5NH2, C5H5N
−1 ions N3, NO2, Br
−2 ions SO3,
Soft C2H4, C6H6, CO, R3P(RO)3P, R3As, R2S
−1 ions H, CN, SCN, RS, I.
−2 ions S2O3
Fe2+, Co2+, Ni2+, Cu2+, and Zn2+ are borderline, and, therefore, their complexing
trends are less easily predicted. For F−, we find the order of log β1 to be
Cu > Zn ∼ Fe > Mn > Ni > Co
whereas for thiocyanate, SCN−, it is
Cu >> Ni > Co > Fe > Zn > Mn
For borderline metals and/or borderline ligands, the order is determined by other
factors such as dehydration (removal of hydrate water), steric effects, etc.
Table 3.3 shows some stability constants for hard and soft types of metal
ligand complexes. The constants are experimental values [1,2] and, as such,
contain some uncertainties. Nevertheless they support well the HSAB principle.
The HSAB principle is useful in solvent extraction, as it provides a guide
to choosing ligands that react strongly (high log β) to give extractable com-
plexes. For example, the actinide elements would be expected to, and do, com-
plex strongly with the β-diketonates, R3C−CO−CH2−CO−CR3 (where R = H
100 Choppin
Table 3.3 Stability Constants for Hard and Soft Types of Metal Complexes
and Halide Ions
Hard type ion: In3+ Soft type ion: Hg2+
Liganda log K1 log K2 log K3 log K4 log K1 log K2 log K3 log K4
−
F (hardest) 3.70 2.56 2.34 1.10 1.03
Cl− ↓ 2.20 1.36 6.74 6.48 0.85 1.00
Br− ↓ 1.93 0.67 9.05 8.28 2.41 1.26
I − (softest) 1.00 1.26 12.87 10.95 3.78 2.23
a
For In3+ the ionic medium is 1M NaClO4 + F−, Cl−, and Br−, and 2M NaClO4 + I−. For Hg2+
the ionic medium is 0.5M Na(X,ClO4).
or an organic group), as the bonding is through the oxygens (hard base sites) of
the enolate isomer anion. The formation of a chelate structure by the metal–
enolate complex results in a relatively strong complex, and if the R groups on
the β-diketonates are hydrophobic, the complex would be expected to have good
solubility in organic solvents. In fact, many β-diketonate complexes show excel-
lent solvent extraction properties.
The solvent extraction of the β-diketonate system can be used to illustrate
the importance of the high coordination numbers of the actinides and lanthan-
ides (see section 3.2) in which hydrophobic adducts are involved. The charge
on a trivalent actinide or lanthanide cation is satisfied by three β-diketonate
ligands (each is −1). However, the three β-diketonate ligands would occupy
only six coordination sites of the metal while the coordination number of these
metal cations can be 8 or 9. Thus, the neutral ML3 species can coordinate other
ligand bases. The alkyl phosphates, (RO)3PO, are neutral hard bases that react
with the ML3 species to form ML3ⴢSn (n = 1 to 3). Tributyl phosphate is widely
used as such a neutral adduct-forming ligand of actinides and provides en-
hanced extraction as the ML3ⴢSn species is more hydrophobic than the hydrated
ML3 and, thus, more soluble in the organic phase. However, the role of coordina-
tion sphere saturation by hydrophobic adducts is not limited to synergic systems.
In the Purex process for processing irradiated nuclear fuel, uranium and pluto-
nium are extracted from nitric acid solution into kerosene as UO2(NO3)2 ⴢ (TBP)2
and Pu(NO3)4 ⴢ (TBP)2. In these compounds the nitrate can be bidentate and so
fills four or fewer of the coordinate sites for the uranium and eight or fewer for
the plutonium. Uranium (U) in UO22+ usually has a maximum coordination num-
ber of 6, while that of Pu4+ can be 9 or even 10. The addition of the two TBP
adduct molecules in each case causes the compound to be soluble and extract-
able in the kerosene solvent.
3.3.4 Quantitative Models of Acid-Base Reaction

The major disadvantage of the HSAB principle is its qualitative nature. Several
models of acid-base reactions have been developed on a quantitative basis and
have application to solvent extraction. Once such model uses donor numbers
[8], which were proposed to correlate the effect of an adduct on an acidic solute
with the basicity of the adduct (i.e., its ability to donate an electron pair to the
acidic solute). The reference scale of donor numbers of the adduct bases is
based on the enthalpy of reaction, ∆H, of the donor (designated as B) with SbCl5
when they are dissolved in 1,2-dichloroethane solvent. The donor numbers, des-
ignated DN, are a measure of the strength of the B−SbCl5 bond. It is further
assumed that the order of DN values for the SbCl5 interaction remains constant
for the interaction of the donor bases with all other solute acids. Thus, for any
donor base B and any acceptor acid A, the enthalpy of reaction to form B:A is:
∆H B:A = a DN B:SbCl5 + b (3.14)
From Eq. (3.14), it is possible to calculate −∆HB:A for a base B and an acid A
if −∆HB:A has been measured for two other donors. For example, the interaction
of UO2(HFA)2 (HFA = hexafluoroacetone) with pyridine and with benzonitrile
has been measured in CHCl3 solvent. The equilibrium constants were deter-
mined to be log KB:A 0.88 in pyridine and −2.79 in benzonitrile, assuming that
log KB:A ⬀ −∆HB:A. With the use of the donor number for these ligands in Table
3.4, the values of a and b are calculated to be 0.04 and −4.85, respectively. In
turn, with these values and the donor number that is given for acetonitrile
(AcN), 59.0 in Table 3.4, the value of log K for the reaction
Table 3.4 Donor Numbers (kJⴢmol1−)

Donor Donor
Donor Number Donor Number
Benzene 0.42 Diphenyl phosphonic chloride 93.7

Nitromethane 11.3 Tributylphosphate 99.2
Benzonitrile 49.8 Dimethoxyethane 100.4
Acetonitrile 59.0 Dimethylformamide 111.3
Tetramethylene sulfone 61.9 Dimethylacetamide 116.3
Dioxane 61.9 Trimethylsulfoxide 124.7
Ethylene carbonate 63.6 Pyridine 138.5
n-Butylnitrile 69.5 Hexamethyl phosphoramide 162.4
Acetone 71.1 Ethylenediamine 230.2
Water 75.3 t-Butylamine 240.6
Diethylether 80.4 Ammonia 246.9
Tetrahydrofuran 83.7 Triethylamine 255.3
102 Choppin
UO 2 (HFA)2 + AcN = UO 2 (HFA)2 ⋅ AcN

in CHCl3 is estimated to be 3.3. Experimentally, it was measured to be 3.5.
The donor number model has been used for correlating a number of sys-
tems. The success of this approach, to a large extent, is based on its application
to hard systems. It is unreasonable to expect that a single order of reaction
strengths for donors would be applicable to all acidic solutes. Also, the assumption
of constant entropy, which is required if log βB:A ⬀ −∆HB:A, would not be univer-
sally valid. In general, it can be expected to have more limited value for soft
systems than for hard ones. Nevertheless, for oxygen and nitrogen donors and,
particularly, in nitrogen and oxygen solvents, it has proven to be very useful.
3.4 THERMODYNAMICS OF COMPLEXATION

The free energy of complexation reaction, ∆Gn, is defined by (cf. section 2.4):
∆Gn = ∆Gno − RT ln [ MLn ] /[ M ][ L]n (3.15a )
where
∆G°n = −RT ln β n (3.15b)
and βn is the stability constant for the complexation. The deviation from equilib-
rium conditions (defined by βn) determines the direction of reaction that goes
spontaneously in the direction of ∆G°n < O. The free energy term is the differ-
ence between the enthalpy of reaction, ∆Hn, and the entropy changes for the
reaction, ∆Sn; that is
∆G°n = ∆H °n − T∆S°n (3.16)
The free energy is calculated from the stability constant, which can be deter-
mined by a number of experimental methods that measure some quantity sensi-
tive to a change in concentration of one of the reactants. Measurement of pH,
spectroscopic absorption, redox potential, and distribution coefficient in a sol-
vent extraction system are all common techniques.
The enthalpy of complexation can be measured directly by reacting the
metal and ligand in a calorimeter. It can also be determined indirectly by mea-
suring log βn at different temperatures and applying the equation
d ln β n / dT = − ∆H °n / RT (3.17)
The temperature variation method is used often in solvent extraction studies. It
can give reliable values of the enthalpy if care is taken to verify that the change
in temperature is not causing new reactions, which perturb the system. Also, a
reasonable temperature range (e.g., ∆T ⬃ 50°C) should be used so that it is
possible to ascertain the linearity of the ln βn vs. 1/T plot upon which Eq. (3.17)
is based.
3.4.1 Enthalpy–Entropy Compensation

The thermodynamics of complexation between hard cations and hard (O, N
donor) ligands often are characterized by positive values of both the enthalpy
and entropy changes. A positive ∆H value indicates that the products are more
stable than the reactants, i.e., destabilizes the reaction, while a positive entropy
favors it. If T∆S°n > ∆H°,n ∆G °n will be negative and thus logβn positive, i.e., the
reaction occurs spontaneously. Such reactions are termed “entropy driven” since
the favorable entropy overcomes the unfavorable enthalpy.
Complexation results in a decrease in the hydration of the ions which
increases the randomness of the system and provides a positive entropy change
(∆Sdehyd > 0). The dehydration also causes an endothermic enthalpy (∆Hdehyd > 0)
as a result of the disruption of the ion–water and water–water bonding of the
hydrated species. The interaction between the cation and the ligand has a nega-
tive enthalpy contribution (∆Hreaction < 0) due to formation of the cation–anion
bonds. This bonding combines the cation and the anion to result in a decrease
in the randomness of the system and gives a negative entropy contribution
(∆Sreaction < 0). The observed overall changes reflect the sum of the contributions
of dehydration and cation–ligand combination. If the experimental values of ∆H
and ∆S are positive, the implication is that the dehydration is more significant
in these terms than the combination step and vice versa. The reaction steps can
be written (with charges omitted and neglecting intermediates):
1. Dehydration: ∆Hh, ∆Sh
M (H 2O) p + L(H 2O)q = M (H 2O) m + L(H 2O)s + (p + q − m − s) H 2O (3.18a )
2. Combination: ∆Hc, ∆Sc
M (H 2O) m + L(H 2O)s = ML(H 2O) m+s (3.18b)
3. Net reaction: ∆Hr, ∆Sr
M (H 2 O) p + L(H 2 O) q = ML(H 2 O) m +s + ( p + q − m − s) H 2 O (3.18c)
and, therefore:
∆H r = ∆Hh + ∆Hc = ∆H °
∆Sr = ∆Sh + ∆Sc = ∆S° (3.19)
It has been shown that for many of the hard–hard complexation systems,
there is a linear correlation between the experimental ∆H° and ∆S° values. Fig-
ure 3.11 shows such a correlation for actinide complexes. These ∆H and ∆S
data involve metal ions in the +3, +4, and +6 oxidation states and a variety of
both inorganic and organic ligands. Such a correlation of ∆H° and ∆S° has
been termed the compensation effect. To illustrate this effect, reconsider the
104 Choppin
Fig. 3.11 Correlation of ∆H and ∆S of formation of a series of 1:1 complexes at

different ionic strengths.
step reactions above for complexation. The net reaction has the thermodynamic
relations:
∆Gr = ∆Gc + ∆Gh = (∆H c + ∆H h ) − T (∆Sc + ∆Sh ) (3.20)
The compensation effect assumes ∆Gh ⬃ 0; the positive values of ∆Hr (=∆Hc +
∆Hh) and ∆Sr (=∆Sc + ∆Sh) mean *∆Hh* > *∆Hc* and *∆Sh* > *∆Sc*. So Eq. (3.20)
can be rewritten:
∆Gr ≈ ∆Gc ≈ ∆Hr − T ⋅ ∆Sr ≈ ∆Hc + T ⋅ ∆Sc (3.21)
The complexation, interpreted in this fashion, implies that:
1. The free energy change of the total complexation reaction, ∆G°, is related
principally to step 2 [Eq. (3.18b)], the combination subreaction;
2. The enthalpy and entropy changes of the total complexation reaction, ∆H°
and ∆S°, reflect, primarily, step 1 [Eq. (3.18a)], the dehydration subreaction.
This discussion is important to solvent extraction systems, as it provides further
insight into the aqueous phase complexation. It also has significance for the
organic phase reactions. In organic solvents, solvation generally is weaker than

for aqueous solutions. As a consequence, the desolvation analogous to step 1
would result in small values of ∆H(solv) and ∆S(solv) (note that we term the aqueous
phase solvation hydration). Therefore, ∆Hc = ∆Hsolv and ∆Sc = ∆Ssolv, which means
that ∆H° is more often negative, while ∆S° may be positive or negative but
relatively small.
3.4.2 Inner Versus Outer Sphere Complexation

Complexation reactions are assumed to proceed by a mechanism that involves
initial formation of a species in which the cation and the ligand (anion) are
separated by one or more intervening molecules of water. The expulsion of this
water leads to the formation of the “inner sphere” complex, in which the anion
and cation are in direct contact. Some ligands cannot displace the water and
complexation terminates with the formation of the “outer sphere” species, in
which the cation and anion are separated by a molecule of water. Metal cations
have been found to form stable inner and outer sphere complexes and for some
ligands both forms of complexes may be present simultaneously.
Often, it is difficult to distinguish definitely between inner sphere and
outer sphere complexes in the same system. Based on the preceding discussion
of the thermodynamic parameters, ∆H and ∆S values can be used, with cation,
to obtain insight into the outer vs. inner sphere nature of metal complexes. For
inner sphere complexation, the hydration sphere is disrupted more extensively
and the net entropy and enthalpy changes are usually positive. In outer sphere
complexes, the dehydration sphere is less disrupted. The net enthalpy and en-
tropy changes are negative due to the complexation with its decrease in random-
ness without a compensatory disruption of the hydration spheres.
These considerations lead, for example, to the assignment of a predomi-
nantly outer sphere character to Cl−, Br−, I−, ClO−3, NO−3, sulfonate, and trichloro-
acetate complexes and an inner sphere character to F−, IO−3, SO42−, and acetate
complexes of trivalent actinides and lanthanides. The variation in ∆H° and ∆S°
of complexation of related ligands indicates that those whose pKa values are <2
form predominantly outer sphere complexes, while those for whom pKa > 2
form predominantly inner sphere complexes with the trivalent lanthanides and
actinides. As the pKa increases above 2, increasing predominance of inner sphere
complexation is expected for these metals.
3.4.3 Thermodynamics of Chelation

In the preceding section, the positive entropy change observed in many com-
plexation reactions has been related to the release of a larger number of water
molecules than the number of bound ligands. As a result, the total degrees of
freedom of the system are increased by complexation and results in a positive
106 Choppin
Table 3.5 Thermodynamic Parameters of Reaction of Cadmium(II)-Ammonia

Complex with Ethylenediamine
Product log K ∆H o(kJⴢmol−1) ∆S o (JⴢmⴢK−1)
Cd(en)+2 0.9 +0.4 5.4

Cd(en)+2
2 2.2 −3.4 +15.0
[Cd(en) +2
n ][ NH 3 ]
2n
K= +2
[Cd(NH 3 ) 2 n [en ]n
value of the entropy change. In many systems, a similar explanation can be

given for the enhanced stability of chelates. Consider the reaction of Cd(II) with
ammonia and with ethylenediamine to form Cd(NH3)42+ and Cd(en)22+, whose
structures are given in Fig. 3.10. Table 3.5 gives the values of log K, ∆H°, and
T∆S° for the reactions
Cd(NH 3 ) 22 + + en = Cd(en) 2 + + 2 NH 3 (3.22)
2+ 2+
Cd(NH ) 3 4 + 2 en = Cd(en) 2 + 4 NH 3 (3.23)
The enthalpy value of Eq. (3.23) is very small as might be expected if two
Cd−N bonds in Cd(NH3)2+2 are replaced by two Cd−N bonds in Cd(en)2+. The
favorable equilibrium constants for reactions [Eqs. (3.22) and (3.23)] are due to
the positive entropy change. Note that in reaction, Eq. (3.23), two reactant mole-
cules form three product molecules so chelation increases the net disorder (i.e.,
increase the degrees of freedom) of the system, which contributes a positive
∆S° change. In reaction Eq. (3.23), the ∆H is more negative but, again, it is the
large, positive entropy that causes the chelation to be so favored.
As the size of the chelating ligand increases, a maximum in stability is
normally obtained for 5 or 6 membered rings. For lanthanide complexes, oxalate
forms a 5-membered ring and is more stable than the malonate complexes with
6-membered rings. In turn, the latter are more stable than the 7-membered che-
late rings formed by succinate anions.
3.5 SUMMARY
The important role of thermodynamics in complex formation, ionic medium
effects, hydration, solvation, Lewis acid-base interactions, and chelation has
been presented in this chapter. Knowledge of these factors are of great value in
understanding solvent extraction and designing new and better extraction sys-
tems.
REFERENCES
1. Martell, A. E.; Smith, R. M.; Critical Stability Constants, Vol. 1–5, Plenum Press,
New York, (1974–1982).
2. Högfeldt, E. Stability Constants of Metal-Ion Complexes, Part A: Inorganic ligands.
IUPAC Chemical Data Series No. 22, Pergamon Press, New York (1982).
3. Robinson, R. A.; Stokes, R. H.; Electrolyte Solutions, Second Revised Edition, But-
terworth, London (1970).
4. Bjerrum, J.; Metal Amine Formation in Aqueous Solution, P. Haase and Son, Copen-
hagen (1941).
5. Moeller, T.; Inorganic Chemistry, John Wiley and Sons, New York (1952).
6. Lewis, G. N.; Valence and the Structures of Atoms and Molecules, The Chemical
Catalog Co., New York (1923).
7. Pearson, R. G.; Ed. Hard and Soft Bases, Dowden, Huchinson and Ross, East
Stroudsburg, PA, (1973).
8. Gutmann, V.; The Donor-Acceptor Approach to Molecular Interactions, Plenum
Press, New York (1980).
4
Solvent Extraction Equilibria
JAN RYDBERG* Chalmers University of Technology, Göteborg, Sweden
GREGORY R. CHOPPIN* Florida State University, Tallahassee, Florida,
U.S.A.
CLAUDE MUSIKAS* Commissariat à l’Energie Atomique, Paris, France
TATSUYA SEKINE† Science University of Tokyo, Tokyo, Japan
4.1 INTRODUCTION
The ability of a solute (inorganic or organic) to distribute itself between an
aqueous solution and an immiscible organic solvent has long been applied to
separation and purification of solutes either by extraction into the organic phase,
leaving undesirable substances in the aqueous phase; or by extraction of the
undesirable substances into the organic phase, leaving the desirable solute in the
aqueous phase. The properties of the organic solvent, described in Chapter 2,
require that the dissolved species be electrically neutral. Species that prefer the
organic phase (e.g., most organic compounds) are said to be lipophilic (“liking
fat”) or hydrophobic (“disliking water”), while the species that prefer water
(e.g., electrolytes) are said to be hydrophilic (“liking water”), or lipophobic
(“disliking fat”). Because of this, a hydrophilic inorganic solute must be ren-
dered hydrophobic and lipophilic in order to enter the organic phase.
Optimization of separation processes to produce the purest possible prod-
uct at the highest yield and lowest possible cost, and under the most favorable
environmental conditions, requires detailed knowledge about the solute reac-
tions in the aqueous and the organic phases. In Chapter 2 we described physical
factors that govern the solubility of a solute in a solvent phase; and in Chapter
3, we presented the interactions in water between metal cations and anions by
This chapter is a revised and expanded synthesis of Chapters 4 (by Rydberg and Sekine) and 6 (by
Allard, Choppin, Musikas, and Rydberg) of the first edition of this book (1992).
*Retired.
†Deceased.
109
110 Rydberg et al.
which neutral metal complexes are formed. This chapter discusses the equations
that explain the extraction data for inorganic as well as organic complexes in a
quantitative manner; i.e., the measured solute distribution ratio, Dsolute, to the
concentration of the reactants in the two phases. It presents chemical modeling
of solvent extraction processes, particularly for metal complexes, as well as a
description of how such models can be tested and used to obtain equilibrium
constants.
The subject of this chapter is broad and it is possible to discuss only the
simpler—though fundamental—aspects, using examples that are representative.
The goal is to provide the reader with the necessary insight to engage in solvent
extraction research and process development with good hope of success.
4.1.1 The Distribution Law

The distribution law, derived in 1898 by W. Nernst, relates to the distribution
of a solute in the organic and in the aqueous phases. For the equilibrium reaction
A (aq) A (org) (4.1a )
the Nernst distribution law is written
Concentration of Species A in organic phase [A]org
K D, A = = ( 4.1b)
Concentration of Species A in aqueous phase [A]aq
where brackets refer to concentrations; Eq. (4.1) is the same as Eqs. (1.2) and
(2.23). KD,A is the distribution constant (sometimes designated by P, e.g., in
Chapter 2; see also Appendix C) of the solute A (sometimes referred to as the
distribuend ). Strictly, this equation is valid only with pure solvents. In practice,
the solvents are always saturated with molecules of the other phase; e.g., water
in the organic phase. Further, the solute A may be differently solvated in the
two solvents. Nevertheless, Eq. (4.1) may be considered valid, if the mutual
solubilities of the solvents (see Table 2.2) are small, say <1%, and the activity
factors of the system are constant. If the solute is strongly solvated, or at high
concentration (mole fraction >0.1), or if the ionic strength of the aqueous phase
is large (>0.1 M) or changes, Eq. (4.1) must be corrected for deviations from
ideality according to
yA,org [A]org yA,org
K D,0 A = = K D, A ( 4.2)
yA,aq [A]aq yA,org
where y’s are activity coefficients [see Eq. (2.25)]. For aqueous electrolytes, the
activity factors vary with the ionic strength of the solution (see sections 2.5, and
3.1.3, and Chapter 6). This has led to the use of the constant ionic medium
method (see Chapter 3); i.e., the ionic strength of the aqueous phase is kept
constant during an experiment by use of a more or less inert “bulk” medium
Solvent Extraction Equilibria 111
like NaClO4. Under such conditions the activity factor ratio of Eq. (4.2) is as-
sumed to be constant, and KD is used as in Eq. (4.1) as conditions are varied at
a constant ionic strength value. In the following derivations, we assume that the
activity factors for the solute in the aqueous and organic solvents are constant.
Effects due to variations of activity factors in the aqueous phase are treated in
Chapter 6, but no such simple treatment is available for species in the organic
phase (see Chapter 2).
The assumption that the activity factor ratio is constant has been found to
be valid over large solute concentration ranges for some solutes even at high
total ionic strengths. For example, the distribution of radioactively labeled GaCl3
between diethyl ether and 6M HCl was found to be constant (KD,Ga ⬇ 18) at all
Ga concentrations between 10−3 and 10−12 M [1].
In the following relations, tables, and figures, the temperature of the sys-
tems is always assumed to be 25°C, if not specified (temperature effects are
discussed in Chapters 3 and 6, and section 4.13.6). We use org to define species
in the organic phase, and no symbol for species in the aqueous phase (see Ap-
pendix C).
4.1.2 The Distribution Ratio

The IUPAC definition of the distribution ratio, D, is given in the introduction
to Chapter 1 and in Appendix C. For a metal species M it can be written
Concentration of all species containing
M in organic phase [M]t,org
DM = = (4.3))
Concentration of of all species containinng [M]t,aq
M in aqueous phase
When M is present in various differently complexed forms in the aqueous phase
and in the organic phase, [M]t refers to the sum of the concentrations of all M
species in a given phase (the subscript t indicates total M). It is important to
distinguish between the distribution constant, KD, which is valid only for a single
specified species (e.g., MA2), and the distribution ratio, DM, which may involve
sums of species of the kind indicated by the index, and thus is not constant.
4.1.3 Extraction Diagrams

Solvent extraction results are presented typically in the form of diagrams. This
is schematically illustrated in Fig. 4.1a for three hypothetical substances, A, B,
and C. The distribution ratio is investigated as a function of the concentration
of some reactant Z, which may be pH, concentration of extractant in the organic
phase (e.g., an organic acid HA, [HA]org), the extractant anion concentration in
the aqueous phase (e.g., [Cl−]), salt concentration in the aqueous phase, etc. The
112 Rydberg et al.
Fig. 4.1 Liquid-liquid distribution plots. (a) The distribution ratios D for three different
substances A, B, and C, plotted against the variable Z of the aqueous phase. Z may
represent pH, concentration of extractant in organic phase ([HA]org), free ligand ion con-
centration in the aqueous phase ([A−]), aqueous salt concentration, etc. (b) Same systems
showing percentage extraction %E as a function of Z. D and Z are usually plotted on
logarithmic scale.
range of D is best measured from about 0.1–10, though ranges from about
10−5 –104 can be measured with special techniques (see section 4.15).
In many practical situations, a plot like Fig. 4.1a is less informative than
one of percentage extraction, %E, where:
% E = 100 D /(1 + D) (4.4)
Such a plot is shown in Fig. 4.1b for the same system as in Fig. 4.1a. Percentage
extraction curves are particularly useful for designing separation schemes. A
series of such curves has already been presented in Fig. 1.3.
A convenient way to characterize the S-shaped curves in Figs. 1.3 or 4.1b,
where the extraction depends on the variable Z, is to use the log Z value of 50%
extraction, e.g., log[Cl−]50. The pH50-value indicates −log[H+] for 50% extrac-
tion. This is shown in Fig. 4.1 for distribuends A and B.
Very efficient separations are often needed in industry, and a single ex-
traction stage may be insufficient. The desired purity, yield, etc. can be achieved
by multiple extractions, as discussed in Chapter 7 (see also section 1.2). In
the design of separation processes using multistage extractions, other extraction
diagrams are preferred. Only single stage extraction is discussed in this chapter,
while multistage extraction is discussed in the second part (Chapters 7–14) of
this book.
4.2 THERMODYNAMICS OF EXTRACTION SYSTEMS

Extraction from aqueous solutions into organic solvents can be achieved through
different chemical reactions. Some may seem very complicated, but usually oc-
cur through a number of rather simple steps; we assume this in making a model
of the system. The subdivision of an extraction reaction into its simpler steps is
useful for understanding how the distribution ratio varies as a function of the
type and concentration of the reagents. Often these models allow equilibrium
constants to be measured.
As solute, we consider both nonelectrolytes (abbreviated as A or B, or-
ganic or inorganic), and electrolytes (e.g., as metal-organic complexes, metal
ions rendered soluble in organic solvents through reactions with organic anions
A− and with adduct formers B). The system of equations shown later is only
valid as long as no species are formed other than those given by the equations,
all concentrations refer to the free concentrations (i.e., uncomplexed), and activ-
ity factors and temperatures are constant. Further, we assume that equilibrium
has been established. It may be noted that the use of equilibrium reactions mean
that the reactions take place in the aqueous phase, the organic phase or at the
interface, as is illustrated in the next examples, but do not show any intermedi-
ates formed; this information can be obtained by kinetic studies, as described in
Chapter 5, or by “fingerprinting” techniques such as molecular spectroscopy.
Before a detailed analysis of the chemical reactions that govern the distri-
bution of different solutes in solvent extraction systems, some representative
practical examples are presented to illustrate important subprocesses assumed to
be essential steps in the overall extraction processes.
4.2.1 Case I: Extraction of Uranyl Nitrate

by Adduct Formation
This is a purification process used in the production of uranium. The overall
reaction is given by
2 + 2 HNO 3 + 2 TBP(org) → UO 2 ( NO 3 ) 2 (TBP) 2 (org)

UO 2+ (4.5)
where TBP stands for tributylphosphate. The organic solvent is commonly kero-
sene. In Table 4.1 this extraction process is described in four steps. In Table
114 Rydberg et al.
Table 4.1 Schematic Representation of the Hypothetical Steps in U(VI) Extraction

by TBP and Their Associate ∆G°i of Reaction
Second
First step step Third step Fourth step
Organic phase TBP UO2(TBP)2(NO3)2
(TBP + diluent) ↓ ↑
Aqueous solution 2 HNO3 + UO2+2 → ↓ UO2(NO3)2 + 2TBP ↑
(HNO3 + UO2+2 + UO2(NO3)2 + 2H+ TBP → UO2(TBP)2(NO3)2 UO2(TBP)2(NO3)2
H2O)
Start → Final ∆G°1 > 0 ∆G°2 > 0 ∆G°3 ≠ 0 ∆G°4 Ⰶ 0
∆G°ex < 0
4.1, the sign of the free energy change, ∆G0, in each step is given by qualita-
tively known chemical affinities (see Chapter 2). The reaction path is chosen
beginning with the complexation of U(VI) by NO−3 in the aqueous phase to form
the uncharged UO2(NO3)2 complex (Step 1). Although it is known that the free
uranyl ion is surrounded by water of hydration, forming UO2(H2O)2+ 6 , and the
nitrate complex formed has the stoichiometry UO2(H2O)6(NO3)2, water of hydra-
tion is not listed in Eq. (4.5) or Table 4.1, which is common practice, in order
to simplify formula writing. However, in aqueous reactions, water of hydration
can play a significant role. As the reactive oxygen (bold) of tributylphosphate,
OP(OC4H9)3, is more basic than the reactive oxygen of water, TBP, which
slightly dissolves in water (Step 2), replaces water in the UO2(H2O)6(NO3)2 com-
plex to form the adduct complex UO2(TBP)2(NO3)2. This reaction is assumed to
take place in the aqueous phase (Step 3). Adduct formation is one of the most
commonly used reactions in solvent extraction of inorganic as well as organic
compounds. (Note: the term adduct is often used both for the donor molecule
and for its product with the solute.) The next process is the extraction of the
complex (Step 4). Even if the solubility of the adduct former TBP in the aqueous
phase is quite small (i.e., DTBP very large), it is common to assume that the
replacement of hydrate water by the adduct former takes place in the aqueous
phase, as shown in the third step of Table 4.1; further, the solubility of the
adduct UO2(TBP)2(NO3)2 must be much larger in the organic than in the aqueous
phase (i.e., DUO2 (TBP)2(NO3)2 Ⰷ 1), to make the process useful. Other intermediate
reaction paths may be contemplated, but this is of little significance as ∆G0ex
depends only on the starting and final states of the system. The use of such a
thermodynamic representation depends on the knowledge of the ∆G0i values as
they are necessary for valid calculations of the process.
The relation between ∆G0ex and Kex is given by
∆Gexo = Σ∆Gio = −RT ln K ex ( 4.6)
Omitting water of hydration, the equilibrium constant for the net extraction pro-
cess in Eq. (4.5) is Kex, where
[UO 2 (NO3 )2 (TBP)2 ]org
K ex = ( 4.7)
[UO 22 + ][HNO3 ]2 [TBP]2org
The extraction constant, Kex, can be expressed as the product of several equilib-
rium constants for other assumed equilibria in the net reaction:
−2
K ex = ΠK i = β2, NO3 K DR β2, TBP K DC (4.8)
where β2,NO3 is the complex formation constant of UO2(NO3)2, and β2,TBP the
formation constant of the extractable UO2(NO3)2(TBP)2 complex from from
UO2(NO3)2 and TBP. KDR and KDC are the distribution constants of the un-
charged species, the reagent and the extractable complex, respectively.
Kex determines the efficiency of an extraction process. It depends on the
“internal chemical parameters” of the system, i.e., the chemical reactions and
the concentration of reactants of both phases. The latter determine the numerical
value of the distribution factor for the solute, which for our example is
[U]tot,org  UO 2 ( NO3 )2 ( TBP )2 

DU = =
org
(4.9a )
 UO 2  + UO 2 ( NO3 )n 
2−n
[U]tot,aq 2+ 
 
In the aqueous phase we have included the UO2(NO3)n2−n complexes but excluded
the UO2(NO3)2(TBP)2 complex, because the concentration of the last complex
in the aqueous phase is negligible compared to the other two. In dilute solutions,
the nitrate complex can be negleted compared to the free UO2+ 2 concentration.
In the latter case the U distribution equals
DU = K ex [HNO 3 ] [TBP]org
2 2
( 4.9 b)
Of the reaction steps, only the first three have values of ∆G > 0; however,
0
the large negative value of the fourth step makes the overall reaction ∆G0ex nega-
tive, thus favoring the extraction of the complex. The first step can be measured
by the determination of the dinitrato complex in the aqueous phase. The second
is related to the distribution constant KD,TBP in the solvent system. Also, the
formation constant of the aqueous UO2(NO3)2(TBP)2 can be measured (for ex-
ample by NMR on 31P of TBP in the aqueous phase). Thus, ∆G04 can be derived.
4.2.2 Case II: Synergistic Extraction of Uranyl Ions

by Chelation and Adduct Formation
Solvent extraction is a powerful technique in research on metal complexes. Con-
sider a metal complexed by a chelate compound (see Chapter 3), where the
chelate is a weak organic acid. For example, the uranyl ion can be neutralized
116 Rydberg et al.
by two TTA− (Appendix D:5e) anions to form the neutral UO2(TTA)2(H2O)2

complex. This complex is extractable into organic solvents, but only at high
concentrations of the TTA anion.
A large adduct formation constant increases the hydrophobicity of the
metal complex and thus the distribution ratio of the metal. This is commonly
referred to as a synergistic effect. Figure 4.2 illustrates the extraction of the
UO2(TTA)2 complex from 0.01 M HNO3 into cyclohexane. Because the linear
O−U−O group is believed to have five to seven coordination sites, where only
Fig. 4.2 Synergistic extraction: Distribution of U(VI) between 0.01 M HNO3 and mix-
tures of thenoyltrifluoroacetone (TTA) and tributylphosphate (TBP), or tributylphos-
phineoxide (TBPO), at constant total molarity ([TTA]org plus [TBP]org or [TBPO]org =
0.02 M) in cyclohexane. (From Ref. 2.)
four are occupied in this complex, the uranyl group is coordinativaly unsatu-
rated. At the left vertical axes of Fig. 4.2, the free coordination sites are occu-
pied by water and/or NO−3, only; and the U(VI) complex is poorly extracted, log
DU about −1. When TBP or TBPO (tributylphosphine oxide*) [both indicated
by B] are added while [HTTA] + [B] is kept constant, the DU value increases to
about 60 for TBP and to about 1000 for TBPO. At the peak value, the complex
is assumed to be UO2(TTA)2B1 or 2. The decrease of DU at even higher [B] is due
to the corresponding decrease in [TTA−], so that at the right vertical axes of
Fig. 4.2 no U(VI)—TTA complex is formed. For this particular case, at much
higher nitrate concentrations, the U(VI) is complexed by NO−3 and is extracted
as an adduct complex of the composition UO2(NO3)2 B1–2, as discussed earlier
for Case I.
The primary cause for synergism in solvent extraction is an increase in
hydrophobic character of the extracted metal complex upon addition of the ad-
duct former. Three mechanisms have been proposed to explain the synergism
for metal + cheland† + adduct former. In the first suggested mechanism, the
chelate rings do not coordinately saturate the metal ion, which retains residual
waters in the remaining coordination sites and these waters are replaced by other
adduct-forming molecules. The second involves an opening of one or more of
the chelate rings and occupation by the adduct formers of the vacated metal
coordination sites. The third mechanism involves an expansion of the coordina-
tion sphere of the metal ion upon addition of adduct formers so no replacement
of waters is necessary to accommodate the adduct former. As pointed out before,
it is not possible from the extraction constants to choose between these alterna-
tive mechanisms, but enthalpy and entropy data of the reactions can be used to
provide more definitive arguments.
The HTTA + TBP system can serve to illustrate the main points of ther-
modynamics of synergism. The overall extraction reaction is written as:
M n + + n HTTA(org) + p TBP(org) ← +
→ M(TTA) n (TBP) p (org) + p H ( 4.10a )
We assume that the first step in the extraction equation is complexation in the
aqueous phase
M z + + n TTA − ← M(TTA) nz − n (aq) + n H + ( 4.10 b)
→
leading to the formation of the uncharged complex M(TTA)n, which immedi-

ately dissolves in the organic phase due to its high hydrophobicity/lipophilicity
M(TTA) n (aq) ←
→ M(TTA) n (org) (4.10c)
*TBPO = (C4H9)3PO, see Appendix D, example 16, at the end of this book.
†Cheland or chelator is the chelating ligand.
118 Rydberg et al.
The adduct formation reaction in the organic phase (the “synergistic reaction”)
is obtained by subtracting Eqs. (4.10b) and (4.10c) from Eq. (4.10a):
M(TTA) n (org) + pTBP(org) ←
→ M(TTA) n (TBP) p (org) (4.10d )
Thermodynamic data for the extraction reactions of Eqs. (4.10a) and (4.10c)
allow calculation of the corresponding values for the synergistic reaction of Eq.
(4.10d). Measurements of the reaction
UO 2 (TTA) 2 (org) + TBP(org) ←
→ UO 2 (TTA) 2 ⋅ TBP(org) (4.11)
at different temperatures gives log K = 5.10, ∆H0 = −9.3 kJ ⴢ mol−1, T∆S o = 20.0
kJ ⴢ mol−1.
In another experiment, it was found for Th(TTA)4
Th(TTA) 4 (org) + TBP(org) ← Th(TTA) 4 ⋅ TBP(org) ( 4.12)
→
−1
the corresponding values: log K = 4.94, ∆H = −14.4 kJ.mol , T∆S = 13.7 kJ.
o o
mol−1.
Both UO2(TTA)2 and Th(TTA)4 have two molecules of hydrate water
when extracted in benzene, and these are released when TBP is added in reac-
tions Eqs. (4.11) and (4.12). The release of water means that two reactant mole-
cules (e.g., UO2(TTA)2 ⴢ 2H2O and TBP) formed three product molecules (e.g.,
UO2(TTA)2 ⴢ TBP and 2H2O). Therefore, ∆S is positive. Since TBP is more
basic than H2O, it forms stronger adduct bonds, and, as a consequence, the
enthalpy is exothermic. Hence, both the enthalpy and entropy changes favor the
reaction, resulting in large values of log K.
4.2.3 Case III: Maintaining Metal

Coordination Number
A guiding principle for the solvent extraction chemist is to produce an un-
charged species that has its maximum coordination number satisfied by lipo-
philic substances (reactants). For trivalent lanthanides and actinides (Ln and An,
respectively), the thermodynamic data suggest a model in which addition of one
molecule of TBP displaces more than one hydrate molecule:
An(TTA)3 (H 2O)3 TBP
 → An(TTA)3 (TBP)(H 2O)1−2
TBP
 → An(TTA)3 (TBP)1−3 (4.13)
This scheme of steps reflects the ability of some metals, like the trivalent actin-
ides and lanthanides, to vary their coordination number; since the trivalent Ln
and An may go from 9 to 8 and, finally, back to 9. The last step reflects the
operation of the third mechanism proposed for synergism.
Th(TTA)4 can be dissolved in dry benzene without hydrate water. The
values of the reaction of Eq. (4.12) in the system are: log K = 5.46, ∆H o = −39.2
kJ ⴢ mol−1, T∆S o = −8.0 kJ ⴢ mol−1 ⴢ K−1. The negative entropy is understandable

as the net degrees of freedom are decreased (two reactant molecules combine
to form one product molecule). However, the ∆H o value is much more negative.
These equations do not provide complete definition of the reactions that
may be of significance in particular solvent extraction systems. For example,
HTTA can exist as a keto, an enol, and a keto-hydrate species. The metal com-
bines with the enol form, which usually is the dominant one in organic solvents
(e.g., K = [HTTA]enol/[HTTA]keto = ⬃6 in wet benzene). The kinetics of the keto
→ enol reaction are not fast although it seems to be catalyzed by the presence
of a reagent such as TBP or TOPO. Such reagents react with the enol form in
drier solvents but cannot compete with water in wetter ones. HTTA ⴢ TBP and
TBP ⴢ H2O species also are present in these synergistic systems. However, if
extraction into only one solvent (e.g., benzene) is considered, these effects are
constant and need not be considered in a simple analysis.
In section 4.13.3 we return briefly to the thermodynamics of solvent ex-
traction.
4.3 OVERVIEW OF EXTRACTION PROCESSES

Many organic substances as well as metal complexes are less extracted from
aqueous solutions into organic solvents than expected from simple considera-
tions such as the amount of organic matter in the solute or their solubility in
organic solvents. Such substances are hydrated (see Chapter 3). More basic
donor molecules can replace such water, forming adducts. For the most common
oxygen-containing adduct molecules, the efficiency of the replacement depends
on the charge density, also referred to as basicity, of the oxygen atoms. The
sequence in which these donor groups are able to replace each other is
RCHO < R 2CO < R 2O < ROH < H 2O ≈ (RO)3PO′
< R ′′R ′NCOR ≈ (RO) 2 RPO < R 3PO
where R stands for organic substituend. In Chapter 3 the basicity was presented
in form of donor number. The larger the difference between the donor number
of water and the adduct former, the larger the adduct formation constant. Often
the donor property has to be rather strong, which is the case for many phospho-
ryl compounds (like TBP, TBPO, TOPO, etc.), because the concentration of
H2O in the aqueous phase is very large (often >50 M), even though H2O is only
a moderately strong donor.
Table 4.2 gives a survey of the most common extraction processes. In
general, Type I extraction refers to the distribution of nonelectrolytes, without
(A) or with adduct former (B). Type II refers to extraction of (mainly organic)
acids, Type III to the extraction of metal complexes, and Type IV to the special
(but common) use of solvent extraction for evaluation of formation constants
120
Table 4.2 Symbolic Survey of Fundamental Liquid-Liquid Distribution Processesa
Type I-A Nonelectrolyte extractionb

A
Solute A extracted into organic phase (solvent)
↓↑
(Equilibrium governed by the Nernst distribution law)
A
Solute is the nonelectrolyte A in water
Type I-B Nonelectrolyte adduct formation and extractionc

B AB
Adduct AB in organic phase (plus eventually B)
(↓↑) ↓↑
A+B → ← AB
Solute A and adduct former (or extractant) B
Type II-A Extraction of nonadduct organic acids ← H1⁄2H2A2 + . . .

HA →
Acid and dimer (and possible polymers) in organic phase
↓↑
← H+ + A−
HA →
Acid dissociation in aqueous phase
Type II-B Extraction of acid as adduct

HAB ←) B (+) HA
(→
Acid adduct (and acid and adduct former) in organic phase
↓↑ (↓↑) (↓↑)
HAB →
← B + HA → ← H + + A−
Acid dissociation in aqueous phase
Type III-B d Extraction of saturated metal complex

MAz
Neutral, coordinatively saturated metal complex in organic phase
↓↑
← MAz
Mz+ + zA− →
Rydberg et al.
Metal ion Mz+ is complexed by z A− ligands

Type III-C Adduct extraction of unsaturated metal complex
B MAzBb
Coordinatively saturated metal complex in organic phase (and B)
(↓↑) ↓↑
← MAzBb
Mz− + zA− + bB →
Formation of saturated metal complex trough adduct former Be
Type III-D Liquid anion exchange extractions

RNH+L− (RNH+)pML−np
Organic phase with anion exchanger and metal complex
↓↑ ↓↑
← (RNH+)pML−np
Mz+ + nL− + pRNH+L− →
Metal with complexing anions L− and organic amine
Solvent Extraction Equilibria
Type III-E Extraction of ion pairs, and other unusual complexes

C+1A−2
Ion pair C+1A−2 (and counter species) in organic phase
↓↑
← C+1A−2 + C+2A−1
C+1A−1 + C+2A−2 →
Aqueous cation C+1 and anion A−2 associated into ion pair C+1A−2
Type IV Hydrophilic complex formation and solvent extraction

MAz
Coordinatively saturated metal complex in organic phase
↓↑
← MAz + MXnz−n
Mz+ + zA− + nX− →
Formation of extractable and nonextractable complexes
a
The organic phase (solvent, diluent) is assumed to be “inert” (shaded area). The aqueous phase (nonshaded area) is unspecified, but may
contain various salting agents, not considered here.
b
A nonelectrolyte solute is denoted A, an electrolyte solute is assumed to be the cation Mz+ and anion A−, L−, or X−.
c
The extractant (or reactant) is denoted A− (from acid HA), or ligand L−, and by B (for adduct).
d
Type III-A (denoted Class A in first edition of this book) is closely related to and covered by Type I-A.
e
If B is undissociated HA, the self-adduct MAz(HA)x may be formed.
121
122 Rydberg et al.
for hydrophilic complexes. An arrow within parentheses suggests a reaction of

secondary importance. Our three examples are all of Type III-C, but contain
also elements of Type I-A (the distribution of TBP) and II-A (the distribution
of the weak acid HTTA), though the presence of undissociated acid (HTTA) or
the acid adduct (TBP-HTTA) is not discussed. In evaluations of experiments,
all molecular species present and all equilibria must be taken into account, as
demonstrated subsequently for a number of cases.
Solutes containing metals can further be classified according to the type
of ligand; N refers to the maximum coordination number of the metal relative
to the ligand:
Class A: Type MXN. (Note: We generally assume that the ligand is monova-
lent.) A small number of almost purely covalent inorganic compounds that
are extracted by nonsolvating organic solvents. As these complexes are
nonelectrolytes and almost as inert as the solutes of Type I-A, they are
treated jointly in section 4.4.
Class B: Type MAz. Neutral coordinatively saturated complexes formed be-
tween the metal ion and a lipophilic organic acid. This class contains
the large group of metal-organic chelate compounds. For monbasic acids
forming bifunctional chelates, z = N/2. They belong to the extraction Type
III-B, treated in section 4.8.
Class C: Type MAzBb or MLzBb. These Type III-C complexes are discussed in
section 4.2. They are neutral complexes formed between the metal ion and
ligands A− or L−, where the neutral complex MAz or MLz is coordinatively
unsaturated (N > z or 2z) and acts as an acceptor for uncharged organic
compounds (adducts B) containing lipophilic donor groups. If the system
does not contain any donor molecules B, the water of hydration may be
replaced by undissociated HA (assuming the ligand A− to be a dissociation
product of HA), at least at high HA concentrations; the MAz(HA)x com-
plexes are refered to as self-adducts. Both types of complexes are dis-
cussed in section 4.9.
Class D: Ion pairs, consisting of the metal bound in an anionic complex (e.g.,
MLnz−n, where n > z) and one or more large organic (usually monovalent)
cations (symbolized by RNH+); the extracted complex is written (RNH)n−z
MLn. These complexes are treated in section 4.10.
Class E: Metal complexes that do not fit into these categories; e.g., other types
of ion pairs and chlatrate compounds (see section 4.11).
All metal ions in water are hydrated, and at higher pH most of them also
hydrolyze. It can be difficult to distinguish between the hydrolyzed and the
complexed species, as well as their self-adducts. For such systems, plots of DM
against [A−] at various pH and total concentrations of [HA] show three types of
curves: (a) for the simple chelate MAn, (b) for the self-adduct MAn(HA)b, and
(c) for the (mixed) hydroxide MAn(OH)p see Fig. 4.3. It should be noted that the
mixed MAn(OH)p complexes include the MAn + M(OH)p complexes. As mixed
complexes are more difficult to determine, they are less often described. How-
ever, it is important to realize that if metal hydroxy complexes are formed and
not corrected for, the result of the investigation can be misleading. A test of the
system according to Fig. 4.3 rapidly establishes the type of metal complexation.
Because metals differ in size, charge, and electronic structure, no two
metals behave exactly the same in the same solvent extraction system, not even
for the same class of solutes. Nevertheless, there are systematic trends in the
formation and extraction of these complexes, as described in Chapter 3. Here,
the emphasis is on models that give a quantitative description of the extraction
within each type or class.
In the subsequent discussion, the following simplifications are made:
1. The systems behave “ideally,” i.e., the activity factors are assumed to be
unity, unless specifically discussed;
2. The metal extracted is in trace concentration: [M]t Ⰶ [Extractant]t, as this
simplifies the equations;
3. The reactants are at very low concentrations in both phases.
These are great simplifications in comparison with the industrial solvent
extraction systems described in later chapters. Nevertheless, the same basic reac-
tions occur also in the industrial systems, although activity factors must be intro-
duced or other adjustments made to fit the data, and the calculation of free
Fig. 4.3 Extraction curves for various types of metal chelate complexes, when log DM
is plotted against free ligand ion concentration, pA = −log[A−], or against [HA][H+]−1.
From such plots, the general type of metal chelate complex may be identified: (a) type
MAn, (see also Fig. 4.10); (b) type MAn(OH)p(HA)r, (see also Figs. 4.14 and 4.30); (c)
type MAn(OH)p, (see also Fig. 4.19). (From Refs. 3a and 3b.)
124 Rydberg et al.
ligand concentrations are more complex. Some of these simplifications are not
used in later chapters.
4.4 EXTRACTION OF INERT MOLECULES (TYPE I-A)

Here, and in later sections, we begin with same kind of rectangular figure to
indicate the type of extraction: to the right we indicate the distribution of the
solutes in a two-phase system (the organic phase is shaded); the system is also
briefly described by the text to the left, and—of course—in detail in the main
text.
Solute A extracted into organic phase (solvent) A

(Nernst distribution law for regular mixtures and solvents:) ↓↑
The non-electrolyte solute A in water A
If the solute A does not undergo any reaction in the two solvents, except for the
solubility caused by the “solvation” due to the nonspecific cohesive forces in
the liquids, the distribution of the solute follows the Nernst distribution law, and
the equilibrium reaction can be described either by a distribution constant KD,A,
or an (equilibrium) extraction constant Kex:
A(aq) ←
→ A(org); K D,A = K ex = [A]org /[A]aq (4.14)
Kex always refers to a two-phase system. The measured distribution ratio for
the solute A, DA, equals KD,A, and is a constant independent of the concentration
of A in the system. Only “external” conditions influence the KD,A value. In
“external” conditions we include the organic solvent, in addition to physical
conditions like temperature and pressure.
The noble gases and the halogens belong to the same type of stable molec-
ular compounds: RuO4, OsO4, GeCl4, AsCl3, SbCl3, and HgCl2. The simplest
example is the distribution of the inert gases, as given in Table 4.3. The larger
Table 4.3 Distribution Ratios of Some Gases Between Organic Solvents

and 0.01 M NaClO4 at 25°C
Solute
Permittivity
Solvent ε Xenon Radon Bromine Iodine
Hexane 1.91 41 80 14.5 36

Carbontetrachloride 2.24 35 59 28 86
Chloroform 4.90 35 56 37 122
Benzene (π-bonds) 2.57 27 55 87 350
Nitrobenzene 34.8 14 21 41 178
Source: Ref. 4.
Rn is extracted more easily than the smaller Xe, because the work to produce a
cavity in the water structure is larger for the larger molecule. The energy to
produce a cavity in nonpolar solvents is much less, because of the weaker inter-
actions between neighboring solvent molecules. Energy is released when the
solute leaves the aqueous phase, allowing the cavity to be filled by the hydro-
gen-bonded water structure. Thus the distribution constant increases with in-
creasing inertness of the solvent, which is measured by the dielectric constant
(or relative permittivity). The halogens Br2 and I2 show an opposite order due
to some low reactivity of halogens with organic solvents. Very inert solvents
with low permittivity, such as the pure hydrocarbons, extract inert compounds
better than solvents of higher permittivity; conversely, liquids of higher permit-
tivity are better solvents for less inert compounds. Molar volumes should be
used for accurate comparisons; such data are found in Table 2.1 and in Ref. [6].
In benzene, the distribution constant depends on specific interactions be-
tween the solute and the benzene pi-electrons. Table 4.4 shows the importance
of the volume effect for the mercury halide benzene system (Cl<Br<I).
Undissociated fatty acids (HA) behave like inert molecules. Figure 4.4
shows the distribution (DHA = KD,HA) between benzene and 0.1 M NaClO4 of
fatty acids of different alkyl chain lengths (Cn, n = 1 to 5); the distribution con-
stant for an acid with chain length n is given by the expression log KD,HA = −2.6
+ 0.6n. Similar correlations between KD,HA and molecular size or chain length
are observed also for other reagents (e.g., normal alcohols).
For organic solutes, not only the size but also the structure is of impor-
tance. Table 4.5 gives distribution constants for substituted oxines. When the
substitution increases the size of the molecule, the distribution constant in-
creases. The variations within Table 4.5 and position of the substitution in the
oxine molecule reflect structural effects. Table 4.6 shows distribution constants
for β-diketones. The increasing KD with molecular size for the series acetylace-
tone, benzoylacetone, and dibenzoylmethane, reflects the decreasing solubility
in the water phase, mainly governed by the increased energy necessary to over-
come the solute-solvent interactions for the larger extractant molecules in water.
Thenoyltrifluoroactetone has a greater hydrophilic character than the other β-
diketones due to its O− and F− atoms, which interact with the water molecules,
Table 4.4 Distribution Constants for Mercury

Halides Between Benzene and 0.5 M NaClO4
Solute Log KD Solute Log KD
HgCl2 −0.96 HgClBr −0.42

HgBr2 0.15 HgICl 0.28
HgI2 1.79 HgIBr 0.79
Source: Ref. 5.
126 Rydberg et al.
Fig. 4.4 Distribution constants KD,HA of fatty acids as a function of the number n of
carbon atoms in the alkyl chain (C1 is acetic acid) in the system 0.1 M NaClO4/benzene.
(From Ref. 7.)
Table 4.5 Dissociation, Ka, and Distribution, KD,HA,

Constants for Substituted Oxinesa
Reagent pKa log KD,HA
Oxine 9.7 2.7

2-Methyloxine 10.0 3.4
5-Methyloxine 9.9 3.3
5-Acetyloxine 7.8 2.8
4,7-Dichlorooxine 7.4 3.9
7,7-Diiodooxine 8.0 4.2
5-Chloro-7-iodooxine 7.9 3.9
a
Aqueous phase 0.1 M NaClO4; organic phase chloroform at
25°C.
Source: Refs. 8a, b.
Table 4.6 Dissociation Ka and Distribution Constants KD,HA for β-Diketones;

Aqueous Phase 0.1 M NaClO4a
log KDR
Reagent (solute) pKa C6H6 CHCl3 CCl4
Acetylacetone, AA; CH3 R CH3 8.76 0.76 1.36 0.51

Benzoylacetone, BZA; CH3 R C6H5 8.74 3.15 3.60 2.81
Dibenzoylmethane, DBM; C6H5 R C6H5 9.35 5.34 5.40 4.51
Thenoyltrifluoroacetone, TTA; R′R CF3 6.3 1.61 1.84 1.54
R is C CH 2 C R ′ is thenoyl, H 3C 3SC
|| ||
O O
a
Organic phases 0.1 M in solute; 25°C.
Source: Ref. 4.
leading to a reduction in the distribution constant. Thus, either the size effect
related to the water structure or the presence of hydrophilic groups in the solute
determines the general level of its distribution constant.
4.5 EXTRACTION OF ADDUCT-FORMING

NONELECTROLYTES (TYPE I-B)
Adduct AB in organic phase (plus evt. B) B AB

(↓↑) ↓↑
← AB
Solute A and adduct former (extractant) B A + B →
The extraction of a solute A may be improved by its reaction with another
solute (“extraction reagent”, or extractant), B, forming an adduct compound,
AB. This occurs through chemical interaction between A and B.
B←
→ B(org) K D , B = [ B]org /[ B] (4.15a )
A+B←
→ AB K ad = [AB]/[A] [B] (4.15b)
where Kad is the adduct formation constant (in the aqueous phase)
AB ←
→ AB(org) K D, AB = [ AB]org /[ AB] (4.15c)
and KD,AB the adduct distribution constant. The extraction constant for the overall
reaction is
128 Rydberg et al.
A(aq) + B(org) ←
→ AB(org) K ex = [ AB]org /[ A ]aq [ B]org
= K D,AB K ad K D,B (4.15d )
and also
DA = K ex [B]org (4.15e)
For the extraction reaction it may suffice to write the reaction of Eq.
(4.15d), though it consists of a number of more or less hypothetical steps. As
mentioned, equilibrium studies of this system cannot define the individual steps,
but supplementary studies by other techniques may reveal the valid ones. Equa-
tion (4.15) indicates that the reaction takes place at the boundary (interface)
between the aqueous and organic phases. However, it is common to assume that
a small amount of B dissolves in the aqueous phase, and the reaction takes place
in the steps
A(aq) + B(aq) → AB(aq) → AB(org)
These equations allow definition of a distribution constant for the species AB,
KD,AB [see Eq. (4.15c)]. Distribution constants can also be defined for each of
the species A, B and AB (KD,A, etc.) but this is of little interest as the concentra-
tion of these species is related through Kex. A large Kex for the system indicates
that large distribution ratios DA can be obtained in practice. As shown in Eq.
(4.15), the concentration of B influences the distribution ratio DA.
Consider first the extraction of hexafluoroacetylacetone (HFA) by TOPO
by Example 1, and, second, the extraction of nitric acid by TBP (Example 7).
The principles of volume and water-structure effects, discussed for the solute A
in section 4.4, are also important in the distribution of the adducts.
Example 1: Extraction of hexafluoroacetylacetone (HFA) by trioctylphospine
oxide (TOPO).
Abbreviating HFA (comp. structure 5e, Appendix D) by HA, and TOPO
by B, we can write the relevant reactions
HA(aq) ←
→ HA(org) K D = [HA]org /[HA] = D0 (4.16a )
←
HA(org) + B(org) → HAB(org) K ad1 = [HAB]org /[HA]org [ B ]org (4.16b)
HA(org) + 2B(org) ← 2
→ HAB2 (org) K ad2 = [HAB2 ]org /[HA]org [ B ]org (4.16c)
assuming that 2 adducts are formed, HAB and HAB2, the latter containing 2
TOPO molecules. Equation (4.16a) denotes the distribution of “uncomplexed
HA” by Do. Combining these equations yields
D ⋅ D0−1 = 1 + Kad1[B] + Kad 2 [B]2 ( 4.16d )
−1
Figure 4.5 shows the relative distribution, log D ⴢ D , of hexafluoroace-
o
tylacetone as a function of the concentration of the adduct former TOPO. HFA
Fig. 4.5 Relative increase, D/Do in extraction of hexafluoroacetylacetone (HFA) into

hexane from 0.1M NaClO4 at pH = 2, at different concentrations of the adduct trioctyl-
phosphine oxide (TOPO) in the organic phase. The fitted curve is D/Do = 1 + 104.22
[TOPO]org + 107.51[TOPO]2org. (From Ref. 9.)
is a moderately weak acid, while TOPO associates strongly with hydrogen-

bond donors in nonpolar solvents like hexane. The constants were determined
to log Kad1 = 4.22 and log Kad2 = 7.51. Thus even at moderately low TOPO
concentrations, the “dimer adduct” dominates.
4.6 EXTRACTION OF NONADDUCT ORGANIC ACIDS

(TYPE II-A)
Acid and dimer (and possibly ← 1⁄2 H2 A2 + . . .

HA →
polymers) in organic phase ↓↑
← HA →
Acid dissociation in aqueous phase (H+AH) → ← H+ + A−
(and protonation)
Tables 4.5–4.7 and Fig. 4.6 list organic acids commonly used as metal
extractants. When the acids are not protonated, dissociated, polymerized, hy-
drated, nor form adducts, the distribution ratio of the acid HA is constant in a
given solvent extraction system:
HA(aq) ←
→ HA(org) K D ,HA = [HA]org /[HA] = DHA (4.17)
This is shown by the horizontal trends in Fig. 4.6, for which Eq. (4.17) is valid;
i.e., the distribution constant KD,HA equals the measured distribution ratio. When
130 Rydberg et al.
Table 4.7 Physical Properties of Some Commonly Used Acidic Extractantsa

Organic
Acid diluentb Sorg(M)c −log Ka log KDR
Salicylic acid Chloroform 0.17 2.9 0.5

Cupferron Chloroform 0.4 4.2 2.3
8-Hydroxyquinoline (oxine) (OQ) Chloroform 2.63 9.8d 2.66
D:o CCl4 — 9.66 2.18
Acetylacetone (AA) Benzene ∞ 8.85 0.78
D:o Chloroform — — 1.36
D:o CCl4 — 8.67 0.51
Benzoylacetone (BA) CCl4 — 8.39 2.81
Benzoyltrifluoroacetone (BTFA) CCl4 — 6.03 2.39
Thenoyltrifluoroacetone (TTA) Benzene 5.27 6.3 1.6
D:o Chloroform — — 1.84
1-Nitroso-2-naphthol Chloroform 1.35 7.6 2.97
Di(2-ethylhexyl)phosphoric acidf n-Octane ∞ 1.4 3.44
Mono(2-ethylhexyl)phosphoric acid n-Octane ∞ 1.3e —
Dinonyl naphthalene sulfonic acid
a
The aqueous phase is mostly 0.1 M NaClO4 at 25°C.
b
The choice of organic diluent only affects the distribution constant, not the acid dissociation
constant.
c
Sorg is solubility in M in organic solvent.
d
Log KaH = 5.00.
e
Ka1.
f
Dimerization constant: log K = 4.47; see also section 4.6.3.
HA is used for the extraction of a metal, KD,HA is abbreviated KDR, for the distri-
bution constant (of the unmodified) reagent (or extractant).
Figure 4.7a shows the effect of aqueous salt concentrations on the DHA
value of acetylacetone at constant total HA concentration and pH. The salt has
two effects: (1) it ties up H2O molecules in the aqueous phase (forming hydrated
ions) so that less free water is available for solvation of HA; and (2) it breaks
down the hydrogen bond structure of the water, making it easier for HA to
dissolve in the aqueous phase. Figure 4.7 shows that the former effect dominates
for NH4Cl while for NaClO4 the latter dominates. We describe the increase of
the distribution ratio with increasing aqueous salt concentration as a salting-out
effect, and the reverse as a salting-in effect.
Figure 4.7b shows DHA for the extraction of acetylacetone into CHCl3 and
C6H6 for two constant aqueous NaClO4 concentrations at pH 3, but with varying
concentrations of HA. Acetylacetone is infinitely soluble in both CHCl3 and
C6H6; at [HA]org = 9 M, about 90% of the organic phase is acetylacetone (Mw
100), so the figure depicts a case for a changing organic phase. Figure 4.7b also
Fig. 4.6 Distribution ratios calculated by Eq. (4.22) for acetylacetone (HAA); benzoyl-
acetone (HBA); bezoyltrifluoroacetone (HBTFA); and oxine (8-hydroxyquinoline,
HOQ), in the system 0.1 M NaClO4 /CCl4, using the following constants. (From Refs.
8a, b.)
HAA HBA HBTFA HOQ
log KD 0.51 2.81 2.39 2.18

log Ka 8.67 8.39 6.03 9.66
log KaH — — — 5.00
indicates different interactions between the acetylacetone and the two solvents.
It is assumed that the polar CHCl3 interacts with HA, making it more soluble in
the organic phase; it is also understandable why the distribution of HA decreases
with decreasing concentration (mole fraction) of CHCl3. C6H6 and aromatic sol-
vents do not behave as do most aliphatic solvents: in some cases the aromatics
seem to be inert or even antagonistic to the extracted organic species, while in
other cases their pi-electrons interact in a favorable way with the solute. For
acetylacetone, the interaction seems to be very weak. The salting-in effect is
shown both in Figs. 4.7a. and 4.7b.
4.6.1 Dissociation
Acids dissociate in the aqueous phase with a dissociation constant Ka
132 Rydberg et al.
HA ←
→ H +A
+ −
K a = [H + ][A − ]/[HA] (4.18)
The distribution ratio incorporates the Ka for extraction of acids, HA, as:
DA = [HA]org ([HA] + [A − ]) −1 = K D ,HA (1 + K a [H + ]−1 ) (4.19)
Index A indicates that the distribution ratio refers to the concentration of all
species of A in the organic and in the aqueous phase. In Fig. 4.6 the distribution
of the β-diketones is constant in the higher hydrogen ion concentration range
(lower pH) where they are undissociated. In the higher pH region, DA becomes
inversely proportional to the hydrogen ion concentration due to increase in the
concentration of the dissociated form of the acid A−, in agreement with Eq.
(4.19.)
The free ligand concentration, [A−], is an important parameter in the for-
mation of metal complexes (see Chapter 3 and section 4.8). In a solvent extrac-
tion system with the volumes V and Vorg of the aqueous and organic phases,
respectively, [A−] is calculated from the material balance:
log [A − ] = log Ka − log[H + ] + log ( mHA, t / Vorg ) − log F ( 4.20a )
where
−1 −1
F = K D ,HA + V Vorg + K a [H + ] V Vorg (4.20b)
mHA,t is the total amount (in moles) of HA (reagent) added to the system. Often
−1 o
mHA,t V org is abbreviated [HA]org, indicating the original concentration of HA in
the organic phase at the beginning of the experiment (when [HA]aq = 0). When
Vorg = V and pH Ⰶ pKa, F = 1 + KD,HA. From Eq. (4.20) it can be deduced that
[A−] increases with increasing pH, but tends to become constant as the pH value
approaches that of the pKa value. In the equations relating to the extraction of
metal complexes, HA is often identical with reagent R; the indexes may be
changed accordingly, thus e.g., KD,HA ≡ KDR. (Note: Various authors use slightly
different nomenclature; here we follow reference Appendix C.)
4.6.2 Protonation
At low pH, some organic acids accept an extra proton to form the H2A+ com-
plex. This leads to a decrease in the DA value at pH < 6, as shown in Fig. 4.6:
䉳
Fig. 4.7 Distribution ratio DHA of undissociated acetylacetone. (a) Distribution between
benzene and aqueous phase containing different inorganic salts; 25°C. (b) Distribution
between CHCl3 (upper curves) or C6H6 (lower curves) and aqueous phase 0.1 and 1.0 M
in NaClO4 as a function of [HA]org. The uncertainty at the lowest D values is ±1 for
CHCl3 and ±0.2 for C6H6. (From Ref. 10.)
134 Rydberg et al.
HA + H + ← H2A+ K aH = [HA][H + ]/[H 2 A + ] (4.21)

→
+
Because H2A is ionic, it is not extracted into the organic phase, and thus the
distribution ratio becomes
DA = [HA]org ([H 2 A + ] + [HA] + [A − ]) −1
−1
= K D,HA ( K aH [H + ] + 1 + K a [H + ]−1 ) −1 (4.22)
as illustrated in Fig. 4.6 for oxine.
4.6.3 Dimerization
Figure 2.1 illustrates a number of orientations by which two linear acids may
form a dimer. The partial neutralization of the hydrophilic groups leads to in-
creased solubility of the acid in the organic solvent, but is not observed in the
aqueous phase. The dimerization can be written as:
2HA(org) ←
→ H 2 A 2 (org) K di = [H 2 A 2 ]org /[HA]org
2
(4.23)
The distribution ratio for the extraction of the acid becomes:
DA = ([HA]org + 2[H 2 A 2 ]org )([HA] + [A − ]) −1
= K D ,HA (1 + 2 K di K D ,HA [HA]aq (1 + K a [H]−1 ) −1 (4.24)
The last term can be expressed in several different ways. Because the distribu-
tion ratio DA reflects the analytical concentration of A in the organic phase, the
dimer concentration is given as 2[H2A2], although it is a single species (one
molecule). Figure 4.8 illustrates how the dimerization leads to an increase of
acid distribution ratio with increasing aqueous acid concentration. For propionic
acid logKa = −4.87, logKD,HA = −1.90 and logKdi = 3.14 in the system. The ex-
traction increases as the size of the acid increases. A dimeric acid may form
monobasic complexes with metal ions, as is illustrated by the formulas in Ap-
pendix D:14 b–d, for the M(H(DEHP)2)3 and UO2((DEHP)2)2 complexes. The
situation may be rather complex. For example, at very low concentrations in
inert solvents, dialkylphosphates (RO)2POOH act as a monbasic acid, but at
concentrations >0.05 M they polymerize, while still acting as monobasic acids
(i.e., like a cation exchanger). The degree of dimerization/polymerization de-
pends on the polarity of the solvent [11b].
4.6.4 Hydration
In solvent extraction, the organic phase is always saturated with water, and the
organic extractant may become hydrated. In the extraction of benzoic acid, HBz
(Appendix D:2), it was found that the organic phase contained four different
Fig. 4.8 Distribution ratios (from bottom to top) of acetic (C2 ●), propionic (C3 O),
butyric (C4 ∆), and valeric (C5 ▲) acids (carbon chain length Cn) between carbon tetra-
chloride and water as a function of the acid concentration in the aqueous phase, [HA]aq.
(From Ref. 11a.)
species: the monomer HBz, the monomer hydrate HBz ⴢ H2O, the dimer H2Bz2,
and the dimer hydrate H2Bz2(H2O)2. Only by considering all these species is it
possible to explain the extraction of some metal complexes with this extractant.
4.7 EXTRACTION OF ACIDS AS ADDUCTS

(TYPE II-B)
← B + HA)
Acid adduct (and acid and adduct HAB (→
former) in organic phase ↓↑ (↓↑) (↓↑)
Acid HA dissociating and forming HAB →← B + HA →← H + + A−
adduct in aqueous phase
The solubility of organic acids in water is due to the hydrophilic oxo- and
hydroxo-groups of the acid that form hydrogen bonds with water molecules. If
the hydrogen ion of the acid is solvated by a donor organic base, B, in the
136 Rydberg et al.
organic phase, the adduct B–HA is likely to have much greater solubility in the
organic phase.
4.7.1 Weak (Organic) Acids, HA

In extraction of weak organic acids (abbreviated HA) from acidic aqueous solu-
tions, the concentration of undissociated acid, HA, exceeds the concentration of
its dissociated anions, A−, as long as pKa > pH. The acid may then act as adduct-
forming nonelectrolyte; see section 4.5 and Example 1.
4.7.2 Strong (Inorganic) Acids, HL

To avoid confusion with weak organic acids, strong inorganic acids are denoted
by HL. Most strong acids are completely dissociated and both cations and
anions are hydrated in aqueous solutions even at pHs as low as 0. The hydration
makes them lipophobic and almost insoluble in inert organic solvents. The hy-
drogen ion is a Lewis acid (Chapter 3) and is solvated by strong organic (donor
or base) molecules, such as those in Table 4.8 (e.g., alcohols, ethers, ketones,
esters, amines, phosphoryls, etc.). This results in greater lipophilicity, and the
acid becomes more soluble in inert organic solvents. The structure of these
“solvated hydrogen salts” is not well known, but may be represented symboli-
cally by HB+b L−, where B refers to the adduct former or the solvating solvent; b
may have a value of 1–4.
The order of extractability changes with aqueous acidity, but in general
follows the order HClO4 ⬇ HNO3 > HI > HBr > HCl > H2SO4 (see Table 4.8).
Since the hydration energies of the acids follow the opposite order, dehydration
is an essential step in the solvent extraction process. This order of acids has a
practical significance: acids higher can be replaced by the acids lower in the
sequence; e.g., HF and HNO3 are extracted from acidic stainless steel pickling
waste solutions into kerosene by addition of H2SO4 (see Chapter 14 of this book).
The extraction of most acids is accompanied by extraction of water. In
the extraction of HNO3 by TBP into kerosene, many different species have been
identified, several of which involve hydration. The ratio of acid:adduct is not
very predictable. For example, HClO4 apparently is extracted into kerosene with
1–2 molecules of TBP, HCl into ethylether with one molecule of ethylether,
etc. Also, the extracted acid may dimerize in the organic solvent, etc. Example
2 illustrates the complexity of the extraction of HNO3 by TBP into kerosene.
Assuming that B is almost insoluble in the aqueous phase, the equilibrium
reaction can be written in two ways:
1. The interface extraction model assumes that HA reacts with B at the interface.
Thus
HA(aq) + bB(org) ←
→ HABb (org) (4.25a )
Table 4.8 Basicity (Electron Pair-Donating Tendency) of Some Ions

and Molecules (R is an alkyl or aryl group)
Basicity of some common anions
relative to the (hard type) actinide cations
ClO−4 < I − < Br− < Cl− < NO−3 < SCN − < acetate− < F −
Basicity of some organic molecules

Amine compounds R3Na < R2NHb < RNH2c < NH3
Arsine compounds R3As
再
Phosphine compounds R3P
Phosphoryls (RO)3POd < R′(RO)2POe < R2′(RO)POf < R3′POg
Arsenyls R3AsOh
Oxo-compounds Carbonyls RCHO < R2CO (≤R2O < ROH < H2O)i
Sulfuryls (RO)2SO2j < R2SOk2 < (RO)2SOl < R2SOm
Nitrosyls RNOn2 < RNOo
Substitutions causing basicity decrease of oxo compounds
(CH3)2CH— < CH3(CH2)n — < CH3 — < CH3O— < ClCH2 —
a–c
Tertiary, secondary, and primary amines.
d
tri-R phosphate.
e
di-R-R′ phosphonate.
f
R-di-R′ phosphinate.
g
tri-R′ phosphine oxide.
h
arsine oxide.
i
ether and hydroxo compounds.
j
sulfates.
k
sulfones.
l
sulfites.
m
sulfoxides.
n
nitro compounds.
o
nitroso compounds.
Source: Ref. 12.
Since [HA]aq and [B]org are easily measurable quantities, it is common to define
the extraction constant Kex for this model:
K ex = [HABb ]org [HA]−1 [B]org
−b
(4.25b)
2. The organic phase reaction model assumes all reactions take place in the
organic phase. Thus one assumes
HA(org) + bB(org) ←
→ HABb (org) (4.26a )
The equilibrium constant for this reaction is
−1 −b
Kad, bB = [HABb ]org [HA]org [B]org ( 4.26 b)
Rydberg_5063-2_Ch05_R3_02-09-04 11:29:21
138 Rydberg et al.
where Kad,bB is the (organic phase) adduct formation constant. The distribution
ratio of the acid in this system becomes
DA = ([HA]org + [HABb ]org )([HA] + [A − ]−1
= K D ,HA (1 + ΣK ad,bB [B]borg ) /(1 + K a [H + ]) (4.26c)
Equation (4.25b) becomes identical to Eq. (4.26b) if Kex is replaced by KD,HA
Kad,bB. Equilibrium measurements do not allow a decision between the two reac-
tion paths.
Example 2: Extraction of nitric acid by pure TBP.
Many metals can be extracted from nitrate solutions by TBP. In those
systems it is important to account for the HNO3-TBP interactions. The next set
of equations were derived by [13] and are believed to be valid for the extrac-
tion of HNO3 at various nitrate concentrations into 30% TBP in kerosene. Ab-
breviating HNO3 as HL, and TBP as B, and including hydration for all species
without specification, one derives
1. The formation of an acid monoadduct:
H + + L− + B(org) ←
→ HLB(org) (4.27a )
For simplicity, we write the adduct HLB, instead of HB+L−. The extraction
constant is
K ex1 = [HLB]org [H]−1 [L]−1 [B]org
−1
(4.27 b)
2. The formation of a diacid monoadduct:

2 H + + 2 L− + B(org) ←
→ (HL) 2 B(org) (4.28a )
K ex2 = [(HL) 2 B]org [H] [L] [B]
−2 −2 −1
org
(4.28b)
3. Ion pair association:
H + + L− ← + −
→HL (4.29a )
K ass = [H + ] [L− ]/[H + L− ] (4.29b)
This reaction only occurs under strong acid conditions, and the equilibrium
constant may be <1.
4. The distribution of nitric acid is then given by
DNO3 = ([HLB]org + 2[(HL) 2 B]org + . . . )([L− ] + [H + L− ])−1 (4.30)
5. Furthermore, the dimerization of TBP in the organic phase must be taken

into account:
2B←
→ B2 (4.31a )
K di = [B2 ]org /[B] 2
org
(4.31b)
yielding the total concentration of TBP in the organic phase
[TBP]org,t = [B]org + 2 K di [B]org
2
(4.32)
Rydberg_5063-2_Ch05_R3_02-09-04 11:29:21
6. The distribution ratio in terms of only [H+] and monomeric [B]org can then
be expressed by
DNO3 = K ass−1 [B]org ( K ex1 + 2 K ex2 [H + ]2 ) (4.33)
In this equation it is assumed [H+]=[L−] (electroneutrality in the aqueous

phase). Equation (4.33) has been tested, and the results agreed with >2300
experiments under varying conditions, see Fig. 4.9. The example illustrates the
rather complicated situation that may occur even in such “simple” systems as
the extraction of HNO3 by TBP.
4.8 EXTRACTION OF COORDINATIVELY

SATURATED METAL CHELATE TYPE
COMPLEXES (TYPE III-B)
Neutral, coordinatively saturated metal complex MAz

in organic phase ↓↑
Metal ion M z+ is complexed by z A− ligands to ← MAz
M z+ + zA− →
form neutral MAz
Fig. 4.9 Test of the equations in Example 2 for extraction of 0.01–0.5 M nitric acid
with 30% TBP in kerosene at temperatures 20–60oC. (From Ref. 13.)
140 Rydberg et al.
Section 3.2 describes an important class of organic ligands that are able
to complex a metal ion through two or more binding sites of “basic” atoms, like
O, N, or S, to form metal chelates. Table 4.9 presents the various types, their
number of acidic groups, the chelate ring size, and the coordinating atoms. Neu-
tral chelate compounds are illustrated in Appendix D: 5h, 14b and 14d. In Ap-
pendix D:5h the ring size is 6 for the complex between Cu2+ and each of the
two acetylacetonate anions, Aa−, while for the dimeric HDEHP ligand in the
figures in Appendix D:14b and 14c the ring size becomes 7 for the M3+ and
UO2+2 complexes (see the structures in Appendix D). As discussed in Chapter 3,
Table 4.9 Some Organic Compounds Functioning as Polydentate Anions

in Metal Extraction
Chelate Acidic Coordinating
ring size groups atoms Compound group and examples
4 1 O, O Carboxylic acid, RCOOH; e.g., perfluorobutyric

(C3F7COOH), salicylic C6H4(OH)COOH, cin-
namic (C6H5(CH)2COOH) acids
4 1 O, Oa Di(alkyl or aryl)phosphoric and phosphinic
acids, RR′PO(OH); e.g., HDEHPb; correspond-
ing thioacidsa
4 1 S, S Dithiocarbamate, RR′NC(S)SH, xanthate,
ROC(S)SH; e.g., NaDDCc
4 2 O, Oa Mono(alkyl or aryl)phosphoric and phosphinic
acid, RPO(OH)2; e.g., H2MEHPd
5 1 O, O Nitrosohydroxylamine, RN(NO)OH; e.g., cupfer-
ron (R = C6H5); hydroxamic acid, RC(O)NHOH
5 1 O, N 8-Hydroxyquinoline (oxine), C9NH6OH
5 or 6 1 S, N, or N, N Diphenylthiocarbazone (dithizone), C6H5NHNC
(SH)NNC6H5
6 1 O, O β-Diketone, RC(O)CHC(OH)R′; e.g., acetylace-
tone (R = R′ = CH3), HTTAe (R = C4SH3, R′ =
CF3)
6 1 O, O l-Nitroso-2-naphthol, C10H6(NO)OH
6 2 O, O Di(alkyl or aryl)pyrophosphate, RP(O)(OH)
OP(O)(OH)R′; e.g., dioctylpyrophosphate (R =
R′ = C8H17O)
>5 or 2 × 4 2 O, O Dicarboxylic acids, R(COOH)2
a
O, S, or S, S for the corresponding thioacids.
b
Di(2-ethylhexyl)phosphoric acid.
c
Sodium diethyldithiocarbamate.
d
Mono(2-ethylhexyl)phosphoric acid.
e
Thenoyltrifluoroacetone.
Source: Ref. 12.
chelation provides extra stability to the metal complex. The formation and extrac-
tion of metal chelates are discussed extensively also in references [14–16].
In Chapter 3 we described how an uncharged metal complex MAz is
formed from a metal ion Mz+ (central atom) through a stepwise reaction with
the anion A− (ligand ) of a monobasic organic acid, HA, defining a stepwise
formation constant kn, and an overall formation constant βn, where
βn = [MA nz −n ]/[M z + ][A − ]n (4.34)
The MAz complex is lipophilic and dissolves in organic solvents and the distri-
bution constant KDC is defined (index C for complex):
K DC = [MA z ]org /[MA z ] (4.35)
Taking all metal species in the aqueous phase into account, the distribution
of the metal can be written (omitting the index aq for water)
[MA z ]org K DCβz [A − ]z
DM = = (4.36)
Σ [MA nz −n ] Σ β z [ A − ]z
The distribution ratio depends only on the free ligand concentration, which
may be calculated by Eq. (4.20). Most coordinatively saturated neutral metal
complexes behave just like stable organic solutes, because their outer molecular
structure is almost entirely of the hydrocarbon type, and can therefore be ex-
tracted by all solvent classes 2–5 of Chapter 2. The rules for the size of the
distribution constants of these coordinatively saturated neutral metal complexes
are then in principle the same as for the inert organic solutes of section 4.4.
However, such complexes may still be amphilic due to the presence of electro-
negative donor oxygen atoms (of the chelating ligand) in the chelate molecule.
In aqueous solution such complexes then behave like polyethers rather than
hydrocarbons. Narbutt [17] has studied such outer-sphere hydrated complexes
and shown that the dehydration in the transfer of the complex from water to the
organic solvent determines the distribution constant of the complex. This is
further elaborated in Chapter 16.
Example 3: Extraction of Cu(II) by acetylacetone.
Simple β-diketones, like acetylacetone (Appendix D: 5d) can coordinate
in two ways to a metal atom, either in the uncharged keto form (through two
keto oxygens), or in dissociated anionic enol form (through the same oxygens)
as shown in Appendix D: 5c, 5h. It acts as an acid only in the enolic form.
Figure 4.10 shows the extraction of Cu(II) from 1 M NaClO4 into benzene at
various concentrations of the extractant acetylacetone (HA) [18]. Acetylace-
tone reacts with Cu(II) in aqueous solutions to form the complexes CuA+ and
CuA2. Because acetylacetone binds through two oxygens, the neutral complex
CuA2 contains two six-membered chelate rings; thus four coordination posi-
tions are taken up, forming a planar complex (Appendix D: 5f). This complex
is usually considered to be coordinatively saturated, but two additional very
142 Rydberg et al.
Fig. 4.10 Extraction of Cu(II) from 1 M NaClO4 into benzene as a function of pH

(large figure) and of free acetylacetonate ion concentration (insert) at seven different
total concentrations of acetylacetone ([HA]aq 0.05–0.0009 M). (From Ref. 18.)
weak bonds can be formed perpendicular to the plane; we can neglect them
here.
The distribution of copper, DCu, between the organic phase and water is
then described by
[CuA 2 ]org
DCu = (4.37)
[Cu 2+ ] + [CuA + ] + [CuA 2 ]
One then derives
K DC β2 [A − ]2 K β [A − ]2
DCu = = DC 2 − n (4.38)
1 + β1 [A ] + β2 [A ]
− − 2
Σ βn [A ]
where KDC refers to the distribution constant of the uncharged complex CuA2.
In Eq. (4.37), log D is a function of [A−], the free ligand concentration,
only, and some constants. In Fig. 4.10, log Dcu is plotted vs. log [H+] (= −pH).
Through Eqs. (4.17) and (4.18) it can be shown that log DCu is a function of
pH only at constant [HA]org (or [HA]aq), while at constant pH the log DCu de-
pends only on [HA]org (or [HA]aq).
In the insert of Fig. 4.10, log D is plotted as a function of log [A−],

where [A−] has been calculated from pH, [HA]0org, phase volumes, and Ka and
KDR (same as KD,HA) for HA by means of Eq. (4.20); it is found that all curves
coincide into one at high pH (high [A−]), as expected from Eq. (4.38). The
distribution curve approaches two asymptotes, one with a slope of 2 and one
horizontal (zero slope). From Eq. (4.38) it follows that, at the lowest [A−]
concentration (lowest pH), the concentration of CuA+ and CuA2 in the aqueous
phase becomes very small; Eq. (4.38) is then reduced to
lim DCu = [CuA 2 ]org /[Cu 2+ ] = K DC β2 [A − ]2 (4.39)
[ A − ]→0
At the highest A− concentrations a horizontal asymptote is approached:

lim DCu = [CuA 2 ]org /[CuA 2 ] = K DC (4.40)
[ A − ]→∞
The horizontal asymptote equals the distribution constant of CuA2, i.e., KDC.
From the curvature between the two asymptotes, the stability constants β1 and
β2 can be calculated.
This example indicates that in solvent extraction of metal complexes with
acidic ligands, it can be more advantageous to plot log D vs. log[A−], rather
than against pH, which is the more common (and easy) technique.
In order to calculate DM from Eq. (4.36), several equilibrium constants as
well as the concentration of free A− are needed. Though many reference works
report stability constants [19, 20] and distribution constants [4, 21], for practical
purposes it is simpler to use the extraction constant Kex for the reaction
M z+ (aq) + zHA(org) ←
→ MA z (org) + zH (aq)
+
(4.41a )
z-n
in which case the MA complexes in the aqueous phase are neglected. The
n
relevant extraction equations are
K ex = [MA z ]org [H + ]z [M z+ ]−1 [HA]org
−z
(4.41b)
and
DM = K ex [HA]org
z
[H + ]− z (4.41c)
Thus only one constant, Kex, is needed to predict the metal extraction for given
concentrations [H+] and [HA]org. Tables of Kex values are found in the literature
(see references given).
Equation (4.41) is valid only when the complexes MAnz−n can be neglected
in the aqueous phase. Comparing Eqs. (4.37b) and (4.41c), it is seen that no
horizontal asymptote is obtained even at high concentrations of A−, or HA and
H. Thus, for very large distribution constant of the uncharged complex (i.e.,
Ⰷ1000) a straight line with slope −z is experimentally observed, as in the case
for the Cu(II)-thenoyltrifluoroacetone (HTTA) system (Appendix D: 5g).
144 Rydberg et al.
Example 4: Extraction of Cu(II) by Thenoyltrifluoroacetone.

Figure 4.11a shows the distribution of Cu(II) between three organic sol-
vents and water in the presence of isopropyltropolone (HITP) or thenoyltriflu-
oroacetone (HTTA) as a function of pH [22]. The straight line of slope −2 in
the pH-plot fits Eq. (4.41c); thus z = 2. It indicates that the aqueous phase does
not contain any significant concentrations of the complexes CuA+ or CuA2, yet
CuA2 must be formed in considerable concentrations, otherwise there would be
no extraction of Cu(II). The line also corresponds to the asymptote Eq. (4.39),
or (i.e., [A−] ⬀ [H+]−1). Thus the conclusion is that the aqueous phase is com-
pletely dominated by Cu2+, while the organic phase contains only CuA2. This
leads to the copper distribution ratio
DCu = [CuA 2 ]org /[Cu 2+ ] (4.42)
which is valid for the reaction
Fig. 4.11 Distribution of Cu(II) between hexone (∆), carbon tetrachloride (●), or chlo-
roform (䊊) and 0.1 M NaClO4 in the presence of isopropyltropolone (IPT) or thenoyl-
trifluoroacetone (TTA); (a) as a function of −log[H+] at constant [TTA]org = 0.1 M; (b)
as a function of [TTA]org at constant [H+] = 0.1M. (From Ref. 22.)
Cu 2+ + 2HA(org) ←
→ CuA 2 (org) + 2 H
+
(4.43a )
The extraction constant is
K ex = [CuA 2 ]org [H + ]2 /[Cu 2+ ] [HA]org
2
(4.43b)
and combining with Eq. (4.42)

DCu = K ex [HA]org
2
[H + ]−2 (4.44)
+
From Eq. (4.44) a plot of log D against −log[H ] should yield a straight
line with slope +2, as in Fig. 4.11a, and a plot against log [HA]org as in Fig. 4.11b
should also yield a straight line with slope +2. Continued in Example 14.
This example illustrates a case of considerable analytical importance, es-

pecially for the determination of complex formation constants for hydrophilic
complexes, as discussed in section 4.12, when the equilibrium constants for
the stepwise metal-organic complexes are of secondary interest. Kex values are
tabulated in several reference works. Kex is a conditional constant and only valid
provided no other species are formed besides the extracted one.
The distribution constant of the neutral complex MAz, KDC, has been re-
ferred to several times. In favorable cases, when both the organic and the aque-
ous phases are dominated by the same uncharged complex over a larger concen-
tration region, KDC can be directly measured, as is the case for most of the data
in Table 4.10 [22–23b]. Otherwise KDC can be estimated or calculated from Kex
data combined with βn, Ka, and KDR [see Eqs. (4.8) and (4.46)].
4.9 EXTRACTION OF METAL COMPLEXES

AS ADDUCTS (TYPE III-C)
Coordinatively saturated metal adduct B MAzBb

complex in organic phase (and B) (↓↑) ↓↑
Formation neutral complex cordinatively M z+ + zA− + bB ←
→ MAzBb
saturated by adduct former B
If the neutral metal complex is coordinatively unsaturated, it forms MAz
(H2O)x in the aqueous phase, where 2z + x (A being bidentate) equals the maxi-
mum coordination number. In the absence of solvating organic solvents, this com-
plex has a very low distribution constant. Obviously, if water of hydration can be
replaced by organic molecules B, the result is a more lipophilic adduct complex
MAzBb; many adduct formers are listed in Appendix D and several tables. De-
pending on the ligand, several types of such adducts exist: (i) type MAzBb, where
A and B are different organic structures; (ii) type MXzBb, where MXz is a neutral
inorganic compound (salt); and (iii) type MAz(HA)b, where A and HA are the
basic and neutral variant of the same molecule (so-called self-adducts).
146 Rydberg et al.
Table 4.10 Distribution Constants for Acetylacetone (HA) and Some Metal
Acetylacetonates Between Various Organic Solvents and 1 M NaClO4 at 25°C
log KDC
ε log KDR
Organic solventa Solvent HA ZnA2 CuA2 NpA4
n-Hexane (1) 1.88 −0.022 −1.57 — 0.5

Cyclohexane (3) 2.02 0.013 −1.16 −0.04 0.8
Carbon tetrachloride (4) 2.24 0.52 −0.39 0.85 2.7
Mesityleneb (6) 2.28 0.44 — 0.43 —
Xylene (7) 2.27 0.57 −0.47 0.80 —
Toluene (8) 2.38 0.66 −0.37 0.85 —
Benzene (9) 2.28 0.77 −0.21 1.04 3.3
Dibutylether (2) 3.06 — −1.05 — —
Methylisobutylketone (5) 13.1 0.77 −0.15 0.61 —
Chloroform (10) 4.9 1.38 0.83 2.54 —
Benzonitrile (11) 25.2 0.21 — —
a
Numbers in parentheses refer to Figs. 4.23 and 4.26.
b
1,3,5-trimethylbenzene.
Source: Refs. 22–23.
4.9.1 Metal-Organic Complexes with Organic Adduct

Formers, Type MAzBb
The extraction of the metal complex adduct can be written
M z + (aq) + zHA(org) + bB(org) ← MA z Bb (org) + zH + (aq) ( 4.45a )
→
The extraction constant is defined by

[MA z Bb ]org [H + ]z
K ex = (4.45b)
[M z+ ] [HA]org
z b
[B]org
or
−z −b
K ex = DM [H + ]z [HA]org [B]org (4.45c)
Thus the distribution of the metal, DM, is shown to depend on the concentrations
of H+ and HAorg to the power z of the charge of the metal ion, and on the
concentration of B to the power of b (i.e., number of adduct formers in the
extracted complex).
It can be shown that
−z
K ex = K az βz K DR K DC K ad,bB (4.46)
where (omitting ionic charges) Ka = [H][A]/[HA] is the acid dissociation con-

stant [see Eq. (4.18)], KDR = [HA]org/[HA] is the distribution constant for the
undissociated acid HA [see Eq. (4.17)], and
K ad,bB = [MA z Bb ]org /[MA z ]org
b
(4.47)
is the formation constant for the adduct MAzBb in the organic phase [see Eqs.
(4.15), (4.16), and (4.26)]. The five parameters Ka, βn, KDR, KDC, and Kad,bB are
in principle unrelated, even though it may not always be possible to change one
without affecting the others, as each molecular species may take part in several
equilibria. Without considering the independent parameters, it is often difficult
to understand why Kex varies in the fashion observed, and it may be impossible
to predict improvements of the system. A good example is the extraction of
Zn(II) by β-diketones and TBP:
Example 5: The extraction of Zn(II) by β-diketones and phosphoryl adduct
formers.
Figure 4.12 illustrates the extraction of Zn(II) from 1 M NaClO4 into
carbon tetrachloride by β-diketones (HA) in the presence of the adduct formers
Fig. 4.12 Enhancement of Zn(II) extraction, D D−o1, from 1 M NaClO4 into carbon
tetrachloride containing the complexing extractants acetylacetone (䊊), trifluoroacetone
(∆), or hexafluoroacetone (䊐) as a function of the concentration of the adduct former
trioctyl phosphine oxide (B). The curves are fitted with Eq. (4.50) using the constants
log Kad1 = 3.07 (AA), 6.70 (TFA), 7.0 (TFA), and Kad2 = 4.66 (AA), nil (TFA), 11.6
(HFA). (From Ref. 24.)
148 Rydberg et al.
TBP and TOPO (B) [12]. The extracted neutral complex is ZnA2Bb. The distri-
bution ratio becomes
[ ZnA 2 ]org + [ ZnA 2 B]org + [ ZnA 2 B2 ]org + L
DZn = (4.48a )
[ Zn] + [ ZnA] + [ ZnA 2 ] + L
To analyze these systems, the overall extraction reaction must be broken into
its partial reactions, or by introducing Eq. (4.47), to obtain
K DCβ2 [A]2 (1 + K ad,1[B]org + K ad,2 [B]org
2
+ L)
DZn = (4.48b)
Σ βn [A]n
where the adduct formation constant is defined by
K ad,b = [ ZnA 2 Bb ]org [ZnA 2 ]org
−1 −b
[B]org (4.49)
In the absence of any adduct former, DZn is given as a function of the free
ligand concentration by Eq. (4.36), i.e.,0 the parentheses in Eq. (4.48b) equals
1; denoting this DZn-value as Do, and introducing it into Eq. (4.48) gives
DZn′ = Do (1 + K ad ,1[B]org + K ad,2 [B]org
2
+ L) (4.50)
DZn′ (instead of DZ′ n) indicates that this expression is valid only at constant
[A−], or, better, constant [H+] and [HA]org [see Eqs. (4.36) and (4.41c)]. In Fig.
4.12, log DZ′ n D−o1 is plotted as a function of log [B]org. The distribution ratio
proceeds from almost zero, when almost no adduct is formed, towards a limit-
ing slope of 2, indicating that the extracted complex has added two molecules
of B to form ZnA2B2. From the curvature and slope the Kad,b-values were deter-
mined (see section 4.10). The calculation of the equilibrium constants is further
discussed under Example 13.
Tables 4.11–4.13 presents adduct formation constants according to Eq.
(4.47). For the alkaline earths TTA complexes in carbon tetrachloride in Table
4.11, the TBP molecules bond perpendicular to the square plane of the two TTA
rings, producing an octahedral complex. The higher the charge density of the
Table 4.11 Adduct Formation Constants

for the Reaction M(TTA)2 + bTBP X
M(TTA)2(TBP)b in CCl4 Showing Effect
of Charge Density
Mz+ (r pm) log Kad1 log Kad2
Ca2+ (100) 4.11 8.22

Sr2+ (118) 3.76 7.52
Ba2+ (135) 2.62 5.84
Source: Ref. 4.
Table 4.12 Adduct Formation Constants for the Reaction EuA3(org) +

bTBP(org) X EuA3(TBP)b(org), Eq. (4.26a), Where org = CHCl3,
and HA Substituted Acetylacetone Ligands
Ligand (A) Log Ka Log Kad1 Log Kad2
Acetylacetone, AA 8.76 1.90 —

Benzoylacetone, BZA 8.74 1.60 —
Trifluoroacetone, TFA — 3.32 4.64
Benzoyltrifluoroacetone, BTA — 3.64 5.28
2-Furoyltrifluoroacetone, FTA — 3.50 5.00
Thenoyltrifluoroacetone, TTA 6.3 3.34 5.28
Source: Ref. 4.
central atom, the stronger is the adduct complex with TBP. Note: Charge density
refers to the electrostatic interaction between ions of opposite charge according
to the Born equation [see (2.13), and (3.13)] (based on the Coulomb interac-
tion). It is mostly given as the ratio between the ionic charge, z+ (or z+z−) and
the ionic radius r+ (or r+r−). Table 4.12 compares the adduct formation of the
europium β-diketone complexes with TBP in chloroform. In Eu(TTA)3 the
TTAs only occupy six of the eight coordination positions available; the two
empty positions have been shown to be occupied by water. Though the tendency
is not strong, the stronger the acid (i.e., the larger its electronegativity), the
larger is the adduct formation constant. Table 4.13 compares the adduct forma-
tion tendency of the Eu(TTA)3 complex with various adductants. The basicity
of the donor oxygen atom increases in the order as shown in Table 4.13, as does
the adduct formation constant; this is in agreement with the order of basicity in
Table 4.8 and the donor numbers of Table 3.3; see also section 4.2.
Table 4.13 Constants for Formation of Eu(TTA)3Bb Adducts

According to Eq. (4.26a)
Chloroform Carbon tetrachloride
Adduct forming
ligand (B) log Kad1 log Kad2 log Kad1 log Kad2
TTA(self-adduct) 0.56 — >0.5 —

Hexone 1.16 1.52 1.71 2.34
Quinoline 3.29 — 3.48 5.16
TBP 3.63 5.40 5.36 8.96
TOPO 5.40 7.60 7.49 12.26
Source: Ref. 4.
150 Rydberg et al.
The difference between the two solvent systems is likely a result of CHCl3
solvating the Eu complex to some extent, while CCl4 is inert. This has two
effects: the KDC value increases due to the solvation by CHCl3 (not shown in the
table), while the adduct formation constant Kad decreases as the solvation hinders
the adduct formation; the more inert solvent CCl4 causes an opposite effect, a
lower KDC and a larger Kad.
4.9.2 Metal Inorganic Complexes with Organic

Adduct Formers, Type MXzBb
The extraction of mineral salts is generally less complicated than the extraction
of mineral acids. Metal salts with monodentate univalent anions like Cl−, ClO−4,
SCN−, and NO−3 are strongly hydrated in the aqueous phase and have quite small,
if any, solubility in inert solvents. In order to extract these acids, they must
either form an adduct with a strongly basic extractant like TBP or TOPO, or be
in solvating solvents such as ethers, ketones, alcohols, or esters. Examples of
extracted metal salt adducts are: Br3(EtO)b, PaCl3(MIBK), UO2(NO3)2(TBP)2,
Co(ClO4)2(octanol)b, etc. (b is uncertain). Tables of the extraction of a large
number of metal salts by solvating solvents or commercial adduct formers dis-
solved in kerosene are given (see Ref. [5]).
It has been shown [25] that monomeric metal hydroxides can be extracted
by strong donor molecules; e.g., in the form of the adducts Ln(OH)3(TOPO)b
into CHCl3, where b is 2–3. Under favorable conditions, the DLn value may
exceed 1, though the fraction of hydroxide is quite low.
When the solvent is a good solvater, the determination of the solvation
number b is difficult, unless the dependence of the extractant concentration on
the solvent can be obtained. Solvation numbers can be obtained in mixtures of
a solvating extractant and an inert diluent like hexane. Further, in these systems
the extraction of the metal commonly requires high concentrations of salt or
acid in the aqueous phase, so the activity coefficients of the solutes must be
taken into account.
Example 6: Extraction of Zn(II) thiocyanate complexes by TOPO.
Figure 4.13 shows the extraction of Zn(II) from aqueous thiocyanate
(L−) solutions into 0.001 M TOPO (B) in hexane; the aqueous phase is 1.0 M
Na(SCN −, ClO−4 ) at a pH around 5 [26]. Zn(II) is known to form a number of
weak Zn(SCN)n−2 n complexes in the aqueous phase. The uncharged one is as-
sumed to accept TOPO to form the adducts Zn(SCN)2(TOPO)b, where b = 1
or 2.
Assume that the reaction between the neutral complex and the solvating
molecule takes place at the interface (to assume the reaction to take place in
the organic phase would be unrealistic, as the zinc thiocyanate is insoluble in
hexane); thus the etxraction reaction is
Fig. 4.13 Distribution ratio of Zn(II) when extracted from 1 M Na(SCN−,ClO−4 ) into
0.001 M TOPO in hexane, as a function of aqueous SCN− concentration. The following
equilibrium constants were obtained with Eq. (4.53): β1 3.7, β2 21, β3 15, Kex 2.5 107 for
b = 2. (From Ref. 26.)
ZnL 2 + b B(org) ←
→ ZnL 2 Bb (org) (4.51)
for which we may define an equilibrium constant Kex,bB.
Because more than one solvated species may be extracted, the distribution ratio
becomes
[ ZnL 2 B]org + [ ZnL 2 B2 ]org + L
DZn = (4.52)
[ Zn] + [ ZnL] + [ ZnL 2 ] + [ ZnL3 ] + L
Inserting the partial equilibrium constants [see Eq. (4.46)],
β2 [L]2 Σ K ex,bB [B]borg
DZn = (4.53)
1 + Σ βn [L]n
where both summations are taken from 1. The solvation number b can be deter-
mined from the dependence of D on [B]org while [L] is kept constant. From the
slope of the line in Fig. 4.13 at low SCN − concentrations, it follows that b =
2; thus only one adduct complex is identified: Zn(SCN)2(TOPO)2. The authors
were able to calculate the formation constants βn from the deviation of the
curve from the straight the line at constant [B]org, assuming b constant. With
152 Rydberg et al.
these equilibrium constants, the line through the points was calculated with Eq.
(4.53).
4.9.3 Self-Adducts, Type MAz(HA)b

Hydration only occurs in neutral complexes that are coordinatively unsaturated
by the organic ligand. The hydrate water reduces the extractability of the com-
plex. In the absence of strong donor molecules, which can replace this hydrate
water, there is still a chance for the undissociated acid to replace the water,
leading to a self-adduct according to the reaction
MA z (H 2O) w + bHA ←
→ MA z ( HA)b + wH 2 O (4.54a )
This competition between the formation of an adduct with HA or with H2O is
observed as an increased extraction with increasing HA concentration. However,
stoichiometrically the complexes MAZ(HA)b and HbMAZ+b are equivalent. For-
mally the former is a self-adduct and the latter is an ion pair. Thus Eq. (4.54a)
could be written
MA z (H 2O) w + bHA ← −b
→ M(H 2 O) w−2 A z+b + b H + w − 2b H 2 O
+
(4.54b)
assuming the HA can replace 2 H2O through its bidentate structure. All of the
hydrate water can be displaced by the organic ligand, with formation of nega-
tively charged chelate complexes.
Chemical equilibrium experiments, e.g., distribution ratio measurements,
cannot distinguish between these two types of complexes; however, they may
be identified by fingerprinting techniques like NMR, IR, or x-ray structure de-
terminations. Existence of similar adducts like MAzBb support the existence of
self-adducts. The case of promethium(III) acetylacetone is an interesting illus-
tration of this problem.
Example 7: Extraction of Pm(III) by acetylacetone.
Figure 4.14 shows the distribution ratio log DPm for promethium in the
system acetylacetone (HA), benzene/1 M NaClO4 as a function of −log[A−] =
pA at various starting concentrations, [HA]oorg, of HA in the organic phase. The
coordination number of Pm(III) with respect to oxygen is reported to be 8 or
9. In the aqueous phase all stepwise complexes up to PmA−4 can therefore be
expected to be formed. The vacant coordination sites may be filled with water
(forming a hydrate) or undissociated acetylacetone (forming a self-adduct).
The last assumption is suppported by the fact that a large number of adducts
of type LnA3Bb are known.
Omitting water of hydration, the distribution ratio becomes
[PmA 3 ]org + [PmA 3 HA]org + [PmA 3 (HA) 2 ]org + L
DPm = (4.55a )
[Pm] + [PmA] + [PmA 2 ] + L + [PmA 4 ]
which is abbreviated, using the earlier relations, to
Fig. 4.14 Extraction of Pm(III) by acetylacetone (HAa) from 1 M NaClO4 into ben-
zene at three different original concentrations of HAa in the organic phase. pA =
−log[Aa−] is calculated according to Eq. (4.20). The analysis of the system yielded the
constants log β1 5.35, log β2 9.20, log β3 13.22, log β4 14.06, Kad1 7, Kad2 3, and KDC
0.008, shown for Pm in Fig. 4.15. (From Refs. 27a,b.)
K DCβ3 [A]3 (1 + K ad1[HA]org + K ad2 [HA]org

2
+ L)
DPm = (4.55b)
Σβ3 [A]n
where KDC refers to the distribution of uncharged PmA3 between the benzene
and the aqueous phase
K DC = [PmA 3 ]org /[PmA 3 ] (4.56)
Kad,b is the equilibrium constant for the self-adduct formation in the organic
phase, i.e.,
PmA 3 B(org) + b HA(org) ←
→ PmA 3 (HA) b (org) (4.57a )
K ad,b = [PmA 3 (HA) b ]org [PmA ]−1
3 org
−b
[HA]
org
(4.57 b)
For pedagogic reasons we rewrite Eq. (4.55b)
154 Rydberg et al.
K DC (1 + K ad,1[HA]org + K ad,2 [HA]org

2
+ L)
DPm = (4.58)
(β3 [A]3 ) −1 Σβn [A]n
Note that the denominator only refers to species in the aqueous phase.
Thus at constant [HA]org, the curvature of the extraction curve and its position
along the [A−] axes is only caused by varying [A−], i.e., the aqueous phase
complexation. The numerator refers to the organic phase species only, and is
responsible for the position of the extraction curve along the D-axis; thus at
three different constants [HA]org, three curves are expected to be obtained, with
exactly the same curvatures, but higher up along the D-scale with higher [HA]org
concentration; see Fig. 4.14.
The asymptote with the slope of −3 at high pA (i.e., low [A−]), fits Eq.
(4.58), when Pm3+ dominates the denominator (i.e., the aqueous phase), while
the asymptote with slope of +1 fits the same equation when the aqueous phase
is dominated by PmA−4. Between these two limiting slopes, the other three
PmAn complexes are formed in varying concentrations. A detailed analysis of
the curves yielded all equilibrium constants Kex, Kn, Kad,b and KDC (see section
4.14.3), which are plotted in Fig. 4.15. The curves in Fig. 4.14 have been
calculated with these constants. Kn is defined by
βn = Π K n (4.59)
according to Eq. (3.5).
The maximum distribution ratio for the Pm-HA system (DPm about 0.1)
is reached in the pH range 6–7. It is well known that the lanthanides hydrolyze
in this pH region, but it can be shown that the concentration of hydrolyzed
species is <1% of the concentration of PmAn species for the conditions of Fig.
4.14, and can thus be neglected.
Self-adducts are rather common, and have been identified for complexes of Ca,
Sr, Ba, Ni, Co, Zn, Cd, Sc, Ln, and U(VI) with acetylacetone, thenoyltrifluoro-
acetone, tropolone, and oxine. Table 4.14 lists some self-adduct constants. The
fact that the constants vary with the organic solvent indicates that the self-adduct
䉴
Fig. 4.15 The system La(III) acetylacetone (HA) − 1M NaClO4/benzene at 25°C as a
function of lanthanide atomic number Z. (a) The distribution ratio DLn (stars, right axis)
at [A−] = 10−3 and [HA]org = 0.1 M, and extraction constants Kex (crosses, left axis) for
the reaction Ln3+ + 4HA(org) X LnA3HA(org) + 3H+. (b) The formation constants, Kn,
for formation of LnA3n−n lanthanide acetylacetonate complexes (a break at 64Gd is indi-
cated); circles n = 1; crosses n = 2; triangles n = 3; squares n = 4. (c) The self-adduct
formation constants, Kad, for the reaction of LnA3(org) + HA(org) X LnA3HA(org) for
org = benzene. (A second adduct, LnA3(HA)2, also seems to form for the lightest Ln
ions.) (d) The distribution constant KDC for hydrated lanthanum triacetylacetonates, LnA3
(H2O)2−3, between benzene and 1M NaClO4. (From Ref. 28.)
Rydberg_5063-2_Ch05_R3_02-09-04 11:29:22
156 Rydberg et al.
Table 4.14 Constants for the Formation of Some Metal Self-Adducts

According to MAz(org) + b HA(org) X → MAz(HA)b
Metal ion Mz+ Ligand and adduct former HA Organic solvent Log Kad
Zn(II) Isopropyltropolone, IPT CHCl3 1.72

Zn(II) Isopropyltropolone, IPT CCl4 1.9
La(III) Acetylacetone, AA C6H6 2.6
Eu(III) Thenoyltrifluoroacetone, TTA CHCl3 0.56
Eu(III) Thenoyltrifluoroacetone, TTA CCl4 0.5
Eu(III) Isopropyltropolone, IPT CHCl3 2.1
Eu(III) Isopropyltropolone, IPT CCl4 2.0
U(VI) Acetylacetone, AA CHCl3 1.16
U(VI) Thenoyltrifluoroacetone, TTA C6H6 0.50
Source: Ref. 4.
reaction probably occurs in the organic phase. Table 4.14 contains both very
inert (e.g., CCl4), polar (CHCl3), and pi-bonding solvents (C6H6).
4.10 METAL EXTRACTION BY LIQUID ANION

EXCHANGERS (TYPE III-D)
Organic phase with anion RNH+L− (RNH+)p ML−np

exchanger and metal complex ↓↑ ↓↑
Aqueous phase: metal with ← (RNH+)pML−np
M z+ + nL− + pRNH+L− →
complexing anions L−
and amine
Metals that react with inorganic ligands to form negatively charged com-
plexes as described in Chapter 3 can be extracted into organic solvents with
large organic cations in a process referred to as liquid anion exchange.
4.10.1 Two Industrially Important Cases

Extraction of metals from chloride solutions into kerosene containing trioctyl
amine (TOA), as illustrated by Fig. 1.3, has found application in the production
of pure Co and Ni metals from their ores, as discussed in Chapter 11. The
aqueous phase contains the Co and Ni in addition to other metals in chloride
solution, where the chloride concentration goes to 300 g/L (i.e., about 8.5 M).
At this high aqueous ionic strength, a rigorous mathematical treatment of the
system would require the use of activities in order to explain the extraction
diagram; however, here it suffices to use concentrations. An interesting feature

in Fig. 1.3 is the differences between metals in various oxidation state, notably
Fe(II) and Fe(III), the latter being much more easily extracted than the former
because Fe(III) forms much stronger complexes with Cl− ions. The requirement
for extraction of metals by an aminelike TOA, or by its amine salt, TOAH+Cl−,
is the formation of negatively charged metal chloride complexes in the aqueous
phase. The complex formation from Mz+ to MClnz−n, where z − n ≤ −1, proceeds
in the stepwise manner described in Chapter 3. For example, Fe(III) forms the
complex FeCl−4, while cobalt forms CoCl42−. The extraction reaction for Co can
be written
CoCl 24− + 2 RNH +Cl − (org) ←
→ (RNH) 2CoCl 4 (org) + 2 Cl
−
(4.60a )
Amine extraction is used also in another important industrial process, the
extraction of uranium from sulphuric acid leached ores, which uses trilauryl
amine (TLA). In that case, the extraction reaction is
UO 2 (SO 4 ) 24− + 2 (RNH + ) 2SO 42− (org)
←
→ (RNH) 4 UO 2 (SO 4 )3 (org) + SO 42− (4.60b)
These processes are referred to as liquid anion exchange, because the
aqueous anionic metal complex replaces the anions Cl− or SO42− of the large
organic amine complex. The amine complex is an ion pair, or salt, RNH+ L−,
where R is a large hydrocarbon group. The amine salt is highly soluble in the
organic (e.g., kerosene) phase, and almost insoluble in the aqueous phase. The
reaction is presumed to occur either at the interface or in the aqueous phase with
a low concentration of dissolved RNH + L−. Before discussing these reactions in
terms of molecular species, it is necessary to consider more specific aspects of
liquid anion exchangers.
4.10.2 Properties of Liquid Anion Exchangers

As mentioned earlier, hydrogen ions can be solvated by strong donor molecules
like TBP, sometimes leading to the extraction of a solvated hydrogen salt,
HB+bL−; in Example 2, TBP extracts nitric acid, HNO3, in the form of the adduct
complex TBP ⴢ HNO3, which could be considered as the complex HTBP+NO−3
ion pair. The amines form stronger adducts with hydrogen ions; in fact, they are
so strong that they remain protonated while exchanging the anion. A classical
reaction is the formation of the ammonium ion NH+4 when NH3 is dissolved in
water.
The organic amines (the amine base RN) have a nitrogen atom N attached
to a large organic molecule R usually containing >7 aliphatic or aromatic carbon
atoms. They are highly soluble in organic solvents (diluents) and almost insolu-
ble in water. In contact with an aqueous phase containing HL, the amine base
158 Rydberg et al.
RN reacts with the acid HL to form RNH + L−, but extracts with an excess
amount of acid HL (over the 1:1 HL:RN ratio) into the organic solvent, and
also with additional water. The practical concentration of amine in the organic
solvent is usually less than 20%; at higher concentrations the amine salt solu-
tions become rather viscous. The amine salts dissociate in highly polar solvents,
while in more inert diluents they easily polymerize to form micelles, and at
higher concentrations a third phase. For example, in xylene (TLA ⴢ HBr)n aggre-
gates with n = 2, 3, and 30 have been identified, and in other systems aggrega-
tion numbers above 100 have been reported. TLA ⴢ HNO3 is mainly trimeric in
m-xylene at concentrations 0.002–0.2 M, but larger aggregates are formed at
higher amine concentrations. These aggregates seem to behave like monofunc-
tional species, each extracting only one anionic metal complex. The aggregation
can be reduced, and the third phase formation avoided, by using aromatic diluents
and/or by adding a modifier, usually another strong Lewis base (e.g., octanol or
TBP). Such additions often lead to considerable reduction in the Kex value.
Four types of organic amines exist, as shown in Table 4.8: primary amines
RNH +3, secondary R2NH +2, tertiary R3NH +, and quaternary R4N +(Appendix D).
The hydrocarbon chain R is usually of length C8−C12, commonly a straight
aliphatic chain, but branched chains and aromatic parts also occur. In general
the amines extract metal complexes in the order tertiary > secondary > primary.
Only long-chain tertiary and—to a smaller extent—quarternary amines are used
in industrial extraction, because of their suitable physical properties; trioctylam-
ine (TOA, 8 carbons per chain) and trilauryl amine (TLA, 12 carbons per chain)
are the most frequently used. For simplicity we abbreviate all amines by RN,
and their salts by RNH + L−.
The tertiary and quaternary amine bases are viscous liquids at room tem-
perature and infinitely soluble in nonpolar solvents, but only slightly soluble in
water. The solubility of the ion-pair RNH + L− in organic solvents depends on
the chain length and on the counterion, L−: the solubility of TLA ⴢ HCl in wet
benzene is 0.7 M, in cyclohexane 0.08 M, in CHCl3 1.2 M, and in CCl4 0.7 M.
Nitrate and perchlorate salts are less soluble, as are lower molecular weight
amines.
The formation of the ion pair salt can be written
←
RN(org) + H + + L− → RNH + L− (org) K ex,am (4.61)
Table 4.15 gives the equilibrium constants Kex,am (for extraction, amine) for this
reaction with trioctylamine in various solvents. Although the ion pairs are only
slightly soluble in water, they can exchange the anion L− with other anions, X−,
in the aqueous phase. (Note that we use L− to indicate any anion, while X− is
used for (an alternative) inorganic anion.)
RNH + L− (org) + X − ← NH + X − (org) + L− K ex,ch (4.62)
→
Table 4.15 Equilibrium Constants Kex,am for Formation

of Trioctyl Amine Salts [Eq. (4.61)]
Anion Solvent Log Kex,am
F− Toluene 3.0
Cl− Toluene 5.9
Cl− Carbontetrachloride 4.0
Cl− Benzene 4.1
Br− Toluene 8.0
NO−3 Toluene 6.6
NO−3 Carbontetrachloride 5.0
SO24− a Carbontetrachloride 6.7
SO24− a Benzene 8.3
a
The equilibrium constant refers to the formation of (R3NH)2SO4.
Source: Ref. 5.
The equilibrium constant (Kex,ch for extraction, exchange) for this reaction in-
creases in the order ClO−4 < NO−3 < Cl − < HSO−4 < F −. From Chapter 3 it follows
that the formation constants for the metal complex MLnz−n usually increases in
the same order. Therefore, in order to extract a metal perchlorate complex, very
high ClO−4 concentrations are required; the perchlorate complex is easily re-
placed by anions higher in the sequence.
All negatively charged metal complexes can be extracted by liquid anion
exchangers, independent of the nature of the metal and the complexing ligand.
Liquid anion exchange has extensive industrial application, and many examples
are given in later chapters. Lists of extraction values are found in other works
[e.g., 5, 21].
4.10.3 Equations for Liquid Anion

Exchange Extractions
The discussion on the properties of the amine salts serves to explain why, in
metal amine extractions, it is difficult to obtain simple mathematical relations
that agree well with the experimental data. While there is no difficulty, in princi-
ple, in obtaining reasonably good values for the formation of the negatively
charged metal complexes in the aqueous phase, there is a major problem in
defining the organic phase species, which may consist of free amine [RN], mono-
meric [RNHL], and polymeric [RNHL]n amine salt, and several extracted metal
complexes [(RNH+)p(MLnp−)]. A contributing difficulty in practice is the need to
use high ligand concentrations, [L−], in the aqueous phase in order to obtain the
negatively charged complexes (see Fig. 1.3).
160 Rydberg et al.
In general the metal Mz+ reactions with a monobasic anion L− can be

written
←
M z+ + n L− → MLzn−n βn = [MLzn−n ]/[M z+ ] [L− ]n (3.5, 4.34)
When z − n = p is negative, a negatively charged metal complex has been
formed, which can be extracted
ML−np + p RNH + L− (org) ← −p
→ (RNH ) p ML n (org) + p L
+ p+ −
(4.63a )
We define the extraction constant by
−p − p
[(RNH) +p
p ML n ]org [L ]
K ex = (4.63b)
[ML−np ] [RNHL]org
p
A priori, it must be assumed that the aqueous phase contains all the stepwise
complexes MLnz−n. Thus the distribution ratio is
−p
[(RNH) +p
p ML n ]org βp [L− ]p [RNHL]org
p
DM = = K ex (4.64)
Σ [MLzn−n ] 1 + Σβp [L− ]p
The distribution of M depends on both the free amine salt in the organic phase
and the concentration of free L− in the aqueous phase until all metal in the
aqueous phase is bound in the ML−np complex. At constant amine concentration,
Eq. (4.64) indicates that a plot of DM vs. [L−] would have a linear slope p if the
denominator of Eq. (4.64) is Ⰶ1; i.e., the metal species in the aqueous phase
are dominated by the uncomplexed metal ion Mz+. At higher [L−] concentrations,
where the ML−np complex begins to dominate in the aqueous phase, the DM value
p
becomes equal to Kex [RNHL]org . Equations (4.64) and (4.4) show that S-shaped
curves result for metals with large Kex values. In a plot of DM vs. [RNHL]org a
straight line of slope p is obtained only at constant [L−]. From such measure-
ments both p, Kex and βp can be evaluated. The following example illustrates
this.
Example 8: Extraction of Trivalent Actinides by TLA.
In an investigation of the extraction of trivalent actinides, An(III) from
0.01 M nitric acid solutions of various LiNO3 concentrations into o-xylene
containing the tertiary amine salt trilaurylmethylammonium nitrate, TLMA
HNO3, Van Ooyen [29] found that the amine was monomeric only at very low
concentrations (≤0.1M in the organic phase) but at higher concentration formed
both dimers and trimers.
Using trace concentrations of Ce(III) and An(III) a log-log plot of DM
against the nitrate ion activity, mγ± = [LiNO3]1/2, had a slope of approximately
3, Fig. 4.16b. From Eq. (4.64) this slope corresponds to the p-value of 3 when
the aqueous phase is dominated by the free metal ion, which is not an unrea-
sonable assumption at low nitrate concentrations.
Fig. 4.16 Distribution ratio of M3+ ions between the trilaurylmethyl ammonium nitrate
(TLMA) in o-xylene and aqueous phases of varying LiNO3 concentrations. (a) As a
function of TLMANO3 concentration at 1–7 M, 2–5 M, 3–3 M LiNO3. (b) Extraction
of Eu(III) and tree actinide(III) ions at 0.1 M TLMANO3 in o-xylene and varying aque-
ous salt concentrations. (From Ref. 29.)
162 Rydberg et al.
In plots of log DAm against [TLMA ⴢ NO3]org at different nitrate concen-

trations, the curves in Fig. 4.16a had straight slopes of 1.5–1.8 at low concen-
trations of TLMA NO3, but bending at higher concentrations, was explained
by the formation of polymeric amine species. If Eq. (4.64) is valid, these slopes
correspond to the number of TLMA HNO3 groups attached to the extracted
Am species. Thus Van Ooyen described his complex as {(TLMA NO3)n}2
Am(NO3)3, for which n = 1 only at the very low amine concentrations. Thus
for n = 1 the complex could as well be written (TLMA)2Am(NO3)5. The Ce
and other An complexes would have similar configurations.
4.11 OTHER EXTRACTABLE METAL COMPLEXES

(TYPE III-E)
Ion pair (and possibly “counter species”) (C +X − and Y +A−) C +A−

in organic phase ↓↑
Aqueous cation C + and anion A− associated ← C +A −
C + + A− →
into ion pair C +A−
There are a few types of complexes that do not fit well into the previous
classifications: monovalent metals that form extractable complexes with large
organic monobasic anions, and, conversely, monovalent inorganic anions that
form extractable complexes with large organic cations. Though the amine-type
liquid anion exchangers could fit into the latter group, it is simpler to treat them
as a separate homogenous group (class III-D). The large monovalent ion pair,
tetraphenylarsonium tetraphenylborate (Ph4As+ Ph4B−), which has been suggested
as a reference in solvent extraction (see Chapter 2), also belongs to this class.
Of some importrance is the extraction of alkali ions by tetraphenylborate
and by crown ethers (Appendix D: 21), of fluoride by tetraphenylantimonium
ions, and of perrhenate by tetraphenylarsonium ions. Because most of the vol-
ume of these complexes is taken up by organic groups, the complexes are highly
lipophilic and, therefore, extractable into organic solvents without additional
solvation. Though these systems have limited applications, the crown ethers offer
some interesting extraction systems. The ethers form cage compounds (“clath-
rates”); i.e., the metal cation fits into a cage formed by the crown structure. Be-
cause the cage can be designed to fit almost any ion of a certain size, rather
selective extractions are possible with this system, as described in Chapter 15.
Example 9: Extraction of K+ by tetraphenylborate.
Consider the extraction of K+ from aqueous solution into the organic
solvent nitrobenzene by addition of NaPh4B. Abbreviating Ph4B− by A−, the
extracted complex is the ion pair K+A−. In an inert solvent this is a stable ion
pair, but in a highly polar solvent like nitrobenzene, the ion pair may dissociate.
The organic phase may thus contain both solvated K+ and K+A− species, while
the only potassium species in the aqueous phase is free K+.

The extraction equilibria may be written
K+ + A− ← + −
→ K A (org) Equil. const. K ex (4.65)
while for the reaction in the organic phase

K + A − (org) ←
→ K (org) + A (org) Equil. const. K ass
+ −
(4.66)
The distribution ratio of K+ becomes

DK = ([K + A − ]org + [K + ]org ) [K + ]−1 = K ex [A − ](1 + K ass /[A − ]org ) [K]−1 (4.67)
When A− is added into the aqueous phase as a Na+ salt in large excess to K+,
the dissociation in the organic phase becomes negligible and Eq. (4.67) is re-
duced to
DK = K ex [A − ] (4.68)
Figure 4.17 shows the distribution ratio of K+ when a large excess of
Na+A− is added to the system. Although the extracted complex should be com-
Fig. 4.17 Distribution ratio of potassium(I) between nitrobenzene and water as a function
of initial aqueous tetraphenylborate concentration in 0.1M NaClO4. (From Refs. 30a,b.)
164 Rydberg et al.
pletely dissociated in the nitrobenzene, the ionic concentration of the organic

phase is large enough to suppress dissociation, and the distribution ratio thus
becomes proportional to the concentration of the extractant in the aqueous phase.
4.12 STUDIES OF HYDROPHILIC COMPLEXATION

BY SOLVENT EXTRACTION (TYPE IV)
Coordinatively saturated metal MAz

complex in organic phase ↓↑
Formation of extractable and ← MAz + MX nz−n
M z+ + zA− + nX − →
nonextractable complexes
Solvent extraction has become a common technique for the determination of
formation constants, βn, of aqueous hydrophilic metal complexes of type MXn,
particularly in the case when the metal is only available in trace concentrations,
as the distribution can easily be measured with radioactive techniques (see also
section 4.15). The method requires the formation of an extractable complex of
the metal ion, which, in the simplest and most commonly used case, is an un-
charged lipophilic complex of type MAz. The metal-organic complex MAz
serves as a probe for the concentration of metal Mz+ ions in the aqueous phase
through its equilibrium with the free Mz+, section 4.8.2. This same principle is
used in the design of metal selective electrodes (see Chapter 15). Extractants
typically used for this purpose are β-diketones like acetylacetone (HAA) or
thenoyltrifluoroacteone (TTA), and weak large organic acids like dinonyl naph-
talene sulphonic acid (DNNA).
The pertinent distribution equation is
[MA z ]org
DM = (4.69)
[M]{1 + Σβn [A]n + Σβp [X]p }
(summed from 1 to n or p) where the formation constants for MAn are βn and
those for the MXp complexes βp and summations are taken from 1. In the ab-
sence of MXp complexes, the distribution ratio of M is defined by DMo:
DMo = [MA z ]org /[M] (1 + Σβn [A]n ) (4.70)
If the conditions are chosen so that the aqueous complexes of HA can be ne-
glected [see Eq. (4.41) and Example 4], DMo of Eq. (4.70) equals DM of Eq.
(4.41c); i.e.,
DMo = DM [of Eq. (4.41c)] = K ex [HA]org
z
[H + ]− z (4.41c, 4.71)
for which linear correlations are obtained in plots of log DM against log [HA]org
at constant pH, or against pH at constant [HA]org. Dividing Eq. (4.69) by Eq.
(4.70) gives
D o /D = 1 + β1[X] + β2 [X]2 + β3[X]3 + L (4.72)

Thus a plot of D /D against [X] provides a polynomial from which the βps can
o
be calculated (see section 4.14).

Example 10: Determination of formation constants for Be(II) oxalate com-
plexes.
Figure 4.18 shows the extraction of beryllium with HTTA into hexone
as a function of the concentration of oxalate ions, Ox2−, in the aqueous phase.
Here the conditions have been chosen so that the aqueous complexes of TTA
can be neglected as has been described earlier, and thus Eq. (4.71) is valid,
and the relation DMo = Kex [HA]2org [H]−2 so that
log D / D o = log D K ex−1 [HA]org

−2
[H]2
= − log (1 + 103.55 [Ox 2− ] + 105.40 [Ox 2− ]) 2 (4.73)
The parameter values on the right-hand side of the equation, are the βn values,
and were determined by a curve fitting technique (section 4.14). Introducing
these values in Eq. (4.71) gives the solid line through the points in Fig. 4.18.
The advantage of this “strange” ordinate is that the curvature only reflects the
oxalate complexation.
Hydrolysis of some metals begins at very low pH (e.g., Zr at pH ⬃ 0.4, Th(IV)
and U(VI) at pH 1–2, the lanthanides at pH 3–5, etc.). However, if complex
Fig. 4.18 Decrease in distribution ratio of Be(II) as a function of oxalate ion (Ox2−)
concentration due to formation of aqueous BeOx2n−2n complexes. The extraction system is
0.03 M TTA in methylisobutylketone and 1.0 M Na(0.5 Ox2−, ClO−4 ). See Eq. (4.72) for
ordinate function. (From Ref. 31.)
166 Rydberg et al.
formation strongly dominates over hydrolysis, the latter can be neglected. In the
overview in section 4.3, it was shown in Fig. 4.3 that the general shape of the
extraction curves indicates the type of complexes formed. Thus, the extraction
of gallium(III) with acetylacetone, Fig. 4.19, indicates a behavior according to
Fig. 4.3c. in the plot of log D vs. free ligand concentration, [A−].
Example 11: Formation of hydrolyzed Ga(III) acetylacetonates.
Figure 4.19 shows the extraction of Ga(III) by acetylacetone into ben-
zene at various concentrations of total acetylacetone, [HA]t, and constant ionic
strength, using the AKUFVE technique (see section 4.15.3). By comparing
with Fig. 4.3, one may guess that hydroxy complexes are formed (diagram c).
If the general complex is designated GamAn(OH)p, a first investigation showed
Fig. 4.19 The distribution of Ga(III) between benzene and 1 M NaClO4 as a function
of pA = −log[A−] at 6 different acetylacetone (HA) concentrations (0.06–0.0006 M).The
different curves are due to different extent of hydrolysis in the aqueous phase. The con-
centration of HA decreases from the upper left corner, where [HA]aq is 0.06 M (▲),
toward the right lower corner, where it is 0.0006 M (䉫). (From Ref. 32.)
that no polynuclear complexes were formed, i.e., m = 1, as the extraction

curves were independent of the Ga(III) concentration. This was also expected
as the total concentration of Ga(III) ranged from 10−4 to 10−6 M. Next, using a
computer program, the formation constants for the following complexes were
determined: GaA2+, GaA+2, GaA3, Ga(OH)3 and Ga(OH)−4 [32]. It was concluded
that no mixed complexes with both A− and OH− were formed. The organic
phase contained only GaA3 (which was hydrated).
In using this technique accurately, it is necessary to consider possible
sources of errors. A rather extensive summary of this technique applied to hy-
drophilic actinide complexation by practically all inorganic ligands, as well as
a considerable number of weak organic acids, has recently been presented
[33a,33b].
4.13 EFFECTS AND CORRELATIONS

Because of the many parameters involved in solvent extraction, chemical as
well as physical, it is a difficult task to draw reliable conclusions about the
reactions that are responsible for the observed distribution values. In the intro-
ductory part of this chapter, section 4.2, the extraction reaction described by Kex
was shown to be a product of a number of parameters related to the different
steps involved in the formation of the extracted complex. Sections 4.4–4.11
have described these various subreactions, which are described by Eq. (4.46):
−z
K ex = K az βz K DR K DC K ad,bB (4.46, 4.74)
Examples of how Kex data may be interpreted in more fundamental terms follow.
In the lack of information on the independent parameters of Eq. (4.74), tenable
conclusions can be drawn about their size and effects from known correlations
between the parameters. This section explores the effect of such correlations.
4.13.1 Effects of the Dissociation Constant

of Acidic Ligands and Formation Constants
of the Corresponding Metal Complexes
4.13.1.1 The Ligand (Acid) Dissociation Constant Ka
According to Eq. (4.74) Kex increases to the z power of the dissociation constant,
Ka, for the extractant acid HA. Thus it appears desirable to use stronger acids
as extractants, i.e., those with large Ka (or small pKa). Thus factors that affect
the dissociaton constant of an acid are of importance for extraction behavior.
The acidity of organic acids increases (i.e., pKa decreases) by electrophilic
substitution in the carbon chain. For example, replacing an H atom in the CH3
group of acetic acid (pKa 4.75) by I, Br, Cl, and F atoms yields monohalide
acetic acids with pKa of 3.0, 2.9, 2.8, and 2.7, respectively. The mono-, di-,
and trichloroactic acids have pKa 2.8, 1.48, and about 1.1, respectively. Other
168 Rydberg et al.
electrophilic groups which can increase acidity are HO−, NC−, NO2−, etc.
Conversely, nucleophilic substitution by aliphatic or aromatic groups usually
has little effect on the pKa, though it may affect the distribution constant KDR
(see section 4.13.2): e.g., the addition of a CH3− or C6H5− group in acetic acid
changes the pKa to 4.9 (propionic acid) and 4.3 (phenyl acetic acid). The further
the substitution is from the carboxylic group, the less the effect: e.g., while pKa
for benzoic acid is 4.19, a Cl in orthoposition yields pKa 2.92 and in the parapos-
ition, 3.98.
These general rules hold rather well for the acidic organic extractants and
can be used to extrapolate from related compounds to new ones as well as to
develop new extractants. The effect of various substituends on pKa is extensively
discussed in textbooks on organic chemistry [e.g., 14, 34].
4.13.1.2 The Complex Formation (or Stability)
Constant βn
Kex also increases with increasing formation constant of the uncharged metal
complex, βz. Thermodynamic factors and geometrical aspects that influence the
size of βn are discussed in Chapter 3. Some further observations follow.
To achieve the highest possible D value, the concentration of the extracted
uncharged complex (MAz, MAzBb or BbMLz+b) must be maximized. A large
value for the formation constant of the neutral complex favors this goal. In
Chapter 3 it was pointed out that for hard acids the complex formation constant
increases with the charge density of the metal ion, provided there is no steric
hindrance. This is seen for the metal fluoride complexes in Fig. 3.4 (no steric
hindrance). For the lanthanide acetylacetonates in Fig. 4.15b, the increase in βn
from 57La to 64Gd is due to a reduction in the ionic radius and, accordingly, to
the increased charge density. It should be noted that the diminishing lanthanide
size leads to a successive diminution of the coordination number (from 9 to 8
for H2O) for La → Gd; for Gd → Lu the coordination number is probably con-
stant, as is inferred from Fig. 4.15c; see also [35].
4.13.1.3 Correlations Between Ka and βn
Because the H+ ion acts as a hard metal ion, one should expect a close correla-
tion between the formation constants for HA and for MAz. Since Ka is defined
as the dissociation constant of HA, while βz is the formation constant of MAz,
a correlation is therefore expected between pKa (= −logKa) and logβz. The corre-
lations in Figs. 3.5 and 3.6 show that large logβn values are usually observed
for organic acids with large logKa. Since Kex is directly proportional to both βz
and K za [see Eq. (4.74)], and a large βz value is likely to be accompanied by a
small Ka value (large pKa); in Eq. (4.74) the two parameters counteract each
other. Thus in order to obtain high free ligand concentrations, which favors the
formation of the MAz complex, [H+] must be low, i.e., pH must be high. How-
ever, at high pH the metal may be hydrolyzed. A careful balance must be struck
between βz, Ka, and pH to achieve an optimum maximum concentration of the
uncharged complex.
4.13.2 Effects of Molecular Volume and Water

Structure Upon KD
Chapters 2 and 3 factors are discussed which affect the distribution of uncharged
species between the organic and aqueous phases, the most important being (1)
the interaction of the species (complex or extractant) with the solvents (solva-
tion); and (2) energy released or required when an uncharged species leaves its
original phase (cavity closure) and that required when it enters the new phase
(cavity formation). Sections 4.4 and 4.5 discuss the distribution of inert species,
including undissociated acid HA, KDR. From Eq. (4.74) it is seen that Kex is
negatively influenced by a large distribution constant for the extractant reagent,
but favored by a large KDC for the extraction of the metal complex.
4.13.2.1 Metal Complexes, KDC
There are rather limited data on systematics for the distribution constants KDC
of metal complexes. Table 4.10 gives distribution constants for Zn(II), Cu(II),
and Np(IV) acetylacetonates in various solvents. The zinc complex has an octa-
hedral coordination: four oxygen atoms from the ligands and two oxygen atoms
from water molecules can occupy six positions in an almost spherical geometry
around the zinc atom. The copper complex has a square planar configuration
with four oxygen atoms from the acetylacetones in the plane; a fifth (and possi-
bly sixth) position perpendicular to the plane and at a longer distance from the
copper atom can be occupied by an oxygen from a water molecule. The zinc
complex has a larger degree of hydration in the aqueous phase and consequently
a lower distribution constant. The NpA4 structure is likely a square antiprism
with no hydrate water. It has a larger molecular volume than the two other
complexes, and its larger distribution constant is in agreement with this, reflect-
ing the release of more water structuring energy when leaving the aqueous
phase.
Two further observations can be made from Table 4.10: (1) the distribu-
tion constant seems to increase with increasing polarity of the solvent; (2) when
the reagent acetylacetone has a large KDR, the metal complexes also show a large
KDC value.
The lanthanide ions, Ln3+, are known to contract with increasing atomic
number (Z), from La3+ with a hydrated radius of 103 pm to Lu3+ of 86 pm
(lanthanide contraction). Thus one expects that the neutral LnA3 complex be-
comes smaller with increasing atomic number, and consequently that β3 should
increase and KDC decrease with increasing Z. Figure 4.15d shows that measured
170 Rydberg et al.
KDC values for the lanthanide acetylacetonates are the reverse from that expected
for the size effect. The explanation is likely that the neutral complex with the
formula LnA3 is coordinatively unsaturated, which means that a hydrated com-
plex exists in the aqueous phase (possibly also in the organic phase). The more
coordinatively unsaturated lanthanide complexes (of the larger ions) can accom-
modate more water and thus are more hydrophilic. The result is a KDC several
orders of magnitude lower for the lightest, La, than for the heaviest, Lu.
The tetravalent metal acetylacetonates [36] have all coordination numbers
(CN) 8 or 9 in the neutral complexes. For Th (radius 109 pm at CN 9) there is
often one molecule of water in ThA4, whereas for the corresponding complexes
of U(IV) and Np(IV) (radius 100 and 98 pm, respectively, CN = 8), there seems
to be none. The measured distribution constants (log KDC) are for Th 2.55 and
for U and Np 3.52 and 3.45, respectively, in 1 M NaClO4/benzene. This agrees
with the greater hydrophilic nature of the ThA4 ⴢ H2O complex.
4.13.2.2 Correlations Between KDR and KDC
Qualitatively, the distribution constants of the complexing reagent and the corre-
sponding neutral metal complex are related due to their similar outer surface
toward the solvents. Thus it is reasonable that there should be good correlation
between KDR and KDC for homologous series of reagents or complexes. This is
indeed the case, as is seen when KD,R is plotted vs. KD,C, in Fig. 4.20 for a
complex with a variation of the solvent, and in Fig. 4.21 for complexes of
related reagents with the same solvent. This correlation can be related to the
solubility parameter concept discussed later.
4.13.3 The Solubility Parameter

In Chapter 2 we learned that the cohesive forces that keep molecules together
in a solution can be described by
δ 2 = (∆HV − RT ) / V (2.1; 4.75a )
where δ2 is the cohesive energy density, and δ, the solubility parameter, V, the
molar volume and ∆HV, the molar heat of vaporization. This holds for regular
solutions [37] that show ideal entropy effects in mixing solute and solvent, and
no interactions occur besides the cohesive forces between the solute and solvent
molecules. [Note: Regular solutions exhibit heat changes when mixed, while for
ideal solutions, the heat of mixing is zero (see section 2.4.1).] There is no
change of state in association or in orientation. Thus the work required to pro-
duce a cavity, ∆Gcav, by the solute in the solvent is given by
∆Gcav = Acav VB δ 2 (2.10; 4.75b)
where Acav is a proportionality factor and VB is the molar volume of the solute B.
Fig. 4.20 Distribution constants KDC for uncharged metal complexes MAz, vs. distribu-
tion constants KDR for the corresponding undissociated reagent, the acid ligand HA, for
various organic solvents and 1 M NaClO4: open circles Cu(II), solid circles Zn(II). Num-
bers refer to solvents listed in Table 4.10. (From Ref. 36.)
Pioneering work on the application of this theory for correlating and pre-
dicting distribution ratios was done in the 1960s [38a–40c]. Several reviews on
the use of this theory for two-phase distribution processes are also available
[41,42]. Recently this theory has been refined by the use of Hansen solubility
parameters [6,43,44], according to which
δ 2total = δdispersion
2
+ δ polar
2
+ δ hydrogen
2
(4.75c)
where interactions between the solute and the solvent are described by contribu-
tions from the various types of cohesive forces. In general, the dispersion (or
London) forces dominate. In Fig. 4.22 the measured distribution ratios of an
americium complex between an aqueous nitrate solution and various organic
solvent combinations are compared with distribution ratios calculated from tabu-
lated Hansen parameter values [45].
Few solubility parameters are available for the metal-organic complexes
discussed in this chapter. Another approach is then necessary. The distribution
constant for the reagent (extractant), R, can be expressed as:
172 Rydberg et al.
Fig. 4.21 Distribution constants, KDC, for uncharged metal complexes MAz vs. distribu-
tion constants KDR for the corresponding undissociated acid ligand HA; solid circles
Zn(II), solid triangles Co(III). Variation with ligand composition: HFA hexafluoroace-
tylacetone, TFA trifluoroacetylacetone, AA acetylacetone, FTA 2-furoyltrifluoroacetone,
TTA 2-thenoyltrifluoroacetone, PTA pivaloyltrifluoroacetone, BFA benzoyltrifluoroace-
tone, BZA benzoylacetone. (From Ref. 36.)
Fig. 4.22 Comparison of measured and calculated distribution ratios DAm of americi-
um(III)-terpyridine-decanoic acid complexes between 0.05 M HNO3 and various organic
solvent combinations. The calculated values are obtained with the Hansen partial solubil-
ity parameters. (From Ref. 45.)
RT ln K DR = VR [(δaq − δ R ) 2 − (δorg − δ R ) 2 ] + RT VR (1 / Vorg − 1 / Vaq ) (4..76)

where KDR is given in terms of mole fractions. The solubility parameter of water
has to be empirically estimated. The thermodynamic value is 48, but, since the
system may be sensitive to the choice of δaq, it is better to select a “reference δaq”
in each set of experiments, assuming constant values for the aqueous systems in
the two different two-phase systems. Molar volumes and solubility parameter
values are listed in Table 2.1 and Refs. [6,44].
The overall enthalpy and entropy changes for the distribution reaction
(i.e., transfer of the metal complex from the aqueous to the organic phase) can
be obtained from the temperature dependence of KDR according to
R T ln K DR = −∆H + T ∆S (4.77)
An enthalpy term corresponding to −VR[(δaq − δR)2 − (δorg − δaq)2] and an entropy
term corresponding to R VR(1/Vorg − 1/Vaq) can be obtained formally from Eq.
(4.76). If the distribution constant for a standard system is used as a reference,
changes in enthalpy and entropy relative to this standard system can be assessed
by replacing δaq and Vaq with the corresponding data for the organic diluent in
the reference system, δref ref
org and V org.
174 Rydberg et al.
Thus from solubility parameters, which are specific for the various solutes
and solvents, and molar volumes, values for KDR can be estimated, or deviations
from regularity can be assessed. These deviations can be estimated quantita-
tively and, in individual systems, can be ascribed to specific reactions in either
of the phases, e.g., hydration, solvation, adduct formation, etc.
From Eq. (4.76) the relation
log K DC = (VC / VR ) log K DR + const. (4.78)
can be derived; the subscript C refers to the neutral metal complex. The molar
volume ratio is close to or smaller than z, where z is the number of singly
charged anionic species attached to the central metal ion Mz+. From Fig. 4.23 a
ratio of ca. 1.5 (for z = 2) is obtained. This equation is useful for estimating KDR
values in extraction systems where the corresponding KDC value is known.
Rearrangement of Eq. (4.75) shows that log KDR/(δaq − δorg) vs. δ, where
δ = δorg RT (δaq − δorg ) −1 (Vorg
−1
− Vaq−1 ) (4.79)
should yield a straight line with slope VR / RT ln10. A plot of this relation in
Fig. 4.23 demonstrates a satisfactory correlation between distribution constants
and solvent parameters, indicating the usefulness of the solubility parameter
concept in predicting KDR as well as KDC values.
Fig. 4.23 Application of the regular solution theory for correlation of distribution con-
stants for ZnA2 and CuA2 with solvent properties (solubility parameters); the numbers
refer to the solvents listed in Table 4.10. (From Ref. 22.)
4.13.4 Effects of the Solvent (Diluent) Composition

The composition of the organic phase solvent (diluent) influences the distribu-
tion of both the neutral metal organic complex MAz and the complexing reagent
HA, through similar interactions ranging from that of cavity formation for very
inert diluents like hexane, through dipole-dipole interactions, pi electron interac-
tion, and hydrogen bonding for the more reactive solvents. The expected degree
of interaction can be judged from values of permittivity and solubility parame-
ters. Previous tables have illustrated some of these aspects: Table 4.1 on the
distribution of inert solutes (hole formation), Table 4.7 on distribution of organic
acids (polarity), and especially Table 4.10 on distribution of metal acetylaceto-
nates (solvent permittivity), discussed in section 4.13.2.1. These cases involve
the aqueous phase, so some phenomena may be attributed to the interaction with
water molecules.
Figure 4.24 shows a decrease in KD for acetylacetone with increasing num-
ber of methyl groups in the substituted benzene solvent. Since acetylacetone is
more soluble in aromatics than in aliphatics, as illustrated by the KDR sequence:
Fig. 4.24 Distribution constants KDR of acetylacetone between methyl-substituted ben-

zenes at different aqueous ionic strengths: (a) 0.001 M, (b) 0.1 M, and (c) 1.0 M NaClO4
at 25°C. The number of methyl groups, n, changes the aromatic character at n = 0 to
more aliphatic at n = 3. (From Ref. 46.)
176 Rydberg et al.
hexane 0.95, cyclohexane 1.03, ethylbenzene 3.31, and benzene 5.93 (aqueous
phase 1 M NaClO4), the decrease observed in Fig. 4.24 may simply be an effect
of the increasing aliphatic character of the solvents.
4.13.5 Effects of the Aqueous Medium

The effect of the water activity of the aqueous phase, discussed in Chapters 2,
3, and 6, is determined by the total concentration and nature of the salts. Gener-
ally, the distribution constant for a neutral metal complex would increase with
increasing ionic strength, as illustrated in Fig. 4.25. This salting-out effect is
often ascribed to a reduction in free water available for hydration. On the other
hand, the salt also breaks down the water structure, which could reduce the
energy to form a hole in the phase.
4.13.6 Effects of the Temperature

The introduction of a neutral complex into a solution phase involves a number
of processes that can be associated with large changes in enthalpy (solvation
processes) and in entropy (solvent orientation and restructuring), leading to con-
siderable temperature effects; see Chapter 2. In general, the distribution of a
neutral metal complex increases with increasing temperature for complexes with
significant hydrophobic character. The metal acetylacetonate systems in Fig.
4.26 illustrate this effect. In these particular systems, the free energy of distribu-
tion between the organic and aqueous phases is dominated by the enthalpy term.
Fig. 4.25 The distribution constant KDC for zinc diacetylacetonate, ZnA2, between ben-
zene and various concentrations of NaClO4 in the aqueous phase; 25°C. (From Ref. 36.)
Fig. 4.26 Distribution constant KDC for metal diacetylacetonates, MA2, between or-
ganic solvents and 1 M NaClO4 as a function of temperature: Cu(II) (open circles) and
Zn(II) (solid circles) with various solvents (numbers refer to the solvents given in Table
4.10). (From Ref. 36.)
From measurements of the temperature dependency of the equilibrium

constant, thermodynamic parameters may be deduced (section 3.4). Very few
enthalpy and entropy constants have been derived for the distribution reaction
MAz(aq) →← MAz(org) of neutral complexes; such investigations give informa-
tion about hydration and organic phase solvation.
4.13.7 Further Structural Considerations

In Chapters 2 and 3 several physicochemical factors of importance to solvent
extraction have been described, and many of their effects have been illustrated
in this chapter. Here, we summarize some observed regularities. The effect of
various structures on the bond strengths in metal organic complexes has been
extensively treated in other publications [47–49].
4.13.7.1 Chelate Ring Size
Formation of metal-organic chelate complexes results in stronger complexation
(i.e., larger βn values) compared to interaction with monodentate ligands (Chap-
ter 3). The common types of bidentate ligands are presented in Table 4.9; the
chemistry of these complexes has been extensively discussed in the literature
[14,47]. Chapter 3 presents the most important factors in the formation of such
complexes: (1) the type of binding atom; (2) the chelate ring size (or “bite”);
178 Rydberg et al.
and (3) the number of chelate rings formed (mono- or polydentate). The follow-
ing diagram illustrates the general structure of UO22+ and Am3+ diamide chelate
complexes: two oxygen binding atoms; ring size is 5 + n, 2–3 chelate rings per
complex; the −O−M−O− part is commonly referred to as the “claw.”
| |
N Cn H 2n N
/ \ / \ / \
CH CH
O O
\ /
M
The extractants TBOA, TBMA, and TBSA are very similar, but their structural
differences (see formulas in Table 4.16) allow the formation of only one type
of metal chelate complex: 5-, 6-, and 7-membered rings, respectively. Similar-
ily, the reactants DMDOMA and DMDOSA form only 6- and 7-membered
chelates. Table 4.16 shows that extraction (i.e., largest Kex value) is favored by
6-membered rings. This is not unexpected as the Kex values in this case reflect
the stability constants βn acc. to Eq. (4.74).
The electron shells of the M3+ elements with unfilled or partially filled 3d,
4d, and 5f orbitals contract as these shells are being filled with electrons, in-
creasing the charge density (z/r) of the cation, leading to increasing stability
constants for the MA3 complexes with increasing atomic number. For bidentate
ligands the f-electron (lanthanide) MA3 complexes are coordinatively unsatu-
rated, i.e., only six of eight available coordination sites are filled. Therefore, the
two remaining sites are occupied by H2O or HA or some other donor molecules
B, leading to the self-adduct MA3HA−1 or adduct MA3B1–2. It has been shown
that the adduct formation constants for the reaction MA3 + B → MA3B decrease
Table 4.16 Consultants for Extraction of Actinide Amides

from Nitrate Solutions into an Organic Solvent Showing the Effect
of Chelate Ring Size
Ring
Metal ion Extractanta Kex size Ref.
UO22+ TBOA 0.005 5 [50]

UO22+ TBMA 10.6 6 [50]
UO22+ TBSA 9.3 7 [50,51]
Am3+ DMDOMA 0.47 6 [52]
Am3+ DMDOSA <0.0002 7 [52]
a
TBOA (C4H9)2NCO2; TBMA [(C4H9)2NCO]2CH2; TBSA [(C4H9)2NCO]2(CH2)2;
DMDOMA (CH3C8H17NCO)2CH2; DMDOSA (CH3C8H17NCO)2(CH2)2.
in the same order [27a–28], see Fig. 4.15; i.e., the stronger the complex, the
weaker the adduct.
The increase, by adduct formation, in the coordination number of the cen-
tral atom is possible not only for weak but also for strong chelates, provided
their ligand bites are relatively short. To make room in the inner sphere of the
metal ion for another adduct-forming ligand, three chelating ligands of small
bite angle can easily be shifted away without significant energy-consuming dis-
tortion of the chelate rings [53a,b]. The electronic structure of the central atom
is also of key importance for synergism, as illustrated by the easy adduct forma-
tion of metal ions with unfilled or partially filled d or f orbitals, contrary to,
e.g., p-block elements. Inner- and outer-sphere complexation is further discussed
in the next two paragraphs and in Chapter 16.
4.13.7.2 Donor Ligand Effects
Table 4.17 shows extraction constants for some metal ions with three alkyl
phosphates substituted by 0, 1, and 2 sulfur atoms, but with almost identical
aliphatic branchings. Section 3.3 discusses hard and soft acids and bases (HSAB
theory). According to this theory, hard acids form strong complexes with hard
bases, while weak acids form strong complexes with weak bases. In Table 4.17,
the metals are ordered in increasing hardness, the sub II.b group being rather
soft (Table 3.2). Presumably the Kex values reflect this pattern, as they are pro-
portional to K az and βz. In Table 4.17, the acidity of the acids increases (i.e., Ka
increases) in order R ⴢ PSSH < R ⴢ POSH < R ⴢ POOH (consult Tables 4.8 and
4.9), as sulfur is less basic than oxygen. In general, the Kex increases for for the
dialkyl phosporic acid (hard donor ligand) with increasing metal charge density
within each group as predicted in Chapter 3. For the soft metals, Kex also in-
creases with increasing softness of the ligand, while the opposite effect is seen
for the hard Ln-metal ions. The divalent subgroup II.b metals prefer to bind to
sulfur rather than to oxygen because they have a rather soft acceptor character,
while the hard metals III.b prefer to bind to the hard O-atom of the ligand.
Table 4.17 Comparison of the Extraction Constants Kex for the IIb-Subgroup Divalent
Ions, and IIIb Lanthanide Ions, with Sulfur or Oxygen Dialkyl Phosphoric Acidsa
Metal ion Hg2+ Cd2+ Zn2+ 57 La3+ 63 Eu3+ 71 Lu3+
Charge density Z2/r 3.6 4.1 5.4 7.83 8.74 9.68

log Kex for RⴢPOOH −2.20 −1.80 −1.20 −2.52 −0.44 2.9
log Kex for (C4H9O)2 ⴢ POSH 5.40 3.70 0.70 −4.78 −4.23 0.34
log Kex for (C4H9O)2 ⴢ PSSH 4.40 3.49 2.40 −8.28
a
Values estimated from Refs. 5, 20, 54, 55, and 56. R is (C4H9CH(C2H5)CH2O)2.
180 Rydberg et al.
4.13.7.3 Inner and Outer Sphere Coordination

The extractabilities of metal-organic complexes depend on whether inner or
outer sphere complexes are formed. Case I, section 4.2.1, the extraction of ura-
nyl nitrate by TBP, is a good example. The free uranyl ion is surrounded by
water of hydration, forming UO2(H2O)62+, which from nitric acid solutions can
be crystallized out as the salt UO2(H2O)62+(NO3)22−, though it commonly is written
UO2(NO3)2(H2O)6. Thus, in solution as well as in the solid salt, the UO22+ is
surrounded by 6 H2O in an inner coordination sphere. In the solid nitrate salt,
the distance dU-O(nitrate) between the closest oxygen atoms of the nitrate anions,
(O)2NO, and the U-atom is longer than the corresponding distance, dU-O(water),
to the water molecules, OH2, i.e., dU-O(nitrate) > dU-O(water); thus the nitrate
anions are in an outer coordination sphere.
When the adduct former TBP is added, the OP(OC4H9)3 groups are closer
to the U-atoms than the OH2’s, dU-O(TBP) < dU-O(water), two waters are ex-
pelled, although the OP- bond only occupies one coordination site. The charge
distribution around the UO2-group allows the nitrate group to enter the inner
coordination sphere of the uranyl complex. The resulting configuration is shown
in Fig. 4.27. Thus, the extraction reaction can more correctly be written
UO 2 (H 2O)62+ + 2 NO3− + 2 TBP(org)
← UO 2 (TBP) 2 (NO3 ) 2 (org) + 6 H 2O (4.5b)
→
In many systems it is difficult to distinguish between inner and outer

sphere complexation. In that case, knowledge about the thermodynamic parame-
ters ∆H and ∆S may be of help. For inner sphere complexes, the hydration is
disrupted more extensively and the net enthalpy and entropy changes are posi-
tive. In Eq. (4.5b) it is noted that the number of entities increases from five to
seven, an entropy increase. This entropy effect (i.e., −T∆S ) is obviously larger
than the enthalphy effect, as the overall energy change ∆G is negative. In outer
sphere complexes, the dehydration is less disrupted, and the net enthalpies and
entropies are negative owing to the complexation with its decrease in random-
ness without a compensary disruption of the hydration spheres.
Migration experiments have shown that the hydrated cations not only
carry with them the water in the inner coordination sphere, but also one or more
shells of additional water molecules, for typical total values of 10–15. When the
metal ion leaves the aqueous phase in the solvent extraction step, this ordered
coordinated water returns to the bulk water structure, contributing an additional
factor to consider in evaluating the thermodynamics of extraction.
4.13.7.4 Steric Hindrance
The closer the reactive ligand atoms get to the central metal ion, the stronger
the bond, and, consequently, the greater the formation constant. This can be
Fig. 4.27 Structure of the UO2(NO3)2 ⴢ 2(C2H5O)3PO. The uranyl oxygens are situated
perpendicular to the plane shown around the central atom. (From Ref. 12.)
182 Rydberg et al.
studied by making substitutions in the organic ligand, which structurally inter-

feres with the formation of the complex. For example, Dyrssen [8] substituted
oxines (8-hydroxyquinolinols), Appendix D.6 in various positions, Table 4.5,
and measured their distribution constants, KDC. Although the pKa and KD,HA in
Table 4.5 did not vary much with the substitution, the KDC for the ThA4 complex
with unsubstituted oxine was 425, and for the 5-methyl substituted complex
around 1000, while the 2-methyl substituted complex was not extractable. This
can be attributed to the 2-methyl group blocking the complex formation.
Table 4.18 lists Kex values for UO22+ and Pu4+ with four organophosphorous
extractants, TBP and DOBA, and TiBP and DOiBA, the iso-forms being more
branched. The data in the upper half of the table shows little change in Kex when
TBP is replaced by TiBP. Thus the branching has little effect due to the free
rotation of the substituents around the phosphorous atom. In the lower group,
the amide DOBA, and its branched isomer DOiBA, the nature of R, R′ and R″
in R″R′NCOR is important due to the molecular rigidity of the amide group. The
branched substituent strongly depresses the extraction constant, which likely is
due to a much lower stability constant for the branched complex. This effect is
more pronounced for Pu(IV) than for UO22+. In a search for selective systems
for separating trivalent actinides form fission lanthanides, Spjuth et al. [59] in-
vestigated the extraction of trivalent ions from aqueous HNO3 solutions contain-
ing 2-bromodecanoic acid (HBDA) by means of terphenyl and triazine deriva-
tives dissolved in the organic diluent tert-butylbenzene (TBB). Figure 4.28
shows the distribution ratio of the system as a function HBDA in TBB with and
without the terpyridine (formula in Fig. 4.29). This is an illustration of the
Table 4.18 Constants for Extraction of Actinide Nitrate Adducts with

Phosphoric Acid Trialkylesters (Free Rotation of the P Substituents) and
with N,N-Dialkylamides (Restricted Rotation Around the Amide Moiety)
Showing the Effect of Steric Hindrance
Complex in Extraction
Metal ion Extractant organic phasea constant Kex Ref.
UO22+ TBP UO2(NO3)2(TBP)2 28.3 [57]

UO22+ TiBP UO2(NO3)2(TiBP)2 26 [57]
Pu4+ TBP Pu(NO3)4 (TBP)2 11.7 [57]
Pu4+ TiBP Pu(NO3)4 (TiBP)2 8.9 [57]
UO22+ DOBA UO2(NO3)2(DOBA)2 5.75 [58]
UO22+ DOiBA UO2(NO3)2(DOiBA)2 0.55 [58]
Pu4+ DOBA Pu(NO3)4 (DOBA)2 0.235 [58]
Pu4+ DOiBA Pu(NO3)4 (DOiBA)2 0.003 [58]
a
TBP (C4H9)3PO; TiBP [(CH3)2CHCH2O]3PO; DOBA [C4H9CH(C2H5)CH2]2NCOC3H7;
DOiBA [C4H9CH(C2H5)CH2]2NCOCH(CH3)2.
Fig. 4.28 Extraction of Eu(III) and Am(III) from K,HNO3 with 0.02 M oligopyridine
or triazine and 1 M 2-bromodecanoic acid in tert-butylbenzene. (From Ref. 59.)
necessity to achieve synergistic effects by using adduct formers for extraction

of metal ions that otherwise are too strongly hydrated to be extractable. The
structures of the terphenyls (adduct formers) tested are listed in Fig. 4.29 with
the distribution ratios achieved for Am(III), DAm. For example, in the system
with DODOXY, DAm is 4.2, while with TADPTZ it is 359. The triazine TAD-
PTZ is more branched (the terphenyl DODOXY more linear) and has a slightly
larger molar volume, but the difference may also be attributed to the greater
basicity (i.e., smaller Ka) compared to terphenyl. However, the 100 times larger
184 Rydberg et al.
Fig. 4.29 Structure of various oligopyridine and triazine adducts, and distribution ra-
tios (inserts) for Am(III) complexes with these oligopyridines (0.02 M) and 2-bromode-
canoic acid (0.00025 M) in tert-butylbenzene and 0.01 M HNO3. (From Ref. 59.)
Rydberg_5063-2_Ch05_R3_02-09-04 11:29:22
extraction by the triazine cannot be explained by the HSAB theory; various

explanations are suggested [59]. Steric hindrance can be predicted by modeling
the molecular structures and chemical reactions. Advanced programs require
mainframe computers, but useful programs are available for desktop computers
[49].
4.14 CALCULATION OF EQUILIBRIUM CONSTANTS

Earlier sections presented chemical models for the extraction of acids and metals
into organic solvents, and show that these models, expressed mathematically,
agree with experimental data at trace metal concentrations and at constant activ-
ity coefficients. These models provide a rationale for understanding the chemi-
cal principles of solvent extraction.
From plots of the distribution ratio against the variables of the system—
[M], pH, [HA]org, [B], etc.—an indication of the species involved in the solvent
extraction process can be obtained from a comparison with the extraction curves
presented in this chapter; see Fig. 4.3. Sometimes this may not be sufficient,
and some additional methods are required for identifying the species in solvent
extraction. These and a summary of various methods for calculating equilibrium
constants from the experimental data, using graphical as well as numerical tech-
niques is discussed in the following sections. Calculation of equilibrium con-
stants from solvent extraction is described in several monographs [60–64].
4.14.1 Identification of Species

An initial step in data analysis is to develop an equation that represents the
experimental data reasonably. Although previous sections dealt with this issue,
the approach assumed that certain species are formed. Two alternatives to this
procedure are discussed here, both yielding the approximate stoichiometry of
the complexes formed in the system. The most elementary is referred to as Job’s
method, while the ligand number method, developed by J. Bjerrum, is slightly
more advanced.
4.14.1.1 Job’s Method
When a metal M with a ligand A forms an extractable complex of the chemical
form MmAa, the extraction is at maximum when the molar ratio of these two in
the system is m:a. Similarily, if the extracted species is MAaBb, a plot of the
mole ratios of A and B yields a maximum at the ratio a:b. This is illustrated in
Fig. 4.2 for the extraction of U(VI) in the TTA-TBP (or TBPO) system. The
curves are almost symmetrical around the mole fractions 0.5:0.5, indicating that
the extracted species has a 1:1 ratio of TTA to TBP. Because the complex must
contain 2 TTA (for electroneutrality), there should also be 2 TBP (or TBPO)
molecules in the complex, i.e., MA2B2.
186 Rydberg et al.
This method is useful when only one species is extracted, but it has little value
for the study of solvent extraction systems that contain several complex species.
4.14.1.2 Ligand Number Method
This method [65–67] is useful for identifying the average composition of the
metal species in the system. Consider Eq. (4.36a) for the extraction of MAz, and
assume—for the moment—that only one species, MAn, exists in the aqueous
phase. Taking the derivative of the logarithm of Eq. (4.36) yields
d log D / d log [A] = z − n (4.80)
In 1941, J. Bjerrum [65] developed the useful concept of average ligand num-
ber, n̄, defined as the mean number of ligands per central atom:
n = Σ n [MA n ]/[M]t (4.81)
It can be shown [66a–67] that n̄ equals n in Eq. (4.79), which can be rewritten
n = z − d log D / d log [A] (4.82)
For example, in Fig. 4.10 (Example 3), the slope of the plot of log DCu vs. log
[A−] can be used to conclude what species dominate the system at a given [A−]
value. The relation in Eq. (4.82) indicates an asymptote of slope 2, so the aque-
ous phase is dominated by uncomplexed Cu2+ (n = 0), while for slope 0 the
neutral complex CuA2 dominates the system (n = 2). Equation (4.82) shows that
any tangent slope of the curve (i.e., d log D / d log [A]) yields the difference
z − n̄ in these simple systems.
Example 12: Extraction of Th(IV) by acetylacetone.
Figure 4.30 shows a smoothed curve of measurements of the distribution
of Th(IV) from 0.1 M NaClO4 into chloroform containing the extractant acetyl-
acetone (HA) [66a,b]. Taking the derivative of this curve according to Eq.
(4.79) the average ligand number is derived as shown in the lower insert.
Th(IV) is successively complexed by A− forming ThA3+, ThA22+, ThA+3 and un-
charged ThA4, which is extracted. At pA > 8.5 the n̄-value is zero, i.e., Th is
uncomplexed, while at pA < 3.5 the average ligand number of 4 is reached,
i.e., Th is fully complexed as ThA4. See also Example 15.
The average ligand number can be used to obtain approximate equilibrium con-
stants, as described by [65], assuming that at half integer n̄-values the two adja-
cent complexes dominate: e.g., at n̄ = 0.5 the species Mz+ and MAz−1 dominate,
while at n̄ = 1.5 MAz−1 and MA2z−2 dominate, etc.. The following expression for
the stepwise formation constant is approximately valid at
n = n − 0.5, log K n ≈ −log [A]n (4.83)
This method of obtaining an estimate of the formation constants was done as a
first step in the Th(IV)-acetylacetone system in Fig 4.30, where in the lower
Fig. 4.30 Upper curve: the distribution of Th(IV) between benzene and 0.1 M NaClO4
as a function of aqueous acetylacetonate ion concentration; pA = −log[A−]; the asymp-
totes have slopes 0 and −4. Lower curve: The average number of ligands per central
atom, n̄, in same system, as obtained from a derivation of the log D(pA) curve. Using
the ligand number method, the following equilibrium constants were estimated (with
values from graphical slope analysis within parenthesis): logK1 8.0 (7.85), logK2 7.6
(7.7), logK3 6.4 (6.3), and logK4 5.1 (5.0). Log KD4 2.50 is obtained from the horizontal
asymptote. (From Refs. 66a,b.)
figure, n̄ is plotted against −log [A−], yielding the preliminary log Kn values
given in Fig. 4.3. In a similar manner, the adduct formation constants can be
determined:
Example 13: The extraction of Zn(II) by β-diketones and phosphoryl adduct
formers (cont. of Example 5).
Consider the formation of adducts of the type MAzBb, as described in
Example 5 (Fig. 4.12). The derivative of Eq. (4.48b) with respect to [B] at
constant [A] yields
d log D / d log [B] = b (4.84)
where b is the average number of adduct forming molecules in the molecule
at given [B] value. In Fig. 4.12 the asymptote has a slope of 2, indicating that
a maximum of two molecules of TBP (or TOPO) bind to the neutral metal
complex. From the lower slopes of the curve, the average number, b, can be
estimated according to Eq. (4.83).
188 Rydberg et al.
4.14.2 Graphic Slope Analysis

For a rational application of slope analysis it is important to measure the varia-
tion of the distribution ratio D with one component x at a time while the others,
C, are kept constant. The solvent extraction equation can then be expressed in
the form of a simple polynomial of type
y = a0 + a1 x + a2 x 2 + L (4.85)
where y is a function of the distribution ratio Dconst.C and x a function of the
variable (e.g., pH, the free ligand ion concentration [A−], the concentration of
free extractant HA or adduct former B in the organic phase, etc.).
4.14.2.1 Linear Plots
When the distribution equation can be expressed in the form y = a0 + a1 x, from
a plot of y vs. x the intercept on the y axis yields the a0 parameter and the slope
the a1 parameter. This treatment is referred to as the limiting value method.
Example 14: The extraction of Cu(II) by HTTA (cont. from Example 4).
Example 4 (Fig. 4.11a), is represented by Eq. (4.44), which in logarith-
mic form is written
log DCu = log K ex + 2 log [HA]org + 2 pH (4.86)
For constant [HA]org (Fig. 4.11a), a plot of log DCu vs. pH yields a line of slope
2, which intercepts pH = 0 at log DCu = log Kex + 2log [HA]org. Another plot of
log DCu against log [HA]org at constant pH yields a line that intercepts [HA]org
= 0 at log DCu = log Kex + 2pH (Fig. 4.11b). In either case, the HTTA system
yields log Kex = −1.25 for the CHCl3, and −1.08 for the CCl4 system. The corre-
sponding values for the IPT system are 1.60 and 0.95, respectively.
Example 15: Extraction of Th(IV) by acetylacetone (cont. from Example 12).

In Example 12, it was concluded from Fig. 4.30 that the aqueous phase
contained all ThAn-complexes with 0 < n < 4, and the organic phase only the
uncharged ThA4 complex. We can therefore write the distribution of Th(IV)
between chloroform and water
DTh = [ThA 4 ]org / Σ [ThA n ] = K DC β4 [A]4 / Σ βn [A]n (4.87)
This is rearranged to yield
DTh−1 = F0 = (1 + a1 x −1 + a2 x −2 + a3x −3 + a4 x −4 ) / K DC (4.88a )
where KDC is the distribution constant of ThA4, x equals [A], and a4−n = βn/β4.
In a plot of F0 against x−1, the intercept becomes K −1
DC and the slope is a1 /KDC.
In the next step the function F1 is calculated
F1 = ( K DC D −1 − 1) x = a1 + a2 x −1 + a3 x −2 + L (4.88b)
A plot of F1 vs. x−1 yields a1 at the intercept and a2 as the slope. In third step,
F2 = (F1 − a1) x = a2 + a3 x−1 + a4 x−2 is calculated, yielding a2 and a3, etc. Using
this technique all four an values are obtained, from which one can deduce the
βn values [actually the Kn values in Eq. (3.5)] and the KDC value.
Dyrssen and Sillén [68] pointed out that distribution ratios obtained by
conventional batchwise techniques are often too scattered to allow the determi-
nation of as many parameters as used in Examples 15 and 16. They suggested
a simplified graphic treatment of the data, based on the assumption that there is
a constant ratio between successive stability constants, i.e., Kn/Kn+1 = 102b, and
that all distribution curves can be normalized so that N −1 log βN = a, where N is
the number of ligands A− in the extracted complex. Thus, the distribution curve
log DM vs. log[A−] is described by the two parameters a and b, and the distribu-
tion constant of the complex, KDC. The principle can be useful for estimations
when there is insufficient reliable experimental data.
4.14.2.2 Nonlinear Plots
It is not possible to obtain simple linear relations between D and the variables
when both the aqueous or organic phase contain several metal species. Instead
a double polynomial such as
b0 + b1 + b2 y 2 + b3 y 3 + L Σ bi y i
D =Y = = (4.89)
a0 + a1 x + a2 x 2 + a3 x 3 + L Σ an x n
is obtained for which there is no simple solution.
Example 16: Extraction of U(VI) by acetylacetone.
Fig. 4.31 shows the extraction of U(VI) by acetylacetone (HA) from
nitric acid solutions, DU, as a function of pA = −log [A−] at six different total
concentrations of HA. A comparison with Fig. 4.3 indicates that we can expect
complexes of the type MAn(OH)p(HA)r, which formally is equivalent to
Hr+pMAn+r. The extraction reaction is of the type Eq. (4.86), as one expects a
series of self-adduct complexes MAn(HA)r in the organic phase, and one or
two series of stepwise complexes MAn and M(OH)p in the aqueous phase. Thus
the extraction reaction is of the type Eq. (4.89) with a ratio of several polyno-
mials. Even in this case, a graphic extrapolation technique was useful by deter-
mining a set of intermediary constants for each constant [HA]org from logDU
vs. pA. Plotting these intermediate constants vs. [HA]org, a new set of constants
were obtained, from which both the stepwise formation constants βn,p and the
adduct formation constants Kad, 1 and Kad,2 for the reaction UO2A2(H2O)2–3 +
nHA → ← UO2A2(HA)1–2 in the organic phase. A more detailed analysis also al-
lowed the determination of formation constants for UO2(OH)p (p = 1 and 2).
There may be other explanations to the extraction results. However, with the
model used and the constants calculated, log DU could be correctly predicted
over the whole system range [U]tot 0.001–0.3 M, pH 2–7, and [HA]org 0.01–1.0
M. See also Example 17.
190 Rydberg et al.
Fig. 4.31 Extraction of U(VI) by acetylacetone (HA) from 0.1 M NaClO4 into chloro-
form as a function of pH at different total concentrations of HA. (From Ref. 69.)
Rydberg_5063-2_Ch05_R3_02-09-04 11:29:22
4.14.3 Numerical Methods

The most complicated equations presented in this chapter are of the types shown
by Eq. (4.89), where a series of complexes are formed in the organic phase and,
at the same time, one or a series of complexes in the aqueous phase. More
complicated equations would be obtained if mixed complexes and/or polynu-
clear complexes are present, and if varying activity factors are introduced, but
these cases are not discussed here. Thus the solvent extraction equations can be
expressed by Eq. (4.89), where x and y are independent variables and an and bn
are unknown independent parameters. The polynomial in the denominator refers
to the formation of aqueous phase complexes; in the case that the metal forms
several series of complexes with different ligands A−, L−, etc., the denominator
contains several polynomials: c0 + c1z + . . ., etc. The polynomial in the numera-
tor always refers to the formation of organic phase complexes.
There is no exact numerical solution to Eq. (4.89) when both x and y > 0.
In practice, therefore, one variable must be kept constant (or zero), while the
value of the other changes. This was described for Example 16. Thus, if x is
kept constant, the double polynomial is reduced to a simple one
y = Cx Σβi y i (Cx−1 = a0 + a1 x + a2 x 2 + L) (4.90)
where Cx is a constant at constant x. The problem is therefore reduced to the
determination of the parameters of a simple polynomial, to which many numeri-
cal methods have been applied with varying degrees of success. The main re-
quirement is that all computed parameters should be positive numbers (or zero).
In order to solve an algebraic system of n parameters, only n equations
are needed (a minimum with no error estimates). When the solvent extraction
reaction can be described by Eq. (4.90), there are as many equations as there
are experimental points. Commonly, in solvent extraction 10–50 points are
needed to cover the whole concentration range of interest, while the number of
unknown parameters in simple cases is <5. In evaluating the parameters, it is
important to use the complete suite of experimental data, as that gives greater
significance to the an or bn values.
There are several ways to solve a large number of polynomials like Eq.
(4.90) for the an parameters, e.g., by minimizing the sum of the residuals, the
least square method (LSQ) being the standard procedure. It is an objective re-
gression analysis method, which yields the same results for a given chemical
system, independent of method of investigation, provided there are no errors in
the technique or fundamental chemical assumptions. Because the method may
be applied in somewhat different ways, it may sometimes give slightly different
results. Therefore, in judging a set of equilibrium constants, it is useful to learn
how an author has applied the minimization technique.
In Eq. (4.91) x is usually the free ligand concentration [A−], or concentra-
tion [B] of the adduct former, while y is a simple function of the distribution
192 Rydberg et al.
ratio D. The least square method requires that S (the weighted squared residu-
als) in expression Eq. (4.91) is minimized
2
L
 N 
S = ∑ wi  ∑ (an xin ) − yi  (4.91)
i =1  n =0 
where N + 1 is the number of parameters, and L the number of experimentals
points; N + L − 1 is referred to as the number of degrees of freedom of the
system; each point has a value xi / yi. Because experiments are carried out over
a large range of D, [A−] and/or [B] values, the points carry different algebraic
weight (e.g., the value 1000 obscures a value of 0.001). Therefore, in order to
use the LSQ technique properly, each point must be correctly weighted, wi. This
can be done in several ways, the most common being to weight it by yi, or by
σ−i 2, or by a percentage value of yi ; σi is the standard deviation in the measure-
ment of yi. The difference an xni − yi (the residual) is not zero, because the differ-
ence is to be taken between a measured value, yi (meas.), and the corresponding
calculated value, yi (calc.), by the an x ni function [i.e., yi(calc.) − yi(meas.)], using
the actual an values at the time of the operation.
The principle of the LSQ technique is to compute the set of positive an
values that give the smallest sum of the residuals; Eq. (4.91) reaches the Smin
value. If the residuals equal zero (which rarely occurs in practice), Smin would
be zero, and there would be a perfect fit between the experimental points and
the calculated curve.
There are several mathematically different ways to conduct the minimiza-
tion of S [see Refs. 70–75]. Many programs yield errors of internal consistency
(i.e., the standard deviations in the calculated parameters are due to the devia-
tions of the measured points from the calculated function), and do not consider
external errors (i.e., the uncertainty of the measured points). The latter can be
accommodated by weighting the points by this uncertainty. The overall reliabil-
ity of the operation can be checked by the χ2 (chi square) test [71], i.e., Smin/(L
+ N − 1) should be in the range 0.5–1.5 for a reasonable consistency between
the measured points and the calculated parameters.
Example 17: Extraction of Pm(III) by acetylacetone.
In Example 7, it was concluded that a number of self-adducts PmA3
(HA)b were formed in the organic phase (0 < b < 2) in addition to the PmAn
complexes in the aqueous phase (0 < n < 4); some extraction curves are given
in Fig. 4.14. Equation (4.55) is of the same form as Eq. (4.89); rewriting Eq.
(4.55b)
Y = DPm Σβn [A]n / β3 [A]3 = K DC (1 + K ad1[HA]org + L) (4.92)
which can be subdivided into

Y[A] = C1 K DC (1 + K ad1[HA]org + L) (4.93a )
and
Y[HA]org = C2 DPm Σβn [A]n / β3 [A]3 (4.93b)
Y[HA]org contains the measured DPm, the [A] values, and the βn values. It can thus
in principle be treated as an Example 16 to yield the βn values and the C2 values
(one for each [HA]org), remembering that a0 = 1. Because Y[A] /C1 = Y[HA]org/C2
= Y′, a second plot of Y′ against [HA]org yields the KD3 and Kadb values. The
analysis of the system yielded KDC = 0.008, Kad,1 = 7 and Kad,2 = 3. In these cal-
culations, a SIMPLEX program was used [75].
In comparisons of equilibrium constants collected from the literature (e.g.,
Fig. 4.22 or [47]), or correlations of data for a large number of systems (e.g.,
Figs. 4.20–4.23), it is desirable to present both the statistical uncertainty of each
“point,” which is often given by the standard deviation (one or several σ’s) of
the point, and the general reliability (statistical significance) of the whole corre-
lation [76], for which the chi-square test offers a deeper insight into the reliabil-
ity of the experimental results [77]. More advanced statistical tests for systems
of our kind have been described by Ekberg [78].
4.15 EXPERIMENTAL DETERMINATION

OF DISTRIBUTION RATIOS
The studies described above have the purpose of identifying the reacting species
in a solvent extraction process and developing a quantitative model for their
interactions. The fundamental parameter measured is the distribution ratio, from
which extraction curves are derived. Solvent extraction work can still be carried
out with simple batchwise (or point-by-point) technique, but continuous on-line
measurements give faster and more accurate results.
4.15.1 Stirred Cell Semicontinuous Techniques

Each point-by-point experiment requires a complete set of mixing, separation,
sampling, and analysis. This usually leads to scattered results, though it may
not be critical, if the D values cover a limited range from 0.1–10. However, the
more the D values deviate from 1, the more accurate must be the measurements;
also the number of points required for a reliable extraction curve usually in-
creases. To reduce the uncertainty and labor involved with the batch technique,
the stirred cell technique has become popular.
Figure 4.32 shows a typical thermostated stirred cell. The liquid volumes
are commonly 50 + 50 mL. The stirrer may consist of a single paddle at the
interface, or a double paddle, one in the center of each phase. The cell may
contain baffles, etc., to improve the mixing. Stirring may be violent, completely
destroying the interface and producing very small (Ⰶ1 mm) droplets, or slow
194 Rydberg et al.
Fig. 4.32 A thermostated double jacket (1) cell for solvent extraction studies (heavier
phase 8, lighter phase 9) under nonoxidizing conditions, using a hydrogen gas inlet (6)
to a Pd-black catalyst (11), pH glass electrode (3), magnetic stirrer (10), connections for
additions (5). Alternative constructions contain rotating paddles and fixed pipings con-
nected to the two phases for frequent sample withdrawals.
in order not to destroy the interface. The stirring rate is optimized to the time
for reaching equilibrium and complete phase separation. The experiments are
either carried out with intermittent violent stirring, in which case samples are
withdrawn after each complete phase separation, or with mild stirring during
which it is possible to continously withraw samples. Equal volumes are sampled
each time, commonly <1 mL. The simple stirred cell has been improved by
introducing phase discriminating membranes in the sampling outlet [79]. This
is particulary advantageous for kinetic experiments and is further described in
Chapter 5. The sampling of the stirred cell can be automated, so that at regular
intervals pH and temperature are recorded, and samples withdrawn for automatic
analysis of concentration of interesting species in more or less standard fashion.
It is also possible to use ion-sensitive probes in one of the phases instead of
sampling.
4.15.2 Centrifugal Extraction-Separation Systems

A different approach to solvent extraction experimentation, referred to as the
AKUFVE principle, was developed in the 1960s [80a–d]. The AKUFVE is
illustrated in Fig. 4.33. Efficient mixing of the two phases and their additions is
achieved in the mixing vessel and at the inflow into the phase separator, which
consists of a continuous flow centrifuge, which in a special separation chamber
and at very high rotational velocity (5,000–50,000 rotations per minute) sepa-
rates the mixture into two very pure phases, containing <0.01% of entrainment
of one phase in the other phase. The H-centrifuge may be made of Pd-stabilized
Fig. 4.33 The AKUFVE solvent extraction apparatus: Efficient mixing is achieved in
the separate mixing vessel, from which the mixture flows down into the continuous
liquid flow centrifugal separator (the H-centrifuge, hold-up time <1 s). (From Refs.
83a,b.) The outflow from the centrifuge consists of two pure phases, which pass on-line
detectors, AMXs, for on-line detectors or continuous sampling. (From Refs. 80a–80d,
81.)
196 Rydberg et al.
Ti, or PEEK (polyether ether ketone) to allow measurements under very corro-
sive conditions. The separated phases pass AMX gadgets for on-line detection
(radiometric, spectrophotometric, etc.) or phase sampling for external measure-
ments (atomic absorption, spectrometric, etc.), depending on the system studied.
The aqueous phase is also provided with cells for pH measurement, redox con-
trol (e.g., by reduction cells using platinum black and hydrogen, metal ion deter-
mination, etc.) and temperature control (thermocouples).
The AKUFVE technique allows a large number of points (50–100) to be
determined in a one-day experiment over a D-range of better than 103 to 10−3,
not counting time of preparation. In a special version of this technique (LISOL
for LIquid Scintillation On Line) [82], the D range 10−5 to 104 has been accu-
rately covered, as for the Pm-acetylacetone system, Example 7 and Fig. 4.14.
The centrifugal separator of the AKUFVE system is also used for phase
separation in the SISAK technique [84]. SISAK is a multistage solvent extrac-
tion system that is used for studies of properties of short-lived radionuclides,
e.g., the chemical properties of the heaviest elements, and solvent extraction
behavior of compounds with exotic chemical states. In a typical SISAK experi-
ment, Fig. 4.34, radionuclides are continuously transported from a production
Fig. 4.34 A typical SISAK setup used for studies of α-decaying nuclides, e.g., trans-
actinides. (From Ref. 84.)
site in an accelerator or a research reactor to the SISAK equipment via a gas-

jet transport system. The nuclides are dissolved in an aqueous phase that is fed
into a centrifuge battery comprising 1–4 solvent extraction steps. The product
solution leaving the last step is pumped, e.g., to a nuclear radiation detection
system. The transport time from the target site to the detection system depends
on the centrifuge size, number of centrifuge steps, and flow rate. For a one-step
chemistry, i.e., dissolution step and a single centrifuge extraction, and maximum
flow rates, the overall transport time is around 2.5 s. This fast transport has
allowed detailed γ-spectroscopic studies of radionuclides with half-lives around
1 s. Recently the SISAK equipment was succesfully applied to studies of the
heaviest elements, and solvent extraction data were obtained for element 104 Rf
[85]. Fast chemical separation systems have been developed for quite a large
number of elements [86].
Centrifugal extractors have been designed for a number of industrial uses
(see succeeding chapters). In some cases they have been scaled down to labora-
tory size but mainly been used for developing industrial multistage processes.
4.15.3 Liquid Partition Chromatography

In liquid partition chromatography a solute distributes itself between a mobile
liquid phase and an immobile solvent attached to a solid matrix (in the earliest
experiments a claylike kieselguhr). It is outside the scope of this chapter to
discuss this technique, which, however, is briefly described in Chapter 15.
4.16 CONCLUSION AND FINAL COMMENTS

The solvent extraction process is usually described by a single net reaction,
defined by the extraction constant Kex. Variations in Kex caused by modifications
of the solvent system, such as changes in the temperature or aqueous ionic
strength, or by replacing one solvent by another, or making substitutions in the
extractant molecule, may be explained by careful consideration of the parame-
ters of the system. However, such studies are difficult and not always sufficient
for predicting new systems. A better foundation for understanding the extraction
process is to consider the steps in the process contributing to the net extraction
reaction, particularily when these steps are governed by regularities. Knowledge
of these regularities helps in interpreting systems as well as in predicting new
ones. Extension of these distribution data to thermodynamic constants is likely
to give benefits in increased chemical knowledge of the behavior of solutes in
different solvent systems. Advanced quantum chemistry calculations and com-
puter modeling of extraction processes can help us in designing new, selective
solvent extraction systems, as well as in interpreting extraction phenomena (see
Chapter 16).
198 Rydberg et al.
REFERENCES
1. Grahame, D. C.; Seaborg, G. T. J. Am. Chem. Soc. 1938, 60:2524.
2. Irving, H. p. 91, ISEC’66, Solvent Extraction Chemistry Dyrssen, D.; Liljenzin,
J. O.; Rydberg, J. Eds.; North Holland, Amsterdam (1967).
3a. Rydberg, J. Arkiv Kemi 1954, 8:101.
3b. Rydberg, J. Rec. Trav. Chim. 1956, 75:737.
4. Sekine, T.; Hasegawa, Y. Solvent Extraction Chemistry, Marcel Dekker, New
York, 1977.
5. Marcus, Y.; Kertes, S. Ion Exchange and Solvent Extraction of Metals, Wiley-
Interscience, Chichester, 1969.
6. Barton, A. F. M. CRC Handbook of Solubility Parameters and Other Cohesion
Parameters, CRC Press, 1991.
7. Kojima, I.; Yoshida, M.; Tanaka, M. J. Inorg. Nucl. Chem. 1970, 32:987.
8a. Dyrssen, D. Svensk Kem. Tidskr. 1956, 68:212.
8b. Dyrssen, D. Svensk Kem. Tidskr. 1952, 64:213.
9. Sekine, T. et al. Bull. Chem. Soc. Jap. 1983, 56:700.
10. Rydberg, J. Svensk Kem. Tidskr. 1950, 62:179.
11a. Sekine, T.; Isayama, M.; Yamaguchi, S.; Moriya, H. Bull. Chem. Soc. Jap. 1967,
40:27.
11b. Dyrssen, D. Private communication. See also Acta Chem. Scand. 1960, 14:1091.
12. Ahrland, S.; Liljenzin, J. O.; Rydberg, J. The Chemistry of the Actinides. In Com-
prehensive Inorganic Chemistry, Pergamon Press, London, 1973.
13. Jassim, T. M.; Fridemo, L.; Liljenzin, J. O. in T. M. Jassim, Co-Extraction of
Pertechnetate with some Metal Nitrates in TBP-Nitric Acid Systems. Diss. Chal-
mers Techn. Univ., Gothenburg (1986).
14. Martell, A. E.; Calvin, M. Chemistry of the Metal Chelate Compounds, Prentice-
Hall, Englewood Cliffs, N.J., 1952.
15. Stary, J. The Solvent Extraction of Metal Chelates, Pergamon Press, New York,
1964.
16. Zolotov, Yu. Extraction of Chelate Compounds, Humprey Sci. Publ., Ann Arbor,
London, 1970.
17a. Narbutt, J. J. Inorg. Nucl. Chem. 1981, 43:3343.
17b. Narbutt, J. J. Phys. Chem. 1991, 95:3432.
17c. Narbutt, J.; Fuks, L. Radiochimca Acta 1997, 78:27.
18. Liljenzin, J. O.; Stary, J.; Rydberg, J. ISEC’68, Solvent Extraction Research,
Kertes, A. S.; Marcus, Y., Eds., J. Wiley and Sons, New York, 1969.
19. Sillén, L. G.; Martell, A.; Högfeldt, E., Eds., Stability Constants of Metal-Ion Com-
plexes. The Chemical Society Spec. Publ. No. 25, Burlington House, London, 1971.
20. Martell, A. E.; Smith, R. M. Critical Stability Constants, Vol. 1–5, Plenum Press,
New York 1974–1982.
21. Marcus, Y.; Kertes, A. S.; Yanir, E. Equilibrium Constants of Liquid-liquid Distri-
bution Reactions, Part 1, Butterworths, London, 1974.
22. Sekine, T.; Dyrssen, D. J. Inorg. Nucl. Chem. 1964, 26:1727.
23a. Allard, B.; Johnsson, S.; Rydberg, J. ISEC’74, Proc. Int. Solvent Extraction Conf.,
Lyon, 1974.
23b. Allard, B.; Johnsson, S.; Narbutt, J; Lundquist, R. ISEC ‘77, CIM Special Volume
21, 1977.
24. Sekine, T.; Ihara, N. Bull. Chem. Soc. Jap. 1971, 44:2942.
25. Cecconie, T.; Freiser, H. Anal. Chem. 1990, 62:622.
26. Moryia, H.; Sekine, T. Bull. Chem. Soc. Jap. 1971, 44:3347.
27a. Rydberg, J.; Albinsson, Y. J. Solv. Extr. Ion Exch. 1989, 7:577.
27b. Albinsson, Y.; Rydberg, J. Radiochim. Acta 1989, 48:49.
27c. Albinsson, Y. Acta Chem. Scand. 1989, 43:919.
28. Albinsson, Y. Development of the AKUFVE-LISOL Technique. Solvent Extraction
Studies of Lanthanide Acetylacetonates. Diss. Chalmers Univ. Techn., Gothenburg
(1988).
29. Van Ooyen, J.; Dyrssen, D.; Liljenzin, J. O.; Rydberg, J. (Eds.) Solvent Extraction
Chemistry (ISEC 1966), North-Holland Publ. Co., Amsterdam, 1967.
30a. Sekine. T.; Dyrssen, D. Anal. Chim. Acta 1969, 45:433.
30b. Sekine, T.; Fukushima, T.; Hasegawa, Y. Bull. Chem. Soc. Japan 1970, 43:2638.
31. Sekine, T.; Sakairi, M. Bull. Chem. Soc. Jap. 1967, 40:261.
32. Liljenzin, J. O.; Vadasdi, K.; Rydberg, J. Coordination Chemistry in Solution Hög-
feldt, E. Swedish Nat. Res. Council; Trans.Roy. Inst.Techn., 280, Stockholm, 1972.
33a. Rydberg, J. Rev. Inorg. Chem. 1999, 19:245.
33b. Rydberg, J. Min. Pro. Ext. Met. Rev. 2000, 21:167.
34. Carey, F. A.; Sundberg, R. J. Advanced Organic Chemistry, Third Edition, Plenum
Press, New York, 1990.
35. Fuks, L.; Majdan, M. Min. Pro. Ext. Met. Rev. 2000, 21:25.
36. Allard, B. The Coordination of Tetravalent Actnide Chelate Complexes with β-
Diketones, Diss. Chalmers Univ. Techn. Gothenburg, 1975.
37. Hildebrand, J. H.; Scott, R. C. Solubilities of Nonelectrolytes, Dover Publishers,
1964.
38a. Siekierski, S. J. Inorg. Nucl. Chem. 1962, 24:205.
38b. Siekierski, S.; Olszer, R. J. Inorg. Nucl. Chem. 1963, 24:1351.
39. Buchowski, H. Nature 1962, 194:674.
40a. Wakahayashi, T.; Oki, S.; Omori, T.; Suzuki, N. J. Inorg. Nucl. Chem. 1964, 26:
2255.
40b. Omori, T.; Wakahayashi, T.; Oki, S.; Suzuki, N. J. Inorg. Nucl. Chem. 1964, 26:
2265.
40c. Oki, S.; Omori, T.; Wakahayashi, T.; Suzuki, N. J. Inorg. Nucl. Chem. 1965, 27:
1141.
41. Irving, H. M. N. H. Ion Exchange and Solvent Extraction, Vol. 6, (Marinsky, J.;
Marcus, Y. Eds.) Marcel Dekker, New York, 1973.
42. Siekierski, S. J Radioanal. Chem. 1976, 73:335.
43. Hansen, C.; Skaaup, K. Dansk Kemi 1967, 48:81.
44. Hansen, C. Hansen Solubility Parameters: A Users Handbook, CRC Press, 2000.
45. Nilsson, M. Influence of Organic Phase Composition on the Extraction of Trivalent
Actinides, Dipl. work, Dept. Nucl. Chem., Chalmers Univ. of Techn., Gothenburg,
2000.
46. Johansson, H.; Rydberg, J. Acta Chem. Scand. 1969, 23:2797.
47. Hancock, R. D.; Martell, A. Chem. Rev., 1989, 89:1875–1913.
200 Rydberg et al.
48. Hancock, R. D. Prog.Inorg.Chem. 1989, 36:187–290.

49. CambridgeSoft Corp., 100 Cambridge Park Drive, Cambridge, Mass. (1999).
50. Siddall, T. H., III; Good, M. L. J. Inorg. Nucl. Chem. 1967, 29:149.
51. Charbonnel, M. C.; Musikas, C. Solv. Extr. Ion Exch. 1989, 7:1007.
52. Musikas, C.; Hubert, H. Solv. Extr. Ion Exch. 1987, 5:877.
53a. Narbutt, J.; Krejzler, J. Inorg. Chim. Acta 1999, 286:175.
53b. Narbutt, J.; Czerwinski, M.; Krejzler, J. Eur. J. Inorg. Chem. 2001, 3187 pp.
54. Handley, T. H.; Dean, J. A. Anal. Chem. 1962, 35:1312.
55. Patee, D.; Musikas, C.; Faure, A.; Chachaty, C. J. Less Common Metals 1986, 122:
295.
56. Handley, T. H. Anal. Chem. 1963, 35:991.
57. Schultz, W. W.; Navratil, J. (Eds.), Science and Technology of TBP, Vol. 1, CRC
Press, Boca Raton, (1984).
58. Condamines, N. Commissariat l’Energie Atomique, Report R5519, 1990.
59. Spjuth, L. Solvent Extraction Studies with Substituted Malonamides and Oligopyri-
dines, Diss. Chalmers Univ. Techn., Gothenburg, 1999.
60. Rossotti, F. J. C.; Rossotti, H. Determination of Stability Constants, McGraw Hill,
New York, 1961.
61. Froneus, S. Technique of Inorganic Chemistry (Jonassen, H. B.; Weissberger, A.,
Eds.) Wiley-Interscience, Chichester, 1963.
62. Beck, M. T. Chemistry of Complex Equilibria, Van Nostrand Reinhold, New York,
1970.
63. Hartley, F. R.; Burgess, C.; Alcock, R. Solution Equilibria, Ellis Hoorwood, John
Wiley and Sons, New York, 1980.
64. Meloun, M.; Havel, J.; Högfeldt, E. Computation of Solution Equilibria, John Wi-
ley Sons, New York, 1988.
65. Bjerrum, J. Metal Amine Formmation in Aqueous Solutions, Diss. Copenhagen,
1941.
66a. Rydberg, J. Acta Chem. Scand. 1950, 4:1503.
66b. Rydberg, J. Arkiv Kemi 1953, 5:413.
67. Irving, H.; Williams, R. J. P.; Rossotti, F. J. C. J. Chem. Soc. 1955, 1955:1906.
68. Dyrssen, D.; Sillén, L. G. Acta Chem. Scand. 1953, 7:663.
69. Rydberg, J. Arkiv Kemi 1954, 8:113.
70. Deming, W. E. Statistical Adjustment of Data, John Wiley and Sons, NewYork,
1948.
71. Fisher, R. A. Statistical Methods for Research Workers, Hafner Publishing, New
York, 1958.
72a. Sullivan, J. C.; Rydberg, J.; Miller, W. F. Acta Chem. Scand. 1959, 13:2023.
72b. Rydberg, J. Acta Chem. Scand. 1960, 14:57.
72c. Rydberg, J. Acta Chem. Scand. 1961, 15:1723.
73a. Sillen, L. G. Acta Chem. Scand. 1962, 16:159.
73b. Liem, D. H. Acta Chem. Scand. 1971, 25:1521.
73c. Liem, D. H. Acta Chem. Scand. 1979, A33:481.
74. Dixon, L. C. W. Nonlinear Optimization, English University Press, 1972.
75. Albinsson, Y.; Mahmood, A.; Majdan, M.; Rydberg, J. Radiochim. Acta 1989, 48:49.
76. Meinrath, G.; Ekberg, C.; Landgren, A.; Liljenzin, J. O. Talanta, 2000, 51:231.
77. Rydberg, J.; Sullivan, J. Acta Chem. Scand. 1959, 13:2057.
78. Ekberg, Ch. Uncertainties in Actinide Solubility Calculations Illustrated Using the
Th−OH−PO4 System, Diss. Chalmers tekniska högskola, Göteborg, 1999.
79. Watari, H.; Cunningham, L.; Freiser, H. Anal. Chem. 1982, 54:2390.
80a. Reinhardt, H.; Rydberg, J. Solvent Extraction Chem. (ISEC) 1966, North-Holland
Publ. Co. Amsterdam, 612 pp.
80b. Reinhardt, H.; Rydberg, J. Chem. Ind. 1970, 488 pp.
80c. Rydberg, J., et al. Acta Chem. Scand. 1969, 23:647.
80d. Rydberg, J., et al., Acta Chem., 1969, 2773, 2781, 2797 pp.
81. MEAB Metal Extraction Co Ltd, V. Frölunda, Sweden.
82. Albinsson, Y.; Ohlsson, L. E.; Persson, H.; Rydberg, J. Appl. Radiat. Isot. 1988,
39:113.
83a. Rydberg, J.; Reinhardt, H. Centrifuge for complete phase separation of two liquids,
U.S.Patent 3,442,445, 1966.
83b. Reinhardt, H. En kontinuerlig separator för två-fas vätskeblandningar, Diss. Chal-
mers Univ. Techn., Gothenburg, 1973.
84. Persson, H.; Skarnemark, G.; Skålberg, M.; Alstad, J.; Liljenzin, J. O.; Bauer, G.;
Heberberger, F.; Kaffrell, N.; Rogowski, J.; Trautmann, N. Radiochim. Acta 1989,
48:177.
85. Skarnemark, G., et al., J. Nucl. Radiochem. Sci. 2002, 3:1.
86. Skarnemark, G. J. Radioanl. Nucl. Chem. 2000, 243(1):219.
5
Solvent Extraction Kinetics
PIER ROBERTO DANESI International Atomic Energy Agency,
Vienna, Austria
5.1 GENERAL PRINCIPLES

The kinetics of solvent extraction is a function of both the various chemical
reactions occurring in the system and the rates of diffusion of the various species
that control the chemistry of the extraction process.
5.1.1 Rate-Controlling Role

of the Chemical Reactions
The dependence of the kinetics on the chemical reactions is easily understood
by considering that the final products of any extraction process are usually in a
chemical state different from the initial unreacted species. This is true even for
the simple partition of neutral molecules between two immiscible liquid phases,
where the chemical change is in the solvation environment of the extracted
species. More drastic chemical changes take place in the extraction of a metal
cation from an aqueous solution by a chelating extractant dissolved in an organic
diluent. In the extraction some of the solvation water molecules can be removed
from the metal ion, and a new coordination compound, soluble in the organic
phase, is formed with the chelating group of the extractant. In addition, the
extracting reagent can undergo an acid dissociation reaction and, together with
the extractant-metal complex, can undergo changes in aggregation in the organic
phase. Consequently, whenever at least one of the chemical steps of the overall
reaction mechanism is slow enough, compared with the diffusion rate, the kinet-
ics of extraction would depend on the rate of the slow chemical reactions.
203
204 Danesi
For solvent extraction systems, we have two additional complications.

First, the chemical reactions can take place, at least in principle, in two bulk
phases, since we are dealing with two immiscible liquid layers. Second, the
chemical reactions can occur in the two-dimensional region called the liquid–
liquid interface, that separates the two immiscible liquids, or in a thin volume
region very close to it. When interfacial chemical reactions are important, the
situation is analogous to that describing the kinetics of the chemical reactions
encountered in heterogeneous catalysis and in some electrode processes. Al-
though a relatively large number of sophisticated techniques are available for
studying chemical reactions at solid–fluid interfaces, very few tools have been
developed to investigate chemical changes occurring at liquid–liquid interfaces.
Our knowledge of such interfacial reactions, therefore, is still limited and is
based largely on indirect evidence and speculations.
5.1.2 Rate-Controlling Role

of the Diffusional Processes
To understand the dependence of the extraction kinetics on the rate of diffusion
of the various chemical species that participate in the extraction reaction, we
first have to distinguish between diffusion in the bulk phases and diffusion
through the thin layers adjacent to the interface. In most solvent extraction pro-
cesses of practical interest, both the aqueous and the organic phase are effi-
ciently stirred. It follows that transport of material from the bulk of the phases
up to a region very close to the interface can be considered instantaneous and
that diffusion in the bulk of the phases can be neglected. Nevertheless, diffu-
sional processes can still have an appreciable influence on the solvent extraction
kinetics. For example, even when the two phases are vigorously stirred, it is
possible to describe interfacial diffusion by assuming the existence of two stag-
nant thin layers of finite thickness located on the aqueous and organic side of
the interface. This model of the interface, often referred to as the two-film the-
ory [1,2], is extremely useful for describing extraction kinetics that are con-
trolled by diffusion occurring in proximity to the interface. The two-film model
is used throughout our treatment. Other models and theories of higher complex-
ity exist for describing diffusional transport in proximity to the interface. These
theories (penetration, surface renewal, boundary layer) are described in detail in
more advanced books [3,4].
The two films are schematically described in Fig. 5.1, in which the pres-
ence of an interfacially absorbed layer of extractant molecules is also shown. δo
and δw represent the thickness of the organic and aqueous films, respectively. In
these layers the liquid phases are considered completely stagnant (i.e., no move-
ment of the fluids takes place in spite of the mechanical energy that is dissipated
in the two-phase system to provoke mixing of the aqueous and organic phases).
Solvent Extraction Kinetics 205
Fig. 5.1 Interfacial diffusion films. δo and δw are the thickness of the organic and
aqueous films, respectively. The presence of an adsorbed layer of extractant molecules
at the interface is also shown.
These two thin liquid films, which are also called diffusion films, diffusion
layers, or Nernst films, have thicknesses that range between 10−2 and 10−4 cm
(in this chapter centimeter-gram-second (CGS) units are used, since most pub-
lished data on diffusion and extraction kinetics are reported in these units; com-
parison with literature values is, therefore, straightforward).
The description of the diffusion films as completely stagnant layers, having
definite and well-identified thicknesses, represents only a practical approxima-
tion useful for a simple mathematical description of interfacial diffusion. A
206 Danesi
more realistic physical description should consider that, starting at a given dis-
tance from the liquid interface, the renewal of the organic and the aqueous
fluids becomes progressively less as the interface is approached.
The thickness of the diffusion films is, up to a certain limiting value, an

inverse function of the mechanical energy dissipated into the liquid system to
stir it. This means that δo and δw depend on the stirring rate of the two phases,
on the geometry of the solvent extraction equipment, and of the stirrers used, as
well as on the viscosity and density of the two liquids. For given liquid phases
and a fixed apparatus, the thickness of the diffusion films initially decreases
with increase in the rate of mixing. However, the decrease does not go beyond
a certain limiting value, and the thickness of the diffusion films never seems to
go down to zero, with limiting thickness of about 10−3 –10−4 cm eventually
reached with all systems. The exact limiting value depends on the specific physi-
cochemical properties of the liquids and the specific hydrodynamic conditions.
Even when dealing with the most efficiently stirred systems, the chemical spe-
cies that have to be transported to or away from the interface for the extraction
reaction to take place always have to diffuse through the diffusion films. The
time required for this often can be comparable or longer than that required for
the actual chemical changes in the extraction. As a consequence, diffusion
through these films controls the overall kinetics of extraction. Experience has
shown that film diffusion is the predominating rate-controlling factor in many
practical extraction processes. Although these diffusion processes occur through
very thin films, the films must be considered macroscopic, since the diffusion
distances involved exceed the molecular dimensions by several orders of magni-
tude.
5.2 DIFFUSION
Diffusion is that irreversible process by which matter spontaneously moves from
a region of higher concentration to one of lower concentration, leading to equal-
ization of concentrations within a single phase.
The laws of diffusion correlate the rate of flow of the diffusing substance
with the concentration gradient responsible for the flow. In general, in a multi-
component system, the process is described by as many diffusion equations as
the number of chemical species in the system. Moreover, the diffusion equations
are intercorrelated. Nevertheless, when dealing with solutions containing ex-
tractable species and extracting reagents at concentrations much lower than
those of the molecules of the solvents (dilute solutions), diffusion is sufficiently
well described by considering only those species for which the concentrations
appreciably change during the extraction reaction.
The diffusion flow, J, of a chemical species is defined as the amount of
matter of this species passing perpendicularly through a reference surface of
unit area during a unit time. The dimensions of J are those of mass per square
centimeter per second (mass cm−2 s−1). When the concentrations of the diffusing
species (c) are expressed in moles per cubic centimeter (mol cm−3), the correla-
tion between the flux across a unit reference area located perpendicularly to the
linear coordinate x (along which diffusion occurs) and the concentration gradi-
ent, ∂c /∂x, will be given by the Fick’s first law of diffusion, that is,
J = − D∂c / ∂x (5.1)
D is the diffusion coefficient of the species under consideration. Its dimen-
sions are given in cm2 s−1. Equation (5.1) implies that D is independent of con-
centration; it is a constant, characteristic of each diffusing species, in a given
medium, at constant temperature. This is only approximately true, but holds
sufficiently well for most diffusing species of interest in solvent extraction sys-
tems. The values of D for the majority of the extractable species fall in the
range 10−5 –10−6 cm2 s−1 in both water and organic solutions. When dealing with
viscous phases or bulky organic extractants, or with complexes that may un-
dergo polymerization reactions, the D values can drop to 10−7 or 10−8 cm2 s−1.
Equation (5.1) is extremely useful to evaluate the flux whenever the con-
centration gradient can be considered constant with time (i.e., when we can
assume a steady state). When a steady state cannot be assumed, the concentra-
tion change with time must also be considered. The non–steady-state diffusion
is expressed by Fick’s second law of diffusion:
∂c / ∂t = D∂ 2c / ∂x 2 (5.2)
Diffusion is a complex phenomenon. A complete physical description in-
volves conceptual and mathematical difficulties associated with the need to in-
volve theories of molecular interactions and to solve complicated differential
equations [3–6]. Here and in sections 5.8 and 5.9, we present only a simplified
picture of the diffusional processes, which is valid for limiting conditions. The
objective is to make the reader aware of the importance of this phenomenon in
connection with solvent extraction kinetics.
For steady-state diffusion occurring across flat and thin diffusion films,
only one dimension can be considered and Eq. (5.1) is greatly simplified. More-
over, by replacing differentials with finite increments and assuming a linear
concentration profile within the film of thickness δ, Eq. (5.1) becomes
J = − D dc / dx = − D(c2 − c1 ) / δ = (c2 − c1 ) / ∆ (5.3)
where
∆ = δ/ D (5.4)
is a diffusional parameter dependent on the thickness of the diffusion film and the
value of the diffusion constant of the diffusing species. The units of ∆ are cm−1 s.
208 Danesi
Some simple considerations about the variation of concentrations with

time that can be expected in diffusion-controlled processes, can be made by
considering the imaginary one-liquid-phase system shown in Fig. 5.2. The imag-
inary system consists of two hypothetical well-stirred aqueous reservoirs of vol-
umes V1 and V2, containing different concentrations of the same solute, in con-
tact through a nonstirred aqueous liquid film across which the solute moves
only under the influence of molecular diffusion. Since c1 and c2 are allowed to
vary with time, diffusion through the film of thickness δ tends to equalize c1
and c2. This means that J, c1 and c2 are a function of time. If a linear concentra-
tion gradient is always assumed and the amount of diffusing matter inside the
film is negligible relative to that in the reservoirs, the flux equation J can be
integrated to yield a simple correlation between c1, c2 and time (t). By consider-
ing that the correlation between the flux and the concentration variations within
the reservoirs is
−V1 Q −1dc1 / dt + V2 Q −1 dc2 / dt = J (5.5)
where Q is the area of the diffusion film (equal on both sides), and, assuming,
for simplicity, that V1 = V2 = V, Eqs. (5.3) and (5.5) can be combined to give
VQ∆dc1 / dt = c1 − c2 (5.6)
For the initial conditions at t = 0, c2 = 0, and c1 = c , the integration of Eq.
0
1
(5.6) is straightforward. Separation of the variables leads to
Qdt / V ∆ = dc1 (c10 − 2c2 ) (5.7)
Fig. 5.2 Concentration profile of a solute diffusing across an unstirred liquid film of
thickness δ in contact with two well-stirred reservoirs. The same liquid phase is assumed
throughout this imaginary model system.
Upon integration and insertion of Eq. (5.4), one obtains

ln{(2c1 − c10 ) / c10 } = −2 Q Dt / V δ (5.8)
Two conclusions follow from Eq. (5.8):
1. For t = ∞ , (2c1 − c10) = 0, that is
c1 (at t = ∞) = c2 (at t = ∞) = c10 / 2. (5.9)
This means that at equilibrium the concentrations in the two reservoirs are
equal.
2. The condition of equal concentrations is approached with a rate that is di-
rectly proportional to the contact area (Q) and the diffusion coefficient (D),
and inversely proportional to the volume of the reservoir (V) and the thick-
ness of the diffusion layer (δ).
Although these conclusions are intuitive and trivial and are based on sev-
eral crude approximations introduced into an imaginary model system consisting
of only one liquid phase, the simple mathematical treatment developed helps
the reader better appreciate the fundamental role that film diffusion plays in
controlling the rate of solvent extraction. When two different liquid phases
(aqueous and organic) are in contact, two liquid films can be assumed to exist
on the two opposite sides of the interface, and diffusion through both must be
considered. The presence of two phases at which species can be unevenly dis-
tributed, and with interfacial chemical reactions of varying degree of complex-
ity, will in fact appear in the extraction rate equations only as modifications of
this simple diffusion equation.
5.3 RATE LAWS AND MECHANISMS

5.3.1 Rate Laws and Reaction Mechanisms
For a general treatment of the various rate laws, the reader is referred to several
specialized books existing on the subject [7,8]. However, care must be exercised
in extrapolating the information obtained from the study of reactions in solutions
to solvent extraction systems. Solvent extraction deals with two solutions in
contact and in the presence of the interface, which has to be crossed by a solute
during the extraction, a fact that requires special consideration.
In this section, we briefly summarize the fundamental concepts that are
useful for better understanding those specific solvent extraction rates and mech-
anisms, which are presented in some detail in subsequent sections.
The immediate result of a kinetic study is a rate law. For a general reaction
with the stoichiometric equation
xX + yY = zZ + wW (5.10)
210 Danesi
the rate is expressed by

−d [X] −d [Y] + d [ Z] + d [ W]
, , or (5.11)
dt dt dt dt
where t indicates time and the square brackets concentrations. The rate of the
reaction is given as function of the composition of the system:
rate = − d [X]/dt = f {[X], [Y], [Z], [W]} (5.12)
The first goal of any kinetic study is to devise experiments that establish
the algebraic form of the rate law and to evaluate the rate constants. Rate laws
of the form
−d [X]/dt = k [X]x [Y]y [ Z]z [ W]w (5.13)
are relatively easy to treat mathematically when x, y, z, and w are small inte-
gers.
Rate laws can be derived by measuring concentration variations as func-
tion of time or the initial rates as function of the initial concentrations. Unfortu-
nately, there is no general method for finding the rate law and the reaction order
from measurements of concentration vs. time or other types of measurements.
Usually, a trial-and-error procedure is used, based upon intelligent guesses.
Once a rate law has been defined and the rate constants evaluated, the
next step is to correlate it with the most likely mechanism of the extraction
reaction.
Since the experimental kinetic data refer to a reaction rate and how this is
affected by variables, such as concentration, temperature, nature of the solvent,
presence of other solutes, structural variations of the reactants, and so forth, the
assignment of a mechanism is always only indirectly derived from primary data.
Therefore, it is not surprising that more than one mechanism has often been
proposed to explain the same rate law and that reaction mechanisms, which
were once consistent with all experimental information available on a system,
have later on been considered erroneous and have been disregarded, or drasti-
cally modified, as long as new experimental evidence was accumulated. In gen-
eral, the stoichiometry of the reaction, even when this is a simple one, cannot
be directly related with its mechanism, and when the reaction occurs through a
series of elementary steps, the possibility that the experimental rate law may be
interpreted in terms of alternative mechanism increases. Therefore, to resolve
ambiguities as much as possible, one must use all the physicochemical informa-
tion available on the system. Particularly useful here is information on the struc-
tural relations between the reactants, the intermediate, and the reaction products.
Table 5.1 presents the mathematical description of selected very simple
rate laws.
Irreversible first-order reactions [Eq. (5.14) in Table 5.1] depend only on
Table 5.1 Common Rate Expressions

Irreversible first-order reaction
Reaction X + other reactants (OR) → products (P) (5.14a)
Rate expression −d[X]/dt = k[X] (5.14b)
Solution [X] = [X]0e−kt (5.14c)
Conditions OR = const.; reverse rate: zero; [X] = [X]0 at t = 0
Reversible first-order reactions
k1
Reaction X + OR Y + OP (other products) (5.15a)
k2
Rate expression −d[Xl/dt = k1[X] − k2[Y] (5.15b)

−kt
Solution [X] = [X]0(k1 + k2)−1 (k2 + k1 e ) (5.15c)
Conditions k = k1 + k2; OR and OP = const.; at time t = 0, (5.15d)
[Y] = 0 and [X] = [X]0
Series first-order reactions
k1 k2
Reaction X→Y→Z (5.16a)
Rate expression d[Y]/dt = k1[X]0 e − k t − k2 [Y]
1
(5.16b)
Solution [Y] = [X]0 k1 (k2 − k1 ) −1 (e − k t − e − k t )
1 2
(5.16c)
Conditions no Y or Z initially present; [X] = [X]0 at t = 0
the concentration of a reactant X, and the rate of the reverse reaction, rate
(reverse), is always equal to zero. Simple distribution processes between immis-
cible liquid phases of noncharged components characterized by very large val-
ues of the partition constant can be formally treated as first-order irreversible
reactions.
Reversible first-order reactions are given by Eq. (5.15) of Table 5.1.
Equation (5.15c) can be rewritten in a more elegant form by considering that at
equilibrium (t = ∞ ) the net rate is equal to zero, that is
k1[X]eq = k2 [Y]eq (5.17)
where subscript eq indicates equilibrium concentrations. The equilibrium con-
stant of the reaction is
K = k1 / k2 = [Y]eq /[X]eq (5.18)
Equation (5.15c) can then be reformulated as
ln{([X] − [X]eq )([X]0 − [X eq ]) −1} = −(k1 + k2 )t (5.19)
Equation (5.19) shows that the rate law is still first-order, provided the
quantity ([X] − [X]eq) is used instead of [X]. A plot of ln([X] − [X]eq) vs. t will
then be a straight line of slope -(k1 + k2). The individual rate constants of the
reaction can still be evaluated from the slope of such a plot, providing the
212 Danesi
equilibrium constant [Eq. (5.18)] is also available. Many distribution processes

between immiscible liquid phases of noncharged species, as well as extraction
of metal ions performed at very low metal concentrations, can be treated as
first-order reversible reactions when the value of the equilibrium (partition) con-
stant is not very high.
Equations (5.16) of Table 5.1 refer to series first-order reactions. Of inter-
est for the solvent extraction kinetics is a special case arising when the concen-
tration of the intermediate, [Y], may be considered essentially constant (i.e.,
d[Y]/dt = 0). This approximation, called the stationary state or steady-state ap-
proximation, is particularly good when the intermediate is very reactive and
present at very small concentrations. This situation is often met when the inter-
mediate [Y] is an interfacially adsorbed species. One then obtains
[Y] = [X]0e − k1 t k1 / k2 (5.20)
and
[ Z] = [X]0{1 − (1 + k1 / k2 )e − k1 t } (5.21)
k2 Ⰷ k1 means a very reactive intermediate present at very low concentra-
tion; t Ⰷ 1/k2 indicates that the induction period of the reaction has been passed,
that is, the initial phase of the reaction, when the rate of formation of Z is slow,
is immeasurably short in comparison with the time response of the measuring
technique.
5.4 SOME KINETICS AND MECHANISMS

WITH COMPLEX IONS
5.4.1 Introduction
In the solvent extraction of metal ions, the composition of the coordination
sphere often changes, either because of the formation of complexes between the
metal ions and a complexing reagent, preferentially soluble in an organic phase,
or because of the replacement of a ligand in the aqueous phase metal complex
with another more lipophilic one in the organic phase. Complexes that rather
readily exchange ligands or water molecules with other ligands are termed la-
bile. Inert complexes do not exhibit ligand exchange or do so very slowly. The
terms labile and inert refer to the velocity of the reactions and should not be
confused with the terms stable and unstable, which refer to the thermodynamic
tendency of species to exist under equilibrium conditions. Although, in many
cases, thermodynamic stability parallels the inertness of a complex, there are
many examples of complex ions that are thermodynamically stable, but kinetic-
ally labile, or the reverse.
Kinetic studies of ligand exchange reactions seek to elucidate the mecha-
nisms of such reactions. In the following section, information on the mech-

anisms that control solvent exchange and complex formation is reported. Al-
though the information has been derived from studies of one-phase liquid
systems, much of it can be safely extrapolated to solvent extraction systems
whenever interfacial film diffusion processes can be neglected as the rate-
determining processes.
5.4.2 Mechanisms
Solvent exchange and complex formation are special cases of nucleophilic sub-
stitution reactions.
Generally, when a coordination bond is broken in a metal ion complex,
the electron pair responsible for the bond accompanies the leaving group (com-
monly, the organic ligand). It follows that reactions involving solvent exchange
and complex formation have as reactive centers electron-deficient groups. There-
fore, they can all be considered as nucleophilic substitutions (in symbols SN).
The basic classification of nucleophilic substitutions is founded on the
consideration that when a new metal complex is formed through the breaking
of a coordination bond with the first ligand (or water) and the formation of a
new coordination bond with the second ligand, the rupture and formation of the
two bonds can occur to a greater or lesser extent in a synchronous manner.
When the rupture and the formation of the bonds occur in a synchronous way,
the mechanism is called substitution nucleophilic bimolecular (in symbols SN2).
On the other extreme, when the rupture of the first bond precedes the formation
of the new one, the mechanism is called substitution nucleophilic unimolecular
(in symbols SN1). Mechanisms SN2 and SN1 are only limiting cases, and an
entire range of intermediate situations exists.
Inorganic chemists prefer a slightly more detailed classification and subdi-
vide the range of possible mechanisms into the following four groups.
1. D mechanism. This is the limiting dissociative mechanism, and a transient
inter-mediate of reduced coordination number is formed. The intermediate
persists long enough to discriminate between potential nucleophiles in its
vicinity. Here we are dealing with an SN1 limiting process, since the dissoci-
ation of the metal-ligand bond fully anticipates the formation of the new
bond. The vacancy in the coordination shell that occurs as a result of the
dissociation is then taken by the new ligand.
2. Id mechanism. This is the dissociative interchange mechanism and is similar
to the previous one in the sense that dissociation is still the major rate-
controlling factor. Therefore, we are still dealing with an SN1 process. Nev-
ertheless, differently from the D mechanism, no experimental proof exists
that an intermediate of lower coordination is formed. The mechanism in-
volves a fast outer sphere association between the initial complex and the
214 Danesi
incoming nucleophile (the new ligand), and when the complex metal ion
and the incoming ligand have opposite charges, this outer sphere associate
is an ion pair. As a result of the formation of this new outer sphere associ-
ate, the new group is now suitably placed to enter the primary coordination
sphere of the metal ion as soon as the outgoing group has left.
3. Ia mechanism. This is the associative interchange mechanism and is similar
to the Id mechanism, in the sense that here, also, no proof exists that an
intermediate of different coordination is formed. However, differently from
Id, we have significant interaction between the incoming group and the
metal ion in the transition state. In this instance, the process is partially of
the SN2 type.
4. A mechanism. This is the pure associative mechanism, and here we are
dealing with a process entirely of the SN2 type. The rate-determining step
of the mechanism is the association between the complex metal ion and the
entering ligand, leading to the formation of an intermediate of increased
coordination number that can be experimentally identified. This mechanism
is often operative in ligand-displacement reactions occurring with planar
tetracoordinated complexes.
In practice, there is a continuous gradation of mechanisms from D,
through Id and Ia to A, depending on the extent of interaction between the metal
cation and the incoming group in the transition intermediate. Moreover, the
diagnosis of the mechanism is not a straightforward and unambiguous process.
A variety of methods have to be used, depending on circumstances, since an
approach based only on rate-law consideration can easily lead to false conclu-
sions.
5.4.3 Water Exchange and Complex Formation

from Aquoions
The rate at which solvent molecules are exchanged between the primary solva-
tion shell of a cation and the bulk solvent is of primary importance in the kinet-
ics of complex formation from aquocations. In both water exchange and com-
plex formation, a solvent molecule in the solvated cation is replaced with a new
molecule (another water molecule or a ligand). Therefore, strong correlations
exist between the kinetics and mechanisms of the two types of reactions.
The basic mechanism for the formation of complexes from aquocations
[9] is based on the following observations:
For a given metal ion the rates and the activation parameters for complex
formation are similar to those for water exchange, with the complex formation
rate constants usually about a factor of 10 lower than those for water exchange,
For a given metal ion the rates show little or no dependence on the
identity of the ligand.
This means that, at least as a first approximation, the complex formation

mechanism can be described by the rapid equilibrium formation of an ion pair
(for oppositely charged species) or of an outer sphere association complex be-
tween the aquometal ion and the ligand, followed by a rate-determining inter-
change step (D or Id mechanism). Here, the ligand bonds to the complex cation
only after a water molecule has been released from the primary coordination
sphere. For a six-coordinated single-charged cation, the mechanism can be ex-
pressed as
M(H 2O)6+ + L− → (L− , M(H 2O)6+ ) → kM(H 2O)5 + H 2O
fast outer sphere slow
associate (5.22)
The rate law corresponding to such a mechanism is:
rate = {K ass k[M + ][L− ]}/{1 + K ass [L− ]} (5.23)
In Eqs. (5.22–5.23), coordination water molecules are omitted and Kass
indicates the ion pair (or outer sphere association) constant. In general Kass is a
small number (≤10) and Eq. (5.23) simplifies to
rate = K ass k[M + ][L− ] (5.24)
Rate Eq. (5.23) indicates that the process is second-order, even if the rate-
controlling step is unimolecular. Furthermore, since the constant Kass differs only
slightly from species to species, depending mostly on the charge density, only
a small dependence on the nature of the incoming ligand can be expected.
Once the value of Kass is estimated, either from theoretical considerations
or from separate experimental measurements, the rate constants of the water
exchange rate-controlling step can be calculated. The values of k are in good
agreement with the independently evaluated rate constants of the water ex-
change process:
M(H 2O)6+ + H 2O* → M(H 2O)*(H 2O)5 +H 2O (5.25)
In Eq. (5.25) H2O* represents a water molecule initially present outside the
coordination sphere of the metal ion, which, as a result of the exchange, has
entered the first coordination sphere. It follows that the degree of kinetic reactiv-
ity of aquometal ions with complexing agents parallels their kinetic lability to-
ward water exchange. Moreover, since the water exchange rate constants of
most metal ions are known, predictions on the rate of complex formation of
aquometal ions can be made.
Rates of water exchange at metal ions vary tremendously with the nature
of the cation. The range is from almost 1010 s−1, for Cu2+ or Cr2+, down to less
than 10−7 s−1 for Rh3+. The values for many metal aquocations are shown in Fig.
216 Danesi
5.3 on a logarithmic scale. They have been obtained by nuclear magnetic reso-
nance (NMR) and sound absorption, for the rapidly exchanging ions, and by
isotopic labeling for the more inert ones.
Generally, in solvent extraction, for the simplest case of an organic com-
plexing reagent (ligand) reacting with an aquometal ion, when the water ex-
change rate constant is <102 s−1, the reaction can be considered slow enough that
the complex formation rate may compete with the rate of diffusion through the
interfacial films in controlling the overall extraction kinetics. In exceptional
cases, dealing with very efficiently stirred phases (in which the thickness of the
diffusion films may even be reduced to about 10−4 cm), complexation reactions
with rate constants as high as 106 s−1 can be rate limiting in solvent extraction.
In all other cases, the ions can be considered as reacting instantaneously relative
to the rate of film diffusion. The rate of diffusion also is the predominating rate-
controlling process when the aquocations are extracted into the organic phase
with their entire coordination sphere of water molecules. These considerations
do not hold when other slow chemical processes, such as hydrolysis reactions
of the metal cation, polymerizations in the organic phase of the extractant or
the metal complex, adsorption-desorption processes at the interphase, keto-enol
Fig. 5.3 Logarithm of water exchange constants (s−1) for aquometal ions at 25°C.
(From Ref. 26.)
tautomerizations, or others, control the chemistry of the solvent extraction sys-

tem. In these cases, it is possible that the slow chemical control predominates
over the faster diffusional process. In such systems, solvent extraction rate data
can still provide information on the rate and mechanism of the chemical pro-
cesses.
5.4.4 Ligand Displacement Reactions

For ligand displacement reactions occurring in octahedral complexes, very few
generalizations can be made, since the reaction mechanisms tend to be specific
to each chemical system. This is unfortunate because many solvent extraction
processes are indeed ligand displacement reactions of octahedral complexes in
which L is the extractant [see Eq. (5.22)]. However, it has been experimentally
observed, at least for aqueous-phase reactions, that the variation of the rates with
the identity of the ligand correlates well with the variation in the thermodynamic
stability of the complex. Therefore, whenever the complex is not extracted un-
changed into the organic phase, thermodynamically very stable complexes can
be expected to react slowly with the extractant. An example is the extraction of
trivalent lanthanide or actinide cations from aqueous solutions by diethylhexyl
phosphoric acid (HDEHP). If the aqueous medium contains only weakly com-
plexing ligands (Cl− or NO−3), the extraction kinetics are very fast. On the other
hand, in the presence of polyaminocarboxylic acids, such as EDTA that are
powerful complexing agents, the extraction reaction proceeds only slowly.
For planar tetracoordinated complexes, the mechanistic problem is more
straightforward. A large amount of experimental evidence on ligand substitution
reactions for complexes of transition metals with electronic configuration d8,
such as Pt(II), Pd(II), Au(III), Ni(II), and Rh(I), has been accumulated. The
transition state is a five-coordinated species having the structure of a trigonal
bipyramid on which the entering group lays on the trigonal plane. Ligand-substi-
tution reactions with planar tetracoordinated complexes are often slow in com-
parison with the rate of diffusion through the interfacial diffusional films. It
follows that the kinetics of solvent extraction of planar tetracoordinate com-
plexes in a well-stirred system are often entirely controlled by the rate of the
chemical process. Studies on the rate of extraction can then be directly used
to obtain information on the rate of the biphasic ligand-substitution reactions,
neglecting the complications from diffusional contributions.
5.5 THE LIQUID–LIQUID INTERFACE

As mentioned in section 5. 1, when the two immiscible aqueous and organic
phases are in contact and transfer of chemical species occurs from one phase to
the other, the diffusional transport in proximity of the interface can be described
218 Danesi
by assuming the presence on either side of the interface of two stagnant films
(the diffusional films). Although these films are thin when compared with the
bulk phases, their thickness is still macroscopic (about 10−3 or 10−4 cm) in com-
parison with the molecular dimensions and the range of action of the molecular
and ionic forces. Thus, these films have a finite volume, and the concentrations
of chemical species within them must always be considered as volume concen-
trations. Moreover, the physicochemical properties of the liquids within these
films, such as density, viscosity, dielectric constant, charge distribution, are the
same as those of the bulk phases. However, the situation changes when the
distance from the interface approaches the order of magnitude of the molecular
dimensions and the range of action of the molecular and ionic forces. In this
region, generally defined as the interface, the physicochemical properties differ
from those of the bulk phases and, therefore, any chemical change occurring
herein is likely to be affected by the different environment. The difference will
be particularly enhanced by the presence on the interface of adsorbed layers of
polar or ionizable extractant molecules. The extractant, because of its simultane-
ous hydrophobic-hydrophilic (low water affinity–high water affinity) nature,
necessary to maintain a high solubility in low dielectric constant diluents, and
selective complexing power relative to water-soluble species, tends to orient
itself with the polar (or ionizable) groups facing the aqueous side of the inter-
face. The rest of the molecule, having a prevalent hydrophobic character, will
be directed instead toward the organic phase.
Unfortunately, little direct information is available on the physicochemical
properties of the interface, since real interfacial properties (dielectric constant,
viscosity, density, charge distribution) are difficult to measure, and the interpre-
tation of the limited results so far available on systems relevant to solvent ex-
traction are open to discussion. Interfacial tension measurements are, in this
respect, an exception and can be easily performed by several standard physico-
chemical techniques. Specialized treatises on surface chemistry provide an ex-
haustive description of the interfacial phenomena [10,11]. The interfacial ten-
sion, γ, is defined as that force per unit length that is required to increase the
contact surface of two immiscible liquids by 1 cm2. Its units, in the CGS system,
are dyne per centimeter (dyne cm−1). Adsorption of extractant molecules at the
interface lowers the interfacial tension and makes it easier to disperse one phase
into the other.
Interfacial tension studies are particularly important because they can pro-
vide useful information on the interfacial concentration of the extractant. The
simultaneous hydrophobic-hydrophilic nature of extracting reagents has the re-
sulting effect of maximizing the reagent affinity for the interfacial zone, at
which both the hydrophobic and hydrophilic parts of the molecules can mini-
mize their free energy of solution. Moreover, as previously mentioned, a prefer-
ential orientation of the extractant groups takes place at the interface. Conse-
quently, most solvent extraction reagents are interfacially adsorbed and produce
a lowering of the organic-aqueous interfacial tension. The extent of this adsorp-
tion is a function of the chemical nature and structure of the hydrophilic and
hydrophobic groups, of the extractant bulk concentration, and of the physico-
chemical properties of the diluent. The extractant interfacial concentration can
be extremely high, depending on the way molecules can pack themselves at the
interface, and even moderately strong surfactants can form an interface fully
covered with extractant molecules when their bulk concentration is as low as
10−3 M or less. It follows that, when an extractant is a strong surfactant and
exhibits low solubility in the water phase, the interfacial zone is a region where
a high probability exists that the reaction between an aqueous soluble species
and an organic soluble extracting reagent can take place. This is why a strong
surface activity of an extractant can sometimes lend support to extraction mech-
anisms involving interfacial chemical reactions.
Interfacial concentrations can be evaluated from interfacial tension mea-
surements by utilizing the Gibbs equation
d ∏ / d ln c = nikT (5.26)
In Eq. (5.26), Π is the interfacial pressure of the aqueous-organic system,
equal to (γ0 − γ) [i.e., to the difference between the interfacial tensions without
the extractant (γ0) and the extractant at concentration c (γ)], c is the bulk organic
concentration of the extractant, and ni is the number of adsorbed molecules of
the extractant at the interface. The shape of a typical Π vs. ln c curve is shown
in Fig. 5.4; ni can be evaluated from the value of the slopes of the curve at each
c. However, great care must be exercised when evaluating interfacial concentra-
tions from the slopes of the curves because Eq. (5.26) is only an ideal law, and
many systems do not conform to this ideal behavior, even when the solutions
are very dilute. Here, the proportionality constant between dΠ/d ln c and ni is
different from kT. Nevertheless, Eq. (5.26) can still be used to derive infor-
mation on the bulk organic concentration necessary to achieve an interface
completely saturated with extractant molecules (i.e., a constant interfacial con-
centration). According to Eq. (5.26), the occurrence of a constant interfacial
concentration is indicated by a constant slope in a Π vs. ln c plot. Therefore,
the value of c at which the plot Π vs. ln c becomes rectilinear can be taken as
the bulk concentration of the extractant required to fully saturate the interface.
Many extractants reach a constant interfacial concentration at bulk organic
concentrations far below the practical concentrations that are generally used to
perform extraction kinetic studies. This means that when writing a rate law for
an extraction mechanism that is based on interfacial chemical reactions, the
interfacial concentrations can often be incorporated into the apparent rate con-
stants. This leads to simplifications in the rate laws and to ambiguities in their
interpretation, which are discussed in later sections.
220 Danesi
Fig. 5.4 Typical interfacial pressure (Π) vs. logarithm of bulk organic concentration
(log c) plotted for an extractant exhibiting surface-active properties.
It is interesting to observe that alkylammonium salts, alkylarylsulfonic

acids, hydroxyoximes, alkylphosphoric acids, and alkylhydroxamic acids, as
well as neutral extractants such as crown ethers and tributyl phosphate, all form
water-organic interfaces saturated with extractant molecules when their bulk
organic concentration is larger than 10−3 M.
The liquid–liquid interface is not a sharply defined surface, but rather an
ill-defined region the boundaries of which extend above and below the layer of
the interfacially adsorbed extractant molecules. Nevertheless, it is often conve-
nient to assume a mathematical dividing surface located where the physico-
chemical properties of the two-phase system experience the sharpest discontinu-
ity. This imaginary surface of zero thickness, which we refer to as the interface,
is useful for defining the concentrations of interfacially adsorbed species, which
can be expressed in moles per square centimeter (mol cm−2). On the other hand,
when in later sections we refer to interfacial concentrations, we mean volume

concentrations in that region which is immediately adjacent to this imaginary
dividing surface (i.e., at the extreme limit of the diffusional films).
The physical depth of the microscopic interfacial region can be estimated
to correspond to the distance over which interfacial molecular and ionic forces
exert their influence. Although molecules or ions experience no net forces in
the interior of the bulk phases, these forces become unbalanced as the ions or
molecules move toward the interface. On the aqueous side of the interfaces,
where monolayers of charged molecules or polar groups can be present, these
forces can be felt over several nanometers. On the organic side of the interface,
where van der Waals forces are mainly operative, the interfacial region generally
extends for tens of nanometers. The van der Waals forces decrease with the
seventh power of the intermolecular distances, so molecules experience essen-
tially symmetric forces, once they are a few molecular diameters away from the
interface.
Although an experimentally verified molecular model of the interface is
not yet available, a qualitative description of the interfacial zone can be at-
tempted on the basis of speculations that extrapolate to the liquid–liquid inter-
face results that are predicted from existing and verified physicochemical theo-
ries. Considering that extracting reagents can be present at the interface with
their polar or ionizable groups facing the water side, a more structured water
can be expected in the interfacial zone. The polar heads of the extractant tend
to polarize the water molecules close to the interface, producing a more struc-
tured and viscous interfacial water, as depicted in Fig. 5.5. This interfacial water,
having the possible depth of several molecular layers, is the result of the water
penetration of the hydrophilic groups of the extractant, which hydrogen bond
with the water molecules close to the interface. As a consequence, interfacial
water should represent a more viscous environment to diffusing species than
ordinary bulk water. The aqueous side of the interface can then be envisioned, at
the microscopic level, as a very particular medium that still retains the chemical
properties of bulk water, but for which the physical properties are much more
like those of a strongly structured solvent, such as glycerol [12]. This interfacial
zone should be characterized by lower fluidity, lower dielectric constant, higher
stiffness, and be more compact than bulk water. The extent of these effects is a
function of the surface-active properties of the extractant for a given organic
diluent–aqueous electrolyte solution system. They can slow ligand exchange
reactions, with octahedral aquoions occurring at the interface, since the more
viscous interfacial environment slows the associative component of the overall
interchange mechanism that controls this type of reactions. The associative step
is, by contrast, very fast and, therefore, not rate controlling for most ligand-
substitution reactions of octahedral aquoions occurring in bulk water.
222 Danesi
Fig. 5.5 Simplified picture of the interfacial structuring effect caused by an ionizable
extractant adsorbed at the organic-water interface.
5.6 RATE-CONTROLLING EXTRACTION REGIMES

5.6.1 Definition of the Extraction Regimes
Section 5.1 describes how, in a stirred system, solvent extraction kinetics can
be controlled only by slow chemical reactions or only by diffusion through the
interfacial films. An intermediate situation can also occur whereby both the rates
of chemical reactions and of interfacial film diffusion simultaneously control

the rate of extraction.
When one or more of the chemical reactions is sufficiently slow in com-
parison with the rate of diffusion to and away from the interface of the various
species taking part in an extraction reaction, such that diffusion can be consid-
ered instantaneous, the solvent extraction kinetics occur in a kinetic regime. In
this case, the extraction rate can be entirely described in terms of chemical
reactions. This situation may occur either when the system is very efficiently
stirred and when one or more of the chemical reactions proceeds slowly, or
when the chemical reactions are moderately fast, but the diffusion coefficients
of the transported species are very high and the thickness of the two diffusion
films is close to zero. In practice the latter situation never occurs, as diffusion
coefficients in liquids generally do not exceed 10−4 cm2 s−1, and the depth of the
diffusion films apparently is never less than 10−4 cm.
If, however, all chemical reactions in the biphasic system are very fast in
comparison with the diffusional processes, the solvent extraction kinetics are
said to occur in a diffusional regime. In this situation, the rate of extraction can
be described simply in terms of interfacial film diffusion. Chemical engineers
define this as “mass transfer with instantaneous chemical reaction.” A diffu-
sional regime may control the extraction rate also in the presence of relatively
slow chemical reactions whenever the degree of stirring of the two phases is so
poor that thick diffusional films can be assumed on the two sides of the inter-
face. When both chemical reactions and film diffusion processes occur at rates
that are comparable, the solvent extraction kinetics are said to take place in a
mixed diffusional-kinetic regime, which in engineering, is often referred to as
“mass transfer with slow chemical reactions.” This is the most complicated case,
since the rate of extraction must be described in terms of both diffusional pro-
cesses and chemical reactions, and a complete mathematical description can be
obtained only by simultaneously solving the differential equations of diffusion
and those of chemical kinetics.
The unambiguous identification of the extraction rate regime (diffusional,
kinetic, or mixed) is difficult from both the experimental and theoretical viewpoints
[12,13]. Experimental difficulties exist because a large set of different experimental
information, obtained in self-consistent conditions and over a very broad range of
several chemical and physical variables, is needed. Unless simplifying assump-
tions can be used, frequently the differential equations have no analytical solu-
tions, and boundary conditions have to be determined by specific experiments.
5.6.2 Identification of the Extraction Regimes

The experimental identification of the regime that controls the extraction kinet-
ics is, in general, a problem that cannot be solved by reference to only one set
224 Danesi
of measurements. On the contrary, in some systems, a definite situation cannot

be obtained even when the rate of extraction is studied as a function of both
hydrodynamic parameters (viscosity and density of the liquids, geometry of the
apparatus, geometry of stirrers, or stirring rate) and concentrations of the chemi-
cal species involved in the extraction reaction. This difficulty is because some-
times extraction rates may show the same dependence on hydrodynamic and
concentration parameters, even though the processes responsible for the rate are
quite different. For a correct hypothesis on the type of regime that controls the
extraction kinetics, it is necessary to supplement kinetic investigations with
other information concerning the biphasic system. This may be the interfacial
tension, the solubility of the extractant in the aqueous phase, the composition
of the species in solution, and so forth.
The criteria that are most often used to distinguish between a diffusional
regime and a kinetic regime are
1. Comparison of the heat transfer and the mass transfer coefficients
2. Evaluation of the activation energy of extraction
3. The reference substance method
4. Dependency of the rate of extraction on the rate of stirring of the two
phases.
Criteria 1–3 are difficult to apply and have been used only occasionally
to evaluate the extraction regime.
1. According to this criterion, when, in the same apparatus, the same depen-
dence of the heat transfer coefficient and the mass transfer coefficient on
the stirring rate of the phases is observed, the conclusion can be reached
that the extraction occurs in a diffusional regime.
2. The rationale behind the use of this method is that, when the rate is con-
trolled by a chemical reaction, the activation energy is generally higher than
that expected for a diffusion-controlled process (diffusion coefficients vary
only slightly with temperature). However, this criterion is not always very
meaningful, since many chemical reactions occurring in solvent extraction
processes exhibit activation energies of only a few kilocalories per mole,
i.e., have the same order of magnitude as those of diffusional processes.
3. The reference substance method is based on the addition to the solution,
containing the species for which the transfer rate is going to be investigated,
of another inert component for which the rate of extraction is known to be
controlled only by diffusion. By following the simultaneous transfer of the
species of interest and of the reference component as function of the hydro-
dynamic conditions in the extraction apparatus, a diffusional regime will be
indicated by a similar functional dependence, whereas a kinetic regime is
indicated by a sharply different one.
4. This criterion is the simplest, as proved by its widespread use. A typical
curve of rate of extraction vs. stirring rate is shown in Fig. 5.6. Such curves
are generally obtained both when constant interfacial-area-stirred cells
(Lewis cells) and vigorously mixed flasks are used (see section 5.10 for a
description of the various techniques). In general, a process occurring under
the influence of diffusional contributions is characterized by an increase of
the rate of extraction, as long as the stirring rate of the two phases is
increased (see Fig. 5.6, zone A). When, on the other hand, the rate of ex-
traction is independent of the stirring rate (see Fig. 5.6, zone B), it is some-
times possible to assume that the extraction process occurs in a kinetic
regime.
The rationale behind this criterion is that an increase in stirring rate produces
a decrease in thickness of the diffusion films. Moreover, since at low stirring
rates the relationship between stirring rate and the inverse thickness of the
diffusion films (at least in a constant interfacial-area-stirred cell) is approxi-
mately linear, the first part of such a plot will usually approximate a straight
line. In any case, the rate of extraction will increase with the rate of stirring,
as long as a process is totally or partially diffusion controlled. When, eventu-
ally, the thickness of the diffusion films is reduced to zero, only chemical
reactions can be rate controlling, and the rate of extraction becomes indepen-
dent of the stirring rate. Unfortunately, this kind of reasoning can lead to erro-
Fig. 5.6 Typical extraction rate vs. stirring rate for constant interfacial–area-stirred cell.
226 Danesi
neous conclusions with both constant interfacial-area-stirred cells and vigor-

ously stirred flasks.
In the case of constant interfacial–area-stirred cells, although zone A is
certainly an indication that the process is controlled by diffusional processes,
the opposite is not true for zone B. In fact, in spite of the increased stirring rate,
it may happen that the thickness of the diffusion films never decreases below
a sufficiently low value to make diffusion so fast that it can be completely
neglected relative to the rate of the chemical reactions. This effect, some-
times called “slip effect,” depends on the specific hydrodynamic conditions
of the apparatus in which the extraction takes place and simulates a kinetic re-
gime.
As far as vigorously stirred vessels are concerned, strong ambiguities are also
faced when this criterion is applied. Unless the overall interfacial area is known
and taken into consideration, an increase in the rate of extraction can take place
with the increase in stirring rate when the system is in a kinetic regime deter-
mined by interfacial chemical reactions. The increase observed in zone A indi-
cates here an increase in the number of droplets of the dispersed phase (propor-
tional to the overall interfacial area) and not a decrease in the thickness of the
diffusion film. Moreover, the plateau region of zone B does not necessarily
prove that the extraction occurs in a kinetic regime. Although at high stirring
rates the number of droplets of the dispersed phase eventually becomes con-
stant, since the rate of drop formation equals the rate of drop coalescence, lack
of internal circulation and poor mixing can occur inside the droplets of the
dispersed phase. This is particularly true with small droplets and in the pres-
ence of extractants that are strong surfactants. Therefore, here also, a plateau
region may simulate a diffusion-controlled regime. It is then apparent that crite-
rion 4 can also lead to ambiguous or misleading conclusions.
Finally, it has to be emphasized that both the hydrodynamic parameters
and the concentrations of the species involved in the extraction reaction simulta-
neously determine whether the extraction regime is of kinetic, diffusional, or
mixed diffusional-kinetic type. It, therefore, is not surprising that different in-
vestigators, who studied the same chemical solvent extraction system in differ-
ent hydrodynamic and concentration conditions, may have interpreted their re-
sults in terms of completely different extraction regimes.
5.7 KINETIC REGIME

In a kinetic regime system, the kinetics of solvent extraction can be described
in terms of chemical reactions occurring in the bulk phases or at the interface.
The number of possible mechanisms is, in principle, very large, and only the
specific chemical composition of the system determines the controlling mecha-
nism. Nevertheless, some generalizations are possible on considerations based
on the nature of the reagents involved in the extraction reaction. Since, in many
metal extractions, ligand substitution reactions take place, rate laws similar to
those for complexation reactions in solution may be expected. During most ex-
traction processes, coordinated water molecules or ligands are substituted in part
or wholly by molecules of a more organophilic ligand (the extractant) or of the
organic diluent. When the extractant shows little solubility in the aqueous phase
and is a strong surfactant, the ligand substitution reaction may take place at the
interface. The highest concentration of extractant molecules is at the interface,
so the probability of reaction is higher than in the bulk phases.
In this section, we describe three simple cases of rates and mechanisms
that have been found suitable for the interpretation of extraction kinetic pro-
cesses in kinetic regimes. These simple cases deal with the extraction reaction
of a monovalent metal cation M+ (solvation water molecules are omitted in the
notation) with a weakly acidic solvent extraction reagent, BH. The overall ex-
traction reaction is
M + + BH(org) → MB(org) + H + (5.27)
with equilibrium extraction constant, Kex equal to
K ex = [MB]eq [H + ]eq /[M + ]eq [BH]eq (5.28)
Ideal behavior of all solute species will be assumed.
Case 1: The rate-determining step of the extraction reaction is the aque-
ous phase complex formation between the metal ion and the anion of the extract-
ing reagent. Even if at very low concentration, BH will always be present in
the aqueous phase because of its solubility in water. The rate-determining step
of the extraction reaction is as follows:
k1
M + + B− ↔ MB (slow) (5.29)
k−1
with a reaction rate expression:

rate = −d [M + ]/ dt = k1[M + ][B− ] − k−1[MB] (5.30)
To derive an expression for the rate in terms of easily measurable quanti-
ties (initial concentrations), the following (instantaneously established) equilib-
ria have to be included (cf. Chap. 4):
1. BH ↔ BH; K DB = [BH]/ [BH] (5.31)
−
2. BH ↔ B +H ; K a = [B ][H ]/[BH]
+ + +
(5.32)
3. MB ↔ MB(org); K DM = [MB]/[MB] (5.33)
By substituting Eqs. (5.31–5.33) into Eq. (5.30), we obtain the rate ex-
pression
228 Danesi
−1
rate = k1K a / K DB [M + ][BH][H + ]−1 − k−1K DM
−1
[MB] (5.34)
In the right-hand side of Eq. (5.34), the first term represents the forward
rate of extraction and the second term the reverse rate of extraction.
The values of the apparent rate constants of the extraction reaction,
−1 −1
k1K a K DB and k−1K DM (5.35)
permit the evaluation of the rate constants of the aqueous complex formation
reaction only if KDB, KDM, and Ka are known. Here, comparisons can be made
with literature values of the rate constants for reaction in aqueous solutions
between the same metal ion and aqueous soluble ligands containing the same
complexing groups as the extracting reagent. Agreement between the values
supports, but does not prove, the validity of the proposed mechanism.
The apparent values of the rate constants of the solvent extraction reaction
are usually evaluated by measuring the rate of extraction of M+ as function of
[BH] (at [H+] constant), of [H+] (at [BH] constant), and of [MB] (at [H+] and
[BH] constant). The experimental conditions are usually chosen in such a way
that the reaction can be assumed pseudo-first-order for [M+]. The apparent rate
constants are evaluated from the slope of the straight lines obtained by plotting
ln ([M + ] − [M + ]eq )/([M + ]0 − [M + ]eq ) vs. t (5.36)
Subscript 0 indicates initial concentrations in the aqueous phase.
It must be observed that when this mechanism holds, the rate of the extrac-
tion reaction is independent of the interfacial area, Q, and the volume of the
phases, V. The expected logarithmic dependency of the forward rate of extrac-
tion on the specific interfacial area (S = Q/V), the organic concentration of the
extracting reagent and the aqueous acidity, is shown in Fig. 5.7, case 1.
Case 2: The rate-determining step of the extraction reaction is the inter-
facial formation of the complex between the metal ion and the interfacially
adsorbed extracting reagent. Here, the rate-determining step of the extraction
reaction can be written as
k2
M + + B− (ad) ↔ MB(ad) (slow) (5.37)
k−2
where (ad) indicates species adsorbed at the liquid–liquid interface. The rate of
the reaction that follows is
rate = −d [M + ]/ dt = k2 S w [M + ][B− ]ad − k2 So [MB]ad (5.38)
where Sw and So represent the specific interfacial areas [i.e., the ratios between
the interfacial area Q (cm2) and the volumes of the aqueous (Vw) and organic
(Vo) phase, respectively]. Throughout the following treatments we will always
assume, for simplicity, that Vw = Vo = V, and
Fig. 5.7 Logarithm of the forward extraction rate (normalized to the extractable metal
concentration) vs. log interfacial area (a), log [BH], and log [H+] for some exemplary
kinetic regimes.
230 Danesi
Q / V = S = SW = So (5.39)
To derive a rate of extraction in terms of easily measurable quantities, we
express the concentration of the reagent adsorbed at the liquid–liquid interface
as function of its bulk organic concentration. This can be done by utilizing the
Langmuir’s adsorption law: that is,
[BH]ad = (α 2 [BH]/ γ 2 ) /(1 + [BH]/ γ 2 ) (5.40)
In Eq. (5.40), α2 and γ2 are Langmuir adsorption constants. Their values
are characteristic of each specific system. We can then distinguish two regions
of adsorption. One defined as ideal, where 1 Ⰷ [BH]/ γ2, yielding
[BH]ad = (α 2 [BH]/ γ 2 ) = α[BH] (5.41)
and one of complete interface saturation with extractant molecules, where 1 Ⰷ

[BH]/ γ2, yielding
[BH]ad = α 2 (5.42)
With extractants exhibiting strong surface active properties, Eq. (5.42)
holds true in the entire concentration range, which is generally of interest in
practical studies, and the interface becomes fully saturated with extractant mole-
cules when their bulk concentration is as low as 10−3 M.
The following, instantaneously established, equilibria then have to be con-
sidered:
1. BH(org) ↔ BH(ad) (5.43)
adsorption of the extractant molecules at the interface described by Langmuir’s
law [see Eq. (5.40)]
2. BH(ad) ↔ B− (ad ) + H + ; K a* = [B− ]ad [H + ][BH]ad
−1
(5.44)
interfacial dissociation of the extractant (involving no interfacial adsorption of
the dissociated protons)
3. MB(ad) + BH(org) ↔ MB(org)+BH(ad);
K ex0 = ([MB][BH]ad ) /([MB][BH] (5.45)
fast replacement at the interface of the interfacially adsorbed metal complex
with bulk molecules of the extractant. Equations (5.40), (5.44), (5.45) can be
inserted into Eq. (5.38) to obtain:
rate = S k2α 2 K a*[M + ][H + ]−1 − S k−2 K 0 ex −1 α 2 [MB]−1[BH] (5.46)
which holds for a fully saturated interface, and
rate = S k2 α K a*[M + ][BH][H + ]−1 − S k−2 K 0 ex −1 α[MB] (5.47)
which holds for ideal adsorption.

The equilibrium constant of the extraction reaction [see Eq. (5.28)] is
equal to
K ex = k2 K a* K ex0 k−−21 (5.48)
for both ideal adsorption and for a saturated interface. The expression for Kex
is derived from either Eq. (5.46) or Eq. (5.47), since at equilibrium, rate (for-
ward) = rate (reverse).
Equations (5.34) and (5.47) have the same dependence on the concentra-
tion variables. The only difference is that for the interfacial reaction the apparent
rate constants are directly proportional to the ratio S = Q/V. This dependency
can be experimentally verified by measuring the apparent rate of extraction as
a function of the interfacial area and the volume of the phases. A plot of the
apparent rate constant of the forward rate of extraction vs. S must yield a straight
line through the origin of the axes when case 2 holds.
Equation (5.46) shows a further difference relative to Eq. (5.34). Equation
(5.46) indicates a zero reaction order relative to [BH] in the forward extraction
rate, reflecting the complete saturation of the interface with the extracting re-
agent.
The determination of the true rate constants, k1 and k−2, is generally more
difficult than for case 1, since information on α, α2, K*a and K0ex can be difficult
to obtain.
Case 3: There are two interfacial rate-determining steps, consisting of:
(1) formation of an interfacial complex between the interfacially adsorbed mole-
cules of the extractant and the metal ion; and (2) transfer of the interfacial
complex from the interface to the bulk organic phase and simultaneous replace-
ment of the interfacial vacancy with bulk organic molecules of the extractant.
For this mechanism, we distinguish two possibilities. The first (case 3.1) de-
scribes the reaction with the dissociated anion of the extracting reagent, B−(ad).
The second (case 3.2) describes the reation with the undissociated extractant,
BH(ad).
Case 3.1: The first mechanism is represented by the following equa-
tions:
k3
M + + B− (ad) ↔ MB(ad) (slow) (5.49)
k−3
k4
MB(ad) + BH(org) ↔ BH(ad) + MB(org) (slow) (5.50)
k−4
The rate equations holding for the two slow steps are:
rate 1 = S k3 [M + ][B− ]ad − S k−3 [MB]ad (5.51)
232 Danesi
rate 2 = S k4 [MB]ad [BH] − S k−4 [BH]ad [MB] (5.52)

When the metal concentration is sufficiently low and the concentration of
the interfacially adsorbed metal complex is consequently low, for the stationary
condition at [MB]ad
rate 1 = rate 2 (5.53)
Equation (5.53) can then be solved for [MB], after having introduced the substi-
tution:
[B− ]ad = K a* [BH]ad [H + ]−1 (5.54)
and considering that, for a fully saturated interface [BH]ad = α2. The expression
for [MB]ad found in this way is
[MB]ad = {k3 K a* α 2 [M + ][H + ]−1 + k−4 α 2 [MB]}/{k−3 + k4 [BH]} (5.55)
Substituting the value of [MB]ad into either Eq. (5.51) or Eq. (5.52) gives
rate = {S k3k4 K a*α 2 [BH][M + ][H + ]−1}{k−3 + k4 [BH]}−1
− {S k−3k−4α 2 [MB]}{k−3 + k4 [BH]}−1 (5.56)
An expression for the equilibrium constant of the extraction reaction can be
obtained from Eq. (5.56) by considering that, at equilibrium, one can set rate
(forward) = rate (reverse). Thus
K ex = K 3 k4 K a* k−−31 k−−41 (5.57)
Comparison of Eqs. (5.34), (5.47), and (5.56) shows that if [BH] is suffi-
ciently small to allow introduction of the approximation k4 [BH] Ⰶ k−3 into the
denominator of Eq. (5.56) (i.e., for a low and restricted concentration range of
the extractant), the three rate equations have exactly the same dependence on
[M+], [BH], and [H+], although the extraction mechanisms are characterized by
different rate-determining steps:
1. A homogeneous reaction in the aqueous phase with the anion of the extract-
ant [see Eq. (5.34)].
2. An interfacial reaction with the anion of the extractant, which is ideally
adsorbed at the interface [see Eq. (5.47)].
3. Two sequential interfacial reactions: the first one being the reaction with
the anion of the extractant that saturates the interface; the second one being
the slow desorption of the interfacial complex from the interface [see Eq.
(5.56)].
The difference among these mechanisms can be seen only by measuring
the dependence of the rate of extraction on the specific interfacial area, S, and
by using the broadest possible concentration range of the reactants. If a depen-
dence on S exists, case 1 can be excluded. If in addition the extraction rate is

initially first order and then becomes zero order when [BH] increases, case 2
can also be ruled out.
Case 3.2: When the interfacially reactive species are the undissociated
molecules of the extractant adsorbed at the interface [i.e., the first rate-determin-
ing step of the two-step mechanism is the reaction between the metal ion M+
and HB(ad)], the following equations will hold:
M + + BH(ad) ↔ MB(ad) + H + (slow) (5.58)
MB(ad) + BH(org) ↔ BH(ad) + MB(org) (slow) (5.59)

The rate equations for the slow steps [see Eqs. (5.58) and (5.59)] are
rate 1 = k3* S [M + ][BH]ad − k−*3 S [MB]ad [H + ] (5.60)
rate 2 = k4* S [MB]ad [BH] − k−*4 S [BH]ad [MB] (5.61)

By assuming that the stationary condition, (rate 1 = rate 2), can be applied
and by introducing the same substitutions as those used to derive Eq. (5.56), we
obtain (for a fully saturated interface):
[MB]ad = {α 2 k3* [M + ] − α 2 k−4 [MB]}{k−*3 [H + ] + k4 [BH]}−1 (5.62)
The insertion of Eq. (5.62) into Eq. (5.61) gives
rate = rate (forward) − rate (reverse) =
={S k3*k4α 2 [BH][M + ]}{k−*3[H + ] + k4 [BH]}−1
−{S k−*3k−4α 2 [MB][H + ]}{k−*3 + k4 [BH]}−1 (5.63)
Equation (5.63) has the same dependence on [M+], [BH], and [H+] as Eqs.
(5.34) and (5.47) only when the simplification k4[BH] Ⰶ k−3 [H+] can be intro-
duced in the denominator.
Figure 5.7 shows the logarithmic dependencies expected between the nor-
malized rates of forward extraction, rate (forward)/[M+], and S, [BH] and [H+],
for selected cases.
Although the three described cases represent rates and mechanisms that
have often been used in describing solvent extraction systems that were claimed
to occur in the absence of diffusional contributions, we have to emphasize that
many other mechanisms and rates are possible [12]. However, the three forego-
ing cases are useful in showing the type of ambiguities often faced when trying
to assign a reaction mechanism on the basis of an experimentally determined
solvent extraction rate law. These ambiguities are only partly reduced by study-
ing the extraction rate in a very broad concentration range of the reagents and
by varying S. Much additional information on the species present and on the
234 Danesi
physicochemical properties of the system may be necessary to identify and sup-

port a mechanism. For example, an appreciable water solubility and weak ad-
sorption at the interface of the extracting reagent generally favor cases like 1.
Alternatively, highly water-insoluble and a strongly interfacially adsorbed che-
lating extractant increase the possibility that cases 2 and 3 may describe the
mechanism of extraction.
The simplified chemical system (M+, BH, B−) used here to derive the
equations of cases 1–3, although useful to shown similarities, differences, and
ambiguities that can be encountered when dealing with the rates and mecha-
nisms of solvent extraction processes, only approximately describes the very
complex and diverse chemical systems that have been reported in the literature.
With minor changes and modifications, it is not difficult to derive other equa-
tions suitable for describing many other more complex situations. For example,
with the help of case 1, the equations describing the kinetic of extraction of
divalent metal ions, such as Zn2+, Ni2+ and Co2+, by dithizone and its derivatives
can be derived. Minor modifications also have to be introduced in cases 2 and
3 to obtain the equations that describe the extraction of metal ions, such as Cu2+
and Fe3+, by hydroxyoximes; Fe3+ by hydroxamic acids; and trivalent lanthanide
and actinide cations by alkylphosphoric acids. Other chemical systems in which
neutral or negatively charged metal species are extracted by neutral extractants
or by alkylammonium salts have been described by mechanisms involving aque-
ous-phase complexation or interfacial reactions (one-step or two-step mecha-
nisms) as rate-determining steps. Examples are the extraction of UO2(NO3)2 by
tributyl phosphate (TBP), or FeCl−4 and Pu(NO3)4 by trialkylammonium salts.
The equations reported in cases 1–3 can be easily modified to adapt them to the
specific chemical features of the reagents involved [13]. In most instances, these
modifications involve writing the correct charge for the extracted species (in-
cluding zero and negative charges), eliminating the acid dissociation step of the
extractant whenever this is appropriate, including additional equilibrium equa-
tions, and additional simple rate-determining steps. However, examples of com-
pletely different mechanisms and rates, which in no way can be related to cases
1–3, are abundant in the literature.
5.8 DIFFUSIONAL REGIME

Let us now consider the systems in a diffusional regime. Given the simplified
picture of the interface described by the two-film theory (see Fig. 5.1), we can
now visualize that any species going from one phase to the other, crossing the
liquid–liquid interface, will encounter a total resistance R, which is the sum of
three separate resistances. Two of these resistances are of a diffusional nature
and depend on the fact that diffusion through the stagnant interfacial films can
be a slow process. These resistances are indicated in Fig. 5.1 as Rw and Ro. Rw
includes the diffusional contributions on the water side of the interface, assumed
to occur entirely through the stagnant film of thickness, δw. Ro includes the
contributions on the organic side, where the thickness of the stagnant film is δo.
The resistance associated with the crossing of the interface is indicated as Ri.
As a consequence, the total transfer resistance, R, is
R = Rw + Ri + Ro (5.64)
This equation holds at the steady state. In a diffusional regime and in ab-
sence of rigid interfacial films, Ri is generally negligible relative to Rw and Ro.
Consider two simple cases of extraction processes in which kinetics are
controlled by interfacial film diffusion (the solutions are always considered
stirred). The two cases are treated with the simplifying assumptions introduced
in section 2 (i.e., steady-state and linear concentration gradients throughout the
diffusional films).
Case 4: The interfacial partition between the two phases of unchanged
species is fast. The rate is controlled by the diffusion to and away from the
interface of the partitioning species. In the absence of an interfacial resistance,
the partition equilibrium of A between the aqueous and organic phase, occurring
at the interface, can be always considered as an instantaneous process. Here, A
is any species, neutral or charged, organic or inorganic. This instantaneous parti-
tion process (interfacial equilibrium) is characterized by a value of the partition
coefficient equal to that measured when the two phases are at equilibrium.
The following equations describe the extraction process:
A ↔ A(org) (5.65)
with an extraction equilibrium constant KDA, reached at t = ∞ ;
A i ↔ A i (org) (fast) (5.66)
interfacial equilibrium, holding at any time (subscript i indicates species in con-
tact with the interface); thus
K DA = [A]eq /[A]eq = [A]i /[A]i (5.67)
The symbol [ ]i indicates concentrations at the extreme limit of the diffu-
sional film (i.e., volume concentrations in the region in direct contact or very
close to the liquid-liquid interface). Because of the fast nature of the distribution
reaction, local equilibrium always holds at the interface.
The diffusional process can be treated by applying the first Fick’s diffu-
sion law to the two diffusional processes occurring on both sides of inter face,
under the simplifying assumptions of steady-state and linear concentration gra-
dients shown in Fig. 5.8. The diffusional fluxes through the aqueous, Jw and
organic, Jo, diffusion films will then be
J w = − DA ∂ [A]/ ∂x = − DA ([A]i − [A]) / δ w (5.68)
236 Danesi
Fig. 5.8 Linear concentration profiles for the partition of A between an aqueous and
an organic phase.
J o = − DA ∂ [A]/ ∂x = − DA ([A] − [A]i ) / δo (5.69)

DA and DĀ are the diffusion coefficients of the distributing species in the
aqueous and organic layers, respectively. By considering that, at the steady state,
Jw = Jo = J and by setting
∆ w = δ w / DA , ∆ o = δo / DA (5.70)
Equations. (5.68) and (5.69) can be solved for the interfacial concentrations to
yield
[A]i = [A] − J ∆ w (5.71)
[A]i = J ∆ o + [A] (5.72)

By inserting Eqs. (5.71) and (5.72) into Eq. (5.67), solving for J, and
considering the correlation between flux and rate [see Eq. (5.5)], we obtain
rate = −d [A]/ dt = rate (forward) − rate (reverse)=
=S K DA [A]/( K DA ∆w + ∆ o ) − S[A]/( K DA ∆ w + ∆ o ) (5.73)
The diffusion-controlled, extraction kinetics of A, therefore, can be de-
scribed as a pseudo-first-order rate process with apparent rate constants
k1 = K DA [A]/( K DA ∆ w + ∆ o ) and k−1 = 1 /( K DA ∆ w + ∆ o ) (5.74)
Rydberg_5063-2_Ch05_R3_02-09-04 11:11:10
The rate equation is indistinguishable from that of an extraction process

occurring in a kinetic regime, which is controlled by a slow, interfacial partition
reaction:
k1
A ↔ A(org) (slow) (5.75)
k−1
The diffusion-controlled process [see Eqs. (5.66), (5.68), and (5.69)] can
be experimentally differentiated from the process occurring in the kinetic regime
[see Eq. (5.75)] only by measuring the variations of kl and k−1 with δw and δo .
Otherwise, the identical rate laws will not permit one to distinguish between the
two mechanisms.
Equation (5.73) can be integrated to obtain
ln ([A]o − [A]eq ) /([A] − [A]eq ) = S ( K DA + 1)t /
( K DA ∆ w + ∆ o ) = S (k1 + k−1 )t (5.76)
Equations (5.74) and (5.76) show that when KDA is very high (i.e., the
partition of A is all in favor of the organic phase):
k1 ~ DA / δ w and k−1 ~ 0
and the extraction rate is controlled only by the aqueous-phase diffusional resis-
tance. On the other hand, when KDA is very small, it is
k1 ~ 0 and k−1 ~ DA / δo
and the rate is only controlled by the organic-phase diffusional resistance.
Case 5: A fast reaction between the metal cation and the undissociated
extracting reagent occurs at the interface (or in proximity of the interface). The
rate is controlled by the diffusion to and away from the interface of the species
taking part in the reaction.
The overall extraction reaction (extraction stoichiometry) is represented
by the equation:
M + + BH(org) = MB(org) + H + (5.77)
with equilibrium extraction constant Kex chemical reaction occurring at or near
the interface is considered to be always at equilibrium, although bulk phases
reach this condition only at the end of the extraction process. The condition of
interfacial (local) equilibrium is expressed by
M i+ + BH i (org) ↔ MBi (org) + H i+ (fast) (5.78)
with interface equilibrium constant Ki, which alternatively can be written as the
sum of the two equilibria.
M i+ + Bi− (org) ↔ MBi (org), K1 = [MB]i /[M + ]i [B− ]i (5.79)
238 Danesi
BH i (org) ↔ H i+ + Bi− (org), K 2 = [B− ]i [H + ]i /[BH]i (5.80)

with
K ex = K1K 2 = ([MB]eq [H + ]eq ) /([BH]eq [M + ]eq )
= ([MB]i [H + ]i ) /([BH]i [M + ]i ) = K i (5.81)
Following the previous line of reasoning, the rate of extraction can be
obtained by considering the diffusional fluxes through the diffusion films of M+,
H+, MB, and BH, expressed through Fick’s law, and introducing the chemical
reaction [see Eq. (5.78)] as boundary condition to the differential equations
describing diffusion through the aqueous and organic diffusion films. Under the
usual simplifying assumptions of steady-state and linear concentration gradients
and referring to Fig. 5.9, where the concentration gradients of the four species
M+, H+, BH, and MB are schematically indicated, the rate of extraction of M+
can be derived. Equations (5.82a–d) describe the diffusional fluxes of M+, H+,
BH, MB through the stagnant films:
Fig. 5.9 Linear concentration profiles for the extraction of M+ by BH(org) in presence
of an interfacial chemical reaction.
J M+ = − DM+ ∂ [M + ]/ ∂x = − DA+ ([M + ]i − [M + ]) / δ w (5.82a )

J H+ = − DH+ ∂ [H + ] / ∂x = − DH+ ([H + ]i − [H + ]) / δ w (5.82b)
J MB = − DMB∂ [MB]/ ∂x = − DAB ([MB] − [MB]i ) / δo (5.82c)
J BH = − DBH ∂ [BH]/ ∂x = − DBH ([BH] − [BH]i ) / δo (5.82d)
Furthermore, the four fluxes are not independent and, at the steady state,
are related by the condition:
J M+ = J MB = − J H+ = − J BH (5.83)
+
indicating that, although M and MB diffuse in the positive direction of the x
axis, H+ and BH diffuse in the opposite one. Moreover,
−d [M + ]/ dt V / Q = J = J M+ (5.84)
By solving Eqs. (5.82a–d) for the interfacial concentrations [ ]i after tak-
ing into account the correlation between the fluxes [see Eqs. (5.83) and (5.84)]
and introducing the simplification that the diffusion coefficients depend only on
the nature of the phase, that is,
∆ w = δ w / DM+ = δ w / DH+ (5.85a )
∆ o = δo / DMB = δo / DBH (5.85b)
we obtain
[M + ]i = [M + ] − J ∆ w (5.86a )
+ +
[H ]i = [H ] + J ∆ w (5.86b)
[MB]i = [MB] + J ∆ o (5..86c)
[BH]i = [BH] − J ∆ o (5.86d)
Equations (5.86a–d) can then be inserted into Eq. (5.81) to yield
K ex = [MB] + J ∆ o )([H + ] + J ∆ w ) /([M + ] − J ∆ w )(BH] − J ∆ o ) (5.87)
Equation (5.87) can be solved for J to derive an expression for J (and the
extraction rate) as function of the four concentration variables in the bulk
phases, the equilibrium constant of the extraction reaction, and the two diffu-
sional parameters ∆w and ∆o.
A simple analytical solution for J (and the extraction rate) can be derived
from Eq. (5.87) by neglecting the term in J2 (very low fluxes) and isolating J:
rate = rate (forward) − rate (reverse) = −d [M + ]/ dt = JS =
= {S K ex [M + ][BH]}{∆ w ([MB] + K ex [BH]) + ∆ o ([H + ] + K ex [M + ])}−1 −
− {S [MB][H + ]}{∆ w ([MB] + K ex [BH]) + ∆ o ([H + ] + K ex [M + ])}−1 (5.88)
240 Danesi
In the limiting condition of very low metal ion concentrations, the terms con-
taining [M+] and [MB] can be neglected in the denominator of Eq. (5.88). This
means that [BH] and [H+], in practice, are constant with time, and their initial
and equilibrium concentrations are essentially the same. The rate equation then
becomes
rate = {S K ex [M + ][BH]eq }{∆ w K ex [BH]eq + ∆ o [H + ]eq }−1 −
− {S [MB][H + ]eq }{∆ w K ex [BH]eq + ∆ o [H + ]eq }−1 (5.89)
By dividing the numerators and denominators in Eq. (5.89) by [H+]eq and
setting
K ex* = K ex [BH]eq /[H + ]eq (5.90)
where K*e x is the conditional equilibrium constant of equilibrium [see Eq. (5.77)]
at [H+] and [BH] constant, Eq. (5.89) becomes identical with Eq. (5.73) when
KDA is replaced by Kex. As expected, the rate of extraction of species present at
high dilution in a system of fixed composition can be formally treated as the
simple distribution of an unchanged species between two immiscible phases.
The comparison of Eq. (5.89) with Eq. (5.63) also indicates that the rate laws
obtained for a system in a kinetic regime controlled by slow two-step interfa-
cial chemical reactions [see Eqs. (5.58) and (5.59)] and for a system in diffu-
sional regime controlled by slow film diffusion processes coupled to an in-
stantaneous reaction occurring at the interface (or in the proximity of the
interface) [see Eqs. (5.82a–d) and (5.78)], have the same functional depen-
dence on the concentration variables when low fluxes and low metal concen-
trations are involved. Alternatively, this result can be formulated by stating
that, because of diffusional contributions to the extraction rate, a fast extrac-
tion reaction occurring with a simple stoichiometry can mimic a two-step
interfacial chemical reaction occurring in absence of diffusional contributions
[14,15]. This type of ambiguity, in the past, has led solvent extraction chem-
ists (including this author) to sometimes erroneously derive extraction mech-
anisms invoking a series of two slow interfacial chemical reactions as rate
determining steps in systems for which the two phases were insufficiently
stirred. However, a distinction between the two cases can occasionally be
made if it is possible to conduct kinetic experiments in which the values of
δw and δo are varied. When the rate is independent of δw and δo, Eq. (5.63)
rather than Eq. (5.89) of this section describes the rate of extraction. Never-
theless, since the film thicknesses never seem to go to zero, the ambiguity
may not be resolved.
Film diffusion usually is slower than the rate of many ligand-substitution

reactions. Therefore, when rate laws, such as Eqs. (5.63) or (5.89), are found
for the extraction kinetics of metal species, preference should, in general, be
given to an interpretation in which the determining step of the extraction reac-

tion is film diffusion.
When the fast reactions occurring in the system have stoichiometries dif-
ferent from the simple one shown by Eq. (5.78), analytical solutions of the
diffusion equations are difficult to obtain. Nevertheless, numerical solutions can
be obtained by iterative routines, and the results are conceptually similar to
those described. The additional complications introduced by non–steady-state
diffusion and nonlinear concentration gradients can be similarly handled.
5.9 MIXED DIFFUSIONAL–KINETIC REGIME

A mixed diffusional–kinetic regime occurs when one or more of the chemical
steps of the reaction mechanism proceed with a velocity that is comparable with
that of the diffusional processes through the interfacial films. The other two
regimes are thus limiting cases of this more general one. Up to a certain limit,
the diffusional processes in an isothermal system can be accelerated by stirring,
which reduces the thickness of the diffusion films, and the rate of chemical
reactions can be increased by increasing the concentrations of the reactants. The
range of experimental conditions that determine the nature of the extraction
regime is specific for each chemical system and depends on the hydrodynamic
conditions of the apparatus used for the extraction.
The analytical description of a mixed regime is rather complicated, as all
difficulties and ambiguities associated with both the kinetic and diffusional re-
gimes may be present. To fully describe the extraction kinetics, it is necessary
to solve simultaneously the equations of diffusion and of chemical kinetics.
Furthermore, the slow chemical reactions can be of two types: homogeneous,
wherein the chemical changes occur in the entire volume of a phase or in a thin
layer, generally not too far from the interface; and heterogeneous, wherein the
chemical changes occur at the interface. The description of a diffusion regime
coupled to a homogeneous chemical reaction is particularly complicated, since
the differential equations describing the rate of formation of new chemical spe-
cies have to be solved simultaneously with the system of differential equations
describing the diffusional processes. On the other hand, when the slow chemical
reactions are heterogeneous (i.e., interfacial), the rates of the chemical reactions
appear in the boundary conditions of the differential diffusion equations.
A mathematically simple case, that occurs frequently in solvent extraction
systems, in which the extracting reagent exhibits very low water solubility and
is strongly adsorbed at the liquid interface, is illustrated. Even here, the interpre-
tation of experimental extraction kinetic data occurring in a mixed extraction
regime usually requires detailed information on the boundary conditions of the
diffusion equations (i.e., on the rate at which the chemical species appear and
disappear at the interface).
242 Danesi
Thus, the chemical kinetics at the interface must be known from separate
experiments carried out in a pure kinetic regime. In fact, although the problem
has some similarities with that of a diffusional regime associated with a fast
interfacial chemical reaction, for which the equilibrium law was set up as an
interfacial boundary condition to the diffusion of the species, here, the further
complication exists that the interfacial chemical rate laws cannot be a priori
known and have to be themselves derived from kinetic experiments. This diffi-
culty can sometimes be avoided by assuming that the interfacial rate laws are
simply related to the reaction stoichiometry or by trying to derive information
on the interfacial reactions through fitting, to experimental data obtained in a
mixed regime, analytical solutions of differential equations that take into ac-
count both diffusion and chemical reactions. Unfortunately, both of these proce-
dures can lead to erroneous interpretations, as in only a few cases can the rate
laws be correctly derived from stoichiometric considerations, and when too
many variables (e.g., rate constants, rate laws, and diffusional parameters) are
simultaneously adjusted to fit experimental data, many alternative models can
usually satisfy the same equations. Therefore, it is important that the boundary
condition of the diffusion equations (i.e., the interfacial rate laws) be derived
by separate, suitable experiments.
We now treat, in some detail, under the usual assumptions of steady-state
diffusion and linear concentration gradients through the diffusion films, only
one simple case of mixed regime. The case deals with a situation that is an
extension of case 4 described in section 5.8. Here the additional complication
of a slow interfacial reaction is introduced.
Case 6: The interfacial partition reaction between the two phases of un-
charged species is slow. The rate is controlled both by the slow partition reaction
and by diffusion to and away from the interface of the partitioning species.
Although the extraction process is still described by Eq. (5.65), the slow
interfacial chemical reaction and the corresponding interfacial flux, Ji, are
k1
A i ↔ A i (org) (5.91)
k−1
J i = k1[A]i − k−1 [A]i (5.92)

Following the same derivation as for case 4 of section 5.8 and referring
again to Fig. (5.8) (here, however, the concentrations [A]i and [A]i are at equi-
librium only at the end of the extraction process), we obtain for the interfacial
concentrations, equations identical with Eqs. (5.71) and (5.72). By substituting
them into Eq. (5.92), keeping into account that at the steady state
Ji = J w = Jo = J (5.93)
and solving for J, we obtain
rate = −d [A]/ dt = rate (forward) − rate (reverse) =

= S k1[A](k1∆ w + k−1 ∆ o + 1) −1 − S k−1[A](k1 ∆ w + k−1∆ o + 1) −1 (5.94)
A comparison between Eq. (5.94) and Eq. (5.73) indicates that the pres-
ence of a slow interfacial chemical reaction shows up as an additional term in
the denominator of the rate laws. Nevertheless, the same functional dependence
on S, [A], and [A] is exhibited by Eq. (5.73) and Eq. (5.94). This means that
the rate law does not allow choice between partition kinetics controlled only by
diffusion, or only by a slow partition reaction, or by a combination of the two.
When the rate constants are large numbers (fast reactions), both the numera-
tors and the denominators of Eq. (5.94) can be divided by k−1. Since k1/k−1 =
KDA and (k−1)−1 can be neglected relative to the other terms in the sum (KDA ∆w +
∆o + k−1), an equation identical with Eq. (5.73) is obtained for which only dif-
fusion controls the partition rate. At the other extreme, when interfacial film
diffusion is very fast, and the reactions are very slow, we can set 1 Ⰷ (k1∆w +
k−l ∆o) and the rate equation becomes equal to
rate = S k1 [A] − S k−1 [A] (5.95)
that is, the partition rate is controlled only by the slow chemical reaction.
5.10 EXPERIMENTAL TECHNIQUES

Several experimental techniques are available for measuring the kinetics of sol-
vent extraction. They differ in the different efficiency of stirring which can be
achieved in the two phases, and in the control of the hydrodynamic conditions
in proximity to the interface and in the interfacial area. Most of these techniques
can be grouped into five categories.
1. Highly stirred vessels: Those techniques in which the two phases are so
highly stirred that droplets of one phase are dispersed in a continuum of the
second phase (the dispersing phase).
2. Constant interfacial-area-stirred cells: Those techniques in which the two
phases can be stirred to a variable extent, while a known and constant inter-
facial area is maintained.
3. Rotating diffusion cells: These techniques bring the two phases in contact
at the surface of a rotating membrane filter having well characterized hydro-
dynamics.
4. Moving drops: These techniques measure the extraction that takes place
when a single droplet of one phase travels along a vertical tube filled with
the second phase.
5. Short-time phase contact methods: These techniques do not stir the two
phases, and the contact time between them is very short.
244 Danesi
5.10.1 Highly Stirred Vessels

Different versions of highly stirred vessels are used in kinetic studies. The com-
mon feature is that samples are removed from the extraction apparatus at dif-
ferent times to measure concentrations. The extraction occurs under con-
ditions similar to those that are met in practical technological or analytical
extractions.
Figure 5.10 shows an advanced version of this technique [16]. The appara-
tus incorporates a continuous monitoring of the rate and permits instantaneous
data analyses that are accomplished by the use of a Teflon phase separator and
an online microcomputer. The device allows measurement of extraction rates
with half-lives as short as 10 s.
When the two phases are so highly stirred that one phase is dispersed into
the other in the form of minute droplets, it is generally difficult to control the
interfacial area through which the extraction takes place. This constitutes the
major drawback to the use of this technique to derive unequivocal information
Fig. 5.10 Computer-assisted extraction kinetics-measuring apparatus for highly stirred

phases: (A) high-speed stirrer; (B) stirrer shaft; (C) sample inlet; (D) Teflon stirring bar;
(E) Teflon phase separator; (F) water bath; (G) flow-cell; (H) spectrophotometer; (I)
peristaltic pump; (J) chart recorder; (K) A/D converter; (L) clock; (M) minicomputer;
(N) dual-floppy disk drive; (O) printer, (P) plotter. (From Ref. 16.)
on the extraction regime. Typical drop diameters are 200 µm, that is, about
10–20 times larger than the claimed minimum thickness of the diffusion films
usually reached in well-stirred biphasic systems.
5.10.2 Constant Interfacial-Area-Stirred Cells

In this technique, the two phases are stirred while the aqueous and the organic
layers are positioned one on top of the other and the transfer occurs through the
contact area [17–19]. The interfacial area, which coincides with the geometric
horizontal section at which the contact between the two phases is established,
is always kept constant, whereas the degree of stirring of the two phases is
varied over the widest possible range. The major limitations of the technique is
that the motion of the liquids in proximity of the interface and the interfacial
turbulence are not well defined and vary with the equipment. Consequently,
poorly defined diffusion films result. Nevertheless, experience has shown that
by using well-designed equipment, reproducible data can be obtained, and film
diffusion can be treated quantitatively by using the two-film theory. With these
cells, ultimate film thicknesses of about 10−3 cm can be reached. Figure 5.11
shows a version of a constant interfacial-area-stirred cell.
Several different versions of extraction cells are used. The main difference
between the various cells lies in the presence or absence of internal baffles that
modify the internal forced convection, in the shape and dimensions of the stir-
rers that drive the forced convection, and in the presence of grids (or screens)
in proximity to the interface that help to stabilize the interface when the two
Fig. 5.11 Constant interfacial–area-stirred cell (“Lewis-type cell”) for liquid–liquid

extraction kinetic measurements.
246 Danesi
phases are stirred. In presence of these baffles and grids, the range of variation
of the stirring speed of the phases can be quite wide. The grids (or screens)
transform most of the translation kinetic energy of the liquids into turbulent
energy, preventing the rippling of the interface. The cells allow an independent
change of the revolution number of the stirrers in the two phases. In this way,
the different influence of the density and viscosity of the two phases can be
taken into account, and the organic and aqueous phase can be stirred in such a
way to have similar fluid dynamic conditions.
5.10.3 The Rotating Diffusion Cell

In this cell, the aqueous and the organic phase are brought in contact at the
surface of a microporous membrane filter, the pores of which are filled with the
organic phase [20]. The microporous filter is attached to a hollow cylinder filled
with the same organic phase. The cylinder is dipped into the aqueous phase and
is rotated by a motor. A scheme of the rotating diffusion cell is presented
in Fig. 5.12. The cell has a filter with well-defined hydrodynamics on both
Fig. 5.12 Cross section of the rotating diffusion cell: B, bearing; BA, internal baffle;
FM, filter mount; I, inner compartment; L, lid with holes; M, membrane; MA, mounting
rod; O, outer compartment; P, pulley block; S, hollow rod; T, thermostated beaker. (From
Ref. 20.)
sides. This allows description of the diffusional processes of the reactants and
products of the extraction reactions on the aqueous and the organic side of the
interface.
The basic assumption is that the rotating filter creates a laminar flow field that
can be completely described mathematically. The thickness of the diffusion
boundary layer (δ) is calculated as a function of the rotational speed (ω), vis-
cosity, density, and diffusion coefficient (D). The thickness is expressed by the
Levich equation, originally derived for electrochemical reactions occurring at
a rotating disk electrode:
δ = 0.643 ω−1 / 2 D1 / 3ν1 / 6 (5.96)
ν is the kinematic viscosity. The organic phase inside the membrane is con-
sidered at rest, and the membrane thickness and porosity control the diffusion
rate. The major drawbacks of this technique are difficulties in well characteriz-
ing the thickness and porosity of the membrane filter, in utilizing really thin
filters (less than 10−3 cm), and that, although the hydrodynamics inside the
rotating cylinder can be assimilated to those of a rotating disk, no studies have
conclusively shown that the flows inside and outside the cylinder are the same.
A more recent version of this cell [21] makes use of a rotating capillary,
containing a gelified aqueous phase. This eliminates the microporous membrane
filter.
5.10.4 Moving Drops

In the moving drop technique (also described in Chapters 7 and 9), a drop of
the organic or aqueous phase is produced at the end of a vertical column filled
with the other phase. The drop travels along the tube, during which extraction
occurs across the drop surface. By measuring the time of traveling, the drop
size, and from the volume of collected drops, it is possible to evaluate the rate
of extraction (see Chapter 9 for a detailed discussion of drop behavior and mass
transfer).
The technique can be applied to drops that move both upward and downward,
the choice depending on the higher or lower density of the drops relative to
the continuum liquid that fills the column. Although the technique allows one
to control both drop size and interfacial area, there are some disadvantages that
impose the use of great care both in the treatment of the experimental data and
in the design of the apparatus. When the extraction is very rapid, much of the
extraction can take place during drop formation. Further extraction can also
take place in the stagnant pool where the drops are collected. On the other
hand, when the extraction is very slow, very long columns are required to
assure sufficient drop residence times. The hydrodynamic control of the sin-
gle drops can also be very difficult because of lack of internal circulation and
the presence of drop oscillations. When the drops are quite small and the
248 Danesi
extractant adsorbs on the drop surface, the internal circulation can be absent
or strongly reduced, and it is difficult to establish whether the degree of
turbulence of the two phases in the moving drop method can ever be enough
to ensure the measurement of a transfer rate in a kinetic regime or, on the
opposite side, is so low that the extraction rate is always occurring in a diffu-
sional regime.
5.10.5 Short-Time Phase-Contact Method

This method, which has been developed and extensively utilized in Russia
[22,23], deals with nonstirred liquid phases and is based on measurements of
the electrical conductivity of the aqueous phase and of its changes during extrac-
tion. In the experimental device, the two phases are initially kept separated
inside a thermostated cell and then instantaneously contacted through a known
contact area. At this moment, the concentration variation in one phase is re-
corded as a function of time, generally using conductometry.
The aqueous phase is usually contained in a metal capillary. The capillary is
pushed into the lower part of the cell, containing the organic phase, at time
zero of the extraction. The method examines the reaction during very short
time intervals (from 0.05–1 s) and is relatively simple, as the two phases are
not stirred. The consumption of the reactants is very small (about 20 mm3 per
experiment) and permits one to make many determinations in a relatively short
time (about 10 kinetic curves per hour). Unfortunately, the method proves inap-
plicable for concentrated aqueous solutions and when many species are in-
volved in the extraction reaction.
This technique is based on a non–steady-state approach to diffusion. The
non–steady-state diffusion is not treated in our simplified approach to extraction
kinetics; the interested reader is referred to Yagodin and Tarasov [24] and Tara-
sov and Yagodin [25].
SYMBOLS
Parameter Denoted by First shown, in text order
Aqueous phase No symbol, or index w General

Organic phase Symbols org, index o (not to be con- Eq. (5.27)
fused with zero 0) or a bar above
the variable, constant, etc.
J Diffusion flow, mass cm−2 s−1 Eq. (5.1)
D Diffusion coefficient, cm−2 s−1 Eq. (5.1)
c Concentration, mol cm3 Eq. (5.1), Fig. 5.2
x Diffusion coordinate, cm Eq. (5.1)
∂ Derivative Eq. (5.1)

d Also used for derivative Eq. (5.3)
δ Thickness of organic δo or aqueous Eq. (5.3), Figs. 5.1 and 5.2
film δw at interface
∆ Diffusional parameter, =δ/D, cm−1 s Eq. (5.4)
R Film resistance, transfer resistance, for Fig. 5.1, Eq. (5.64)
indexed phase
V Volume of indexed reservoir Eq. (5.5), Fig. 5.2
Q Area of diffusion film, cm2 Eq. (5.5)
γ Interfacial tension, dynes cm−1 Section 5.5
γ2 Langmuir adsorption constant Eq. (5.40)
Π Inerfacial pressure, dynes cm−1 Eq. (5.26)
k Boltzmann constant Eq. (5.26)
T Temperature, kelvin Eq. (5.26)
t Time, seconds, s Eq. (5.5)
index 0 at beginning of experiment Eq. (5.36)
index eq at equilibrium Eq. (5.28), (5.36)
X, Y, Z, W Abbreviations for chemical com- Eq. (5.10)
pounds
k Rate constant; k−1 rate constant of re- Eq. (5.13), (5.29)
verse reaction
M+ Metal cation (central atom in complex Eq. (5.22), (5.27)
MB, or ML)
L− Ligand anion (ligand in complex ML) Eq. (5.22)
BH Extractant Eq. (5.27), (5.32)
B− (Organic) anion (ligand in complex Eq. (5.29)
MB)
K Equilibrium constant (index D for dis- Eq. (5.18)
tribution)
index ass for association Eq. (5.23)
index ex for (overall) extraction Eq. (5.28)
index DB, distribution constant of Eq. (5.31)
extractant BH
index a, acid dissociation constant Eq. (5.32)
of HB
index DM distribution constant of Eq. (5.33)
complex MB
index MB complex formation Chapter 3
constant of MB
ad Index for species adsorbed at the inter- Eq. (5.37)
face
A Any solute distributing between two Eq. (5.65)
phases
250 Danesi
α2 Langmuir adsorption coefficient Eq. (5.40)

S Specific interfacial area, cm−1 Eq. (5.38)
REFERENCES
1. Whitman, W. G. Chem. Metal Eng., 1923, 29(4):146.
2. Whitman, W. G.; Davies, D. S. Ind. Eng. Chem., 1924, 16:1233.
3. Bird, R. B.; Stewart, W. E.; Lightfoot, E. N. Transport Phenomena, John Wiley
and Sons, New York, 1960.
4. Cussler, E. L. Diffusion, Mass Transfer in Fluid Systems, Cambridge University
Press, Cambridge, 1984.
5. Astarita, G. Mass Transfer with Chemical Reaction, Elsevier, Amsterdam, 1967.
6. Frank-Kamenetskii, D. A. Diffusion and Heat Transfer in Chemical Kinetics, Ple-
num Press, New York, 1969.
7. Frost, A. A.; Pearson, R. G. Kinetics and Mechnanism, Wiley-Interscience, New
York, 1956.
8. Moelwin-Hughes, E. A. The Kinetics of Reactions in Solution, Oxford University
Press, London, 1950.
9. Eigen, M.; Wilkins, R. W. Mechanisms of Inorganic Reactions, Vol. 49, Adv.
Chem. Ser., American Chemical Society, Washington, DC, 1965.
10. Adamson, A. W. Physical Chemistry of Surfaces, John Wiley and Sons, New York,
1982.
11. Davies, J. T.; Rideal, E. K. Interfacial Phenomena, Academic Press, New York,
1961.
12. Danesi, P. R.; Chiarizia, R.; Vandergrift, G. F. J. Phys. Chem., 1980, 84:3455.
13. Danesi, P. R.; Chiarizia, R. The Kinetics of Metal Solvent Extraction, Crit. Rev.
Anal. Chem., 1980, 10:1.
14. Danesi, P. R. Solv. Extr. Ion Exch. 1984, 2:29.
15. Danesi, P. R.; Vandegrift, G. E.; Horwitz, P.; Chiarizia, R. J. Phys. Chem., 1980,
84:3582.
16. Watarai, H.; Cunningham, L.; Freiser, H. Anal. Chem., 1982, 54:2390.
17. Lewis, J. B. Chem. Eng. Sci., 1954, 3:218.
18. Nitsch, W.; Kahn, G. Ger. Chem. Eng. Engl. Trans., 1980, 3:96.
19. Danesi, P. R.; Cianetti, C.; Horwitz, E. P.; Diamond, H. Sep. Sci. Technol. 1982,
17.961.
20. Albery, W. J.; Burke, J. E; Lefier, E. B.; Hadgraft, J. J. Chem. Soc. Faraday Trans.,
1976, 72:1618.
21. Simonin, J. P.; Musikas, C.; Soualhia, E.; Turq, P. J. Chim. Phys., 1987, 84:525.
22. Tarasov, V. V.; Yagodin, G. A.; Kizim, N. F. Dokl. Akad. Nauk. SSR, 1971, 45:
2517.
23. Tarasov, V. V.; Yagodin, G. A. VINITI, Khim. Teknol. Neorganich. Khim., 1974, 4.
24. Yagodin, G. A.; Tarasov, V. V. Proc. Int. Solv. Extr. Conf., ISEC 71, Vol. 1,
Society of Chemical Industry, (Gregory, J. G.; Evans, B; Weston, P.C. Eds.), Lon-
don, p. 888, 1971.
25. Tarasov, V. V.; Yagodin, G. A. Interfacial Phenomena in Solvent E. Ytraction. In
Ion Exchange Solvent Extraction, Vol. 10, Ch. 4 (Marinsky, J. A. and Marcus, Y.
Eds.), Marcel Dekker, New York, pp. 141–237, 1988.
26. Eigen, M. Pure Appl. Chem., 1963, 6:97.
6
Ionic Strength Corrections
INGMAR GRENTHE* Royal Institute of Technology,
Stockholm, Sweden
HANS WANNER Swiss Federal Nuclear Safety Inspectorate,
Villingen-HSK, Switzerland
Thermodynamic data always refer to a selected standard state. The defini-

tion given by IUPAC [1] is adopted in the NEA-TDB project as outlined in
the corresponding guideline [2]. According to this definition, the standard
state for a solute B in a solution is a hypothetical solution, at the standard
state pressure, in which mB = mo = 1 mol kg1, and in which the activity
coefficient gB is unity. However, for many reactions, measurements cannot
be made accurately (or at all) in dilute solutions from which the necessary
extrapolation to the standard state would be simple. This is invariably the
case for reactions involving ions of high charge. Precise thermodynamic
information for these systems can only be obtained in the presence of an
inert electrolyte of sufficiently high concentration, ensuring that activity
factors are reasonably constant throughout the measurements. The objec-
tives of this chapter are to describe and illustrate various methods of
extrapolation to zero ionic strength and to describe the methods preferred in
the NEA review along with recommended values of the auxiliary para-
meters. By following these guidelines, the members of the NEA specialist
teams performing the review will be assured of using the same theory con-
sistently for the extrapolation to zero ionic strength.
*Retired.
By kind permission from the OECD NEA-TDB project, reprinted from Report TDB-2,
Guidelines for the extrapolation to zero ionic strength, OECD Nuclear Energy Agency, Data
Bank, F-91191 Gif-Sur Yvette, France. For latest revised version, see Chemical Thermo-
dynamics of Americium Silva, R. J., Bidoglio, G., Rand, M. H., Robouch, P. B., Wanner, H.,
and Puigdomenech, I., Eds., Elsevier/North-Holland 1995.
253
254 Grenthe and Wanner
The activity factors of all the species participating in reactions in

high ionic strength media must be estimated in order to reduce the ther-
modynamic data obtained from the experiment to the standard state (I = 0).
These estimates are based on the use of extended Debye-Hückel equations,
either in the form of ‘‘specific ion interaction methods’’ or the Davies
equation. However, the Davies equation (see section 6.2.4) should in gen-
eral not be used at ionic strengths larger than 0.1 mol kg1. The follow-
ing forms of specific ion interaction methods have been elaborated in the
past:
1. The Brønsted-Guggenheim-Scatchard approach (abbreviated ‘‘B-G-S

equation’’ in this document), (see section 6.1).
2. The Pitzer and Brewer ‘‘B-method’’ (abbreviated ‘‘P-B’’ herein) (see
section 6.2.1).
3. The Pitzer virial coefficient method, see section 6.2.2. Methods 1 and 2 are
equivalent and differ only in the form of the denominator in the Debye-
Hückel term. Method 3 requires more parameters for the description of
the activity factors. These parameters are not available in many cases.
This is generally the case for complex formation reactions.
The method preferred in the NEA Thermochemical Data Base review is the
specific ion interaction model in the form of the Brønsted-Guggenheim-
Scatchard approach.
One may sometimes have access to the parameters required for the
Pitzer approaches, e.g., for some hydrolysis equilibria and for some solu-
bility product data, cf. Baes and Mesmer [3] and Pitzer [4]. In this case,
the reviewer should perform a calculation using both the B-G-S and the
P-B equations and the full virial coefficient methods and compare the
results.
6.1 THE SPECIFIC ION INTERACTION

EQUATIONS
6.1.1 Background
The Debye-Hückel term, which is the dominant term in the expression
for the activity coefficients in dilute solution, accounts for electrostatic,
nonspecific long-range interactions. At higher concentrations, short-range,
nonelectrostatic interactions have to be taken into account. This is usually
done by adding ionic strength dependent terms to the Debye-Hückel ex-
pression. This method was first outlined by Brønsted [5,6], and elaborated
Ionic Strength Corrections 255
by Scatchard [7] and Guggenheim [8]. The two basic assumptions in the
specific ion interaction theory are:
Assumption 1: The activity coefficient gj of an ion j of charge zj in the
solution of ionic strength Im may be described by Eq. (6.1).
X
log 10 j ¼ z2j D þ "ð j;k;Im Þ mk (6.1)
k
D is the Debye-Hückel term:

pffiffiffiffiffi
A Im
D¼ pffiffiffiffiffi (6.2)
1 þ Baj Im
A and B are constants that are temperature dependent, and aj is the
effective diameter of the hydrated ion j. The values of A and B as a function
of temperature are listed in Table 6.1.
The term Baj in the denominator of the Debye-Hückel term has been
assigned a value of Baj = 1.5, as proposed by Scatchard [9] and accepted by
Ciavatta [10]. This value has been found to minimize, for several species, the
ionic strength dependence of e( j,k,Im) between Im = 0.5 m and Im = 3.5 m. It
should be mentioned that some authors have proposed different values for
Table 6.1 Debye-Hückel Constants
t(8C) A B(108)
0 0.4913 0.3247
5 0.4943 0.3254
10 0.4976 0.3261
15 0.5012 0.3268
20 0.5050 0.3276
25 0.5091 0.3283
30 0.5135 0.3291
35 0.5182 0.3299
40 0.5231 0.3307
45 0.5282 0.3316
50 0.5336 0.3325
55 0.5392 0.3334
60 0.5450 0.3343
65 0.5511 0.3352
70 0.5573 0.3362
75 0.5639 0.3371
Source: Ref. 13.

Baj, ranging from Baj = 1.0 [11] to Baj = 1.6 [12]. However, the parameter
Baj is empirical and as such correlated to the value of "ð j;k;Im Þ . Hence, this
variety of values for Baj does not represent an uncertainty range, but rather
indicates that several different sets of Baj and "ð j;k;Im Þ may describe equally
well the experimental mean activity coefficients of a given electrolyte. The
ion interaction coefficients listed later in the chapter in Tables 6.3–6.5 have
to be used with Baj = 1.5.
The summation in Eq. (6.1) extends over all ions k present in solution.
Their molality is denoted mk. The concentrations of the ions of the ionic
medium is often very much larger than those of the reacting species. Hence,
the ionic medium ions will make the main contribution to the value of
log10 gj for the reacting
P ions. This fact often makes it possible to simplify
the summation k "ð j;k;Im Þ mk so that only ion interaction coefficients
between the participating ionic species and the ionic medium ions are
included, as shown in Eqs. (6.5)–(6.9).
Assumption 2: The ion interaction coefficients "ð j;k;Im Þ are zero for ions of the
same charge sign and for uncharged species. The rationale behind this is that
e, which describes specific short-range interactions, must be small for ions of
the same charge, since they are usually far from one another due to elec-
trostatic repulsion. This holds to a lesser extent also for uncharged species.
Equation (6.1) will allow fairly accurate estimates of the activity
coefficients in mixtures of electrolytes if the ion interaction coefficients are
known. Ion interaction coefficients for simple ions can be obtained from
tabulated data of mean activity coefficients of strong electrolytes or from the
corresponding osmotic coefficients. Ion interaction coefficients for com-
plexes can either be estimated from the charge and size of the ion or
determined experimentally from the variation of the equilibrium constant
with the ionic strength. Ion interaction coefficients are not strictly constant
but vary slightly with the ionic strength. The extent of this variation depends
on the charge type and is small for 1:1, 1:2, and 2:1, electrolytes for mol-
alities less than 3.5 m. The concentration dependence of the ion interaction
coefficients can thus often be neglected. This point was emphasized by
Guggenheim [8], who has presented a considerable amount of experimental
material supporting this approach. The concentration dependence is larger
for electrolytes of higher charge. In order to accurately reproduce their
activity coefficient data, concentration-dependent ion interaction coef-
ficients have to be used: (see Pitzer and Brewer [14]; Baes and Mesmer [3];
or Ciavatta [10]). By using a more elaborate virial expansion, Pitzer
and coworkers [4, 15–21] have managed to describe measured activity
coefficients of a large number of electrolytes with high precision over a large
concentration range. Pitzer’s model generally contains three parameters, or

more, as compared to one in the specific ion interaction model. The use of
the model requires the knowledge of all these parameters. The derivation of
Pitzer coefficients for many metal complexes would require a very large
amount of additional experimental work, since no data of this type are
currently available.
The way in which the activity coefficient corrections are performed
according to the specific ion interaction model is illustrated below for a
general case of a complex formation reaction. Charges are omitted for
brevity.
mM þ qL þ nH2 O(1) Ð Mm Lq (OH)n þ nHþ (6.3)
The formation constant of MmLq(OH)p, *bn,q,m, determined in an ionic
medium (1:1 salt NX) of the ionic strength Im, is related to the corre-
sponding value at zero ionic strength, *bon,q,m, by Eq. (6.4).
log 10 n;q;m ¼ log 10 n;q;m
o
þ m log10 M þ q log10 L þ n log10 H2 O
log10 n;q;m n log10 H (6.4)
The subscript (n,q,m) denotes the complex ion MmLq(OH)n. If the concen-
trations of N and X are much greater than the concentrations of M, L,
MmLq(OH)n and H, only the molalities mNPand mX have to be taken into
account for the calculation of the term k "ð j;k;Im Þ mk in Eq. (6.1). For
example, for the activity coefficient of the metal cation M, gM, Eq. (6.5) is
obtained.
pffiffiffiffiffi
z2M 0:5091 Im
log10 M ¼ pffiffiffiffiffi þ "ðM;X;I m Þ mX (6.5)
1 þ 1:5 Im
Under these conditions, Im & mX = mN. Substituting the log10 gj values in
Eq. (6.4) with the corresponding forms of Eq. (6.5) and rearranging leads to
log 10 n;q;m z2 D n log 10 aH2 O ¼ log 10 n;q;m
o
"Im (6.6)
where
z2 ¼ (mzM qzL n)2 þ n mz2M qz2L (6.7)
pffiffiffiffiffi
0:5091 Im
D¼ pffiffiffiffiffi (6.8)
1 þ 1:5 Im
" ¼ "ðn;q;m;N or XÞ þ n"(H;XÞ q"ðN;LÞ m"ðM;XÞ (6.9)
Here (mzMqzLn), zM and zL are the charges of the complex MmLq(OH)n,

the metal ion M and the ligand L, respectively. Equilibria involving H2O(l)
as a reactant or product require a correction for the activity of water, aH2O.

The activity of water in an electrolyte mixture can be calculated as
k mk
log 10 aH2 O ¼ (6.10)
lnð10Þ 55:51
where F is the osmotic coefficient of the mixture and the summation extends
over all ions k with molality mk present in the solution. In the presence of
an ionic medium NX in dominant concentration, Eq. (6.10) can be simpli-
fied by neglecting the contributions of all minor species, i.e., the reacting
ions. Hence, for a 1:1 electrolyte of ionic strength Im & mNX, Eq. (6.10)
becomes
2mNX
log 10 aH2 O ¼ (6.11)
lnð10Þ 55:51
Values of osmotic coefficients for single electrolytes have been compiled

by various authors, e.g., Robinson and Stokes [22]. The activity of water
can also be calculated from the known activity coefficients of the dissolved
species.
In the presence of an ionic medium NX of a concentration much larger
than those of the reacting ions, the osmotic coefficient can be calculated
according to Eq. (6.12) [23].
1¼
A lnð10Þjzþ z j
mNX (1:5) 3 [ pffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffi
1 þ 1:5 mNX 2 log 10 (1 þ 1:5 mNX )

1
1 þ 1:5 mNX ]
pffiffiffiffiffiffiffiffiffiffi þ
1
"
ln (10Þ (N;X)
mNX (6.12)
The activity of water is obtained by inserting Eq. (6.12) into Eq. (6.11). It
should be mentioned that in mixed electrolytes with several components
at high concentrations, it is necessary to use Pitzer’s equation to calculate
the activity of water. On the other hand, aH2 O is near constant (and = 1)
in most experimental studies of equilibria in dilute aqueous solutions,
where an ionic medium is used in large excess with respect to the reactants.
The ionic medium electrolyte thus determines the osmotic coefficient of the
solvent.
In natural waters the situation is similar; the ionic strength of most
surface waters is so low that the activity of H2O(l) can be set equal to unity.
A correction may be necessary in the case of seawater, where a sufficiently
good approximation for the osmotic coefficient may be obtained by con-
sidering NaCl as the dominant electrolyte.
In more complex solutions of high ionic strengths with more than one
electrolyte at significant concentrations, e.g., (Na +, Mg2+, Ca2+) (Cl,
SO42), Pitzer’s equation may be used to estimate the osmotic coefficient; the
necessary interaction coefficients are known for most systems of geochem-
ical interest.
Note that in all ion interaction approaches, the equation for mean
activity coefficients can be split up to give equations for conventional single
ion activity coefficients in mixtures, e.g., Eq. (6.1). The latter are strictly
valid only when used in combinations that yield electroneutrality. Thus,
while estimating medium effects on standard potentials, a combination of
redox equilibria with H + + e) * 12 H2(g) is necessary (see Example 3).
6.1.2 Estimation of Ion Interaction Coefficients

6.1.2.1 Estimation from Mean Activity Coefficient Data
Example 1: The ion interaction coefficient "ðHþ ;Cl ) can be obtained from
published values of g ± ,HCl vs. mHCl.
2 log 10 ;HCl ¼ log 10 þ;Hþ þ log 10 ;Cl

¼ D þ "ðHþ ;Cl Þ mCl D þ "ðCl ;Hþ Þ mHþ
log 10 ;HCl ¼ D þ "ðHþ ;Cl Þ mHCl
By plotting log10g ± ,HCl + D vs. mHCl, a straight line with the slope "ðHþ ;Cl Þ
is obtained. The degree of linearity should in itself indicate the range of
validity of the specific ion interaction approach. Osmotic coefficient data
can be treated in an analogous way.
6.1.2.2 Estimations Based on Experimental Values of
Equilibrium Constants at Different Ionic Strengths
Example 2: Equilibrium constants are given in Table 6.2 for the reaction
þ
2 þ Cl Ð UO2 Cl
UO2þ (6.13)
The following formula is deducted from Eq. (6.6) for the extrapolation to
I = 0:
log 10 1 þ 4D ¼ log 10 1o "Im (6.14)
The linear regression is done as described in the NEA Guidelines for
the Assignment of Uncertainties [24]. The following results are obtained:
log 10 1o ¼ 0:170 0:021

"(13) ¼ 0:248 0:022
Table 6.2 The Preparation of the Experimental

Equilibrium Constants for the Extrapolation to I = 0 with
the Specific Ion Interaction Method, According to
Eq. (6.13)
Im log10b1(exp)a log10b1,mb log10b1,m + 4D

0.1 0.17 ± 0.10 0.174 0.264 ± 0.100
0.2 0.25 ± 0.10 0.254 0.292 ± 0.100
0.26 0.35 ± 0.04 0.357 0.230 ± 0.040
0.31 0.39 ± 0.04 0.397 0.220 ± 0.040
0.41 0.41 ± 0.04 0.420 0.246 ± 0.040
0.51 0.32 ± 0.10 0.331 0.371 ± 0.100
0.57 0.42 ± 0.04 0.432 0.288 ± 0.040
0.67 0.34 ± 0.04 0.354 0.395 ± 0.040
0.89 0.42 ± 0.04 0.438 0.357 ± 0.400
1.05 0.31 ± 0.10 0.331 0.491 ± 0.100
1.05 0.277 ± 0.260 0.298 0.525 ± 0.260
1.61 0.24 ± 0.10 0.272 0.618 ± 0.100
2.21 0.15 ± 0.10 0.193 0.744 ± 0.100
2.21 0.12 ± 0.10 0.163 0.774 ± 0.100
2.82 0.06 ± 0.10 0.021 0.860 ± 0.100
3.5 0.04 ± 0.10 0.021 0.974 ± 0.100
a
Equilibrium constants for Eq. (6.13) with assigned uncertainties,
corrected to 258C where necessary.
b
Equilibrium constants corrected from molarity to molality units,
according to the procedure described in Ref. 2.
Note: The Linear Regression of this Set of Data is shown in
Fig. 6.1.
The experimental data are depicted in Fig. 6.1, where the area between the
dashed lines represents the uncertainty range that is obtained by using the
results in log10b1o and De and correcting back to I = 0.
Example 3: When using the specific ion interaction theory, the relationship
between the normal potential of the redox couple UO22+/U4+ in a medium
of ionic strength Im and the corresponding quantity at I = 0 should be
calculated in the following way. The reaction in the galvanic cell
Pt; H2 jHþ kUO2þ þ

2 ; U ; H jPt
4þ
(6.15)
is
þ
2 þ H2 (g) þ 2H Ð U
UO2þ þ 2H2 O
4þ
(6.16)
For this reaction

!
aU4þ a2H2 O
log 10 K ¼ log 10
o
(6.17)
aUO2þ a2Hþ fH2
2
log 10 K ¼ log 10 K þ log 10 U4þ log 10 UO2þ 2 log 10 Hþ

o
2
log 10 f ;H2 þ 2 log 10 aH2 O ; (6.18)
fH2 pH2 at reasonably low partial pressure of H2(g), aH2 O 1, and
log 10 U4þ ¼ 16D þ "ðU4þ ;ClO4 Þ mClO4 (6.19)
log 10 UO2þ ¼ 4D þ "ðUO2þ ;ClO4 Þ mClO4 (6.20)

2 2
log 10 Hþ ¼ D þ "ðHþ ;ClO4 Þ mClO4 (6.21)
Fig. 6.1 Plot of log10b1 + 4D vs. Im for Eq. (6.13). The straight line shows the result of
the weighted linear regression, and the area between the dashed lines represents the
uncertainty range of log10b1o and De.
Hence,
log 10 K o ¼ log 10 K 10D þ ð"ðU4þ ;ClO4 Þ "ðUO2þ ;ClO4 Þ
2
2"ðHþ ;ClO4 Þ ÞmClO4 (6.22)

The relationship between the equilibrium constant and the standard poten-
tial is
nF
ln K ¼ E (6.23)
RT
nF o
ln K o ¼ E (6.24)
RT
E is the standard potential in a medium of ionic strength I, E o is the cor-
responding quantity at I = 0, and n is the number of transferred electrons in
the reaction considered. Combining Eqs. (6.22)–(6.24) and rearranging them
leads to Eq. (6.25).

RT lnð10Þ RT lnð10Þ
E 10D ¼ E o "mClO4 (6.25)
nF nF
For n = 2 in the present example and T = 298.15 K, Eq. (6.25) becomes
E[mV] 295:8D ¼ E o [mV] 29:58"mClO4 (6.26)
where
" ¼ ð"ðU4þ ;ClO4 Þ "ðUO2þ ;ClO4 Þ 2"ðHþ ;ClO4 Þ Þ (6.27)
2
6.1.3 On the Magnitude of Ion Interaction

Coefficients
Ciavatta [10] made a compilation of ion interaction coefficients for a large
number of electrolytes. Similar data for complexation reactions of various
kinds were reported by Spahiu [25] and Ferri, et al. [26]. These and some
other data have been collected and are listed in Tables 6.3–6.6. It is obvious
from the data in these tables that the charge of an ion is of great impor-
tance for the magnitude of the ion interaction coefficient. Ions of the same
charge type have similar ion interaction coefficients with a given counte-
rion. Based on the tabulated data, it is judged possible to estimate, with an
error of at most ± 0.1 in e, ion interaction coefficients for cases where there
are insufficient experimental data for an extrapolation to I = 0. The error
made by this approximation is estimated to be ± 0.1 in De in most cases,
based on comparison with De values of various reactions of the same charge
type.
Table 6.3 Ion Interaction Coefficients ej,k for Cations j with k = Cl, ClO4, and NO3a
#
j k!
Cl ClO4 NO3
H+ 0.12 ± 0.01 0.14 ± 0.02 0.07 ± 0.01
NHþ 4 0.01 ± 0.01 0.08 ± 0.04b 0.06 ± 0.03b
ZnHCOþ 3 0.2c - -
CdCl + - 0.25 ± 0.02 -
CdI + - 0.27 ± 0.02 -
CdSCN + - 0.31 ± 0.02 -
HgCl + - 0.19 ± 0.02 -
Cu + - 0.11 ± 0.01 -
Ag + - 0.00 ± 0.01 0.12 ± 0.05b
UOþ 2 - 0.26 ± 0.03d -
UO2OH + - 0.06 ± 3.7d 0.51 ± 1.4d
ðUO2 Þ3 ðOHÞþ5 0.81 ± 0.17d 0.45 ± 0.15d 0.41 ± 0.22d
UFþ3 0.1 ± 0.1e 0.1 ± 0.1e -
UO2F + 0.04 ± 0.07f 0.29 ± 0.05d -
UO2Cl + - 0.33 ± 0.04d -
UO2 ClOþ 3 - 0.33 ± 0.04e -
UO2Br + - 0.24 ± 0.04e -
UO2 NOþ 3 - 0.33 ± 0.04e -
UO2SCN + - 0.22 ± 0.04e -
NpOþ 2 - 0.25 ± 0.05f -
PuOþ 2 - 0.17 ± 0.05f -
AlOH2 + 0.09g 0.31g -
Pb2 + - 0.15 ± 0.02 0.20 ± 0.12b
Zn2 + - 0.33 ± 0.03 0.16 ± 0.02
ZnCO2þ 3 0.35 ± 0.05c - -
Cd2 + - - 0.09 ± 0.02
Hg2 + - 0.34 ± 0.03 0.1 ± 0.1b
Hg2þ
2 - 0.09 ± 0.02 0.2 ± 0.1b
Cu2 + 0.08 ± 0.01 0.32 ± 0.02 0.11 ± 0.01
Ni2 + 0.17 ± 0.02 - -
Co2 + 0.16 ± 0.02 0.34 ± 0.03 0.14 ± 0.01
FeOH2 + - 0.38h -
FeSCN2 + - 0.45h -
Mn2 + 0.13 ± 0.01 - -
UO2þ2 0.21 ± 0.02i 0.46 ± 0.03 0.24 ± 0.03i
UF2þ
2 - 0.3 ± 0.1e -
USO2þ 4 - 0.3 ± 0.1e -
UðNO3 Þ2þ 2 - 0.49 ± 0.14j -
Mg2 + 0.19 ± 0.02 0.33 ± 0.03 0.17 ± 0.01
Ca2 + 0.14 ± 0.01 0.27 ± 0.03 0.02 ± 0.01
Ba2 + 0.07 ± 0.01 0.15 ± 0.02 0.28 ± 0.03
(continued )
Table 6.3 (continued )
#
j k!
Cl ClO4 NO3
Al3 + 0.33 ± 0.02 - -
Fe3 + 0.56 ± 0.03 0.42 ± 0.08
Cr3 + 0.30 ± 0.03 - 0.27 ± 0.02
La3 + 0.22 ± 0.02 0.47 ± 0.03 -
La3 + ? Lu3 + - 0.47 ? 0.52h -
UOH3 + - 0.48 ± 0.08j -
UF3 + - 0.48 ± 0.08e -
UCl3 + - 0.59 ± 0.10j -
UBr3 + - 0.52 ± 0.10e -
UNO3þ 3 - 0.62 ± 0.08j -
Be2OH3 + - 0.50 ± 0.05k -
Pu4 + - 1.03 ± 0.05f -
Np4 + - 0.82 ± 0.05 -
U4 + - 0.76 ± 0.06e -
Th4 + 0.25 ± 0.03 - 0.11 ± 0.02
a
Source: Refs. [10,24] unless specified. Uncertainties are 95% confidence level.
b
Ion interaction coefficients can be described more accurately with an ionic strength-dependent
function, listed in Table 6.5.
c
Ferri et al. [26].
d
Evaluated in NEA-TDB review on uranium thermodynamics [37].
e
Estimated in NEA-TDB review on uranium thermodynamics [37].
f
Riglet et al. [36], where the following assumptions were made: "ðNp3þ ;ClO4 Þ "ðPu3þ ;ClO4 Þ ¼ 0:49 as
for other ðM 3þ ; ClO 4 Þ interactions, and "ðNpO2þ " " ¼ 0:46.
2 ;ClO4 Þ ðPuO2þ
2 ;ClO4 Þ ðUO2þ
2 ;ClO4 Þ
g
Hedlund [35].
h
Taken from Spahiu [25].
i
The coefficients "ðMnþ ;Cl Þ and "ðMnþ ;NO3 Þ reported by Ciavatta [10] were evaluated without taking
chloride and nitrate complexation into account. See section 6.3.
j
Evaluated in NEA-TDB review on uranium thermodynamics [37] using "ðU4þ ;ClO4 Þ ¼ ð0:76 0:06Þ.
k
Taken from Bruno [38], where the following assumptions were made: "ðBe3þ ;ClO4 Þ ¼ 0:30 as for other
"ðM2þ ;ClO4 Þ ; "ðBe2þ ;Cl Þ ¼ 0:17 as for other "ðM2þ ;Cl Þ , and "ðBe2þ ;NO3 Þ ¼ 0:17 as for other "ðM2þ ;NO3 Þ .
6.2. OTHER METHODS FOR IONIC STRENGTH

CORRECTIONS
6.2.1 The Pitzer and Brewer Equation
The P-B equation is very similar to the B-G-S equation. The expression for
the activity coefficient of an ion i of charge zi takes the form
pffiffiffiffiffi
z2 0:5107 Im X
log 10 i ¼ i pffiffiffiffiffi þ B(i; j)mj (6.28)
1 þ Im j
Table 6.4 Ion Interaction Coefficients ej,k for Anions j with k = Li +, Na+, and K+ a
j k!
# Li + Na+ K+
OH 0.02 ± 0.03b 0.04 ± 0.01 0.09 ± 0.01
F - 0.02 ± 0.02c 0.03 ± 0.02
HF2 - 0.11 ± 0.06c -
Cl 0.10 ± 0.01 0.03 ± 0.01 0.00 ± 0.01
ClO 4 0.15 ± 0.01 0.01 ± 0.01 -
Br 0.13 ± 0.02 0.05 ± 0.01 0.01 ± 0.02
I 0.16 ± 0.01 0.08 ± 0.02 0.02 ± 0.01
HSO 4 - -
0.01 ± 0.02
NO 3 0.08 ± 0.01 0.04 ± 0.03b 0.11 ± 0.04b
H2 PO 4 - 0.08 ± 0.04b 0.14 ± 0.04b
HCO 3 - 0.03 ± 0.02 -
SCN - 0.05 ± 0.01 0.01 ± 0.01
HCOO - 0.03 ± 0.01 -
CH3COO 0.05 ± 0.01 0.08 ± 0.01 0.09 ± 0.01
SiOðOHÞ 3 - 0.08 ± 0.03c -
BðOHÞ 4 - 0.07 ± 0.05b -
UO2 ðOHÞ 3 - 0.09 ± 0.05c -
UO2 F 3 - 0.00 ± 0.05c -
ðUO2 Þ2 CO3 ðOHÞ 3 - 0.00 ± 0.05c -
SO2
4 0.03 ± 0.04b 0.12 ± 0.06b 0.06 ± 0.02
HPO2 4 0.15 ± 0.06b 0.10 ± 0.06b
CO23 0.05 ± 0.03 0.02 ± 0.01
CrO2 4 0.06 ± 0.04b 0.08 ± 0.04b
UO2 ðSO4 Þ2 2 0.12 ± 0.06c -
UO2 ðCO3 Þ2 2 0.04 ± 0.09c -
PO3
4 0.25 ± 0.03b 0.09 ± 0.02
P2 O47 0.26 ± 0.05 0.15 ± 0.05
FeðCNÞ4 6 - 0.17 ± 0.03
UðCO3 Þ4 4 0.09 ± 0.10c -
UO2 ðCO3 Þ4 3 0.08 ± 0.11c -
UO2 ðCO3 Þ5 3 0.66 ± 0.14c -
6
UðCO3 Þ5 0.27 ± 0.15c -
a
Source: Refs. [10,34] unless specified. Uncertainties are 95% confidence level.
b
Ion interaction coefficient can be described more accurately with an ionic strength-dependent
function, listed in Table 6.5.
c
Evaluated in NEA-TDB review on uranium thermodynamics [37].

Table 6.5 Ion Interaction Coefficients e(1,j,k) and e(2,j,k) for Cations j with k = Cl, ClO
4 , and NO3 (First Part), and for Anions j with
+ + + a
266
k = Li , Na , and K (Second Part), According to the Relationship e = e1 + e2log10Im
Cl ClO
4 NO
3
j k!
# e1 e2 e1 e2 e1 e2
NHþ 4 0.088 ± 0.002 0.095 ± 0.012 0.075 ± 0.001 0.057 ± 0.004
Ag + 0.1432 ± 0.0002 0.0971 ± 0.0009
Tl + 0.18 ± 0.02 0.09 ± 0.02
Hg2þ
2 0.2300 ± 0.0004 0.194 ± 0.002
Hg2 + 0.145 ± 0.001 0.194 ± 0.002
Pb2 + 0.329 ± 0.007 0.288 ± 0.018
Li + Na + K+
j k!
# e1 e2 e1 e2 e1 e2
OH 0.039 ± 0.002 0.072 ± 0.006
NO 2 0.02 ± 0.01 0.11 ± 0.01
NO 3 0.049 ± 0.001 0.044 ± 0.002 0.131 ± 0.002 0.082 ± 0.006
BðOHÞ 4 0.092 ± 0.002 0.103 ± 0.005
H2 PO 4 0.109 ± 0.001 0.095 ± 0.003 0.1473 ± 0.0008 0.121 ± 0.004
SO2
3 0.125 ± 0.008 0.106 ± 0.009
SO2
4 0.068 ± 0.003 0.093 ± 0.007 0.184 ± 0.002 0.139 ± 0.006
S2 O2
3 0.125 ± 0.008 0.106 ± 0.009
HPO2 4 0.19 ± 0.01 0.11 ± 0.03 0.152 ± 0.007 0.123 ± 0.016
CrO2 4 0.090 ± 0.005 0.07 ± 0.01 0.123 ± 0.003 0.106 ± 0.007
PO24 0.29 ± 0.02 0.10 ± 0.01
a
Source: Refs. [10,34]. Uncertainties are 95% confidence level.
Grenthe and Wanner
where the summation over j covers all anions for the case that i is a cation
and vice versa. Tables of B(I, j) are given by Pitzer and Brewer [14] and by
Baes and Mesmer [3]. The Debye-Hückel term [see Eq. (6.2)] is different
from that in the B-G-S equation. Apart from a slightly different value for A,
the factor Baj has been chosen equal to 1.0 in the P-B equation compared to
1.5 in the B-G-S equation. The B-G-S equation is preferred to the P-B
equation in the critical evaluations of the NEA-TDB Project for the reasons
given in section 6.1.1.
6.2.2 The Pitzer Equations

The following text is only intended to provide the reader with a brief outline
of the Pitzer method. This approach consists of the development of an ex-
plicit function relating the ion interaction coefficient to the ionic strength and
the addition of a third virial coefficient to Eq. (6.1). For the solution of
a single electrolyte MX, the activity coefficient may be expressed by Eq.
(6.29) [15]:
!
3=2
(2M X ) 2 2ðM X Þ

ln ¼ jzM zX jf þ m BMX þ m CMX (6.29)

where nM and nX are the numbers of M and X ions in the formula unit
and zM and zX their charges. m is the molality of the solution and n =
nM + nX. In aqueous solutions at 258C and 105 Pa, the following relations
are given [16]:
pffiffiffiffiffi
Im pffiffiffiffiffi

f ¼ 0:392 pffiffiffiffiffi þ 1:667 lnð1 þ 1:2 Im Þ (6.30)
1 þ 1:2 Im
ð1Þ
MX pffiffiffiffiffi pffiffiffiffi
ð0Þ
BMX ¼ 2MX þ (1 (1 þ 2 Im 2Im )e2 Im ) (6.31)
2Im

CMX ¼ 32 CMX (6.32)
where f g is the Debye-Hückel term extended to include osmotic effects, the

parameters b(0) (1)
MX and b MX define the second virial coefficient (corresponding
to e inPthe B-G-S equation), and CfMX defines the third virial coefficient.
Im = 12 mizi2 is the ionic strength in molal units. In the case of 2:2 elec-
trolytes, one could add to the second virial coefficient terms of the same
form. Pitzer’s equations have been extended to cover electrolyte mixtures
[17], including terms allowing for the interactions of ions of the same charge
sign. Equation (6.33) is an extension of Eq. (6.29) for a anions and c cations
[4].
P Here Pa and a0 P cover all anions, c and c0 cover all cations, and
( mz) = cmczc = ama|za|.
X
2M X X
lnMX ¼ jzM zX j f þ ma BMa þ mz CMa þ Xa
a M
X
2X X M
þ mc BcX þ mz CcX þ Mc
c X
XX
0 1
þ mc ma jzM zX jBca þ ð2M zM Cca þ M Mca þ X caX Þ
c a

1 X X X
0
þ mc mc0 cc0 X þ jzM zX jcc0
2 c c0
1XX M
0
þ ma ma0 Maa 0 þ jz z j
M X aa 0 (6.33)
2 a a0
In Eq. (6.33) f g is the same as in Eq. (6.29); CMX, BMX, and its derivative
with ionic strength, B0 MX, have the forms
ð1Þ pffiffiffiffiffi pffiffiffiffi

ð0Þ
BMX ¼ MX þ (1 (1 þ 2 Im )e2 Im ) (6.34)
2Im
ð1Þ
MX pffiffiffiffiffi pffiffiffiffi
B0MX ¼ 2
( 1 þ (1 þ 2 Im þ 2Im )e2 Im ) (6.35)
2Im

CMX
CMX ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi (6.36)
2 jzM zX j
where b(0), b(1), Cf are the same as for pure electrolytes. In Eq. (6.33) the y
terms summarize the interactions between ions of the same charge sign that
are independent of the common ion in a ternary mixture, and the c terms
account for the modifying influence of the common ion on these interac-
tions. Pitzer [4] points out that higher order electrostatic terms (beyond
the Debye-Hückel approximation) become important in cases of unsym-
metrical mixing, especially if one of the ions has a charge of three or higher.
Higher order electrostatic effects, on the other hand, were found to be
unimportant for cases of symmetrical mixing and for pure unsymmetrical
electrolytes.
Based on the cluster integral method [27], Pitzer [19] divided the dif-
ference terms yMN into two parts
MN ¼ sMN þEMN (6.37)

The EyMN terms may be expressed as:

zM zN 1 1
E
MN ¼ JðxMN Þ JðxMM Þ JðxNN Þ (6.38)
4Im 2 2

E 0 E zM zN 0 1 0 1 0
MN ¼ þ xMN J ðxMN Þ xMM J ðxMM Þ xNN J ðxNN Þ
Im 8Im2 2 2
(6.39)
pffiffiffiffiffi
where xij ¼ 6zi zj 0:392 Im at 258C and 10 Pa and J(x) is an integral that can
5
be approximated as
!1
1 2 X 6
JðxÞ ¼ x ðln xÞe10x þ Ck xk
2
(6.40)
6 k¼1
The first term is in fact important only at very low ionic strengths; thus for
cases used in equilibrium analysis one has

4:118 7:247 4:408 1:837 0:251 0:0164 1
JðxÞ ¼ þ 2 þ 3 þ 4 þ 5 þ (6.41)
x x x x x x6
The syMN parameters can be evaluated from data on mixtures of electrolytes
by calculating the differences between the experimental value of gexp and the
value calculated with the appropriate values for all pure electrolyte terms
and Ey terms but zero values for sy and c terms. For the activity coefficient
of MX in a MX-NX mixture one has Eq. (6.42) and equivalent expressions
for other cases.

1 zM
ln ¼ MN þ
s
mX þ mM MNX (6.42)
2 m
M N 2 zX
This is the equation of a straight line with intercept syMN and slope cMNX,
when the left side of Eq. (6.42) is plotted against a function of the com-
position 12 ðmX þ j zzMX jmM Þ.
The Pitzer equation for single electrolytes with the value of parameters
collected in several publication may be used as a compact source of activity
o
coefficient data. From the values g jk thus obtained, one may calculate, for
example, "ðj;k;Im Þ values using Eq. (6.1).
For the estimation of medium effects on solubility equilibria in mix-
tures of electrolytes involving ions of charge less than three, one may neglect
y and c terms in Eq. (6.33).
In equilibrium analysis studies carried out in the presence of an inert
salt (medium salt NX) and small (trace) concentrations of the reactants,
only the terms involving mNX have to be considered in Eq. (6.33), while
those involving mitrace can be neglected. Nevertheless, as the main difficulty
there still remains the accurate estimation of the parameters of single elec-
trolyte b(0), b(1), and Cf for species such as metal ion complexes.
Equations for single ion activity coefficients [4], osmotic coefficients
[17], and other thermodynamic quantities [28], as well as applications in
different cases (e.g., H2SO4 and H3PO4 solutions) have been given by Pitzer
and coworkers [4,20].
From Tables 6.3 and 6.4 it seems that the size and charge correlations
can be extended to complex ions. This observation is very important because
it indicates a possibility to estimate the ion interaction coefficients for
complexes by using such correlations. It is, of course, always preferable to
use experimental ion interaction coefficient data. However, the efforts nee-
ded to obtain these data for complexes will be so great that it is unlikely that
they will be available for more than a few complex species. It is even less
likely that one will have data for the Pitzer parameters for these species.
Hence, the specific ion interaction approach may have a practical advantage
over the inherently more precise Pitzer approach.
6.2.3 The Equations Used by Baes and Mesmer

Baes and Mesmer [3] use the function F(Im) proposed by Pitzer to express
the ionic strength dependence of the ion interaction coefficient BMX in
Guggenheim’s equations. For a single electrolyte
pffiffiffiffiffi
0:511 Im 2M X
log 10 o MX ¼ jzM zX j pffiffiffiffiffi þ BMX mMX (6.43)
1 þ Im
where
BMX ¼ B1 1
MX þ (BMX BMX )F(Im )
o
(6.44)
pffiffiffiffiffi pffiffiffiffi
1 ð1 þ 2 Im 2Im Þe2 Im
FðIm Þ ¼ ; Fð0Þ ¼ 1; Fð1Þ ¼ 0:
4Im
(6.45)
The Pitzer function linearizes the dependence of the ion interaction coeffi-
cient on the ionic strength quite well, even in the cases of 4:1 and 5:1
electrolytes, where constant e(M,X) values [see Eq. (6.1)] are not obtained at
high ionic strengths. The parameters BoMX and B? MX can be determined from
a single electrolyte activity coefficient datum by calculating first BMX. (Note
that BMX=e( j,k), since the Debye-Hückel term in Eq. (6.43) does not have
the factor 1.5 in the denominator.) By plotting BMX(Im) values against F(Im),
B? o
MX is obtained as the intercept while B MX is obtained from the slope of
the straight line [see Eq. (6.44)]. The equation for a mixture is similar to
Eq. (6.43) and BMX = 0 if M and X are of the same charge sign. In the case
of equilibrium constant measurements, DB values are expressed by equa-
tions similar to Eq. (6.6).
The corresponding DBo and DB? can be obtained together with the bo
values from the system of equations
pffiffiffiffiffiffiffiffi
2 0:511 Im;n
log 10 ðIm;n Þ zi pffiffiffiffiffiffiffiffi q log 10 ðaH2 O Þn
1 þ Im;n
¼ log 10 o þ B1 ½1 FðIm;n ÞIm;n þ Bo FðIm;n ÞIm;n (6:46Þ
where b(Im,n) and ðaH2 O Þn refer to the values of b and aH2 O at ionic strength
Im,n. From the values obtained for DBo and DB? and equations similar to
Eq. (6.9), one may estimate the unknown BoMX and B? MX values.
6.2.4 The Davies Equation

The Davies equation [29] has been used extensively to calculate activity
coefficients of electrolytes at fairly low ionic strengths.
The Davies equation for the activity coefficient of an ion i of charge zi
is, at 258C:
pffiffiffiffiffi
Im
log10 i ¼ 0:5102z2i pffiffiffiffiffi 0:3Im (6.47)
1 þ Im
The equation has no theoretical foundation but is found to work fairly well
up to ionic strengths of 0.1 mol kg1. It should not be used at higher ionic
strengths. The Davies equation has a form similar to the B-G-S equation but
with ion interaction coefficients equal to 0.153zi2, i.e., 0.15, 0.61, and 1.38 for
ions of charge 1, 2, and 3, respectively. These values do not agree very well
with the tabulated e values.
6.3 ION INTERACTION COEFFICIENTS AND

EQUILIBRIUM CONSTANTS FOR ION PAIRS
Two alternative methods can be used to describe the ionic medium depen-
dence of equilibrium constants:
One method takes into account the individual characteristics of the ionic
media by using a medium-dependent expression for the activity coeffi-
cients of the species involved in the equilibrium reactions. The medium
dependence is described by virial or ion interaction coefficients as used in
the Pitzer equations and in the specific ion interaction model.
The other method uses an extended Debye-Hückel expression where
the activity coefficients of reactants and products depend only on the
ionic charge and the ionic strength, but it accounts for the medium-
specific properties by introducing ionic pairing between the medium ions
and the species involved in the equilibrium reactions. Earlier, this
approach has been used extensively in marine chemistry (See Refs. [30–
32, 39]).
It can be shown that the virial type of activity coefficient equations and the
ionic pairing model are equivalent, provided that the ionic pairing is weak.
In these cases, it is in general difficult to distinguish between complex for-
mation and activity coefficient variations unless independent experimental
evidence for complex formation is available, e.g., from spectroscopic data,
as is the case for the weak uranium(VI) chloride complexes. It should be
noted that the ion interaction coefficients evaluated and tabulated by Cia-
vatta [10] were obtained from experimental mean activity coefficient data
without taking into account complex formation. However, it is known that
many of the metal ions listed by Ciavatta form weak complexes with
chloride and nitrate ions. This fact is reflected by ion interaction coefficients
that are smaller than those for the noncomplexing perchlorate ion (see
Table 6.3). This review takes chloride and nitrate complex formation into
account when these ions are part of the ionic medium and uses the value of
the ion interaction coefficient "ðMnþ ;ClO4 Þ for "ðMnþ ;Cl Þ and "ðMnþ ;NO3 Þ . In
this way, the medium dependence of the activity coefficients is described
with a combination of a specific ion interaction model and an ion pairing
model. It is evident that the use of NEA-recommended data with ionic
strength correction models that differ from those used in the evaluation
procedure can lead to inconsistencies in the results of the speciation calcu-
lations.
It should be mentioned that complex formation may also occur
between negatively charged complexes and the cation of the ionic medium.
An example is the stabilization of the complex ion UO2(CO3)35 at high ionic
strength.
6.4 TABLES OF ION INTERACTION

COEFFICIENTS
Tables 6.3–6.5 contain the selected specific ion interaction coefficients used
in this review, according to the specific ion interaction model described in
section 6.1. Table 6.3 contains cation interaction coefficients with Cl,
ClO4, and NO3; Table 6.4 anion interaction coefficients with Li +, with
Na + or NH4+ , and with K + . The coefficients have the units of kg mol1
and are valid for 298.15 K. The species are ordered by charge and appear,
within each charge class, in standard order of arrangement [33].
In some cases, the ionic interaction can be better described by

assuming ion interaction coefficients as functions of the ionic strength rather
than as constants. Ciavatta [10] proposed the use of Eq. (6.48) for cases
where the uncertainties in Tables 6.3 and 6.4 are ± 0.03 or greater.
" ¼ "1 þ "2 log 10 Im (6.48)
For these cases, and when the uncertainty can be improved as compared to
the use of a constant e, the values e1 and e2 given in Table 6.5 should be used.
It should be noted that ion interaction coefficients tabulated in
Tables 6.3–6.5 may also involve ion pairing effects, as described in section
6.3. In direct comparisons of ion interaction coefficients, or when estimates
are made by analogy, this aspect must be taken into account.
6.5 CONCLUSION
The specific ion interaction approach is simple to use and gives a fairly good
estimate of activity factors. By using size/charge correlations, it seems pos-
sible to estimate unknown ion interaction coefficients. The specific ion
interaction model has therefore been adopted as a standard procedure in the
NEA Thermochemical Data Base review for the extrapolation and correc-
tion of equilibrium data to the infinite dilution standard state. For more
details on methods for calculating activity coefficients and the ionic medium/
ionic strength dependence of equilibrium constants, the reader is referred to
Ref. 40, Chapter IX.
REFERENCES
1. Lafitte, M. J. Chem. Thermodyn., 1982, 14, 805.
2. Wanner, H. Standards and conventions for TDB publications, TDB-5.1, Rev. 1,
OECD Nuclear Energy Agency, Data Bank, Gif-sur-Yvette, France, 1989.
3. Baes, C. F., Jr.; Mesmer, R. F. The Hydrolysis of Cations; John Wiley and Sons,
New York, 1976.
4. Pitzer, K. S. Theory: Ion interaction approach. In Activity Coefficients in
Electrolyte Solutions, Pytkowicz, R. M., Ed., Vol. I, CRC Press, Boca Raton,
Florida, 1979; 157–208.
5. Brønsted, J. M. Studies of solubility: IV. The principle of specific interaction of
ions. J. Am. Chem. Soc., 1922, 44, 877–898.
6. Brønsted, J. M. Calculation of the osmotic activity functions of univalent salts.
J. Am. Chem. Soc., 1922, 44, 938–948.
7. Scatchard, G. Concentrated solutions of strong electrolytes, Chem. Rev., 1936,
19, 309–327.
8. Guggenheim, E. A. Applications of Statistical Mechanics; Oxford, Clarendon
Press, 1966.
9. Scatchard, G. Equilibrium in Solution Surface and Colloid Chemistry; Harvard

University Press, Cambridge, MA, 1976.
10. Ciavatta, L. The specific interaction theory in evaluating ionic equilibria. Ann.
Chim. (Rome), 1980, 70, 551–567.
11. Guggenheim, E. A. Phil. Mag. 1935, 19, 588.
12. Vasilév, V. P. Influence of ionic strength on the instability constants of com-
plexes. Russ. J. Inorg. Chem., 1962, 7, 8, 924–927.
13. Helgeson, H. C.; Kirkham D. H.; and Flowers, G. C. Theoretical prediction of
the thermodynamic behavior of aqueous electrolytes at high pressure and
temperatures: IV. Calculation of activity coefficients, osmotic coefficients and
apparent molal properties to 6008C and 5 kb, Am. J. Sci., 1981, 281, 1249–
1516.
14. Pitzer, K. S.; Brewer, L. Thermodynamics, 2nd Ed., McGraw-Hill, New York,
1961.
15. Pitzer, K. S. Thermodynamics of electrolytes: I. Theoretical basis and general
equations. J. Phys. Chem., 1973, 77, 268–277.
16. Pitzer, K. S.; Mayorga, G. Thermodynamics of electrolytes: II. Activity and
osmotic coefficients for strong electrolytes with one or both ions univalent. J.
Phys. Chem., 1973, 77, 2300–2308.
17. Pitzer, K. S.; Kim, J. J. Thermodynamics of electrolytes: IV. Activity and
osmotic coefficients for mixed electrolytes. J. Am. Chem. Soc., 1974, 96, 5701–
5707.
18. Pitzer, K. S.; Mayorga, G. Thermodynamics of electrolytes: III. Activity and
osmotic coefficients for 2–2 electrolytes. J. Sol. Chem., 1974, 3, 539–546.
19. Pitzer, K. S. Thermodynamics of electrolytes: V. Effects of higher-order elec-
trostatic terms. J. Sol. Chem., 1975, 4, 249–265.
20. Pitzer, K. S.; Silvester, L. F. Thermodynamics of electrolytes: VI. Weak elec-
trolytes including H3PO4. J. Sol. Chem., 1976, 5, 269–278.
21. Pitzer, K. S.; Peterson., J. R.; Silvester, L. F. Thermodynamics of electrolytes:
IX. Rare earth chlorides, nitrates, and perchlorates. J. Sol. Chem., 1978, 7, 45–
56.
22. Robinson, R. A.; Stokes, R. H. Electrolyte Solutions; Butterworth, London,
1959.
23. Stokes, R. H. Thermodynamics of solutions. In Activity Coefficients in Elec-
trolyte Solutions, Pytkowicz, R. M., Ed., Vol. I, CRC Press, Boca Raton,
Florida, 1979; 1–28.
24. Wanner, H. Guidelines for the Assignment of Uncertainties. TDB-3, OECD
Nuclear Energy Agency, Data Bank, Gif-sur-Yvette, France, 1989.
25. Spahiu, K. Carbonate complex formation in lanthanoid and actinoid systems;
Ph.D. thesis, Royal Institute of Technology, Stockholm; 1983.
26. Ferri, D.; Grenthe, I.; Salvatore, F. Studies on metal carbonate equilibria: VII.
Reduction of the tris (carbonato) dioxourante(VI) ion, UO2 ðCO3 Þ43 , in carbo-
nate solution. Inorg. Chem., 1983, 22, 21, 3162–3165.
27. Friedman, B. L. Ionic Solution Theory, Interscience Publishers, New York,
1962.
28. Silvester, L. F.; Pitzer, K. S. Thermodynamics of electrolytes: VIII. High-

temperature properties, including enthalpy and heat capacity, with application
to sodium chloride. J. Phys. Chem., 1977, 81, 1822–1828.
29. Davies, C. W. Ion Association; Butterworths, Washington, 1962.
30. Johnson, K. S.; Pytkowicz, R. M. Ion association and activity coefficients in
multicomponent solutions. In Activity Coefficients in Electrolyte Solutions;
Pytkowicz, R. M., Ed., Vol. II, CRC Press, Boca Raton, Florida, 1979; 1–62.
31. Millero, F. J. Effects of pressure and temperature on activity coefficients. In:
Activity Coefficients in Electrolyte Solutions; Pytkowicz, R. M., Ed., Vol. II,
CRC Press, Boca Raton, Florida, 1979, pp. 63–151.
32. Whitfield, M. Activity coefficients in natural waters. In: Activity Coefficients in
Electrolyte Solutions; Pytkowicz, R. M., Ed., Vol. II, CRC Press, Boca Raton,
Florida, 1979; pp. 153–299.
33. Wagman, D. D.; Evans, W. H.; Parker, V. B.; Schumm, R. H.; Halow, I.; Bailey,
S. M.; Churney, K. L.; Nuttall, R. L. The NBS tables of chemical thermo-
dynamic properties; Selected values for inorganic and C1 and C2 organic sub-
stances in S1 units. J. Phys. Chem. Ref. 1982, 11, 1–392.
34. Ciavatta, L. Personal communication, Universita de Napoli, Naples, June
1988.
35. Hedlund, T. Studies of complexation and precipitation equilibria in some
aqueous aluminium(III) systems; Ph.D. thesis, University of Umeå, Umeå,
Sweden, 1988.
þ
36. Riglet, Ch.; Robouch, P.; Vitorge, P. Standard potentials of the (MO2þ 2 =MO2 )
4+ 3+
and (M /M ) redox systems for neptunium and plutonium. Radiochim.
Acta, 1989, 46, 85–94.
37. Grenthe, I.; Fuger, J.; Konings, R. J. M.; Lemire, R. J.; Muller, A. B.; Nguyen-
Trung, C.; Wanner, H. Chemical Thermodynamics of Uranium, Wanner, H.,
Forest, I., Eds., Elsevier Science Publishers, Amsterdam, 1992.
38. Bruno, J. Stoichiometric and structural studies on the Be2+-H2O-CO2(g) sys-
tem; Ph.D. thesis, Royal Institute of Technology, Stockholm, 1986.
39. Pytkowicz, R. M. Activity coefficients, ionic media, and equilibrium in solu-
tions; in: Activity Coefficients in Electrolyte Solutions; R.M. Pytkowicz, Ed.;
Vol. I; CRC Press, Boca Raton, Florida, 1979; 301–305.
40. Grenthe, I.; Puigdomenech, I. Modelling in Aquatic Chemistry; OECD/NEA,
Paris, 1997.
7
Development of Industrial Solvent
Extraction Processes
GORDON M. RITCEY Gordon M. Ritcey & Associates, Inc., Nepean,
Ontario, Canada
7.1 INTRODUCTION
The development of a solvent extraction process from a given aqueous feed
solution may be confined by several restraints. For example, the temperature,
flow rate, acidity, and so forth may be (essentially) fixed, and the solvent
extraction process developed must then be capable of accommodating these
restrictions.
The initial bench-scale experimental investigations into solvent extrac-
tion processes are conducted with small apparatus, such as separating fun-
nels. Following the successful completion of these tests, when the best reagent
and other conditions for the system have been established, small-scale con-
tinuous operations are run, such as in a small mixer-settler unit. The data so
obtained are used to determine scale-up factors for pilot plant or plant design
and operation (see Chapters 7 and 8).
The need for pilot plant operations is considered by one school of
thought to be an unnecessary expense in the development of a solvent pro-
cess. The rationale here is that monies and time spent in piloting the process
can be used to modify, if necessary, the full-scale plant. Thus, if few (or no)
modifications are required, considerable savings can result. This approach
is probably satisfactory if the process is very similar to an existing process
and sufficient data are available on which to base the plant design and
operation. The practical experience of the plant design engineers and oper-
ators is also of considerable importance in such cases. On the other hand,
problems encountered in a plant operation that has been designed without
piloting the process could result in the loss of considerable time and expense.
277
278 Ritcey
Of course, almost any process or plant can be made to work, even though the
basic data and design are poor. The costs, however, may be astronomical! It
should be pointed out, however, that the size of a pilot plant could vary from
a few gallons per minute of total throughput to several hundred gallons per
minute. Normally, one would expect that the size of a pilot plant would be in
direct proportion to the size of the final plant.
This chapter comprises some selected excerpts taken from a com-
prehensive two-volume text on the subject coauthored by the author [1]. It
specifically addresses the extraction of metals, but many of the (general)
conclusions are equally valid for the extraction of organic compounds, with
appropriate modifications.
7.2 INITIAL STUDIES

In the rapid determination of whether an extractant may be useful for
the extraction of a particular substance or to determine the best extractant in
a series of extractants, an approach known as ‘‘screening’’ may be employed.
Much of the usefulness of this approach depends on the investigator’s
knowledge and experience of solvent extraction, since the tests carried out
are minimal and are generally based on the relative properties of an
extractant.
In screening tests, it is common practice to use approximately 0.1 mol
dm3 or 5 vol% solutions of extractant, and extraction tests are conducted at
a phase ratio of 1, with a 5 min contact time. These conditions, of course, may
be varied, but the object is to stay with as simple an approach as possible.
Analysis of only the aqueous raffinate need be carried out, and distribution
data calculated from this.
The initial requirement in the development of a solvent extraction
process for the recovery or separation of metals from an aqueous solution
is knowledge of the solution composition, pH, temperature, and flow rate.
Both pH and temperature can be adjusted, within certain economic limits,
before feeding to the solvent extraction circuit, but only in a few cases can
the leaching or dissolution conditions be dictated by the extraction process.
Consequently, no serious development work on the extraction process can
be carried out before the leaching conditions or the type of feed solution are
established.
Properties of the feed solution and the substance to be extracted will
decrease the number of extractants that may be applicable. For example, in
the extraction of metals, if no anionic metal species are present in the feed
solution, there is little point in considering anionic (amine) extractants.
On the other hand, if anionic metal species are present, then the amine type
best suited to the extraction of the anionic species can be selected, knowing
Industrial Solvent Extraction Processes 279
that primary amines generally extract anionic sulfate complexes best,

whereas tertiary and quaternary amines extract anionic chloro-complexes
best. In this way, a considerable saving in both time and money can be
achieved.
A literature search to determine whether any existing processes are
likely to be applicable should be undertaken. Here, the extent of the search
will depend on the knowledge and experience of the investigator. After
selecting the most suitable class of extractant, the next stage is to determine
the various parameters of the extraction systems including the choice of
diluent and modifier, which with the extractant make up the solvent. These
should include specificity of the selected extractants for the metal (or com-
pound) to be extracted; the requirements for the type of diluent and modifier;
loading and stripping characteristics; pH dependency; kinetics of extraction,
scrubbing, and stripping; concentration of extractant and phase ratio re-
quired; solubility of the solvent in the aqueous phase; and the physical
attributes of the system. Many of these aspects are discussed in some detail in
previous chapters.
7.3 SOLVENT PRETREATMENT OR CONDITIONING

Almost all solvent solutions used in solvent extraction processes require
treatment, either before their initial use or after being stripped of the
extracted species. After preparation of the solvent by mixing appropriate
amounts of extractant, modifier, and diluent, the solution is treated
by contacting with an appropriate aqueous solution before entering the
extraction stage. Such pretreatment is generally referred to as equilibration
of the solvent.
The object of equilibration is to provide a solvent that will effectively
extract the required species, either because of the form of the active con-
stituent of the solvent or by maintaining the necessary extraction pH. As an
example of the former condition, consider the extraction of uranium using a
tertiary aliphatic amine as extractant. Extraction of metal species by such
amines is considered to occur by liquid ion exchange (see Chapters 3 and 4).
For a tertiary amine to act in this manner, it must be first converted to an
amine salt:
R3 NðorgÞ +Hþ +A $ R3 NHþ A

ðorgÞ (7.1)
This is accomplished by contacting the solvent with an appropriate aqueous

acid, for example, sulfuric acid:
2R3 NðorgÞ +H2 SO4 $ (R3 NH)2 SO4ðorgÞ (7.2)

280 Ritcey
Having formed the amine salt, the solvent can be used for the extraction of
uranium. This is shown as an ion exchange process in Eq. (7.3):
2[(R3 NH)2 SO4 ]ðorgÞ +UO2 ðSO4 Þ4
3
$ [(R3 NH)2 UO2 (SO4 )3 ]ðorgÞ +2SO2

4 (7:3)
in which the anionic sulfate moiety of the amine salt (organic phase) is
exchanged for the anionic uranyl sulfate species (aqueous phase), to form a
uranyl amine sulfate complex as the extracted species (anion exchange).
Stripping the uranium from the solvent can be accomplished by using
either acid or alkaline solutions. If an alkali carbonate solution is used, the
stripped solvent then requires equilibrating with sulfuric acid before recy-
cling to the extraction stage. Sulfuric acid stripping obviates the need for
such equilibration.
Acidic extractants that undergo a cationic exchange reaction with
metal ions do not generally require equilibration before extraction, because
the exchange involves protons that are already present in the extractant:
Mnþ +nHXðorgÞ $ MXnðorgÞ +nHþ (7.4)
and, as before, if alkaline stripping is employed, equilibration with acid is

required before recycling the solvent to the extraction stage. These equi-
librium equations are discussed in detail in Chapters 3 and 4.
Problems tend to arise in cationic systems when a particular pH is to be
maintained. The protons resulting from the exchange increase the acidity of
the aqueous phase, which can alter (lower) considerably the extraction of a
metal ion. For example, the separation of cobalt and nickel using di-2-
ethylhexylphosphoric acid (DEHPA) as the extractant is carried out at pH
5–6. This system is discussed extensively in section 11.3, where it is pointed
out that the use of a nonequilibrated solvent results in lowering the pH of the
aqueous phase to such an extent that very poor cobalt extraction is achieved.
It has been shown that if the ammonium salt of DEHPA is used, rather than
the free acid, a pH *5.5 can be easily maintained without having to add
ammonia continuously throughout the extraction stage to maintain the
optimum extraction pH (a messy operation). The exchange of ammonium
ions for metal ions produces an ammonium salt (nitrate, sulfate, or other,
depending on the anion present in the aqueous feed). These salts are almost
neutral in aqueous solution, but they have a buffering effect. Thus, an aque-
ous feed pH as low as 3.5 can be used, since the equilibrium pH is maintained
at between 5 and 6. The reactions involved are:
[(RO)2 PO(OH)]ðorgÞ +NH4 OH $ [(RO)2 PO(ONH4 )]ðorgÞ (7.5)

2[(RO)2 PO(ONH4 )]ðorgÞ +CoSO4

$ [ð(RO)2 POO)2 Co]ðorgÞ +(NH4 )2 SO4 (7.6)
A similar situation occurs using the sodium salt of DEHP.

Stripping of cobalt from the loaded extract is accomplished by various
acids, such as nitric acid:
[((RO)2 POO)2 Co]ðorgÞ +2HNO3

$ 2[(RO)2 PO(OH)]ðorgÞ +Co(NO3 )2 (7.7)
and the solvent phase is then equilibrated with ammonia [Eq. (7.5)] for
recycling to the extraction stage.
Commercially produced extractants are not pure and always contain
some residual starting materials, by-products formed during manufacture, or
degradation products. For example, DEHPA contains some 2-ethylhexanol
(starting material) and some of the monoester acid. Tertiary amines invari-
ably contain primary or secondary amines, and so on. If the impurities in
the extractant used in the solvent are more soluble than the extractant it-
self, which is usually the case, they will be lost from the system to the aqueous
phase after some recycling of the solvent. The concentration and solubility of
such impurities will depend on the time required to remove them from the
system. The problems associated with such impurities become particularly
evident if their effects are not considered during bench-scale investigations.
Thus, their presence in a solvent system can produce cruds, good or poor
phase separation, and enhanced or poor loading characteristics. Such effects
could result in abandoning a particular solvent because of poor chemical and
physical characteristics; conversely, problems could arise when the process is
tried on a continuous scale if the enhancing effects of the impurities result in
poor extraction characteristics as a result of their loss to the raffinate. It is
suggested, therefore, that all solvents be conditioned before use in bench-
scale studies that will ultimately be developed into a continuous system.
Although there is no specific method of conditioning, a rule of thumb
is to treat the freshly prepared solvent with a solution similar to that used in
the test work. This may require several contacts with the aqueous solution,
with stripping between contacts if a metal is extracted.
Inclusion of this practice in all solvent extraction studies will ensure
that the solvent is not discarded as being unsuitable, and that problems in
pilot plant or continuous operations are not the result of impurities that were
present in the bench-scale test work.
It should also be pointed out that extractant quality might vary from
sample to sample. This can be particularly frustrating in both development
282 Ritcey
work and pilot plant operations. It is good practice to test every batch of
extractant before use, to ensure that its characteristics are similar to the
material previously used. Such testing will also provide the investigator with
the knowledge that the reagent is as ordered, and is not another reagent
entirely. It is not uncommon for the wrong material to be shipped.
Similar comments can be applied to diluents and modifiers. The inves-
tigator should be aware of the problems that can arise from impurities in the
components that go to make up a solvent.
7.4 DISTRIBUTION DATA

7.4.1 Extraction
Distribution data can be obtained in two ways. The first employs variation
of the phase ratio of the aqueous and organic phases; the second involves
recontacting the organic phase with fresh aqueous phase until the saturation
loading of the solvent is reached.
In the phase ratio variation method, the aqueous and organic phase are
mixed mechanically until equilibrium is reached at phase ratios between
about 1:10 and 10:1. After allowing the phases to disengage, they are sepa-
rated and analyzed for the metal or metals of interest. It is necessary to
remove any entrained solvent from the aqueous phase, and any aqueous
phase from the organic phase, before analysis. This can be accomplished
by centrifuging the phases or by filtering the aqueous phase through dry,
coarse filter paper, and the organic phase through a silicone-treated phase-
separating paper.
It must be emphasized that the equilibrium pH must be the same in all
tests. One way of ensuring this is to check the pH of the aqueous phase after
equilibration of the phases, adjust if necessary by the addition of acid or
alkali, and then mix the phases again till equilibrium is reached.
A distribution isotherm is then constructed by plotting the metal
concentration in the organic phase against the concentration in the aqueous
phase, as a function of the phase ratio. An example of such an isotherm is
shown in Fig. 7.1, for the extraction of nickel by DEHPA(Na) at pH 6,
showing three different concentrations of extractant [1].
The second method of constructing an extraction isotherm initially
requires the choice of a suitable phase ratio, and the organic and aqueous
(feed) phases are contacted at this ratio until equilibrium is obtained. The
phases are allowed to separate, and the aqueous phase is removed and
analyzed. A small measured portion of the organic phase is also taken for
analysis. Fresh aqueous feed is then added to the remainder of the organic
phase, in an amount to give the same phase ratio as that originally used. The
Fig. 7.1 Distribution isotherms for nickel at pH 6 for three different concentrations
(vol%) of DEHPA(Na), in the kerosene-aqueous sulfate systems.
phases are again contacted until equilibrium is obtained, and the same
procedure performed as just described. This process is repeated until sat-
uration of the solvent with the metal is obtained. Here again, care must be
taken to maintain the same pH throughout the procedure. One drawback
to this method is that a large volume of the organic phase must be used
initially to allow for sampling, especially if numerous contacts are required.
Generally, the first method employing variation in the phase ratio is
recommended.
The extraction isotherms constructed from these data can be used in
the construction of a McCabe-Thiele diagram (see Chapter 8).
The loading capacity of the solvent can be obtained from either of
these two methods for the particular concentration of extractant used. It
cannot be assumed that the loading capacity is linear with increasing
extractant concentration (see Fig. 7.1). If high concentrations (>10 vol%)
are to be used, it is advisable to determine the loading capacities at these
concentrations. This is readily done either by contacting a solvent several
times with fresh aqueous solution, or by using a concentrated solution of
284 Ritcey
the metal with a high (aqueous/organic) ratio, until maximum loading is

achieved.
For most exploratory work, analysis of the organic phase is not
necessary. If no volume change of the phases occurs and no third phase or
crud is formed, the analysis of the aqueous raffinates is sufficient, since the
metal concentration in the solvent can be readily calculated from the initial
metal concentration of the feed solution and the phase ratio used.
The determination of the variation of metal extraction with equilib-
rium pH is carried out in much the same manner as described earlier, and is
described in detail in Chapter 4.
7.4.2 Stripping
Data required for the construction of stripping isotherms are obtained in a
manner similar to those for extraction. A general procedure is as follows:
the loaded solvent is contacted with a suitable strip solution (e.g., acid,
base, etc.) at an appropriate phase ratio until equilibrium is attained. The
aqueous phase is then removed, fresh strip solution is added to the organic
phase, and the procedure repeated. This process is continued until all (or as
much as possible) of the metal has been stripped from the organic phase.
Analysis of the aqueous strip liquors, if no volume changes occur, allows
the calculation of the concentration of metal in the organic phase after
each strip. However, if volume changes do occur, then the organic phases
must also be analyzed. Then the stripping isotherm can be drawn and a
McCabe-Thiele diagram constructed to determine the number of theore-
tical strip stages required at the phase ratio and strip solution concentra-
tion used.
7.4.3 Kinetics of Extraction and Stripping

The time required for equilibrium to be reached in both of these operations
will need to be known to design a plant.
The time required for a system to reach equilibrium can be determined
by shake-out tests, as described in earlier sections. Contact times are varied
between about 0.5 and 15 min, at suitable intervals, and the extraction coef-
ficient for each contact time plotted as a function of time. With this method,
there is a lower practical limit on the contact time of about 0.25 min. These
data will not be directly applicable to a continuous process because the rate
of metal extraction is a function, in part, of the type and degree of agitation.
However, a good idea of whether the extraction rate is sufficiently fast for
the system to be suitable for use in a large contactor can be obtained. For
example, if equilibrium is attained in less than 1 min, almost any type of
contactor may be used.
A procedure has been reported using data on the kinetics of mass

transfer to estimate the Murphree stage efficiencies in continuous counter-
current mixer-settlers so that the number of real contacting stages can be
calculated for a wide range of operating conditions [2]. Extensive test work
can thus be minimized by using a limited set of statistically designed experi-
ments to determine the effect on kinetics of mass transfer of the major
operating variables. Figure 7.2 shows the noticeable effects on the increase
in the rate coefficients as the stirring speed is increased in a Kelex-copper
sulfate system. The dispersion of the solvent resulted in poorer kinetics than
dispersion of the aqueous phase. There was little effect on the kinetics from
change in phase ratio. Data were also presented for the effects on sedi-
mentation and coalescence.
Stripping rates are also of importance in commercial operations,
and these can be determined in a manner similar to those previously des-
cribed.
Fig. 7.2 Effect of agitation speed on the rate coefficient.

286 Ritcey
7.4.4 Scrubbing
The object of scrubbing a loaded solvent is to remove as much as possible of
any unwanted coextracted metal. There are many possibilities of accom-
plishing this by varying the scrub solution. Normally, however, scrubbing is
achieved by water, dilute acid, or base, or by an aqueous solution of a salt of
the metal of primary interest in the solvent phase. Scrubbing tests are carried
out in a manner similar to those described earlier with the loaded solvent
being contacted with the scrub solution at the appropriate concentration, pH,
phase ratio, contact time, and temperature. After separating the phases, the
organic phase is usually analyzed to determine whether the unwanted metals,
and how much of the metal of interest, have been removed. The best scrub
solution and conditions are determined by varying the individual conditions.
The scrub raffinate may contain substantial amounts of valuable
metals; thus, consideration should be given to where in the overall process
this raffinate should be recycled.
Data obtained in scrubbing tests can be presented graphically, such as
by plotting the concentration of metals in the organic phase against A/O
ratio, salt concentration in the scrub solution, temperature, etc. Typical plots
for the removal of nickel from cobalt in a DEHPA-containing solvent are
shown in Figs. 7.3 and 7.4 [3].
7.5 EXTRACTANT CONCENTRATION

The concentration of extractant required to extract a metal at a phase ratio
suitable for a given type of contactor will depend essentially on the metal
concentration in the aqueous feed solution. Accordingly, low metal con-
centrations will not require high extractant concentrations in the solvent,
since the large phase ratio needed in such cases (so that large volumes of
unused solvent will not be cycling through the circuit) may not be compa-
tible with the contactor. On the other hand, high metal concentrations in the
feed would probably require high extractant concentrations in the solvent to
maintain a suitable phase ratio.
The loading characteristics of an extractant can be determined as
described above. With this information and knowledge of the different types
of contactors available, the metal concentration in the feed, the extrac-
tion rate of the metal, the approximate concentration of extractant in the
solvent can be determined. The optimum concentration can be determined
only by operating a continuous circuit.
One problem that should be mentioned here again is that of the
coextraction of a second (unwanted) metal, as discussed in Chapter 8. This
Fig. 7.3 Effect of scrub solution composition on the removal of nickel from cobalt in a
DEHPA-kerosene solvent.
might occur if this second metal has extraction characteristics similar to

those of the metal that is preferentially extracted. To inhibit coextraction, it
is advisable to operate a circuit at as close to saturation loading of the
solvent with the metal of primary interest as is practical; in this way,
coextraction of other metals is depressed. As the extractant concentration in
the solvent in a continuous operation is fixed, if fluctuations in the con-
centration of metal in the feed occur, any decrease in concentration will
result in coextraction of the unwanted metal. Conversely, an increase in
metal concentration would result in a loss of metal to the raffinate. This
problem is usually minimized by using a large feed tank for the solvent
extraction circuit to level out such fluctuations in metal concentration to the
solvent extraction circuit.
288 Ritcey
Fig. 7.4 Effect of temperature on the scrubbing of nickel from a DEHPA-kerosene

solvent.
7.6 DILUENT AND MODIFIER REQUIREMENTS

Probably the first tests that should be carried out to determine the best
diluent for the system are those concerned with coalescence rates of the
phases. These can be done in several ways, but all comparisons should be
made using the same experimental conditions.
One simple, but effective, method is to make up solvents with each of
the diluents to be tested, with the same concentration of extractant (and
modifier if necessary) in each diluent. Place measured volumes of the solvent
and aqueous phases in a graduated beaker or suitable container, and mix for
a given period at the same phase continuities. The degree of agitation should
be the same for all tests and is best carried out with a mechanical shaker or
stirrer [4]. As soon as the desired agitation time has elapsed, start a timer,
and place the container in an upright position. Measure the length of time
required for the primary and secondary phase breaks to occur. The rate of
coalescence can also be determined by timing the rise of the coalescing band
up the container.
The effects of different diluents on the extraction and stripping prop-

erties of a solvent can be determined by similar mixing tests as described
above.
The formation of a third phase during the extraction of a metal by
solvent (extractant and diluent) cannot be tolerated in liquid–liquid extrac-
tion processes for obvious reasons. Elimination of a third phase is usually
accomplished by the addition of a modifier to the solvent, or by increasing
the temperature of the system.
The amount of modifier required to prevent third phase formation
can be determined in the following way. The aqueous and solvent phases
are first contacted, and once the three phases have separated, the lower
aqueous phase is drawn off and discarded. The modifier to be considered is
then added from a burette in small increments to the two organic phases,
and the mixture shaken after each addition. The amount of modifier
required to produce a single organic phase is then used to calculate the
amount required to be added to the solvent. Generally, about 2–5 vol% of
modifier is needed, but more may be necessary if high concentrations of
extractant are used in the solvent. Any effects of modifiers on the kinetics
and equilibria of metal extraction and stripping can be determined by shake-
out tests.
7.7 TEMPERATURE EFFECTS

Temperature can have a considerable effect on both the extraction and
stripping properties of a solvent extraction system relative to equilibrium,
kinetics, and metal separations [5] (see Chapters 3–5). Therefore, it is advis-
able to investigate these effects, especially when the solvent tends to be vis-
cous or high loadings of metal are to be obtained in the solvent (Fig. 7.5).
7.8 STAGEWISE SEPARATIONS

Many of these experimental methods can be used in the development of
systems more complex than the extraction of a single substance, such as
metal separations. Metal separation processes may involve as few as two
metals and as many as 15, as in the rare earths (see Chapter 11).
Consider the separation of two chemically similar metals having very
similar extraction properties, such as cobalt and nickel. Some of the avail-
able commercial extractants show very poor discrimination between these
metals. Obviously, a few stages of extraction will be insufficient to obtain a
good separation. An approximate idea of how many stages are required can
be obtained by contacting a solvent several times with fresh feed containing
290 Ritcey
Fig. 7.5 The effect of temperature on the loading and separation of cobalt and nickel.
Solvent: 15 vol% DEHPA; 5 vol% TBP; 80 vol% kerosene.
both these metals. Analysis of the solvent after each contact stage, and
plotting these concentrations against the number of stages, will give a curve
for each metal similar to those shown in Fig. 7.6. Without completing the
curves, a good approximation of the number of stages required to provide
a given Co/Ni ratio can be obtained by extrapolation.
In the example shown in Fig. 7.6, using an alkylphosphoric acid, about
60 stages are required to give a Co/Ni ratio of about 100. So many stages
would be too many for mixer-settler operation, and other types of con-
tactors would have to be considered. In this particular example, a sieve plate
pulsed column has been shown to be very effective [3]. However, with the
development of the alkylphosphonic and alkylphosphinic acids, the sep-
aration of cobalt and nickel can be achieved in very few stages, owing to the
high rejection of nickel (see Chapter 11).
It is a good plan to begin solvent extraction studies involving several
metals with solutions containing only single metals to obtain the basic
extraction data. It also provides time for any necessary analytical develop-
ment to be carried out. However, synthetic solutions should not be worked
on longer than necessary, as the actual plant liquor will almost certainly be
much different. Another point to be made here is that in leaching, filtering,
precipitation, and flotation, chemicals are added to aid these processes.
These chemicals can occasionally cause adverse effects, such as poor phase
disengagement, during solvent extraction (see section 7.13).
7.9 SCALE-UP DECISIONS

As in any process development work, the data have to be examined critically
to make certain that there is sufficient reason to carry the project on to the
next phase. Thus, if a particular system has poor stripping characteristics, it
is very doubtful whether any improvement can be made by going to a larger
scale, in which case the investigation may have to be terminated. However, if
the bench data have produced sufficient information to draw a conceptual
flow sheet, then a decision can be made on whether to run a small-scale
continuous operation or a pilot plant. Many other factors have to be con-
sidered in making this decision, such as the economics of the process, cost of
Fig. 7.6 Cobalt-nickel separation as a function of number of contact stages: con-

centration of cobalt and nickel 20 g dm3 each in aqueous feed solution.
292 Ritcey
construction, materials and labor, sale of the anticipated product, possible

by-products, environmental aspects, and so on.
In the development of solvent extraction systems, the initial con-
siderations of the chemistry of the system will limit the choice of extractant,
diluent, and modifier. Shake-out tests and a study of extraction and strip-
ping characteristics will further limit the choice. Simple graphic construction
of extraction and stripping isotherms will permit the design of a small-scale
continuous operation in terms of the number of stages and flow rates.
In summary, the following parameters are required before considering
scale-up to a larger system:
The solvent system: extractant, diluent, and, if necessary, modifier
Equilibrium data: extraction and stripping isotherms
Kinetic data, as these will govern, to a large extent, the type of contactor
required
Approximate number of theoretical stages for extraction and stripping
The necessity or desirability of a scrub stage between extraction and stripping
The physical characteristics of the system: i.e., dispersion requirements and
settling rates, emulsion tendencies, viscosity, etc.
Approximate solubility of the solvent in the aqueous phase
Temperature effects
Solvent pretreatment and conditioning requirements
7.10 DISPERSION, COALESCENCE,

AND CONTACTORS
The transfer of a metal from one phase to another phase is only one aspect
of the solvent extraction process; equally important are the physical aspects
of the system. These are primarily concerned with (1) the dispersion of the
phases on mixing, (2) the type of droplet formation, and (3) the rate and
completeness of coalescence. These are important aspects in the choice of
suitable contacting equipment and in the design and operation of the plant.
Because these topics are extensively discussed in Chapter 9, only the
essential points are summarized briefly here.
The rate of mass transfer is a function, among other variables, of the
drop size distribution or interfacial area between the phases. The drop size is
governed by the surface tension, and densities of the two phases and the type
of agitation and design of the contactor. Up to a point, the smaller the drop,
the greater the rate of mass transfer.
Coupled with increasing dispersion, however, is increased coalescence
time and, therefore, the size requirements of the settler. Also, as the size of
the droplets of the dispersed phase decreases, the droplets behave more and
more like rigid spheres, and so the rate of mass transfer decreases. Thus, the
kinetics, along with the proper dispersion and coalescence, will influence the
choice of contacting equipment. In addition to drop size, other factors
affecting coalescence include the temperature, viscosity, and density of the
phases, and the presence of surface-active agents, or solids. The addition of
wetting materials within the contactor or phase separation equipment will
promote the coalescence rate.
Coalescence can be divided into primary and secondary break time.
Primary break time is the time required for the two phases to meet at
a sharply defined interface and provides a measure of the settler or phase
separation requirement of the process. The secondary break concerns the
fine haze of droplets and is the major cause of entrainment losses. The speed
and completeness of phase disengagement have a marked effect on the
capital and operating costs of the plant and are directly linked to the design
and method of operating of the mixer. In general, but only to a point, higher
extraction efficiencies are achieved as the speed of the mixer is increased and
the resulting droplet size of the dispersion is reduced. However, the dis-
persion band thickness increases as the droplet size of the dispersion is
reduced, and a larger settler area (with consequently larger inventories of
solvent) are then required for a given plant throughput. Figure 7.7 shows the
variation of dispersion band depth with the total flow/unit settler area under
varying conditions of mixing, expressed by the term nR3d 2 where nR is the
speed of the impeller (rev/s) and d its diameter (ft) [6].
Increased capital costs will be incurred if the mixer is designed to
produce a high proportion of extremely small droplets, and this cost has to
be balanced against the increased revenue obtained from the improved
extraction efficiency. At upper levels of N3D 2, the problem of organic carry-
over with the aqueous phase may become an important consideration.
In addition to the directly calculable increase in settler costs when operat-
ing with very highly dispersed phases, the possibilities of emulsion forma-
tion and flooding caused by the accumulation of ‘‘crud’’ in the settler must
be considered. Both of these become more likely as the settler duty be-
comes more severe. Many authors have cited design ideas on mixer settlers
[7, 8].
In all practical mixing systems, a range of droplet sizes is produced by
the mixer, and there will always be a proportion of droplets of the dispersed
phase that will not settle out readily and, therefore, will remain entrained in
the continuous phase leaving the settler. Impellers that consist of enshrou-
ded radial vanes create very high shear rates within the mixer, particularly
when they are run close to the bottom of the mixing tank to provide a
pumping as well as a mixing action. These high-shear rates are the prime
cause of the haze of fine droplets produced in many plants. Thus, the design
294 Ritcey
Fig. 7.7 Effect of variation of mixer N3D2 on settler dispersion band depth.
of the impeller blades is important in minimizing organic entrainment in the

aqueous phase [7].
The major emphasis in the development and optimization of con-
tactors has been to increase throughput while maintaining efficiency, and
to increase efficiency without reduction in throughput [1,5]. Many types of
contactors are available for achieving mass transfer, each with its own
particular advantage so that there is no single contactor that is best for all
solvent extraction processes, either technically or economically. A classifi-
cation chart of various contactor types is given in Fig. 7.8 [1].
Mixer-settlers have been the more common type of equipment and,
with the development of hydrometallurgy over the past 20 years, designs
have improved considerably. To select the appropriate equipment, a clear
understanding of the chemical and physical aspects of the process is required.
Also the economics must be considered relative to the type of equipment to
suit particular conditions of given throughput, solution and solvent type,
kinetics and equilibrium, dispersion and coalescence, solvent losses, number
of stages, available areas, and corrosion.
The design procedure for extracting equipment requires the evaluation

of the following:
Number of stages of extraction, scrubbing, and stripping
Stage efficiency
Flow capacity
Droplet size
The operation of equipment is dependent on the dispersion and
coalescence of the phases to achieve mass transfer. For example, systems
with high mass transfer rates do not require the formation of the extre-
mely fine droplets that systems of low rates of mass transfer usually
require.
In stagewise equipment, the design and scale-up is simple and is often
determined from bench data. The stage efficiency is usually high, and the
capacity is determined by the settler design necessary to achieve coalescence
of the dispersed phases. For differential contactors, such as columns, the flow
capacity is determined by the droplet size and the type of internals (see
Fig. 7.8 Contactor classification.

296 Ritcey
Chapter 9). Differential contactors are more susceptible than mixer-settlers

to axial and backmixing. If plug flow can be used, there would be little
problem, but this rarely occurs. Backmixing can be caused by localized high
velocities in the continuous phase, resulting in:
1. Entrained dispersed phase droplets, causing disperse-phase backmixing
2. Turbulent eddies in the continuous phase
3. Nonuniform velocity distribution normal to the direction of the flow
The net result of such deviations from countercurrent plug flow is to decrease
mass transfer as well as the total throughput per cross-sectional area. If no
allowance is made for backmixing, the resulting plant will be considerably
underdesigned.
Some advantages and disadvantages of stagewise and differential
contactors are shown in Table 7.1 [1].
The types of equipment used, which range from stirred tanks and
mixer-settlers to centrifugal contactors and various types of columns, affect
both capital and operating costs [9]. In the decision to build a plant, the
choice of the most suitable contactor for the specific situation is most
important. In some systems, because of the chemistry and mass transfer rates
involved, several alternative designs of contacting equipment are available.
In the selection of a contactor, one must consider the capacity and stage
requirements; solvent type and residence time; phase flow ratio; physical
properties; direction of mass transfer; phase dispersion and coalescence;
holdup; kinetics; equilibrium presence of solids; overall performance; and
maintenance as a function of contactor complexity. This may appear very
complicated, but with some experience, the choice is relatively simple.
A comparison of contactors is not valid unless the same system is being
compared under the same process conditions of phase ratio, temperature,
solution composition, etc. Because not all contactors perform equally for a
given process, comparative tests must be run under identical conditions to
yield reliable physical and economic comparisons. In general, for systems
requiring up to four theoretical stages, mixer-settlers probably offer the most
economic solution; however, for systems in which the partition coefficients
are small, therefore necessitating many stages of extraction, some type of
differential contactor is usually preferred. An analysis of the many factors
will generally result in the selection of the most economic contactor that is
capable of performing the required separation.
7.11 SMALL-SCALE CONTINUOUS TESTING

Commercially available equipment for small-scale continuous test work on a
solvent extraction process is limited. Generally, a series of small mixer-settlers
Table 7.1 Equipment Performance
Nonagitated Agitated
Mixer-settlers differential differential Centrifugal
Advantages Good contacting of Low initial cost Good dispersion Handle low gravity
phases Low operating cost Reasonable cost difference
Handle wide range flow Simplest construction Many stages possible Low holdup volume
ratio (with recycle) Relatively easy scale-up Short holdup time
Low headroom Small space
High efficiency requirement
Many stages Small inventory of
Reliable scale-up solvent
Industrial Solvent Extraction Processes
Low cost
Low maintenance
Disadvantages Large holdup Limited throughout Limited throughout High initial cost
High power costs with small gravity with small gravity High operating cost
High solvent inventory difference difference High maintenance
Large floor space Cannot handle wide Cannot handle emulsifying Limited number of
Interstage pumping may flow ratio systems stages in single unit,
be necessary High headroom Cannot handle high although some units
Sometimes low flow ratio have 20 stages
efficiency Will not always handle
Difficult scale-up emulsifying systems,
except perhaps pulse
column
297
298 Ritcey
having up to about 500 cm3 capacity for each stage has been used. Much of
the early work employed homemade equipment, and even now much of the
equipment is constructed in the laboratory.
Several designs of small-scale mixer-settler units can be obtained com-
mercially and can be assembled to provide several extraction, scrub, and strip
stages. Regular turbine blades are used for pumping and mixing; thus the
mixer-settlers are horizontal and the cascading principle is not necessary. The
major problem with small-scale equipment made of Plexiglas is that it tends
to craze quite rapidly and discolor. Consequently, it becomes difficult to
observe accurately the interface level or the dispersion band in the settler.
However, the ease with which mixer-settler units can be made out of Plexiglas
means that they can be discarded and replaced by new ones when such
conditions occur.
Mixer-settlers are the smallest continuous countercurrent equipment
available. Other types of contactors can be purchased or constructed, but
require larger volumes of solutions for operation. Hence, a Karr column is
available having a 1 inch diameter column. Similarly, a sieve plate pulse
column of similar dimensions can be readily constructed. However, scale-
up from small diameter columns such as these is not too satisfactory
[G. M. Ritcey, unpublished data]. The major problem appears to be that wall
effects in small-diameter columns significantly influence both the physical
and chemical characteristics of a solvent extraction system. Thus, a 2 inch
diameter sieve plate column appears to be the minimum size for which wall
effects have a minimum effect on the system. Other columns, with greater
axial and backmixing, require a larger diameter column for scale-up.
Whether a 2 inch diameter column can be considered small-scale or pilot
plant size is debatable as the total flow capacity of such a column can be up to
0.5 US gal min1 (1.8 dm3 min1). The size of a pilot plant, as we have noted
before, covers a wide range of throughput.
Other small-scale contacting equipment is available, e.g., cen-
trifuges and the ‘‘raining bucket’’ or RTL contactor. The minimum
flow rate of available centrifuges seems to be about 1 gal (US) min1
(3.6 dm3min1). A small-scale raining bucket contactor can be obtained that
is suitable for bench-scale operations. The 4 inch (10 cm) diameter, 30 inch
(0.76 m) long unit would have a maximum throughput of about 1 US gal hr1
(3.6 dm3 hr1).
In-line mixers manufactured by, for example, Kenics, Lightning, and
Sulzer are also applicable for continuous small-scale testing of a solvent
extraction process, and 1 inch diameter models are available. This mixer
system can be used either horizontally or vertically. However, few data are
available for this type of contactor, although they would appear to offer
many possibilities, not only for liquid–liquid systems, but also for use in
solvent-in-pulp processes (in the case of perhaps the Kenics and Lightning
mixers) [G. M. Ritcey, unpublished data].
Because of the diversity of contacting equipment available, it is unlikely
that all these contactors will be available in any one laboratory or pilot plant.
Consequently, unless test work is carried out on similar contactors, the
system may not be optimized. Since mixer-settlers are the easiest to construct,
are simple to operate, and require little room and low-flow rates, these con-
tactors are, in many cases, the only ones used to investigate a continuous
solvent extraction process. This is by no means ideal and may result in
abandonment of a process that, using another type of contactor, could be
found to be entirely satisfactory.
The data obtained from small-scale continuous operations will be
required to determine whether the process should be investigated on a larger
scale, such as a pilot plant. These data should be sufficient to draw a con-
ceptual flow sheet, which will include a number of stages for extraction,
scrubbing [10], and stripping; the flow rates, size and type of equipment, and
the various parameters considered earlier in this chapter. Another important
aspect should also be considered in the continuous test work, and that is
chemical analysis of both the aqueous and organic phases for their various
components.
7.12 PILOT PLANT OPERATIONS

The pilot plant design should be small enough to provide the maximum
amount of data at a minimum cost, while being large enough to provide
sufficiently accurate design data to design the full-sized plant. The actual
size and capacity of the pilot plant depends on many factors; thus, it is
impossible to indicate optimum size-capacity parameters. For example,
pilot plant operations at Duval Corporation, Arizona, used a 75 gal min1
(270 dm3 min1) flow rate, whereas those for the Bluebird operation (Ari-
zona) used a total throughput of 0.2–0.6 gal min1 aqueous at an O/A ratio
of about 3 [total throughput approximately 0.8–2.4 gal min1 (3.0–9.0 dm3
min1)].
Choice of contactors for pilot plant operations will probably have been
made following bench-scale and small-scale test work, during which the
kinetic and separation factors were determined. The design of the mixing
equipment has to take into account the necessary volume throughput and the
required contact time. Contact times can be extended in mixers by partial
recycling. In mixer-settler operations, the settlers are normally designed with
a length/width ratio of between 2 and 4, to minimize channeling effects and
excessive velocity of the solutions in the settler. The area of the settler will
depend on the coalescence rate of the dispersed phase and, therefore, will
300 Ritcey
vary somewhat from one extraction system to another. In copper extraction

circuits, settler area is usually based on a phase disengagement rate of
2 gal min1 ft2 (0.08 m3 m2min1), allowing for a 2 inch (5 cm) dispersion
band. Solvent depth varies up to 1.5 ft (0.45 m).
Equipment chosen for pilot plant operations should closely approx-
imate the commercially available equipment that will be used eventually.
Thus, it would be of little use to operate a pilot plant using stagewise con-
tactors if differential contactors were to be employed in the actual com-
mercial operation.
Metering of flows is a most important consideration in a pilot plant,
and these should be accurately controlled wherever possible. Thus, small
upsets in flow conditions, resulting in a change in O/A ratio, can create
many problems that do not become apparent until analyses are done on the
various solutions.
Pilot plant operations, as we have noted previously, can vary between
extremes of flow rates. It is necessary, therefore, that the feed and reagent
volumes be large enough that the pilot plant may be operated for a sufficient
length of time to obtain meaningful data. For example, if the aqueous feed
to the solvent extraction (SX) circuit is being produced batchwise, variations
between batches are bound to occur. Such variations should be controlled as
much as possible. Batchwise production of the feed solution may be very
different from feed to the actual plant, especially if the plant process involves
continuous production of the feed to the SX circuit.
It must also be remembered that feed solutions that are allowed to
stand for any length of time (such as when they are shipped from one place
to another) may behave in quite different ways from a freshly prepared feed
solution in the SX circuit. Furthermore, solutions shipped in cold climates
may freeze during shipment, and this may result in problems in SX, such as
emulsion formation, that would not occur if freshly prepared solutions were
employed. The preparation and treatment of feed solutions for small-scale
and pilot operations are important considerations in scale-up, and failure to
recognize this may well result in considerable problems in the final plant.
Periodic sampling and analysis are essential to the success of the
operation. With experience, pilot plant operators, in many cases, can quickly
detect upsets without having to rely on analyses. For example, a change in
color of the solvent or raffinate, in many cases, is an indication that a change
in the operating conditions has occurred.
A pilot plant can be relatively inexpensive if construction and instal-
lation is done in-house, using readily available materials and equipment. Of
course, the size of the pilot plant will determine whether in-house con-
struction is feasible, as will the knowledge and experience of the personnel.
On the other hand, it may prove necessary to use one of the several consulting
companies who have experience in this field, or to have the pilot work per-
formed by the manufacturers of the contactors chosen for the operation.
It will be instructive to consider briefly the methodology employed
in the development of solvent extraction processes that have become
operational. The development of the Bluebird Mine operation for the
extraction of copper from dump leaching liquors by solvent extraction and
the subsequent recovery of copper as either copper sulfate or cathode copper
is used as an example. The initial investigations [11] included the following:
1. Bench-scale batch tests to determine the characteristics of the extraction
of copper from sulfuric acid liquors by LIX 64
2. Continuous bench-scale operations to study the extraction of copper
and other metals present in the leach liquor: effects of variables such as
temperature, O/A ratio, solvent composition, extraction kinetics, set-
tling characteristics, etc.
3. Determination of solvent losses from solubility, entrainment, and chem-
ical degradation
4. Stripping characteristics and the production of copper sulfate crystal,
saturated copper sulfate solution, powdered copper, or electrolytic
copper from the strip liquors
Following the successful completion of this test work, pilot plant operations
were initiated, having the following objectives:
1. To establish the economic feasibility of producing wire bar-grade cop-
per by a combined SX–EW (EW=electrowinning) process
2. To develop the engineering data to enable estimates of operating and
capital costs of a process capable of producing approximately 30,000 lb
(13,600 kg) of copper per day
3. To develop the engineering data required for the design and operation
of the commercial plant
4. To produce electrolytic copper for evaluation of product quality
Other objectives of the pilot plant operation include those given for bench-
scale and small-scale continuous studies, together with the following:
1. Determining the effects of impurity build-up on recycle of the extraction

raffinate to the heap leaching operation, and on recycle in the extraction
and electrowinning circuits;
2. Obtaining data on the maximum capacity of settler units
3. Obtaining data on methods for controlling flow, pH, and metal
concentration
4. Measuring the transfer of acid between the strip and extraction circuits
and establishing methods for control
302 Ritcey
5. Establishing control methods between the solvent extraction and elec-

trowinning circuits, for flow rates, acid concentrations, and effect of
recycle liquors on copper purity
6. Establishing the optimum conditions for operation of the electrowin-
ning circuit
This outline of the many factors that have to be investigated for the
design and operation of an SX–EW plant is perhaps typical of the devel-
opment of an SX process on which little or no prior work has been done.
Today, with the knowledge and experience available, processes similar to
those already in operation may be designed and scaled up more econom-
ically. The Bluebird Mines operation took about 4 years from initial bench-
scale tests to plant operation.
7.13 SOLVENT LOSS, RECOVERY, AND

ENVIRONMENTAL CONSIDERATIONS
Any economic advantage of a solvent extraction process over other
separation processes for metals can be lost if the loss of solvent from the
system becomes too high.
Solvent loss can occur in essentially five ways: by solubility in the
aqueous phase; by entrainment; by volatilization; by degradation; and by
loss in crud. In solvent-in-pulp (SIP) processes a further distinction can be
made, that of sorption of the solvent on the solid particles in the system.
Furthermore, there are the additional losses from sampling of the process
and from the spillage that may be found in numerous operating plants. For
most systems, loss of solvent by solubility in the aqueous phase is usually
small, and little can be done about this type of loss. Entrainment losses can
vary considerably, usually because of poor equipment design, instability in
the system, poor diluent choice, etc. Loss of solvent by volatilization will
depend on the temperature of the system and on how well the system is
enclosed. Excessive agitation together with elevated temperatures will cause
another type of loss, often associated with volatilization losses. This is
loss by misting. Where volatilization losses mean the vaporization of
the organic components at that temperature, misting will result in the loss
of all components of the solvent as well as the aqueous phase. Sorption of
solvent components onto solid particles seems to depend largely on the
mineral composition of the solids, and extractant degradation is a function
of the stability of the extractant, composition of the aqueous phase, and
temperature. Crud losses are due to many causes, to be explained later, but
primarily high shear, suspended solids, and composition of the aqueous
phase.
The importance of solvent loss in a solvent extraction process cannot

be overemphasized, because of the constraints imposed by economics and
environmental considerations.
7.13.1 Soluble Losses

Soluble loss of a reagent (extractant, modifier, or diluent) from the solvent
phase is an inherent part of the solvent extraction process, since all organic
compounds are soluble, to some extent, in water. The conditions prevailing
in the system can also promote solubility, which can be a particular problem
if the composition and properties of the aqueous phase are inflexible. For
example, the solubility of alkylphosphoric acid and carboxylic acid extrac-
tants is dependent on temperature, pH, and salt concentration in the aqueous
phase.
Low-molecular-weight extractants can generally be expected to have
uneconomic solubilities in most systems, but where high salt concentrations
prevail, the solubility may be substantially lower and may be economic.
This has been shown to be true for naphthenic and Versatic acids, which
have high solubilities in water but appear to be economically useful when
used in high ammonium sulfate liquors, such as those produced in the
Sherritt-Gordon process for the extraction of cobalt and separation from
nickel [12].
Three causes of extractant solubility in the aqueous phase may be
distinguished: solubility of un-ionized and ionized extractant and metal-
extractant species. For extractants such as acids, amines, and chelating
reagents, their polar character will always result in some solubility in the
aqueous phase over the pH range in which they are useful for metal
extraction. Solubility depends on many factors including temperature, pH,
and salt concentration in the aqueous phase, as discussed in Chapter 2.
Increasing the temperature of a system can be expected to increase the
solubility of all the components of a solvent. Variation of pH is likely to
affect only the polar components of the solvent. By increasing the salt
concentration in the aqueous phase, most solvent components are expected
to exhibit decreased solubility because of salting out. The solubility of a
solvent component in the aqueous phase is also likely to increase with
increase in the concentration of the component in the solvent.
The nature and composition of the aqueous phase can also provide
conditions for solvent degradation or, more particularly, for extractant
degradation. For example, the presence of oxidizing agents, such as nitric
acid, could result in oxidation, especially if the extractant contains a readily
oxidizable bond or functional group. Tertiary amines are usually quite
susceptible to oxidation. Again, the presence of oxidizing agents such as
304 Ritcey
permanganate, vanadate, or chromate can give rise to extractant degrada-

tion. The most thoroughly investigated extractant for this is tributylphos-
phate (TBP), and considerable data have been reported on this reagent [13].
Generally, the degradation products are the dibutyl and monobutyl phos-
phates, which have substantial solubility in the aqueous phase. Such
degradation complicates the analytical determination of extractant losses.
Extractant degradation may not be apparent, and thus a decrease in metal
loading resulting from degradation may be wrongly considered to be due to
extractant loss to the raffinate.
7.13.2 Solubility of Extractants in Aqueous Solution

7.13.2.1 Carboxylic Acids
The solubility of carboxylic acid extractants in aqueous solution is a func-
tion of temperature, pH, and salt concentration. Most of the information on
the solubility of these reagents is limited to their solubility in water, and not
much practical information is available on actual leach liquors and similar
solutions.
Under conditions of low salt concentration in the aqueous phase, and
above a pH *6, the solubility of these acids is economically prohibitive.
Even at lower pH values, the solubility could be considered high from the
point of view of pollution. For a C7 –C9 fraction of aliphatic mono-
carboxylic acids, the solubility in water as a function of concentration in the
solvent phase is shown in Fig. 7.9 [14]. Here again, complete information is
unavailable, since no reference is made to pH or salt concentration. Pre-
sumably, the pH is in the range 4–6. However, it is interesting to note the
rapid increase in solubility in the aqueous phase as the extractant con-
centration is increased. This effect also applies to other extractants.
The solubility of Versatic 911 in water has been studied as a function
of both pH and salt concentration for a fixed (0.5 mol dm3) concentration
of reagent in the solvent [15] and is illustrated in Fig. 7.10. The use of high
salt concentrations (4 mol dm3 ammonium sulfate) has been shown to be
effective in keeping Versatic 911 losses at pH 7–8 to more economic levels
(<100 ppm) in the separation of cobalt from nickel [16].
7.13.2.2 Alkylphosphoric Acids
Available information on DEHPA indicates an effect of salt concentration
similar to that shown for the carboxylic acids [17]. Figure 7.11 indicates a
significant difference between sodium and ammonium salts and solutions
containing sodium and ammonium hydroxides. Moreover, below about
5 wt% salt concentration, the solubility becomes uneconomic. However, all
the data given in Fig. 7.11 are for alkaline systems. For acid systems
Fig. 7.9 Solubility of C7 – C9 aliphatic carboxylic acids in water. (From Ref. 14.)
(H2SO4), the solubility of DEHPA from plant operating data is probably

less than 30 ppm and varies somewhat with pH.
The solubility of the sodium salt of DEHPA in basic (NaOH) solution
has been reported, together with the effect of temperature on the water
solubility of this salt [18], (Figs. 7.12 and 7.13). It is evident that the presence
of salts in the aqueous phase depresses the solubility of this extractant
in water (Fig. 7.11). This has been confirmed in the extraction of cobalt
with DEHPA(Na) at pH 5–6, for which a solubility of the extractant was
found to be <50 ppm. Furthermore, the use of DEHPA in the extraction of
cobalt from an ammoniacal (pH 11) system containing sodium sulfate
showed no apparent loss of extractant after 10 contacts of a DEHPA–ker-
osene solvent with fresh aqueous solution [1]. Operation of pilot plants using
DEHPA(NH4) and DEHPA(Na) for the extraction of cobalt, at pH 5–6 and
at 608C, showed the loss of DEHPA to be less than 50 ppm [3]. Temperature
also has a significant effect on the solubility of DEHPA(Na) (Fig. 7.13).
7.13.2.3 Amines
Aliphatic amines used in solvent extraction operations have low solubility
in acidic aqueous solutions, generally below about 10 ppm. Solubility is
306 Ritcey
Fig. 7.10 Solubility of Versatic 911 (0.50 mol dm3) in ammonium sulfate solution.
(From Ref. 15.)
a function, among other things, of the number of carbon atoms in the

molecule. Thus, low-molecular-weight amines are substantially soluble in
water, whereas those of molecular weight greater than about 200 are essen-
tially insoluble. The degree of chain branching also affects solubility; the
greater the branching, the lower the solubility. These effects are also shown
by other extractant types.
Studies on the solubility of Amberlite LA-2, Adogen 368, and Alamine
336 [19] have shown that in acidic sulfate solutions (pH 1.4 and 1.8), the
solubilities of these three amines tend to a limiting solubility of 10 ppm or
less (Fig. 7.14). The initial high solubilities shown for each of these extrac-
tants are probably not due to solubility of the extractant, but to the presence
of more soluble secondary and primary amines contained in the commer-
cially prepared materials. Presumably, the method of analysis used for the
Fig. 7.11 Solubility of DEHPA in various aqueous solutions. (From Ref. 17.)
Fig. 7.12 Solubility of the sodium salt of DEHPA in NaOH (wt%) solution at 208C.
308 Ritcey
Fig. 7.13 Solubility of the sodium salt of DEHPA in water as a function of temperature.
determination of soluble amine loss provided total (primary, secondary, and

tertiary) losses.
In most commercial operations, amine losses of more than 10 ppm are
reported.
7.13.2.4 Chelating Extractants
Solubility data on the LIX and Kelex extractants indicate that these mate-
rials are poorly soluble in aqueous media. Accordingly, in plant operations at
about pH 1.5, reported losses are approximately <15 ppm, which includes
both soluble and entrainment losses as determined by inventory (detailed
later in Table 7.6). The solubility of LIX 63, LIX 64, and LIX 64N in water at
pH 4.8 has been reported at 5.8, 4.3, and 6.2 ppm, respectively [20].
The solubility of Kelex 100 has been reported to be 4.4 (pH 0.5), 1.4
(pH 1.0), and 1.6 (pH 1.5) ppm, in tests employing 0.5 mol dm3 Kelex 100
in Solvesso 150 diluent for the extraction of copper from solutions contain-
ing about 30 g dm3 of metals (Cu, Ni, and Co) [21]. Furthermore, over
several months no change in the Kelex concentration was found in pilot
plant operations with a 0.5 mol dm3 Kelex solution [21]. At higher pH
values (up to 9) Kelex solubility has been determined at <1 ppm in ammo-
nium sulfate solutions up to 300 g dm3 [22]. It would be expected that the
amphoteric nature of Kelex would show increasing solubility at pH values
above 7, and lower than about 1, similar to the variation in solubility of
8-hydroxyquinoline. Presumably, with high salt concentrations in the aqu-

eous phase, solubility is suppressed. The one result at pH 0.5 (4.4 ppm) would
suggest increasing solubility of this reagent as the pH of the system is
decreased below 1.
7.13.2.5 Tributylphosphate (TBP)
Relative to other extractants, a substantial amount of information is avail-
able on the solubility of TBP in various aqueous media [23]. The solubility of
this extractant in water is a function of temperature [24], and decreases with
increase in temperature.
In aqueous acid solutions, solubility varies depending on the type of
acid and on its concentration. Thus, in halo acids, the solubility increases
markedly with acid concentration, a general order being HI > HBr > HCl.
For nitric and phosphoric acids, solubility decreases up to about 8 mol dm3
acid, and then begins to increase.
The effect of increasing TBP concentration in the organic phase results
in an increase in its solubility in water. Furthermore, the diluent used also
affects the TBP solubility.
7.13.3 Solubility of Modifiers

There are little or no data on the stability or solubility of the various
reagents used as modifiers, except TBP. Neither is the relationship between
solubility losses and concentration in the solvent known nor whether it is a
function of only its solubility in the aqueous phase. Of those alcohols used
as modifiers, the only solubility data appear to have been determined by
Fig. 7.14 Solubility of amines in acidic sulfate solution at (a) pH 1.4 and (b) pH 1.8.
310 Ritcey
Blake et al. [17], Table 7.2. Aqueous solubility decreases with increase in
the number of carbon atoms and with a decrease in concentration in the
organic phase. This observation is general and has been found to apply in
other cases, for example, for C8–C9 carboxylic acids, for which the solubility
is almost a linear function of the carboxylic acid concentration in the
organic phase [25]. (see also Chapter 2).
7.13.4 Solubility of Diluents

Here again, few data are available on the solubility of the various diluents, in
either its use or being studied for use in solvent extraction circuits. Although
this is expected to be low (<2 ppm), what few data are available apply to
water and not to solvent extraction raffinates (see, however Chapter 2).
7.13.5 Reagent Stability and Degradation

Degradation of solvent components can arise from various sources, such as
oxidizing properties of the aqueous phase; too high a temperature of the
system; aging and oxidation by atmospheric oxygen; and general instability
of the component over prolonged periods of use. In this context, it is
apparent that the LIX extractants are susceptible to dehydration resulting
from the adjacent oximato and hydroxyl groups. Indeed, the manufacturer
of the LIX reagents states in various publications that those containing LIX
63 should not be used above about 408C. Undoubtedly, dehydration occurs
and is readily observed. Thus, for example, if LIX 65N is heated, globules of
water are formed. The ease of dehydration of the LIX reagents is also
illustrated by the fact that, even under high vacuum, they cannot be distilled
unchanged. The distillates always show the absence of OH by infrared
spectroscopy [26]. In addition, oxidation has been observed in some
Table 7.2 Solubility of Some Alcohols in Acid Solution

Aqueous phase Equilibrium conc
SO42 Organic phase Aqueous phase

pH (mol dm3) Alcohol O/A ratio (wt/vol%) (ppm)
1 0.5 2-ethylhexanol 300 0.81 40
1 0.2 2-ethylhexanol 400 0.60 36
1.7 0.2 2-ethylhexanol 400 0.60 36
1 0.5 2-ethylhexanol 800 0.48 19
1 0.5 4-ethyloctanol 800 1.76 3
Source: Ref. 17.

extraction circuits [27]. Such degradation was shown to be catalyzed by the

presence of transition metals (Co>Ni>Cu) and can be easily reversed by
treatment with hydroxylamine.
Kelex 100, on the other hand, does not contain adjacent OH groups
and can be distilled unchanged under vacuum. However, heating pure Kelex
100 at atmospheric pressure results in considerable darkening of the reagent,
probably from decomposition at the double bond [28]. However, the reagent
appears to withstand pressure reduction to produce metal powders [personal
communication from A.W. Ashbrook, 1973].
The stability of these reagents may be compared, for example, to
DEHPA, which has been used at 608C for many months without decom-
position [3], and is shortly to be used at 70–808C in a solvent extraction plant
[personal communication from A.W. Ashbrook, 1973]. However, the thio-
phosphinic acids are prone to oxidation; thus the commercial reagents
CYANEX 302 and 301, respectively mono- and dithiophosphinic acids are
both degraded. CYANEX 301 degrades in two stages, the first being the
formation of the disulfide, which can be reversed by treatment with, for
example, zinc powder or nascent hydrogen [29]. CYANEX 302 on the other
hand is degraded in one step to elemental sulfur and CYANEX 272, parti-
cularly in the presence of iron and air.
Tertiary amines are susceptible to oxidation (R3N:O), and some evi-
dence has been found to suggest that oxidation does indeed occur after
several months recycling in a uranium solvent extraction circuit [A. W. Ash-
brook, unpublished data].
Diluent stability has apparently never been seriously considered,
except for the solvent extraction processes involving radioactive materials.
Generally, one would expect that straight-chain saturated aliphatic diluents
would be very stable materials, and that with the introduction of other
elements and bond types, such as oxygen and nitrogen and double bonds,
the stability would be expected to decrease. Thus for a series of different
hydrocarbons, the most stable or inert would be expected to be the n-
paraffins, and the least stable those containing olefinic compounds. The acid
stability of various hydrocarbons has been studied by contacting with
4 mol dm3 nitric acid (25 vol%) for 7 days at 708C. The order of stability of
these conditions is shown in Table 7.3 [28]. However, in the petrochemical
industry the degradation of hydrocarbons in the presence of cobalt has
been observed [30]. This phenomenon has also been found in some cobalt
solvent extraction circuits with the production of carboxylic acids. The
presence of such potential metal extractants may have a deleterious effect
on the overall process [31]. The problem can be minimized by the addi-
tion of an antioxidant, e.g., 2,6-di-tert-butyl-4-methylphenol (BHT) into
the solvent.
312 Ritcey
Table 7.3 Stabilities of Pure Hydrocarbons

Type of hydrocarbon Stability
n-Paraffins Stable
Isoparaffins Variable stability
(low-boiling fraction, stable;
high-boiling fraction, unstable)
Cycloparaffins Moderately stable to unstable
Aromatics Moderately stable to unstable
Mixed cycloparaffins and Unstable
aromatics
Olefins Unstable
Source: Ref. 31.
7.13.6 Volatilization
Volatilization of solvent components can become a problem when the sys-
tem is operated at elevated temperatures or in hot climates. The human
toxicity of solvent components is a generally unknown factor and could be a
problem in a system enclosed in a building.
Generally, one would expect that the most volatile component would
evaporate first, and this would probably be the diluent. In several cases of
operating simulated solvent extraction processes at temperatures up to
708C, it has been noted that the diluent is rapidly volatilized [G. M. Ritcey
and B. H. Lucas, unpublished data]. Problems of volatilization appear not
to have occurred to any great extent in the past (perhaps the losses were not
measured), but any trend to the use of elevated temperatures would require
that this form of solvent loss be thoroughly investigated.
7.13.7 Entrainment
Entrainment of solvent in the aqueous raffinate, in the strip liquor, or in
cruds can be the most serious cause of solvent loss. Here, it is expected that
the ratio of solvent components in the entrained material will be the same as
in the solvent. Entrainment losses can occur because of insufficient settling
area or time allowed for phase disengagement; poorly designed or operated
mixers; too much energy input into the mixing stage; lack of additives
to suppress emulsion formation; poor diluent choice; high extractant con-
centration in the solvent; solids in the aqueous feeds causing crud formation,
etc. Entrainment losses can be minimized by the incorporation of various
devices in the circuit, such as skimmers, centrifuges, coalescers, after-
settlers, flotation, foam fractionation, activated carbon, or others. Although
flotation units are effective for the removal of entrained solvent and are not
affected by the presence of solids, they are expensive to operate as regards
reagents and power, whereas packed coalescence beds, although relatively
inexpensive and effective, can be easily blocked by solids. Baskets contain-
ing polypipe pieces tend to be coated with a thin film in a reasonably short
time, perhaps less than a month, and thus require expensive periodic main-
tenance. Centrifuges, although capable of high throughput, are expensive,
have high power consumption, and may require frequent cleaning.
7.13.8 Crud
Common to all or most solvent extraction operations in the mining industry
is the problem of stable formation of cruds. The crud can constitute a major
solvent loss to a circuit and thereby adversely affect the operating costs.
Because there can be many causes of crud formation, each plant may have
a crud problem unique to that operation. Factors such as ore type, solution
composition, solvent composition, presence of other organic constituents,
design and type of agitation all can adversely affect the chemical and phys-
ical operation of the solvent extraction circuit and result in crud formation
[32–34].
Crud is defined as the material resulting from the agitation of an
organic phase, an aqueous phase, and fine solid particles that form a stable
mixture. Crud usually collects at the interface between the organic and aque-
ous phases. Other names that have been used for the phenomenon are grun-
gies, mung, gunk, sludge, and others.
The following section covers the general aspects of solvent losses by
crud; its formation and characteristics and its treatment and prevention.
Losses attributed to emulsion and crud formation, can in part be related
to (1) nature of feed; (2) reagent choice; (3) equipment selection; and
(4) method of operation, such as the droplet size, continuous phase, ex-
cessive turbulence, etc.
7.13.8.1 Possible Causes of Crud Formation
7.13.8.1.a Nature of Feed
The nature of the feed composition can be a major determining factor as to
whether crud will form in the subsequent extractive operations [33,34]. The
presence and concentration of certain cations, such as Fe, Si, Ca, Mg, or Al,
with sufficient shear in the mixing process can produce stable cruds [32,34].
Solids must be absent from most solvent extraction circuits, and clarification
is usually aimed at achieving about 10 ppm of solids. One of the major causes
of crud is the lack of good clarification of the feed solution, with the result
that solids get through to the solvent extraction circuit. The presence of
314 Ritcey
colloids, such as silica [35], can produce stable emulsions and crud during
the mixing of the phases to achieve mass transfer. Aged feeds can constitute
a greater potential crud problem than fresh leach solutions [36]. In plants
where bacteria have been prevalent because of favorable environmen-
tal conditions, crud has resulted, and expensive circuit modifications
were subsequently required. The elimination of air to such circuits is often
necessary to minimize bacterial and fungal growth [37]. Certain systems may
have hydrolyzed compounds precipitating out of solution, and thus a crud
results. In certain extraction systems, the anionic strength of the aqueous feed
solution may be insufficient, so that stable emulsions occur when the two
phases are mixed. If sufficient agitation is applied over a period, then crud
can result. One other important cause of cruds in solvent extraction plants is
dust from the air if drawn into the agitation in a mixer-settler circuit. Thus,
vessels should be covered to prevent dust accumulation. Organic matter in
the feed, such as lignin or humic acids, may also promote crud formation.
7.13.8.1.b Nature of Solvent
The choice of the extractant and solvent composition is an important aspect
of the successful solvent extraction operation, but the possibility of crud
due to the solvent composition must not be overlooked. Many systems
require a modifier to improve phase disengagement, to assist in solubilizing
the metal-organic species, and to reduce third-phase and emulsion tendency.
If a solvent has the tendency to produce emulsions on mixing with the
aqueous feed solution, which could cause cruds if colloids or suspended
solids are present in the aqueous feed, then the cause may be due to several
factors. Perhaps the system requires the addition of a modifier, a change to a
different modifier, or a higher modifier concentration. Also, possibly the
diluent type and composition may not be compatible with the system. An
aromatic diluent or an aliphatic diluent with some aromatic content may be
more desirable than a completely aliphatic diluent for that particular pro-
cess. Frequently, in the solvent makeup, there are unreacted chemicals from
the manufacturing process or, possibly, impurities from the containers used
to transport the solvent components. The problems associated with such
impurities become particularly evident if their effects are not considered
during bench-scale investigations [1]. Their presence in a solvent system can
produce cruds, good or poor phase separation, and enhanced- or poor-
loading characteristics. Such effects could result in abandoning a particular
solvent because of its poor chemical and physical characteristics.
Degradation of the solvent due to the presence of certain metals in the
feed solution, use of oxidizing agents during stripping, high-temperature
processing, and biodegradability all may result in decomposition products
forming stable emulsions and cruds. Several uranium plants have reported
the degradation of the isodecanol modifier to isodecanoic acid in the amine/

isodecanol system. Naphthenic acid extraction of copper from leach liquors
also showed degradation to an insoluble crud composed of 23% of the ex-
tractant [38]. In the refining of uranium using tributylphosphate in contact
with nitric acid, degradation products of mono- and dibutyl phosphates are
produced. Amines are susceptible to degradation in the presence of oxidizing
agents. Carboxylic acids and DEHPA have been reported to withstand low-
pressure reductions in a process to produce metal powders [39]. The LIX
reagents containing LIX 63 (e.g., LIX 64, LIX 64N, LIX 70, and LIX 73) are
stable up to temperatures of 408C, whereas those not containing LIX 63 (e.g.,
LIX 65N and LIX 71) can safely withstand higher temperatures.
7.13.8.1.c Equipment Selection
There is no universal contacting equipment suitable for all solvent extrac-
tion operations. Even within a plant, it may be completely wrong to select
the same type of contactor for all stages of the extraction process. Therefore,
for each plant the final selection is governed by the nature and composition
of aqueous feed, the type of solvent and its composition, and how their
respective physical characteristics affect the mixing process, flow patterns,
and coalescence [9]. Naturally, the mass transfer efficiency must also be
considered. Thus with an understanding of all the physicochemical variables
that affect the extraction, as well as the minimization of emulsions and crud,
the right equipment for that plant can then be selected [9,38]. That is, if the
tendency to form emulsions and cruds is to be minimized, the type of
equipment and the method of agitation used to achieve mass transfer is of
concern [33]. Degradation of the solvent may also have to be considered in
the equipment choice, depending on the chemical system. Thus in one plant,
centrifugal contactors were chosen over mixer-settlers because of lower
solvent degradation [32].
7.13.8.1.d Method of Operation
In the solvent extraction process, one of the major concerns should be the
technique by which mass transfer is achieved. That is, the physical design and
operation can contribute not only to high solvent losses, such as by
entrainment, but also to the formation of stable emulsions and cruds. These
depend on the nature of the dispersion, the type of droplet formation, and the
rate and completeness of coalescence. These factors are important in the
choice of suitable contacting equipment and in the design and operation of
the plant. Different types of dispersion will be created for a particular system,
depending on the type of contacting equipment selected and the energy input
to the system. Differences will be indicated by the rates of mass transfer, drop
size distribution, wetting of surfaces, sedimentation and coalescence rates,
316 Ritcey
and entrainment. Subject to the physicochemical properties of the two dis-

persed phases such as viscosity, surface tension, and presence of solids or
colloids; with increasing agitation and decrease in drop sizes a region of
instability will be reached, followed by a stable emulsion. If solids or colloids
are also present, then a crud will result. This is demonstrated in Fig. 7.15,
showing the operating regions of pulsed columns [36]. The information is
readily related to excessive turbulence in mixer-settlers, particularly of the
pump-mix design, and in certain agitated columns where backmixing is
severe. Flow patterns during mixing can influence emulsion tendency, which
can be further influenced by the continuous phase. Thus, if solids are present
in mixer-settler operations and excessive turbulence exists, it would be
advisable to use a type of contactor more suited to the physicochemical
characteristics of the system. Centrifugal contractors would also be an
unwise choice if solids are present or crud formation is likely. Equipment
such as the Graesser contactor (RTL contactor), pulse sieve-plate column,
ARD contactor and possibly in-line mixers could be considered. Pulsed
columns have been described in the extraction of uranium from ore leach
slurries and in the presence of crud [36,40].
Choice of the continuous phase, organic, or aqueous, coupled with the
contactor and flow patterns produced during the operation, may reduce the
tendency of emulsification and crud formation if solids are present.
Although it is usually desirable to operate at saturation loading of
the solvent, there are certain situations for which it is necessary to maintain
Fig. 7.15 Operating regions of a pulsed column.

less-than-saturation conditions. For example, in the extraction and separa-

tion of zirconium from hafnium in a nitric acid system, using TBP, the
presence of the solvent and agitation results in a stable emulsion and crud
[40]. Similar crud problems can occur in rare earth circuits using DEHPA,
during which below-saturation loading has to be maintained to prevent crud
formation [36,41].
7.13.8.2 Description of Some Cruds
The chemical and physical aspects of crud can differ for each separate
operation and will vary in inorganic composition, organic content, color,
and density. The composition of many cruds appears to have in common
such constituents as Si, Al, Fe, P, SO4, particles of gypsum, clay, and other
fine particles together with the solvent. Often there is a direct relation
between the feed liquor and the crud compositions, indicating possible
aqueous carryover as well as inefficient clarification before solvent extrac-
tion [33]. Various researchers have reported on the formation of crud and its
characterization in their circuits [42–45].
In some of the South African plants, a tarlike substance is generated
and has been shown to contain aliphatic carboxylic acids. There is evidence
of the presence of isodecanoic acid, from the oxidation of the isodecanol
modifier. More than one plant in North America has experienced some crud
problems when greater than 3% isodecanol was present. Possibly the vortex
created on mixing is a contributing factor to the crud produced by the oxi-
dation of the isodecanol in the presence of ferric iron, as a possible oxidant.
In one early plant, the animal glue flocculent used for filtration of the leach
pulp caused severe fungal growth in the solvent extraction circuit. The
addition of a solvent-soluble fungicide (a derivative of benzothiazole) alle-
viated the crud caused by the fungus. Use of an aromatic diluent, instead of
an aliphatic one, was also effective in minimizing the crud problem caused
by bacteria.
In some scrub and strip circuits, the crud is mainly composed of silica,
as well as inorganic sulfates. Also, if poor pH control is used in the uranium
stripping circuits with ammonium sulfate, then uranium is a major con-
stituent [33,46]. Such crud may be treated with dilute sulfuric acid, and
recirculation through a pump results in the crud breaking down. There is
evidence, in at least a few uranium circuits, that the presence of humic acids
may be a possible cause of the crud problem [34,47]. Lignin appears to be
another cause of crud formation [33,46]. Humic acids contained in the feed
solution have also been implicated in the formation of waxy cruds in plants
extracting uranium from phosphoric acid.
The presence of hydrolyzable compounds, which precipitate because
of excessive agitation and high shear, can cause severe crud problems.
318 Ritcey
Zirconium presents a particular problem and is a major constituent of at least

one crud in a uranium processing circuit. Also, as stated earlier, zirconium
will tend to hydrolyze in an HNO3–TBP circuit given the right conditions of
shear and of energy input (proximity to saturation of the solvent and the
type and construction of contactor, particularly in the coalescing zone). The
use of Teflon plates in columns for such an extraction process definitely
increased coalescence and decreased the tendency for hydrolysis and crud
formation [40].
7.13.8.3 Treatment and Solvent Recovery
There is no universal treatment scheme that would be amenable to all cruds
because of the difference in the causes of crud formation. Because crud
originating in one plant is usually different from that obtained in another; or
in another circuit in the same plant, so the treatment scheme adopted for
solvent recovery will, of necessity, be different. Some cruds are broken down
by settling and skimming, others by filtering [47], some by acidification or
neutralization to an alkaline condition [34,42], and some are broken by
extreme agitation such as cycling through a centrifugal pump. After such
treatment procedures, if successful, the solvent that is released can be readily
decanted for recovery. In some plants, a separate solvent treatment stage is
necessary, with a wash such as Na2CO3 being used.
Thus, with cruds differing in their chemical and physicochemical his-
tory of formation, treatment to break down the crud and recover the solvent
is accomplished only by the testing and evaluation of many possibilities.
Some examples of crud descriptions have been reported [46].
7.13.8.4 Crud Prevention
Because crud is a difficult phenomenon to completely characterize, is often
site-specific, and indeed can vary within an extraction, scrub, and strip-
circuit operation, preventative measures are difficult to cite. The following
are some of the methods that can be suggested for the prevention of crud.
Again, it must be emphasized that because cruds have a different history of
formation, one or more preventive measures may be necessary.
Solids in the feed were mentioned as one of the major causes of sub-
sequent crud formation in the solvent extraction circuit. Good clarification,
therefore, is necessary to minimize crud and thereby operating costs.
Table 7.4 itemizes some benefits of good clarification [48]. In North America
and South Africa, the objective is to obtain approximately 10 ppm suspended
solids in the feed to extraction. This is usually obtained by the use of sand
filters after a countercurrent decantation (CCD) circuit. At least one plant
has reported that crud quickly developed when the sand filters were not in
Table 7.4 Benefits of Good Clarification
Better mixing efficiency

Less crud and, therefore, less solvent loss
Lower organic entrainment
Less difficulty with maintaining continuous phase
Decreased ion transfer to electrolyte (in copper processing)
by suspended solids
Increased efficiency in the tank house
Decreased maintenance
operation [49]. New plants are attempting to achieve this objective by the use
of Enviroclear thickeners, leaf clarifiers, and sand filters. Although sand
filters are used successfully in some plants, other plants have not been able to
achieve their objective. Insufficient frequency of backwashing the sand filters
is a probable cause of poor clarification.
Because dust can cause crud if permitted in a mixer-settler circuit,
particularly if the settlers are located in the open, adequate covers over the
settlers should be provided.
Colloidal silica in some circuits such as TBP–HNO3 in uranium repro-
cessing can be reduced by the addition of gelatin and heating to 808C to
coagulate, followed by centrifuge separation [33]. The addition of certain
surfactants to lower the surface tension may also reduce cruds caused by
colloidal silica [47], as well as break emulsions. Addition of sequestering
agents was successful in eliminating the deposition of calcium–rare earth
precipitate, in a rare earth–DEHPA circuit [50]. However, the addition of
such reagents could also cause solvent degradation with continual cycling,
so the approach must be carefully investigated before adoption. Surfactants
and other compounds used to enhance the liquid–solid separation after
leaching could also enhance emulsion and crud tendency.
Certain organic constituents, such as lignins and humic acids, may be
solubilized in the aqueous feed and thus cause problems in the solvent
extraction circuit [33]. Very little is yet known about this particular area of
crud formation, but it has been found in phosphoric acid circuits that
coagulation of humates with surfactants followed by filtration has been
useful [51]. Lignin can be removed by passing the leach solution through a
bed of activated carbon, and pressure treatment under oxygen at 200–2508C
will destroy the lignin components.
Crud caused by bacterial and fungal growth may be minimized by
elimination of as much air as possible to the system [52]. These growths, in
320 Ritcey
association with any solids present in the feed liquor, lead to the formation of
crud. Fungal growth in some circuits, arising from the isodecanol modifier in
an amine-uranium system, was eliminated by using 35% Solvesso 150 aro-
matic diluent [53,54]. The aromatic diluent acts as a bactericide and fungi-
cide. Commercial bactericides have also been used successfully.
Proper selection of the extractant as well as the other solvent compo-
nents can minimize emulsions and crud formation. Freshly prepared solvents
can often contain impurities that could subsequently cause operational
problems. Therefore, all solvents should be conditioned before use. Having
selected a suitable solvent system, cyclic tests should be performed to deter-
mine whether degradation of any of the solvent components is taking place,
as any degradation products could cause crud formation in the circuit.
There are several items for consideration in the operation of the circuit
to minimize crud formation. In certain systems, solvent saturation can result
in the formation of gelatinous solids, as in the rare earth–DEHPA system
[50,55] and in the zirconium–TBP circuit [56]. The phenomenon is partially
due to the increase in viscosity as loading is reached. An increase in the O/A
ratio thus results in a decrease in crud formation. As the viscosity increases,
excessive agitation can produce stable emulsions.
Flow patterns can be altered by change of the continuous phase and,
accordingly, the tendency for the formation of emulsions or cruds is altered.
At one uranium refinery, the solvent is maintained in the continuous phase
to produce flow patterns to reduce emulsion tendency [57].
Occasionally, the water used for solution makeup to the scrub or strip
circuits, because of impurities, can cause subsequent emulsions and cruds.
At one uranium plant in South Africa, deionized water was used to prepare
the ammoniacal strip solution, rather than normal plant water, which tended
to cause crud formation [53,54].
Equipment selection is important, as is also the proper operation of
the contacting devices. It is generally recognized that high shear is the pri-
mary cause of droplet haze and subsequent emulsion and crud formation.
Thus, the type and amount of agitation (shear) must be optimized for mass
transfer while minimizing emulsion and crud formation. Table 7.5 indicates
items of information that may be required in analysis of crud formation
problems in a plant [32].
7.13.9 Extractant Losses

Table 7.6 shows some typical reported extractant losses occurring in solvent
extraction processes. Most of these data were obtained from inventory
calculation and, hence, represent total loss from the several ways, by which
solvent components may be lost from the system [3,15,16,58–69].
Table 7.5 Possible Information Required for Consideration in Solving Plant Crud
Problems
Ore: mineralogy and analysis
Leach: oxidant and quantity added:
Possible effects on degradation, emulsion production,
or crud stabilization
Liquid–solid separation:
Type of separation (e.g., CCD)
Type and quantity of surfactant added
Use of sand filters or other types of clarifiers (e.g., anthracite) and
frequency of regeneration
Feed solution to solvent extraction:
Suspended solids
Dissolved solids
Solution composition (especially silica, aluminum,
molybdenum, zirconium)
Presence of humic acids, lignin
Solvent extraction:
Extraction circuit:
Extractant
Diluent
Modifier and concentration
Mixer design (e.g., baffling)
Agitation (rpm and design, energy)
Vortex in mixer
Continuous phase
Degradation of diluent, modifier and extractant
(and surfactant from L/S separation)
Settler design (entry to dispersion band,
flow rate design, baffling)
Velocity across settler
Viscosity, surface tension of interfacial tension
Presence of fungus or bacteria
Presence of precipitates
Stripping circuit:
Mixer and settler operations as in ‘‘extraction’’ above
Stripping agent
Viscosity, surface tension, or interfacial tension
Continuous phase
Degradation of diluent, modifiers, and extractant
Presence of fungus or bacteria
Presence of precipitates
322 Ritcey
Table 7.6 Extractant Losses Reported for Some Processes

Metal Extraction Loss
extracted Extractant pH (ppm) Ref.
Ni Naphthenic acid 4.0 90 [58]

Naphthenic acid 6.5 900 [59]
Versatic 911 7.0 900 [15]
300 Shell Chemicals
Co Versatic 911 7.7 100 Shell Chemicals
Rare earths DEHPA 2.0 7 [16]
Co DEHPA 5.5–6.5 30 Personal comm.
U Tert. amines 1.5–2.0 10–40 [3]
Cu LIX 64 1.5–2.0 4–15 [60–63]
U TBP 2.0 25–40 [64–67]
Cu Kelex 100 1–2 1–10 Personal comm.
Hf MIBK 1.5 mol dm3 HCl 20,000 [60–69]
7.13.10 Environmental Considerations

Perhaps the area of environmental pollution most pertinent to solvent
extraction processing is that of water pollution and, consequently, the
toxicity of solvent extraction reagents to aquatic life then become important.
Thus, the primary consideration in the development of water quality criteria
for these reagents is to determine their toxicity toward fish and their bio-
degradability [70].
Since different species of fish may differ widely in their tolerance to the
same reagent, four fish types are usually used, such as the fathead minnow,
bluegill, goldfish, and guppy. These four species have been used by Pickering
and Henderson in the study of tolerance limits for a number of petrochem-
icals [71]. Median tolerance limits (TLm) (LC50) denoting the concentration
at which 50% mortality occurs were computed in different concentrations of
solvent causing 50% mortality of the fish under the experimental conditions
during a period of 96 hr. In general, the toxicity was greater when soft water
was used in the tests. Similar static toxicity test results have been reported
from a study performed in Canada on guppies and minnows in the presence
of various extractants and diluents [G. M. Ritcey and B. H. Lucas, unpub-
lished data; 72]. An ultrasonic probe was used to emulsify the 0.1% solution
of the reagents in the test water to prepare the stock solution for the static
tests. Guppies, each weighing 0.08–0.16 g, and fathead minnows weigh-
ing 0.73–1.89 g were evaluated in the program. The weight/volume ratio
of fish/solution was 0.03 g dm3 for guppies and 0.70 g dm3 for minnows.
The results of these bioassays on extractants and modifiers are shown in

Table 7.7a and for diluents in Table 7.7b. It is interesting to note the wide
range of tolerance limits that are shown for the various extractants. Many
show a toxic effect below their solubility level. Also, the toxicity of the
Solvesso 150 and Isopar L diluents was acute compared with the other
diluents examined.
Table 7.7a Acute Toxicity Static Bioassays on Extractants

and Modifiers (TLm 96 hr)
TLm (ppm)
Guppies Minnows
Aliquat 336 0.18
Primene JMT 0.70
Kelex 100 1.10
LIX 63 4.0 1.6
Alamine 336 10.0 4.0
Isodecanol 12.0 8.4
LIX 64N 15.0 2.7
TBP 18.0 9.6
LIX 70 32.0 15.0
Versatic 911 102.5
DEHPA 173.0
Table 7.7b Acute Toxicity Static Bioassays on Diluents

(TLm 96 hr)
Diluent Effect Concentration (ppm)
Oil-based livestock nontoxic 1000
spray
Napoleum 470 nontoxic 1000
Mentor 29 nontoxic 1000
DX 3641 nontoxic 1000
Isopar L nontoxic 1000
Shell 140 nontoxic 1000
NS-144 total mortality 1000
Solvesso 150 minnows TMm 10.7
Guppies TMm 14.7
Isopar E TMm 10.0
324 Ritcey
Acute toxicity tests are reported for the Cyanamid reagent Cyanex
272, showing TLm values of 4.9 g kg1 and >2.0 g kg1, respectively, for rats
(oral) and rabbits (dermal) [73]. The TLm 96 hr tests for bluegill sunfish and
rainbow trout gave values of 45 and 22 ppm, respectively.
Tests by Mobil Oil Corporation [74] on dibutyl butylphosphonate
showed that its toxic effects on rats were moderate at 3 g kg1 (oral), and
for rabbits it was nontoxic at >5 g kg1 (dermal) exposure. With DEHPA,
acute toxicity was found for rats at concentrations of 1.4 g kg1 (oral).
In most countries, the manufacturer now has to provide such toxicity
data for approval before marketing.
7.14 ECONOMICS
Any successful solvent extraction process depends upon the selection of
inexpensive extractants that can operate at the natural condition of the
solution, with minimum loss of the organic phase to the aqueous solution.
Also, an inexpensive means of recovery of the metal from the organic
phase is necessary. In some cases, the cost of neutralization of the feed
solution before solvent extraction processing or the cost of maintaining a
buffered pH during extraction may prove excessive when coupled with
solvent losses.
The economics of the solvent extraction process are very dependent
upon ‘‘upstream’’ and ‘‘downstream’’ portions of the plant. Integration of
the total processing step is essential to obtain maximum return. Variables,
such as tonnage rates and changes in solution composition, can have a most
significant result on the economics of solvent extraction. Generally, economic
considerations may be divided into two major areas: (1) capital investments
and (2) operating costs.
7.14.1 Capital Investments

Capital investment is primarily related to the size of equipment necessary for
a given total throughput, taking into consideration the number of extraction,
scrubbing, and stripping stages required. Usually, the building that may be
required to house the process is an important cost factor in determining areas
and height demanded by the solvent extraction equipment. The organic
phase is a major cost inventory item and may help determine the choice
of equipment. For example, in one system the solvent inventory was re-
duced from 75,000 to 3,000 gal by using centrifugal contactors instead of
mixer-settlers [75]. In addition, because of the size of the centrifugal con-
tactors, a smaller area was required, resulting in decreased building costs.
Naturally, in some areas of the world, the process requires no building to
house the equipment and, therefore, the capital costs of the process can be
decreased.
Although the capital cost depends on such factors as flow rates,
solvent inventory, equipment, and building, it should not be assumed that
for a given chemical process involving solvent extraction, the capital costs
will be the same in one area of the world as in another. In one area,
something like three plants were installed for an identical process, the
capital cost varying as much as 300% from the lowest to the highest [76].
For any given process, there are many possible designs for the plant,
incorporating different types of equipment and instrumentation that result
in variations in capital cost. Therefore, it must be emphasized that (1)
careful selection of equipment in relation to adherent costs of solvent and
building, and (2) critical evaluation of the engineering contractor’s design
are necessary to optimize the lowest possible capital cost of the solvent
extraction operation.
In summary, capital costs can be influenced by the following:
1. The number of stages and, thus, the number of mixer-settlers or col-
umns required
2. Kinetics and, therefore, the size of the mixing device
3. Coalescence and, therefore, the settling requirements, which dictate
equipment and solvent inventory costs
4. Entrainment and, therefore, the cost of solvent recovery equipment
5. Flow rates and flow ratios
6. Building requirements
7. Ancillary equipment, such as pumps, piping, instrumentation, etc.
8. Equipment to meet environmental regulations
9. Engineering design
7.14.2 Operating Costs

Operating costs may be divided into eight general areas:
1. Preparation of the feed liquor may be required before solvent extrac-
tion. For example, precipitation of an unwanted metal and removal by
filtration, or pH adjustment may be necessary.
2. Preequilibration of the solvent may be required. In some systems, this
cost is minimal, but in others it may be high; for example, in uranium
extraction from sulfate solutions using tertiary amines, the sulfuric acid
preequilibration of the solvent before extraction is a few cents or less per
pound of U3O8 produced. By comparison, in a TBP–HNO3 system for
the recovery of zirconium, the preequilibration costs, using nitric acid,
amount to about 50 cents per pound of Zr produced.
326 Ritcey
3. In some extraction systems it may be necessary to remove undesirable

coextracted metals by scrubbing. Whereas this process step might cost 5
cents per pound of metal produced in one system, in another, scrubbing
may not be necessary.
4. Stripping costs may vary considerably in one process, compared with
others. The choice of the stripping medium is often the major cost dif-
ferential, and this choice will dictate the final product that is produced.
The reagent stripping costs in the TBP–HNO3 system are zero, since
water is used, but by comparison, the costs for recovery of uranium from
amines with sodium carbonate are 1–2 cents per pound of U3O8 pro-
duced. Acids lost in recycling tankhouse acid constitute another oper-
ating cost.
5. Solvent losses are a factor to be considered in every solvent extraction
process. Each process will have a certain loss of reagent by solubility,
because of the pH and salt content. In some cases, loss of solvent by
degradation, such as by high acidity or alkalinity, may occur. Solvent
loss by solubility or degradation can vary considerably from process to
process. In addition to soluble and degradation losses, some plants will
have extreme losses because of entrainment and crud formation. These
losses can be minimized by proper design of the process equipment. The
type of contactor used may actually be causing the high entrainment
losses, as well as losses from crud formation. If the equipment is not
redesigned, and this may be impossible, then expensive solvent recovery
equipment may be necessary. Thus, it is important to carefully select
both the extractant and the diluent, as well as the equipment, when opti-
mizing the process to minimize solvent losses.
6. Labor and maintenance costs can usually be kept to the order of a few
cents per pound of metal produced. However, where a circuit has been
improperly designed and several manual operations are required, then
labor and maintenance costs can be high. Excessive solvent entrainment
losses, mentioned earlier, are added to labor costs. Any crud that forms
and has to be removed can result in additional costs from labor re-
quirements. For example, if incorrectly designed contacting equipment is
chosen, which results in periodic shutdowns necessary for cleaning
because of formation of cruds and precipitates, a major maintenance
problem could arise.
7. Energy costs to the mixing circuits are influenced by the equipment type
and flows.
8. Raffinates containing appreciable amounts of the metal one desires to
recover can also account for increased operating costs. If subsequent
treatment of the raffinate is required to meet environmental regulations,
an additional cost is incurred.
7.15 OPTIMIZATION OF THE SOLVENT

EXTRACTION PROCESS
Much of the optimization of the solvent extraction plant can be achieved in
the pilot plant testing. As noted earlier on the subject of process design, one
must investigate the dependence of the dispersion and coalescence char-
acteristics and their effect on extraction and phase separation. Also, such
variables as metal concentration, equilibrium pH (or free acidity or free
basicity), salt concentration, solvent concentration (extractant, diluent,
and modifier), and temperature have to be studied to determine their effect on
mass transfer. Although many of the variables can be tested in the pilot plant,
many circuits are not optimized until the full-scale plant is in operation.
In the optimization of the solvent extraction process for the recovery
of copper using LIX 64N, Robinson [77] described the cost function in terms
of the sum of the operating and capital costs. The operating costs were taken
as resulting from losses of copper and solvent:
CostðoperÞ =ACc +Sl A (Xl Cx )+ð1 Xl Csol ) (7.8)
where
A is the feed rate of the aqueous phase
Cc is the cost of copper
Xc is the concentration of copper in the raffinate
Sl is the solvent loss per litre
Xl is the volume concentration of LIX 64N
Cxl is the cost of LIX 64N
Csol is the cost of diluent
The capital costs depend on the (1) number of stages of extraction and
stripping, (2) concentration of LIX 64N, (3) size of stages, and (4) ratio of
flow rates.
The size of settlers is determined by the rate of phase disengagement,
which, here, is a function of the solvent concentration of LIX 64N and the
aqueous phase feed rate. The capital cost of extraction equipment is a
function of the number of tanks, the size of the tanks, and the solvent
inventory [9]:
A A
CostðequipÞ =NKl b+ Xl Cxl +(1 Xl )Csol (7.9)
S ex NV
where
N is the number of tanks
Sex is the design settling rate
328 Ritcey
V is the volume of organic phase in each tank

kl is a factor relating tank settling area to cost
b is another cost factor
The costs function for stripping is similar in form to that used for extraction.
Thus the total cost function, Cost(tot) is given by:
CostðtotÞ t = CostðoperÞ +R CostðcapÞ (7.10)
where R is the interest rate on capital, which includes equipment, buildings,

and so forth.
Although it may seem desirable to use the maximum concentration
of solvent mixture and operate at saturation loading, this may be impracti-
cal, or even impossible, in actual plant practice. For the solvent mixture,
although increased loading will result in increased concentration of the ex-
tractant, often the increase will not be a linear relationship [78]. In addition,
in some systems, increasing solvent concentration results in higher solvent
losses in the aqueous phase because of viscosity effects, even at elevated
temperatures. Indeed, as the viscosity increases, excessive agitation can pro-
duce stable emulsions, making the process inoperative. Thus, a compromise
has to be developed between a reasonable solvent concentration and good
operation with equipment that has been selected. Robinson [77] stated that
an increase in concentration of the extractant could result in:
1. Higher loading in the organic phase, resulting normally in reduced strip-
ping costs
2. Higher extraction efficiency for the same number of stages, or fewer
stages for the same extraction efficiency
3. Increased operating costs, associated with solvent losses in the system
4. Increases or decreases in working capital associated with solvent
inventory. The form is dependent on the interactions of Eqs. (7.8) and
(7.9) and the concentration of extractant in the solvent
Some writers have stated that there is no point in maintaining opera-
tion at maximum solvent loading. This may well be true for some situations,
but where a plant can run at saturation loading, the economics will favor this
operation because of maximum utilization of solvent, which is an expensive
inventory item. Generally, a slightly lower than saturation condition is run to
allow changes in solution feed concentration and perhaps surges or change in
flows. However, there are certain situations for which the solvent is main-
tained at considerably less than saturation. Two of these conditions are as
follows:
1. In dump leaching of copper, it is not necessary to use extra stages of

extraction to recover all the copper from the solution, since the raffinate
will be recycled to the dump.
2. In the extraction and separation of zirconium from hafnium in a nitric
acid system, using TBP, the system operates best if run at about 10%
less than saturation [56]. As saturation of the solvent is approached, a
zirconium compound precipitates in the presence of the solvent, causing
cruds and emulsions. This problem is also encountered in rare earth
circuits using DEHPA.
The optimum number of extraction stages is a function of the con-
centration of the active agent, the flow ratio, the value of unextracted metal,
pH, and the interest on the capital investment.
Other variables that may affect extraction efficiency and, therefore,
cost are the following [79]:
1. Viscosity of the organic phase as a function of the diluent (Fig. 7.16)
2. In a mixer-settler, the effect of agitation speed on the extraction or
stripping
3. The effect of organic phase continuous versus aqueous phase con-
tinuous operation
The importance of optimization of as much of the solvent extraction
circuit as early as possible in the flow sheet design has been noted earlier.
Because of the nature of the equipment selected, optimization may well
not be possible until the pilot plant or plant stage. However, if the choice
of equipment favors mixer-settlers, optimization of the process may be
possible early in the developmental stages. Such an optimization was shown
for copper extraction in a joint investigation by the University of Bradford
and CANMET [2]. Procedures for investigating mixer and settler perform-
ance using a batch-stirred tank have been described, with the objective of
reducing experimental work to a minimum. Equations were developed from
a set of statistically designed experiments to determine the effect of major
operating variables on kinetics of mass transfer. These variables were
impeller speed, stirring time, volume fraction of the phase dispersed, and
the type of dispersion (organic or aqueous continuous). The effect of
increase in the extractant concentration at constant dispersed phase holdup
is shown in Fig. 7.17. Knowledge of densities, viscosities, interfacial ten-
sions, and drop sizes was insufficient to explain the relationship of the
measured velocities to each other. Additional factors must affect, therefore,
the coalescence rate. The data obtained from the statistically designed set
of experiments can be used to develop an empirical equation to predict
the degree of copper extraction in a batch-stirred tank as a function of
330 Ritcey
Fig. 7.16 Effect of diluent on viscosity of solvent.
stirring speed, stirring time, and phase ratio. From this data, the stage
efficiency in a flow system may be estimated. Optimization of these para-
meters can result in providing sufficient data for the most economic design
of large-scale mixer-settlers.
The design criteria necessary for optimization of the engineering and
design of the solvent extraction process can probably be summarized as
follows:
1. Clarification of feed liquors
2. pH adjustment and possible filtration
3. Heating requirements
4. Nominal feed flow rate
5. Solution grade expected
6. Production capacity demanded
7. Concentration of solvent
8. Selection of diluent and possible modifier
9. Stage requirements for extraction and, therefore, equipment choice

and design based on total flow requirements
10. Scrubbing: solution type, stages, and equipment
11. Stripping: solution type, stages, and equipment
12. Possible recovery of metal as oxides or powders and expected purity
13. Solvent losses and possible recovery stages and equipment
14. Labor and maintenance
15. Complete utilization of the solvent
16. Dispersion: coalescence (equipment design)
17. Kinetics: choice and size of equipment
Fig. 7.17 Effect of extractant concentration on coalescence.

332 Ritcey
18. Equilibrium: number of stages

19. Possible by-products
20. Environmental considerations
7.16 SOLVENT IN PULP PROCESSING

If substantial reduction in processing costs is to be achieved in the mill, then
the liquid–solids separation must be eliminated. This costly operation
accounts for approximately 50% of the capital and operating costs of the
mill. Such a possibility exists in the use of resin-in-pulp continuous ion
exchange or solvent-in-pulp extraction from the leach slurry [1,80]. There
currently is a renewed activity in several research laboratories to develop
suitable equipment and technology to accomplish such a goal. The Beijing
Research Institute of Uranium Ore Processing in China, for example, is
presently engaged in considerable research in solvent-in-pulp, using sieve
plate pulse columns [81] and other equipment design.
7.17 PLANT SCALE-UP AND OPERATION

Scale-up of a process from pilot plant to plant is essentially an engineering
problem, and is discussed only briefly here. Economics is most important
and that section should be consulted.
Experience in solvent extraction processes has shown that such pro-
cesses can be scaled up from pilot plant-or even bench-scale-data quite
reliably. This is particularly evident in processes employing mixer-settlers.
However, scale-up will only be as reliable as the data on which it is based, and
time spent in obtaining the correct and relevant data will always pay divi-
dends.
One important aspect of pilot plant operations relating to the successful
operation of a commercial process is the training of operators, supervisors,
and analytical personnel. Expertise and knowledge gained can mean several
months saved in getting the plant operation to optimum production.
Several processes that appear to be easy, or relatively so, in pilot plant
testing can prove to be major operations in a large plant. For example, the
following operations should be considered: shutdown and startup; addi-
tional tank capacity for the emptying and refilling of large contactors or
settlers; recovery of solvent spillage, which might seem trivial on a small-
scale operation but could involve considerable losses of expensive extrac-
tant; shutdown and startup of processes upstream or downstream from the
solvent extraction circuit-or tank capacity to allow for shutdown of one
unit process without shutdown of other parts of the process.
In general, the pilot plant should be operated as closely as possible to

the conditions that might prevail in the commercial plant; conversely, the
commercial plant should be based on data and conditions obtained in the pilot
plant operations, and extrapolations to larger units, flows, equipment, and so
on must be given serious consideration if the plant is to be a successful venture.
There are examples of successful operations that have not gone
through the pilot plant stage but have been scaled up directly from bench-
scale data. Thus, one process for the separation of (bulk) rare earths from an
aqueous solution scaled up from only bench-scale information to contactors
capable of flow rates of 500 gal(UK) min1 (2.3 m3 min1). As it happened,
these contactors were ideal for this particular process.
A change in the type of contactor between that used in the pilot plant
to that used in the commercial operation can result in problems. This may
occur because a particular large-scale contactor was not readily available.
For example, in two plants separating chemically similar metals, although
sieve plate pulse columns were found to be excellent contactors for the
required separation, other columns were chosen for the commercial plant
because pulse columns of sufficient size were not readily available. The
columns chosen, at least in one case, resulted in very poor physical opera-
tions and process efficiency.
The answer here seems to be lead time; given sufficient lead time, most
equipment manufacturers are capable of producing the required type of
contactor to suit the process specifications.
Design, scale-up, and operating problems of some solvent extraction
processes have been described in the literature, and make interesting reading
for engineers and chemists involved in such operations [1,3,5,10,11,16,36,
40,59,61,82,83].
REFERENCES
1. Ritcey, G. M.; Ashbrook, A. W. Solvent Extraction, Principles and Application
to Process Metallurgy, Parts 1 and 2; Elsevier Science Publishers, Amsterdam,
(1979, 1984).
2. Slater, M. J.; Ritcey, G. M.; Pilgrim, R. F. Proc. ISEC’74, Lyon, Society of
Chemical Industry, London, 1974, 1, 107–140.
3. Ritcey, G. M.; Ashbrook, A. W.; Lucas, B. H. Development of a solvent
extraction process for the separation of cobalt from nickel, presented at the
Annual AIME Meeting, San Francisco, 1972. CIM Bull., January 1975.
4. Murray, K. J.; Bouboulis, G. J. S/X carrier solvents, Proceedings of AICHE
Symposium on Solvent Ion Exchange, Tucson, Arizona, May, 1973.
5. Lo, T. C.; Baird, M. H. I.; Hanson, C. Eds., Handbook of Solvent Extraction;
John Wiley and Sons, New York, 1982.
334 Ritcey
6. Warwick, G. C. I.; Scuffham, J. B.; Lott, J. B. Proc. ISEC ’71, The Hague,
Society of Chemical Industry, London, 1971, 1373p.
7. Warwick, G. C. I.; Scuffham, J. B. Proceedings of International Symposium
on Solvent Extraction on Metallurgical Processes, Antwerp, 1972, pp. 40–
47.
8. Agers, D. W.; Dement, E. R. Proceedings of International Symposium on
Solvent Extraction on Metallurgical Processes, Antwerp, 1972, pp. 31–39.
9. Ritcey, G. M. Proceedings of AIChE Symposium on Solvent Ion Exchange,
Tucson, Arizona, May 1973.
10. Nelson, R. R.; Brown, R. L. The Design of Metal Producing Processes, Kibby,
R. M., Ed., American Institute of Mining Metallurgy, New York, 1967, p. 324.
11. Miller, A. The Design of Metal Producing Processes, Kibby, R. M., Ed.,
American Institute of Mining and Metallurgy, New York, 1967, p. 337.
12. Kunda, W.; Warner, J. P.; Mackiw, V. N. Trans. CIM, 1962, 65, 21.
13. McKay, H. A. C.; Healy, T. V.; Jenkins, I. L.; Naylor, A. Eds., Solvent Ex-
traction Chemistry of Metals, MacMillan, London, 1966.
14. Gindin, L. M.; Bobikov, P. I.; Rozen, A. M. Russ. J. Inorg. Chem., 1960, 5,
906, 1146.
15. Ashbrook, A. W. J. Inorg. Nud. Chem., 1972, 34, 1721.
16. Ritcey, G. M.; Lucas, B. H. Proc. ISEC’71, The Hague, Society of Chemical
Industry, London, 1971, p. 463.
17. Blake, C. A.; Brown, K. B.; Coleman, C. F. USAEC Rep. ORNL-1903, 1955.
18. Slyusarev, D. F. Zh. Anal. Khim., 1972, 27, 753.
19. Pich, H. C.; Santos, C. S.; Elias, J. T.; Alves, M. F.; Conceicao, E. H.; Viera, H.
X. Proceedings Symposium on Recovery of Uranium, IAEA, Vienna, paper
IAEA-SM-135/17, 1971.
20. Ashbrook, A. W. Anal. Chim. Acta, 1972, 58, 115.
21. Ritcey, G. M. Proceedings Second Annual Meeting, Canadian Hydro-
metallurgists of Metal. Soc., CIMM, Canadian Institute Minerals and Met-
allurgy, October 1972, p. 11.
22. Ritcey, G. M.; Lucas, B. H. CIM Bull., February 1974.
23. Shulz, W. W.; Navratil, J. D.; Kertes, A. S. Science and Technology of Tributyl
Phosphate, CRC Press, Vols. 1–4, 1991.
24. Higgins, C. E.; Baldwin, W. H. Anal. Chem., 1960, 32, 233.
25. Rydberg, J.; Reinhardt, H. Chem. lnd., 1970, p. 488.
26. Ashbrook, A. W. Commercial chelating solvent extraction reagents. 1. o-Hy-
droxyoximes; Purification and isomer separation/extraction. Metallurgy Divi-
sion, Mines Branch, Dept. Energy, Mines and Resources, Canada, Report
EMA 73-10, 1973.
27. Takahashi, M.; Ogata, T.; Okino, H.; Abe, Y. Paper abstracts ISEC’83,
Denver, USA, pp. 359–360, 1983.
28. Ashbrook, A. W. Commercial chelating solvent extraction reagents, II. n-Al-
kenyl-8-hydroxyquinoline; Purification and properties, Metallurgy Division,
Mines Branch, Dept. Energy, Mines and Resources, Canada Report EMA 73–
34 1973.
29. Rickelton, W. A.; Mihaylov, I.; Love, B.; Louie, P. K.; Krause, E., U.S. Patent
5,759,512, June 2, 1998.
30. Rickelton, W. R.; Robertson, A. J.; Hillhouse, J. H., Solv. Extr. Ion Exch.,
1991, 9(1):73–84.
31. Marston, A. L.; West, D. L.; Wilhite, R. N. Solvent Extraction Chemistry of
Metals, McKay, H. A. C.; Healy, T. V.; Jenkins, I. L.; Naylor, A., Eds.,
MacMillan, London, 1966, p. 213.
32. Ritcey, G. M. Hydrometallurgy, 1980, 5, 97.
33. Ritcey, G. M. Proceedings of 14th International Mineral Processing Congress,
CIM, Toronto; Vol. 1, 1982.
34. Ritcey, G. M. International Solvent Extraction Conference, ISEC ’83, Denver,
AIChE, 1983, pp. 88–89.
35. Cao, S.; Sworschak, H.; Hall, A. Proc. ISEC ’74, Lyon. Society of Chemical
Industry, London, 1974, pp. 1453–1480.
36. Ritcey, G. M.; Slater, M. J.; Lucas, B. H. Proceedings of International Hy-
drometallurgy Symposium, AIME, Chicago, February 1973, AIME, New
York, 1973, pp. 419–474.
37. Orth, D. A.; McKibben, J. M.; Scotten, W. C. Proc. ISEC’74, Lyon, Society of
Chemical Industry, London, 1974, pp. 514–433.
38. Fletcher, A. W.; Hester, K. W. A new approach to copper-nickel ore proces-
sing. Paper presented at the Annual AIME Meeting, New York, February
1964.
39. Burkin, A. R. Proceedings of First Hydrometallurgy Meeting, CIM, Ottawa,
October 1971.
40. Ritcey, G. M.; Joe, E. G.; Ashbrook, A. W. Trans. AIME, 1967, 238, 330–334.
41. Huppert, K. L.; Issel, W.; Knoch, W. Proc. ISEC’74, Lyons, 1974, The Hague,
Society of Chemistry Industry, London, 1971, pp. 2063–2074.
42. Young, W. Crud in Gulf Minerals Rabbit Lake solvent extraction circuit, Paper
at AIME Annual Meeting, New Orleans, February 1979.
43. Kowalik, T.; Cantwell, T. Crud problems in solvent extraction and strip circuits
of Conquista uranium, Paper at AIME Annual Meeting, New Orleans, Feb-
ruary 1979.
44. Bakshani, N.; Maurer, E. E.; Alien, M. P.; Degenhart, A. L. Crud at the Sohio
uranium solvent extraction circuit. Paper at AIME Annual Meeting, New
Orleans, February 1979.
45. Moyer, B.; McDowell, W. J. Proceedings of AIME International Hydro-
metallurgy Symposium, Hydrometallurgy-Research, Development and Plant
Practice, Osseo-Asare, K.; Miller, J. D. Eds., Metallurgical Society of AIME,
1983, pp. 503–516.
46. Ritcey, G. M.; Wong, E. W. Proc.ISEC’88, Moscow, 1988, 2:116.
47. Lucas, B. H.; Ritcey, G. M. CIM Bull. June 1975; Canadian Patent No.101759,
September 1977; U.S. Patent 3,969,476, 1976.
48. Rossiter, G. Anamax Twin Buttes oxide plant operating experience-first year.
Presented at the Arizona Section, AIME, Hydrometallurgical Division, Spring,
1976.
336 Ritcey
49. Abramo, J. A.; Lowings, S. W. H. Uranium processing at Exxon’s Highland

operation, Proceedings of AIChE Symposium on Solvent Ion Exchange, Tuc-
son, Arizona, May 1973.
50. Lucas, B. H.; Ritcey, G. M. CIM Bull. January 1975.
51. Hurst, F. J. Recovery of uranium from wet-process phosphoric acid by solvent
extraction, presented at Annual AIME Meeting, Las Vegas, February 1976.
52. Young, W. Davy Power-Gas, Paper at AIME Annal Meeting, New Orleans,
February 1979, p. 59.
53. Meyburgh, R. G. J. S. Afr. Inst. Min. Metall., 1970, October, 54–66.
54. Meyburgh, R. G. J. S. Afr. Inst. Min. Metall., 1971, April, 190–197.
55. Gaudernack, B.; Braaten, O. Occurrence and extraction of rare earths in
Norway, presented at the Ninth Rare Earth Research Conference, Virginia,
October 10–14, 1971.
56. Ritcey, G. M.; Conn, K. Liquid-liquid separation of zirconium and hafnium,
Eldorado Nuclear, R&D Division, Ottawa, Rep. T67-7, 1967.
57. Ryle, B. G. USAEC Rep. TID 5295, 1956.
58. Fletcher, A. W.; Flett, D. S. Solvent Extraction Chemistry of Metals, McKay,
H. A. C.; Healy, T. V.; Jenkins, I. L.; Naylor, A. Eds., MacMillan, London,
1966, p. 359.
59. Fletcher, A. W.; Hester, K. D. Trans. AIME, 1964, 229, 282.
60. Tunley, T. H.; Faure, A. The Purlex process-a description of the pilot plant
that led to the use of the process in South Africa. Paper presented at the Annual
AIME Meeting, Washington, D.C., February, 1969.
61. Tremblay, R.; Bramwell, P. Trans. Can. Inst. Miner. Metall., 1959, 62, 44.
62. Bellingham, A. I. Proc. Australes. Inst. Miner. Metall., 1961, 198, 85.
63. Lloyd, P. J. J. S. Afr. Inst. Miner. Metall., 1961, 62, 465.
64. Dasher, J.; Power, K. L. Eng. Miner. J., 1971, 772(4), 111.
65. Rawling, K. R. World Miner., 1969, 22, 34.
66. McGarr, H. J. Eng. Miner., 1970, 171(10):79.
67. Agers, D. W.; House, J. E.; Swanson, R. R.; Drobnick, J. L. Trans. Soc. Miner.
Eng., 1966, 35, 191.
68. Hartlage, J. A. Kelex 100-a new reagent for copper solvent extraction. Paper
presented at Society of Mineral Engineers, AIME Fall Meeting, Salt Lake City,
September 1969.
69. Flett, D. S. Miner. Sci. Eng., 1970, 2(3), 17.
70. Ritcey, G. M.; Lucas, B. H.; Ashbrook, A. W. Proc. ISEC’74, Lyon, Society of
Chemical Industry, London, 1974, 3, 2873–2884.
71. Pickering, Q. H.; Henderson, C. J.WPCF, 1966, 38(9), 1419–1429.
72. Hawley, J. R. Use, characteristics, and toxicity of mine-mill reagents in On-
tario, Ministry of Environment, Ontario, Toronto, 1974.
73. American Cyanamid Company, Technical data on cobalt-nickel separation
using CYANEX 272 extractant, 1982.
74. Mobil Oil Corporation, Material safety data bulletin, 1981.
75. Chem. Eng. News, 1968, 11, 44–46.
76. Agers, D. W. Proc. First Hydrometallurgy meeting, CIMM Mines Branch,

Ottawa, October 28–29, 1971.
77. Robinson, C. G. Proc. ISEC’71, Hague, Society of Chemical Industry, London,
1971, pp. 1416–1428.
78. Ritcey, G. M.; Ashbrook, A. W.; Lucas, B. H. CIM Bull., January 1975.
U.S.Patent: 3,999,055, 1968.
79. Tunley, T. H.; Birch, C. P. The recovery of copper from sulphate leach liquors
by liquid ion exchange with LIX64N, National Institute of Metallurgy,
Johannesburg, South Africa, Report No. 140904.
80. Ritcey, G. M. Proc ISEC’88, Moscow, 1988, 2, 116.
81. Jintang, Wang; Renli, Zhang. Paper presented at Hydrometallurgy’81, Man-
chester, Society of Chemical Industry, June 1981.
82. Albinsson, Y.; Ohlsson, L. E.; Persson, H.; Rydberg, J. Appl. Radial. Isot.
1988, 39(2), 113.
83. Rydberg, J. Acta Chem. Scand., 1969, 23, 647.
8
Principles of Industrial Solvent
Extraction
PHILIP J. D. LLOYD Energy Research Institute, University of Cape
Town, Rondebosch, South Africa
8.1 INTRODUCTION
The theory of solvent extraction was considered in Chapter 1, and Chapter 7
covered the application of liquid–liquid extraction in industry. The princi-
ples underlying the design of industrial applications are addressed in this
chapter.
At the very simplest level, an aqueous solution contains a valuable
component to be recovered, and a number of other components from which
the desired component should be separated. It is assumed that, as a result of
laboratory studies such as those outlined in previous chapters:
1. A suitable extraction system has been identified that will extract the
desired component selectively from the less desired components.
2. Suitable physical and chemical conditions for carrying out the extrac-
tion have been found.
3. The rates of extraction of the desired and possibly also some of the
undesired components have been determined at least qualitatively.
To design a process to recover the valuable component, a number of
questions which must be answered:
What fraction of the desired component can be recovered?
How much of the extractant must be added to achieve the desired
recovery?
How much of the undesired components are extracted with the desired
component?
339
340 Lloyd
Can anything be done to reduce the concentration of undesired com-

ponents in the extract phase?
Having extracted the desired component, how can it be recovered from
the extract in a useable form?
How, physically, can the extraction be carried out, and what type and
size of apparatus is required?
How much of the extract is lost to the raffinate, and can anything be
done to recover it?
This chapter therefore outlines the methods for answering these types
of questions, so that the solvent extraction process may be applied in practice
and desired components may be recovered in an energy-efficient, environ-
mentally safe, and economical way.
The nomenclature used in solvent extraction has been defined in
Chapter 1 and is illustrated in Fig. 8.1. Not all of the steps shown in this
figure will be found in every extraction process, but equally there may be
occasions where it is necessary to add additional steps; for example, to
recover the extractant from the scrub raffinate. So while Fig. 8.1 is not a
completely general flow diagram it covers most of the processes likely to be
found in practice. Variations of this flow sheet will become apparent during
the remaining chapters.
Fig. 8.1 Basic processes in solvent extraction.

Principles of Industrial Solvent Extraction 341
Table 8.1 Effect of Phase Ratio in Single Stage Extraction

Fraction extracted
Phase ratio Relative purity
Y EA EX EA/EX
0.1 0.500 0.010 50.50
1 0.909 0.091 10.00
10 0.990 0.500 1.98
Note: Assuming DA = 10, DX = 0.1.
8.2 EXTRACTION
8.2.1 Single-Stage Extraction
As described in Chapter 4, if a solution containing a desired component X is
contacted with an immiscible solvent phase, then X distributes itself between
the feed solution and the solvent according to:
DX ¼ [Xe ]=[Xa ] (8.1)
where the subscript e represents the extract (solvent) phase and subscript a
the aqueous phase.
If the phase volume ratio Y = Ve/Va, then the fraction extracted, EX,
is given by:
EX ¼ Dx =(Dx þ 1) (8.2)
that is, the fraction extracted in a single stage is a function of both the dis-
tribution ratio and the phase ratio. This is an important finding. Much of the
work in previous chapters has concentrated on the equilibrium distribution
of species between the extract and raffinate phase. However, as soon as an
answer to the question ‘‘What fraction of the desired component can be
recovered?’’ is sought, the volumetric ratio between the two phases becomes
almost as important as the equilibrium distribution. Consider now another
simple case with a desired component A for which DA = 10 and a con-
taminant X for which DX = 0.1. Table 8.1 shows the effect of varying the
phase ratio on (1) the fractions of A and X which are extracted, and (2) the
ratio of the fractions extracted, which is a measure of the product purity.
This not only shows how the extent of extraction varies with phase
ratio at a constant distribution ratio, but also how varying the phase ratio
affects the relative purity of the product. Increasing the phase ratio by a
factor of 100 nearly doubles the recovery, but drops the relative purity by a
factor of over 25. Note that, in a single stage, it is not possible to achieve
both high recovery and a high degree of separation simultaneously. Also,
342 Lloyd
although the distribution ratios of the two species differ by a factor of 100,
the relative purity is always less than 100.
A second phenomenon of great industrial importance is the effect of
saturation of the solvent on the product purity. Implicit in the derivation of
Eq. (8.2) is the assumption that DX is constant. As discussed in Chapter 2,
DX is only a constant under ideal and constant conditions (usually at trace
concentrations in both phases). It changes markedly as the concentrations
vary in the two phases.
At higher concentrations, so much of an extractable component may
be extracted that an appreciable fraction of the extractant is bonded to the
extracted component, so that in turn the concentration of the free ligand in
the extract phase is significantly reduced.
Industrial practice naturally requires the maximum use of the rela-
tively expensive extractant, so that saturation of the extractant phase with
reduction of the free ligand concentration to a minimum is the general rule.
A model is thus needed to quantify the effect of a reduction in DX as the
concentration of the extracted species in the organic phase increases.
In many extraction systems, DX is proportional to [L]n, where: [L] is
the free ligand concentration, and the exponent n is determined by the
number of ligand molecules per molecule of extracted complex. In these
systems, therefore, we may write:
DX =[Xe ]=[Xa ]=D0 f[Le ]tot m [Xe ]gn (8.3)
in which:
D0 is the distribution ratio of species X at trace concentrations, readily
determined in the laboratory.
[Le]tot is the total ligand concentration in the extract phase.
m approximates the ratio of ligand molecules to extracted molecules close to
saturation.
n approximates the ratio of ligand molecules to extracted molecules close to
infinite dilution.
The term in curved brackets on the right-hand side of the equation is the free
ligand concentration.
It should be noted that m and n in Eq. (8.3) are sometimes equal, but
often differ in value, which underlines the fact that this equation has no
theoretical basis. It is merely a convenient way of representing much ex-
perimental data using three parameters that can readily be determined
experimentally.
The equation can be fitted to a wide range of isotherms with quite
small residual errors over the entire range of concentrations of interest. It is
a liquid–liquid analogue of the Langmuir adsorption isotherm applicable to
vapor–solid interactions, with the extractant concentration taking the place

of the free surface area of the solid.
Where more than one component is extracted, then the free ligand
concentration will be reduced by all components present in the organic
phase. For example, consider the extraction of two components, A and B,
that have similar chemistries of extraction. Assume for both components
m = n = 2, that they have D0 values of 10 and 1 respectively.
The equilibria for the distribution of A and B between the two phases
are described by
DA ¼ [Ae ]=[Ae ] ¼ 10 (1 2[Ae þ Be ])2 (8.4a)
DB ¼ [Be ]=[Ba ] ¼ 1 (1 2[Ae þ Be ])2 (8.4b)

while two mass balance equations
[AF ] ¼ [Aa ] þ [Ae ] (8.4c)
[BF ] ¼ [Ba ] þ [Be ] (8.4d)

where subscript F indicates the feed, provide sufficient equations to solve
simultaneously for the four unknowns [Aa], [Ae], [Ba], and [Be] in terms of
[AF], [BF], and Y. Simple analytical solutions are available only for cases
where n = 1. In the general case, it is easier to solve these sets of equations
using spreadsheets and tools such as Solver in M-S Excel.
The results of one such set of calculations are shown in Fig. 8.2. The
more extractable species A competes strongly for the ligand, so that the
equilibrium curve of A in the presence of B is depressed only slightly below
that for the extraction of A on its own. In contrast, the equilibrium curve for
the extraction of B in the presence of A is depressed markedly relative to the
extraction of free B. The effect of this is to improve the product purity
significantly over what would have been possible at low concentrations of
the extracted species in the extract. Thus, at high concentrations, saturation
effects can improve product purity to a far greater extent than the equilib-
rium isotherms of the individual species would indicate superficially.
This is a key finding in understanding why solvent extraction is widely
adopted in industrial practice. It is possible to achieve a higher purity of
product than would be indicated by the separation between species when
they are extracted individually. Clean separations between species with very
similar extractabilities such as rare earth ions have proved practical by
relying on saturation effects more than on the inherent separability of the
species.
The two phenomena, namely the effect of phase ratio on the fractional
recovery and purity and the effect of saturation on purity, have thus far been
344 Lloyd
Fig. 8.2 Isotherms for the extraction of two species at a phase ratio of 1 and equal
concentrations in the feed.
illustrated using only single-stage batch extraction. In both cases, if better

separation was sought, then the recovery of the desired species was reduced.
This explains why single-stage extraction is rarely adopted commercially.
Multistage operation, outlined in Chapter 1.4.3, offers significant advan-
tages, and in particular permits both high recovery and high purity of the
product.
8.2.2 Repeated Extraction or ‘‘Cross-Flow’’ Systems

Returning to the example shown in Table 8.1, it is intuitively obvious that
the low degree of extraction achieved when using a phase ratio Y = 0.1
could be improved if the raffinate could be reextracted with fresh solvent,
and that if this reextraction were to be carried out at the same phase ratio,
the second extract would show a similar purity. The first and second extracts
could then be combined to give overall a better recovery than could be
achieved in a single stage.
For a single, highly extractable component this argument is entirely
valid. A mass balance over any one stage gives:
Va [AF ]=Ve [AE ] þ Va [AR ] (8.5)

which on rearranging becomes:
([AF ] [AR ])=[AE ] ¼ (8.6)
where
[AF] is the concentration of A in the feed to any one stage,
[AR] is the concentration of A in the raffinate leaving any one stage, and
[AE] is the concentration of A in the extract that is in equilibrium with [AR]
[AE] and [AR] are related via the isotherm, which is known, so the series of
Eq. (8.6) governing each individual stage of extraction can be solved ana-
lytically. The properties of the solution can be readily understood graphi-
cally.
Figure 8.3 is a graphical construction that represents a series of equa-
tions such as Eq. (8.6), for the case Y = 1 and the isotherm of Eq. (8.4). From
[AF]1, a line of slope 1/Y = 1 intercepts the equilibrium curve at [A]1.
A perpendicular from this intercept cuts the x axis at [AR]1 = [AF]2, from
which a further line of slope 1 is drawn. Of course, Y may be varied from
Fig. 8.3 Graphic construction for a four-stage cross-flow cascade.

346 Lloyd
Table 8.2 Extraction of a Single Component in a Cross-Flow Cascade at Y = 1

Aqueous Extract
Stage no. Feed Raffinate [AE] EA ETOT [AE]AVG

1 1.000 0.623 0.377 37.7% 37.7% 0.377
2 0.623 0.292 0.331 53.2% 70.8% 0.354
3 0.292 0.071 0.221 75.7% 92.9% 0.310
4 0.071 0.008 0.063 88.5% 99.2% 0.248
Single 1.000 0.079 0.230 92.1% 92.1% 0.230
stage to stage, but there is no particular advantage in doing so. Four stages
are shown with a feed at a relative concentration of 1.
Table 8.2 gives the stage-by-stage performance of the four stages of
repeated extraction, the cumulative extraction, ETOT, and the average con-
centration of AE, [AE]AVG, in the organic phases mixed together after each
repeated extraction. In addition, Table 8.2 also shows the performance of a
single stage in which the volume of extractant is the same as the total used in
the four stages, i.e., a single extraction at a phase ratio of 4.
This illustrates that even with a moderate distribution ratio (10 in
this case), useful recoveries (>99% as shown by ETOT for Stage 4) can be
achieved in comparatively few stages by repeated extraction. However, as
the number of stages increases, the concentration of the extracted species in
the combined extract, [AE]AVG, drops significantly.
A single-stage extraction using the same total volume of solvent
achieves only 92% extraction, and the extract concentration is only 0.23, vs.
nearly 0.25 for the cross-flow extraction. The use of four cross-flow extrac-
tion stages is clearly preferable to a single extraction. Equally, of course, the
use of more than four extraction stages, each with a proportionately smaller
volume, would improve the performance. In the limit, one would seek a
differential contacting process similar to the Soxhlet extractor employed for
extraction from solid phases, but such a contactor has not found use in
solvent extraction.
The mathematical treatment of a cross-flow cascade is straightfor-
ward. The mass, WE, which is extracted in n stages is given by:
X
WE ¼ VEn [CE ]n (8.7)
The cumulative recovery is given by:
ETOT ¼ WE =([CF ] V) (8.8)
while the average concentration in the combined extracts is given by:
[CE ]AVG ¼ WE =(n VE ) (8.9)

provided the phase ratio is kept the sameP in each repeated extraction; if not,
nVE in Eq. (8.9) must be replaced by VEn.
Now consider the effect of repeated extraction on the degree of
separation that can be achieved from contaminants. Table 8.3 shows how the
performance of the extraction deteriorates when a second component is
present. Comparison with Table 8.2 shows that the percentage extraction of
A in four stages drops to slightly over 95%, while the extraction of the
impurity B rises to 37%. The purity of A, calculated as the ratio of A in the
extract to the sum of A and B in the extract, becomes poorer as more stages of
cross-flow extraction are added. A single stage, using the same total volume
of extractant as used in the four stages of cross-flow extraction, performs
nearly as well as the cross-flow cascade.
The reasons for this poor performance are clear. In stages 3 and 4 of
the cascade, the desired component A is at low concentration in the extract
phase, so there is a relatively high concentration of the free ligand available
and the undesired component B is relatively highly extracted. Indeed, in the
fourth stage, the concentration of A in the extract is lower than that of B
even though A is 10 times more extractable than B.
Clearly, it would be preferable from the point of view of the purity of
the product if there were high concentrations of A in both the extract and
the raffinate of a stage. This can be achieved in countercurrent extraction,
which allows both high recovery and the achievement of high product purity
when properly designed.
8.2.3 Countercurrent Extraction

Consider the extract from the fourth stage of the cascade given in Table 8.3.
The concentration of the desired species is 0.11, so it is clearly not saturated.
Thus in principle it could form the feed to the third stage where the aqueous
Table 8.3 Extraction of Two Components in a Cross-Flow Cascade with Y = 1
Stage [AR] [AE] EA.TOT(%) [BR] [BE] EB.TOT(%) Purity of A(%)

Feed 1.000 - - 1.000 - - -
1 0.662 0.338 33.8 0.951 0.049 4.9 87.4
2 0.371 0.291 62.9 0.882 0.069 11.8 84.2
3 0.158 0.212 84.2 0.778 0.104 22.2 79.1
4 0.047 0.111 95.3 0.630 0.147 37.0 72.0
Single 0.144 0.214 85.6 0.627 0.093 37.3 69.7
348 Lloyd
Fig. 8.4 Countercurrent extraction.
concentration is higher. Systems in which such a strategy is employed are

known as countercurrent extraction systems.
A typical flow arrangement is shown in Fig. 8.4, which clearly shows
how the name arises. A mass balance over any one stage ‘i’ gives:
VE ([AE ]i [AE ]iþ1 ) ¼ VA ([AR ]i1 [AR ]i (8.10)
which on rearrangement yields:
VE ([AR ]i1 [AR ]i )
¼ ¼ (8.11)
VA ([AE ]i [AE ]iþ1 )
If the cascade is calculated by the same techniques as used for the cross-flow
cascade in the previous section, but with the layout shown in Fig. 8.5, the
organic concentration increasing from stage to stage and a phase ratio
Y = 1, the performance is shown graphically in Fig. 8.5. In this case, the
calculation has been carried out from the organic feed end, rather than the
aqueous feed end as in the case of Fig. 8.3, to allow for the increase in
organic-phase concentration. In every other respect the calculation is iden-
tical to that which led to Fig. 8.3. From [AR]3 a vertical line is drawn to
intersect the equilibrium curve, from which point a line of slope 1 is drawn
to reach the point [AR]2 on the axis. A vertical line from this point reaches the
equilibrium curve at point E, and again a line of slope 1 is drawn, but on
this occasion it is only carried to point G, because there is contact between
aqueous phase of concentration [AR]1 and organic phase not of zero con-
centration but of concentration [AE]3. The calculation then continues in the
same way for another stage, making three stages in all.
As the calculation continues, it is found impossible to proceed much
beyond the point marked A on the diagram, at which point the aqueous
concentration is about 0.34. The reason for this is that at a phase ratio
Y = 1 the cascade has too little organic phase for the duty it is being asked
to perform. Consider the rectangles ABCD and CEFG in Fig. 8.5. The
diagonals BD and EG have the same slope. The diagonals AC and CF thus
have the same slope, and AF is therefore a straight line. The point A has the
coordinates [AF]; [AE]1 and the point F the coordinates [AR]2; [AE]3. Thus the
slope of the line
([AE ]1 [AE ]3 )
AF ¼ ¼ 1= (8.12)
([AF ] [AR ]2 )
which is identical to Eq. (8.11) except it represents the mass balance over
two stages.
Thus to achieve an extraction equivalent to that shown in Fig. 8.3
requires an overall mass balance for which, at the first stage, [AE]1 = 0.377
and [AF] = 1.0; while at the last stage [AE]3 = 0 and [AR]3 = 0.008. Then by
Eq. (8.11) over the whole cascade:
(1:00 0:008)
¼ ¼ 2:63
(0:377 0)
or the slope of the mass balance line = 1/2.63 = 0.38.
Mass balance lines such as AF are important and are known as
operating lines in countercurrent cascades. Clearly, an operating line can-
not cross an equilibrium line. Wherever an operating line approaches an
Fig. 8.5 Graphic construction for a countercurrent cascade.

350 Lloyd
equilibrium line, a pinch point is the result and the number of stages needed
to achieve the desired degree of extraction approaches infinity.
In Table 8.4, the calculations behind Fig. 8.5 are repeated for a value
of Y = 2.63. Comparison with the results of Table 8.2 indicates clearly that
the countercurrent cascade offers a similar overall extraction in the same
number of stages, together with an extract of significantly higher con-
centration.
The data of Table 8.4 are also shown in Fig. 8.6, which is shown as an
equilibrium isotherm, an operating line, and a series of steps between the
operating line and the isotherm. These steps are entirely equivalent to the
lines establishing the mass balances for each stage in Figs. 8.3 and 8.5.
For instance, the horizontal line AB represents [AF]1[AR]1, while the ver-
tical line BC represents [AE]1[AE]2.
The graphical construction of an extraction isotherm, an operating
line, and the stepwise evaluation of the number of stages in this manner is
known as a McCabe-Thiele diagram. Historically, it found great application
in a variety of mass transfer operations, from gas adsorption through dis-
tillation to solvent extraction. However, the advent of modern computa-
tional techniques has made it largely redundant, as it is often easier and
certainly more accurate to calculate the cascade directly.
In part also this is because better equilibrium data are now available
through the use of equipment such as AKUFVE, which has brought the
realization that it is not sufficient to view the isotherm as fixed. Quite small
changes in aqueous composition or in phase ratio can change the isotherms
and cause dramatic effects on the performance of a countercurrent cascade.
Even relatively crude models for the isotherms, such as Eq. 8.3, can demon-
strate these effects in cascades.
Consider now the extraction of two species simultaneously in a coun-
tercurrent cascade. To obtain the equilibria requires solving two equations
similar to Eq. (8.4) simultaneously. Determining the concentrations within
the cascade means taking into account a mass balance such as Eq. (8.11) at
Table 8.4 Extraction of a Single Component in a Countercurrent Cascade at Y = 2.63

Aqueous Extract
Stage no. Feed Raffinate [AE] EA ETOT [AE]AVG

1 0.172 0.008 0.062 95.3% 99.2% 0.062
2 0.780 0.172 0.294 77.9% 83.8% 0.294
3 1.030 0.780 0.388 24.2% 26.7% 0.388
4 1.064 1.030 0.401 3.2% 3.2% 0.401
Fig. 8.6 McCabe-Thiele graphic construction for a countercurrent cascade.
each stage. Table 8.5 presents such calculations for the same two equilibria as
used before, with A having D = 10 and B having D = 1.0. It should be noted
that the calculation is not simple, and requires estimation of the equilibrium
at each stage before the next stage is computed.
The behavior of the system is very counterintuitive. The more extract-
able component is relatively well behaved and a recovery of 99.4% is
achieved in four stages. In contrast, the less extractable component behaves
unusually. Its concentration increases in the aqueous phase between the first
and the third stages. In the extract phase, it has a peak concentration in the
extract from the fourth stage. Thus, the less extractable component circulates
between the first and last stages, being ‘‘squeezed out’’ of the organic phase
by the more extractable component, then being extracted back from the
aqueous phase once the concentration of the more extractable component in
the aqueous phase has fallen.
This is caused by the coupling of the two equilibria. The presence of
high concentrations of the less extractable component so disturbs the equi-
libria that the more extractable component has a distribution ratio of only
about 2.5 in the fourth stage of the cascade. If the less extractable component
352 Lloyd
Table 8.5 Calculation of a Countercurrent Cascade with Two Components, Y = 3.0

Aqueous Extract
Stage no. Feed Raffinate Feed Extract

1. The more extractable component
1 1.00 0.686 0.227 0.332
2 0.686 0.265 0.086 0.227
3 0.265 0.058 0.017 0.086
4 0.058 0.006 0.0 0.017
2. The less extractable component

1 1.00 1.208 0.127 0.058
2 1.208 1.491 0.221 0.127
3 1.491 1.496 0.223 0.221
4 1.496 0.826 0.0 0.223
were not present, the distribution ratio would approach the infinite dilution
value of 10.
Because of the high concentrations of the less extractable component
within the cascade, the final separation between the two components is not
particularly good. The final extract is only 85% pure. This could be improved
by reducing the phase ratio, which would have the effect of loading the more
extractable component sooner in the cascade, and thus squeezing out the
less extractable component better. In practice, however, one cannot reduce
the phase ratio too far without running the risk of the cascade becoming
unstable, or requiring, for instance, temperature control. It is preferable to
scrub the extract, as described in the next section.
It should be noted that this was a fairly severe test. Having a feed
containing equimolar quantities of components whose extractability differs
by a factor of only 10 is rare. Usually the differences in extractability are
greater, or the less extractable component is present in low concentrations
relative to the more extractable. In both these cases, good separations can be
achieved in a countercurrent cascade.
The cascade can be drawn graphically in the McCabe-Thiele form, but
the equilibria are so distorted that the exercise is not very valuable. This is a
general finding in industrial practice. Graphical methods are adequate to
give an indication of the number of stages likely to be needed for a partic-
ular duty. They fail when separations between similar species are sought.
Highly selective extractants are desirable, but often show poorer phys-
ical or chemical properties than less selective extractants. As the example
above illustrates, a difference of a factor of only 10 in the distribution ratios
gave a reasonably purified product in four countercurrent extraction stages,

even when the contaminant was present at relatively high concentrations.
One can improve the performance of the cascade if the desired product is
present at a higher concentration than other extractable species. This is the
basis for further purification by means of scrubbing of the final extract.
8.2.4 Extraction with Scrubbing

Where a high-purity product is sought, the same philosophy of saturating the
aqueous phase with the unwanted component is used in scrubbing the
extract. A recycle stream of the pure product is used to ‘‘scrub’’ the impure
extract. In the process, the extract is saturated with the desired product, and
the impurities are removed. The final raffinate from scrubbing is then recy-
cled to the aqueous feed so that the now-contaminated desired product can
be recovered. This is shown in Fig. 8.1.
In Table 8.6, the results of computations identical in principle to those
described previously are given. The final extract from the countercurrent
cascade described in Table 8.5 was contacted with a pure stream of the more
extractable component at a high phase ratio. The high phase ratio is chosen
to minimize the volume of aqueous phase that must be recycled to the feed.
The cascade is fed with an extract of the same composition as that
resulting from the extraction shown in Table 8.5, with EA = 0.332 and
EB = 0.058. In four stages, using a concentrated solution of A at Y = 10,
EA = 0.350 and EB = 0.00014, the purity of the product is increased from
85–99.96%. Note that the composition of the scrub solution is the same as
Table 8.6 Calculation of a Countercurrent Cascade in Scrubbing, Y = 10

Aqueous Extract

1 0.500 0.390 0.339 0.350
2 0.390 0.328 0.333 0.339
3 0.328 0.299 0.329 0.333
4 0.299 0.297 0.332 0.329

1 0.0002 0.0016 0.00014 0.00028
2 0.0016 0.0027 0.00028 0.00039
3 0.0027 0.0035 0.00039 0.00047
4 0.0035 0.0041 0.00047 0.0584
354 Lloyd
the composition of the final extract after scrubbing. In practice, additional

purification can often take place during recovery of the components from
solution after stripping, which allows the scrub solution to be even purer, and
in turn allows the product purity to be higher than shown in this example.
Also of note in Table 8.6 is the composition of the final raffinate, with
AA = 0.297 and AB = 0.004. This can clearly be recycled directly to the feed
to the extraction cascade, where the fact that this stream is purer than the
feed stream will further aid the achievement of product purity.
Techniques for achieving high-purity products by countercurrent
extraction and scrubbing of the extract have proved essential for the pro-
duction of nuclear-grade uranium. They have also found application in the
separation of the rare earths and a number of other difficult separations. A
feature of the operation of these systems is the need for close control of
flow rates and even temperature in order to achieve a consistent product
quality. The product quality is a very nonlinear function of the operating
parameters. However, with modern control systems this disadvantage can be
overcome.
8.2.5 Summary of Extraction

In this section, the first three questions posed in the introduction have been
answered. It has been shown that:
The fraction of the desired component that can be recovered is deter-
mined not only by its inherent extractability but also by the phase ratio
and the presence of competing extractable components.
The volume of solvent needed for a particular duty depends on the
system adopted for the extraction; in general, a single-stage extraction
requires more solvent phase than a cross-flow cascade, which in turn
needs more than a countercurrent cascade.
The purity of the product is determined not only by the inherent selec-
tivity of the solvent system for the component sought, but also on the
phase ratio, the concentration of the contaminants, and the performance
of any scrubbing of the extract.
8.3 STRIPPING
The general principles established for extraction apply to stripping, although
of course distribution ratios are sought that are significantly less than unity in
order to accomplish the strip as efficiently as possible.
Stripping can equally be done in single-stage, cross-flow, or counter-
current systems. To illustrate how the overall concepts remain valid, the per-
formance of a countercurrent cascade accepting as feed the scrubbed extract
of Table 8.6 is calculated assuming that the equilibria involved are described
by:
DA ¼ 0:1(1 2[Ae þ Be ]) (8.13a)
DB ¼ 0:01(1 2[Ae þ Be ]) (8.13b)

that is, as for Eq. (8.4) but with m not equal to n. Further, the cascade should
strip over 95% of the desired component and yield a strip raffinate which is
as concentrated as possible (assuming a solubility of A in the stripping
solution of 2).
As before, the same mass balance limitations on phase ratio apply.
Although this is a stripping cascade, it remains countercurrent in its struc-
ture. For an organic feed concentration of 0.350 (that is, the extract after
scrubbing, as shown in Table 8.6), assuming >95% strip and a final aqueous
of <2, the maximum phase ratio is Ymax = 2/(0.95*0.350) = 6.02. (Max-
imum in this case because stripping is in the reverse direction to extraction.)
Choosing a phase ratio of 5.0 gives the results shown in Table 8.7.
Four stages are needed. The strip raffinate has a concentration of 1.71,
and the stripped organic contains only 2% (0.0073/0.35) of the component A
in the scrubbed extract, so the stripping efficiency is 98%. The less extrac-
table (more readily stripped) component B was completely stripped in three
stages.
Note that the stripped extract still contains traces of the desired com-
ponent. This is a slight nuisance as the stripped extract will, in the normal
course of events be recycled to the extraction stages where it will reduce the
Table 8.7 Calculation of a Countercurrent Cascade in Stripping, Y = 5.0

Aqueous Extract

1 0.000 0.074 0.022 0.0073
2 0.074 0.232 0.054 0.022
3 0.232 0.601 0.128 0.054
4 0.601 1.713 0.350 0.128

1 0.000 0.0000 0.000 0.000
2 0.0000 0.0000 0.0000 0.0000
3 0.0000 0.00003 0.00001 0.0000
4 0.00003 0.00070 0.00014 0.00001
356 Lloyd
extraction efficiency in the first stage. This practice is fairly common in

industrial practice, as it is more economic to lose a little of the desired
component in the extraction stages than to have too dilute a strip raffinate by
seeking a total strip.
The purpose of seeking a concentrated strip solution is to reduce the
energy required to recover the product from the strip solution. In the case
of metal salts, precipitation, electrolysis, direct reduction, and a host of
other techniques may be used to generate the final product. In the case of
the extraction of organic compounds, distillation, crystallization, or similar
separation methods are used. In each case, the more concentrated the strip
solution, the less energy is required to recover the desired components.
It sometimes happens that an undesired component is not stripped. It
then builds up in the organic phase until it interferes seriously with the
extraction. In this case, the solvent is given a more vigorous strip to regen-
erate its performance by removing the contaminant. A small sidestream may
be bled continuously from the recycled organic phase and regenerated, or
the contaminant may be allowed to build up in the extract for some time
before the solvent is regenerated on a batch basis. Because of its vigorous
nature, regeneration of the organic phase can be expensive, but sometimes
the contaminant is very valuable-platinum, gold, and cobalt complexes
have acted as contaminants on occasions, and their recovery has paid for the
regeneration!
8.4 EQUIPMENT FOR CONTINUOUS CONTACT

8.4.1 Stagewise Extraction
Thus far in the discussion of industrial practice, we have referred to ‘‘stages’’
of extraction without defining a stage. Clearly these stages could be scaled-up
versions of separatory funnels, but this is inefficient, because it implies batch
rather than continuous operation. Industry prefers continuous operation
because it is generally simpler to control automatically, and because it makes
better use of labor.
The typical stage might be a mixer-settler. As its name suggests, this
comprises some means for mixing the two phases and an adjoining means for
separating them. Mixers can be stirred tanks, of a sufficient size to retain the
mixed phases long enough to effect transfer of the desired species from one
phase to the other. The mixer usually consists of some form of impeller or
propeller in a tank that usually contains some means to prevent the mixed
liquids from swirling and thus reducing the efficiency of mixing. In circular
cross-section mixers, this usually takes the form of vertical baffles mounted
on the wall of the mixer.
Fig. 8.7 Layout of two stages of a mixer-settler cascade.
Figure 8.7 illustrates two stages of a mixer-settler cascade. The aqueous

feed enters the first stage, where it is mixed with partially loaded extract from
the second stage. The mixed phases pass to the settler, where they are
separated under gravity. The fully loaded extract overflows an upper weir
and passes to scrubbing and/or stripping. The first-stage raffinate overflows a
lower weir and is pumped to the second stage, where it is mixed with the
solvent feed. After settling again, the final raffinate passes to treatment
before discharge as waste.
There are many variants on this simple theme. For instance, many
other methods for mixing have found use. In a design offered by Lurgi, the
phases are mixed in what is essentially an axial flow pump, and then pass
down a relatively long pipe where the turbulence of flow keeps the phases
mixed while the extraction takes place. In another design, the individual
phases are pumped and then join and pass through a static mixer. There are
no particular physicochemical reasons for preferring one type of mixer to
358 Lloyd
another. As long as they provide adequate interfacial area for the extrac-
tion to take place, without creating such small droplets that they will not
settle efficiently, and provided there is sufficient residence time for the desired
degree of extraction to take place, then one mixer will work as well as
another.
There is, however, one physicochemical criterion that is important in
industrial mixing, and that is ensuring that the correct phase is dispersed in
the other. There are several reasons for this:
It is sometimes found that mass transfer is more rapid if one phase is the
dispersed phase rather than the other.
Alternatively, the dispersed phase is chosen because, by definition, it will
not contain droplets of the continuous phase. In this way the dispersed
phase, after settling, will not entrain the continuous phase and entrain-
ment losses from the settler will be reduced.
Whatever the reason for choosing the dispersed phase, it is important to
ensure that the mixer will keep that phase dispersed during operation, as
changes in the dispersed phase, i.e., phase inversion, can cause considerable
operating problems.
Usually the continuous phase is the phase present in greater volume. It
is possible to run for long periods with the greater volume phase dispersed,
but phase inversion is always a risk in such circumstances. To overcome this
risk, where it is desired that the lesser volume phase is continuous, then a
portion of that phase may be recycled from the settler back to the same
mixer to ensure that within each stage it is the greater volume, even if it is
the lesser volume phase overall.
In large-scale operation, the volumetric flow of the phase to be dis-
persed is so large that it becomes necessary to disperse that phase into the
mixed phases. Otherwise ‘‘blobs’’ of the dispersed phase will act locally as
the continuous phase, and the intended continuous phase will be dispersed
in the blobs before the shear forces in the mixer break them up. This can
lead to excessive entrainment losses.
In some mixer-settler designs, the impeller is arranged both to mix the
two phases and to provide the necessary energy to transfer the phases from
one stage to the next, in which case it is known as a pump-mix mixer. The
head required to move a phase from one stage to the next is small, so the
impeller need not be efficient as a pump. Nevertheless, the design of impellers
for the dual purpose of both mixing and pumping is more of an art than a
science. Moreover, in full-scale operation it has been found difficult to start
up cascades of pump-mix mixers and achieve equilibrium rapidly. Accord-
ingly, this design has primarily found use in small-scale applications in the
nuclear and pharmaceutical industries.
Settlers tend to be less varied in their design. They typically comprise

a relatively large, shallow tank, rectangular in shape, with an inlet for the
mixed phases at one end and two outlets for the separated phases at
the other. Various devices are used to introduce the mixed phases gently into
the settler, and to control the flow of the mixed phases while they separate,
but these do not change the basic principle of separation under gravity.
The level of the heavy phase outlet within the settler controls the level
of the interface (Fig. 8.8). At the interface, the static pressure due to light
phase above the interface is determined by the density of the light phase, ro,
and the depth of the light phase, H. Similarly, the static pressure above the
interface in the overflow leg is determined by the density of the heavy phase,
r, and the height h of the weir above the interface. The pressure must be the
same at the same elevation, so hr = Hro, or
H ¼ h ð=o Þ (8.14)
Because the difference between the densities of the two phases is often
small, and may vary from one stage to the next, H can vary strongly with h.
Thus it is often necessary to make the height of the weir adjustable. Similarly,
if there are significant differences in flow rate, then the depth of the liquid
overflowing the weir has an effect on h and thus on the height of the interface.
The importance of these considerations is that the shallower the settler,
the more difficult the interface control. Shallow settlers are desired because
they reduce the inventory of the solvent. However, it is possible to make the
settler so shallow that interface control can be lost. In one case, large, shallow
settlers suffered from the effects of wind pressure, that caused such massive
Fig. 8.8 The control of the level of the interface in a settler.

360 Lloyd
oscillations in the level of the interface that light phase often passed over the
heavy phase weir.
The introduction of the mixed phases into the settler has been found to
be important if clean separation is to be ensured. ‘‘Picket fences,’’ vertical
plates set at an angle to the flow from the mixer into the settler, have been
used to calm the flow and ensure the spreading of the mixed stream across the
width of the settler. Various packings have been employed to aid settling, but
under industrial conditions they are liable to clog with adventitious material
or ‘‘third phases.’’
Baffles placed across the settler, of progressively lower height from
entrance to overflow, have been employed to hold back the mixed phases to
permit them to separate. The mixed phases will spread rapidly right across a
settler unless there is a baffle to hold them back. There have been many
reports of ‘‘wedges’’ of mixed phase in small-scale settlers. These are never
seen in industrial practice because considerations of pressure at a given point,
as were used above to determine the height of the interface, show that a
wedge is inherently unstable.
In one mixer-settler design, the mixed phases flow down a shallow
trough placed over the settler, which gives them an opportunity to coalesce
and separate before entering the settler. In this way, the capacity of the settler
is markedly increased, with a concomitant reduction in the inventory of
solvent required for a given duty.
The manner in which individual mixer-settler stages can be linked
together to form countercurrent cascades is illustrated in Fig. 8.7. If each
stage is on the same level, then some form of pump must be provided to
move each phase from one stage to the next. As indicated earlier, it is
sometimes convenient to use the mixer for this duty. Another arrange-
ment has the individual stages set at different elevations, so that one phase
(usually the phase with the greatest flow rate) can gravitate from one stage to
the next.
8.4.2 Differential Extraction

Thus far we have been concerned with stagewise operations, doing just what
is done in the laboratory when a solvent is mixed with an aqueous phase in a
separatory funnel and allowed to settle before being separated.
While countercurrent cascades can be operated in stages, there is no
need do so. Consider, for instance, pumping the dense phase to the top of a
tower and letting it flow down against drops of the light phase rising upward.
Assuming transfer from the dense to the light phase if the tower were high
enough, at the top there would be saturated light phase and at the bottom,
depleted heavy phase.
Clearly the concept of a stage has no meaning in such a tower. Instead,

we deal with differential transfer units, which are a measure of the change in
concentration per unit of difference in concentration (recall that the rate of
extraction is determined largely by the difference between the actual and the
equilibrium concentration of a solute, or ‘‘driving force’’).
At each point in the tower, a component A has an actual concentration
[A] and an equivalent equilibrium concentration [A]e. Then the number of
transfer units (NTU) required for the extraction is given a first approx-
imation by:
Z½Af
NTU ¼ d[A]=([A] [A]e ) (8.15)
½Ar
where [A]f and [A]r are the feed and raffinate concentrations of A respec-
tively. This is illustrated in Fig. 8.9. The driving force is [A] [A]e and the
inverse of the driving force is to be integrated between the feed and the
raffinate concentrations (not shown in Fig. 8.9).
This integral clearly depends on the slope of the operating line and, as
in the case of stagewise operations, if the operating line approaches the
equilibrium curve too closely, then the driving force approaches zero and
the inverse becomes very large. That is, when the operating line is close to
Fig. 8.9 Illustration of the determination of the driving force from equilibrium and
operating lines.
362 Lloyd
the equilibrium curve and there is a pinch point, the number of transfer units
becomes large.
Equation 8.15 refers to a single phase, and [A]e is related to [A] at each
point via the phase ratio (or operating line, which is the same thing). The
NTU could be calculated for the extract phase instead of the aqueous phase,
that is, either from the difference between [AR] and [AR]e or from that
between [AE]e and [AE].
Where there is an analytical expression for the equilibrium such as
Eq. (8.3), then Eq. (8.15) may be integrated directly. Otherwise it is neces-
sary to perform the integration numerically or graphically.
Once the number of transfer units has been found, the height of the
tower is determined from the product of the number and the height of each
transfer unit (HTU). The HTU is determined by physical parameters such as
the droplet size, the flow patterns in the tower, and the effect of any packing.
These all affect the rate of mass transfer, which is addressed in Chapter 9.
Very often the rate of mass transfer cannot be estimated from first princi-
ples, and it is necessary to estimate the height by determining the number of
transfer units achieved and then dividing the actual height of the column
employed by the number of transfer units, i.e.:
HTU ¼ H=NTU (8.16)
where H is the height of the column employed.
HTU is subjected to the effects of both radial and axial mixing,
and these are not readily quantified, so scale-up of columns of this kind is
often not based on fundamentals, but rather on correlations determined
from detailed studies of several systems in the particular design of column
chosen.
Physically, towers designed for countercurrent contact can be open,
but more usually contain some form of packing or plates. The material of
the packing is chosen so that one phase wets it preferentially, thus increas-
ing the surface area for mass transfer. Similarly, the plates are designed to
breakup droplets and increase the surface area. In addition, the contents of
the tower may be agitated either by an internal agitator or by pulsing the
fluids. The energy imparted by agitation or pulsation breaks up the droplets
of the dispersed phase. Again, further details are given in Chapter 9.
When the equilibrium curve is relatively linear, the driving force does
not vary greatly down the length of the column, and the number of transfer
units approaches the number of McCabe-Thiele theoretical stages. In this
case, it is reasonable to speak of the number of stages in the column, and to
calculate a height equivalent to a theoretical stage (HETS). However, if the
equilibrium curve and the operating line are far from parallel, the number of
theoretical stages becomes a poor measure of the column’s performance,

and the number of transfer units should be used.
8.5 EXTRACTION EFFICIENCY

Solvent extraction is a kinetic process. The key variables in determining the
rate of extraction are (1) the displacement of the system from equilibrium,
also referred to as the driving force; (2) the area through which mass can be
transferred, or the interfacial area; and (3) specific resistances in the inter-
facial region, particularly any slow interfacial reactions.
To a lesser extent, the rate is also affected by diffusion through the
bulk liquids, but in general industrial practice there is sufficient turbulence
to ensure that the bulk phases are well mixed. There is some control over
interfacial area, though not too much flexibility is available because a very
large interfacial area is associated with very fine droplets or very thin films.
This may result in excessive loss of solvent by entrainment. These aspects are
extensively discussed in Chapters 7 and 9.
There is little to be done about the displacement of the system from
equilibrium (although it may be noted that the average displacement is
maximal in a countercurrent cascade). As shown in section 8.4.2, the driving
force can be increased by reducing the slope of the operating line, i.e., in-
creasing the phase ratio y, but this is generally not economical. Very little
can be done about interfacial resistances once the extraction system has been
chosen, although renewal of the surface during bubble coalescence and
dispersion assists in overcoming some forms of this kind of resistance. Thus
industry either has to make the time of contact long enough to ensure that
equilibrium is essentially attained, or it has to accept the inherent ineffi-
ciency of a single stage, and employ more stages than would otherwise be
needed. In practice, it employs the latter strategy.
8.5.1 The Efficiency of a Single Stage

Figure 8.5 gives a graphic construction for a series of equations of the type
given in Eq. (8.11), for a single stage at equilibrium. The same basic equa-
tion governs a nonequilibrium stage, that is:
([AR ]f [AR ]e )=([AE ]e [AE ]f ) ¼ (8.17)
except that the product streams [AR]e and [AE]e are no longer at equilibrium,
but are reduced by the inefficiency to [AR]i and [AE]i. This is illustrated in
Fig. 8.10.
364 Lloyd
8.5.2 The Efficiency of a Cascade

The concepts of section 8.5.1 are readily extended to determine the efficiency
of a cascade. To illustrate this, the results of calculations identical to those of
Tables 8.2 and 8.4, but with an 80% stage efficiency, are given in Table 8.8.
The resultant McCabe-Thiele diagram is given in Fig. 8.11.
Comparison with the earlier data for equilibrium conditions shows that
an extra stage is needed. In spite of this, the final raffinate is significantly
higher than was achieved at 100% extraction efficiency. The influence of the
inefficiency is clearly greater the more dilute the solutions.
Figure 8.11 shows how the equilibrium curve ‘‘shrinks’’ in the presence
of inefficiencies. In multicomponent systems where there is mutual inter-
ference in extraction by several components, the efficency shrinkage comes
on top of the other reductions in the equilibrium curve, and for this reason
there is stress in such systems on achieving high efficiency.
8.6 SOLVENT LOSSES

Throughout, we have made the tacit assumption that the two phases, which
for convenience, we have called the aqueous and the organic phases, are
totally immiscible. For many systems this is a reasonable approximation, but
Fig. 8.10 Illustration of the effect of reduced efficiency of extraction.

Table 8.8 Extraction at 80% Efficiency, Phase Ratio 2.63

Aqueous Extract
Stage no. Feed Equilibrium Inefficient extract Extract Feed

1 0.105 0.004 0.024 0.037 0.006
2 0.402 0.031 0.105 0.150 0.037
3 0.875 0.284 0.402 0.330 0.150
4 1.038 0.834 0.875 0.392 0.330
5 1.064 1.031 1.038 0.402 0.392
for some systems mutual miscibility must be taken into account, particularly
when the primary solute is organic.
The methods for doing so are described in Chapter 9. The basic prin-
ciples remain unchanged-the primary difference is the choice of a consistent
basis for calculation, such as a solvent-free basis. Graphic techniques based
on triangular coordinates provide approximate answers, but modern com-
putational techniques are to be preferred.
Some consideration should be given at this point to the need to pre-
vent loss of the organic phase in the aqueous raffinate. This loss can arise by
either solubility in the aqueous phase or by entrainment of droplets not fully
settled. The solvent lost in this way can offer a finite environmental hazard
and be an economic cost on the process.
Clearly the primary duty is good engineering practice, which is covered
especially in Chapter 9. Often, however, additional security is provided in
the following form:
Additional settler capacity for final raffinate
Extraction of residual organic phase using a third diluent, from which it
is later separated, typically by distillation
Coalescence on a solid wetted preferentially by the organic phase
Flotation with air in the presence of surfactants
8.7 SUMMARY
In this chapter, we have seen the way in which laboratory studies of solvent
extraction are adapted to industrial use. Starting from a batch extraction, it
was shown how both recovery and product purity could be markedly influ-
enced by the volumetric phase ratio, and how it was impossible to achieve
both high recovery and high purity in a single stage. Cross-flow or repeated
extraction was then evaluated, and it was shown how this could improve
366 Lloyd
Fig. 8.11 McCabe-Thiele diagram showing the effect of reduced efficiency of extrac-
tion.
both recovery and purity, yet often resulted in mixed extract solutions that
were too dilute to be processed without further upgrading.
The concept of countercurrent extraction was then introduced, and it
was shown how the minimum phase ratio for a given degree of extraction
was determined. Countercurrent extraction could yield both high recoveries
and concentrated extracts, but studies on two-component extractions soon
showed that product purity suffered.
This led to the discussion of scrubbing as an essential adjunct to
countercurrent extraction where purity was important, and it was shown that
washing the extract with a small amount of aqueous phase could improve
purity markedly. Stripping was shown to follow the same underlying prin-
ciples as extraction for achieving efficient removal of extracted species, and
the need to choose phase ratios carefully to maximize the concentration of
the desired species in the strip solution was stressed.
There followed a brief discussion of equipment for carrying out sol-
vent extraction in industrial practice, both by stagewise and differential
contact. Some of the first principles for the design of differential contactors
were outlined and the part played by the efficiency of extraction in con-
tinuous equipment was discussed. Finally there was an outline of methods
for the control of solvent loss which forms probably the most important
environmental aspect of the application of solvent extraction.
9
Engineering Design and Calculation of
Extractors for Liquid–Liquid Systems
ECKHART F. BLASS* Technische Universität Munich, Munich, Germany
9.1 INTRODUCTION
The previous chapters have demonstrated that liquid–liquid extraction is a
mass transfer unit operation involving two liquid phases, the raffinate and the
extract phase, which have very small mutual solubility. Let us assume that the
raffinate phase is wastewater from a coke plant polluted with phenol. To
separate the phenol from the water, there must be close contact with the
extract phase, toluene in this case. Water and toluene are not mutually soluble,
but toluene is a better solvent for phenol and can extract it from water. Thus,
toluene and phenol together are the extract phase. If the solvent reacts with the
extracted substance during the extraction, the whole process is called reactive
extraction. The reaction is usually used to alter the properties of inorganic
cations and anions so they can be extracted from an aqueous solution into the
nonpolar organic phase. The mechanisms for these reactions involve ion pair
formation, solvation of an ionic compound, or formation of covalent metal-
extractant complexes (see Chapters 3 and 4). Often formation of these new
species is a slow process and, in many cases, it is not possible to use columns
for this type of extraction; mixer-settlers are used instead (Chapter 8).
This chapter explains in detail how this mass separation problem can be
solved. We shall adhere to engineering symbols as listed at the end of this
chapter (see section 9.10) to facilitate comparison with other chemical
engineering texts. Reference to these symbols will be made in the text without
any extensive explanations.
*Retired.
367
368 Blass
9.2 FUNDAMENTAL DESIGN OF EXTRACTORS

Returning to the example of removing phenol from coke wastewater by
solvent extraction, in the first important step, the two liquids must come into
contact. This can be achieved by dispersing one of the liquids into the other.
When the resulting droplets have been in contact long enough with the raf-
finate phase to allow the extractable component to change phases, the two
liquids have to be separated. The most favorable economic way of achieving
this is by separating the drops in the gravity field. This requires that the
liquids possess different densities. Figure 9.1 shows a simple design that
implements the process steps for a continuous phase throughput (i.e., a spray
column). Let us again use the system water–phenol–toluene as an example.
The short table of physical properties in Fig. 9.1 shows that toluene is lighter
than water. Thus, the flow direction of the two liquids in the gravity field is
defined: the water flows from top to bottom, the toluene from bottom to top.
Accordingly, the fluids must be transported to the column and removed by
pumps. If the column is filled with water for the initial operation and if the
water is always kept at the marked height by an overflow control during the
following, stationary operation, the toluene breaks into drops at the feed
point to become the dispersed phase, then rises in the continuous water
phase. When the drops have reached the water interface, they gradually
coalesce and form a continuous toluene phase that floats on the water phase.
This water–toluene boundary is sometimes called the principal interface.
However, the column can also be operated in reverse by filling it with
toluene and adjusting the principal interface at the bottom of the column.
Then water is the dispersed phase and would break into drops at the feed
point at the top of the column. These drops descend in the toluene phase and,
at the bottom of the column, coalesce to a water homophase that is below the
toluene phase. When needed, the principal interface can be adjusted some-
where between top and bottom of the column, whereby the heavier liquid is
dispersed above and the lighter liquid below the principal interface. How is it
decided which of the two liquids should be dispersed? Understanding the
flow and mass transfer processes in the extractor, which are analysed in this
chapter, provides the answer. At this point, only the important factors are
listed; thus, the dispersed phase should be:
The one with the higher viscosity, since a low viscosity of continuous
phase makes a higher phase throughput possible because of the lower
energy of dissipation
The one with the higher flow rate to obtain the maximum mass transfer
area
The one with the lower interfacial tension between liquid and vapor for
easier dispersion
Engineering Design and Calculation of Extractors 369
Fig. 9.1 Spray column for liquid–liquid extraction. The water phase flows from top to
bottom, the toluene as the lighter phase from bottom to top. If the column is always filled
with water at the marked height, the toluene breaks into drops at the feeding point. When
the drops have reached the water interface, they coalesce and form a continuous toluene
phase. The phenol transfers from the water phase to the toluene phase.
The one that is more potentially explosive, more poisonous, and

more expensive, since the holdup of the dispersed phase is only about
10–30%
The extract phase, if the interfacial tension s between the two liq-
uid phases decreases with increasing concentration of the transitional
370 Blass
component x [i.e., (ds/dx<0)] so as not to support the coalescence

between drops by mass transfer (or the raffinate phase, if ds/dx>0)
If the physical and volumetric properties of the two-phase flows change very
much, it may be necessary to adjust the principal interface closer to the
middle of the column. The dispersed phase often cannot be chosen from
theory, but only with the aid of experiments in a pilot column using the real
material system (see section 9.9).
Considering the processes in the simple spray column (Fig. 9.1), the
main questions that a chemical engineer must answer about the design of an
extractor can be listed.
1. Questions concerning the liquid system: the extraction needs a solvent
of suitable and well-known physical properties for the process. The same
is true for the raffinate phase (see Chapter 2).
2. Thermodynamic design: the ratio of extract to raffinate flow rate, the
phase ratio, and the number of theoretical separation stages necessary
for the separation task can be worked out from mass balances in con-
nection with the solution equilibria (see Chapters 2–8).
3. Flow and mass transfer: the transport of the two phases through an
extractor and the production of intensive phase contact are complex
hydrodynamic problems. Mass transfer provides the main dimensions
of the extractor. This chapter is chiefly interested in a suitable extractor
design, but also in the restrictions of the calculation.
4. Design of apparatus: extractors are more or less complicated con-
structions fitted to special tasks. The choice of material, calculation of
stability, and technical production quality of the design are special tasks
of the mechanical engineer. This book does not go into further detail
about these aspects.
9.3 DISPERSION OF A PHASE INTO DROPLETS

9.3.1 Gravitational Extractors Without Input
of Mechanical Energy
To obtain a large transfer area between raffinate and extract phases, one of
the two liquids must be dispersed into drops. Figure 9.2 demonstrates this
process schematically at a single nozzle. Similar to a dripping water tap,
individual drops periodically leave the nozzle when the volumetric flow rate
of the dispersed phase is low. When the flow rate is higher, however, the
liquid forms a continuous jet from the nozzle that breaks into droplets.
Because of stochastic mechanisms, uniform droplets are not formed. If
the polydispersed droplet swarm is characterized by a suitable mean drop
Fig. 9.2 Mechanism of drop formation on a single hole is dependent on the throughput.
Individual drops periodically leave the nozzle when the volumetric flow rate V is low.
When the flow rate is higher, the liquid forms a continuous jet that breaks into droplets.
(From Ref. 2.)
diameter, however, it is possible to describe and calculate the behavior of

such a polydispersed droplet swarm by approximation. The Sauter diameter
d32 can be used for flow and mass transfer calculations. It is defined by the
measured drop size distribution as follows:
P 3
nPi dPi
D32 ¼ Pi (9.1)
n d2
i Pi Pi
where nPi represents the number of drops of the class i of the diameter dP.
Figure 9.3a lists Sauter diameters calculated from drop size measurements at
single nozzles with dN = 1.5 mm for the system toluene (dispersed phase d )–
water (continuous phase c) as a function of the toluene velocity nN in the
nozzle [1]. The results are valid for sieve trays (and perforated plates), if
the distance between the holes is more than twice the Sauter diameter. Sieve
trays are simple and frequently used dispersion devices. Experiments have
shown that the flow rates through all the holes of a sieve tray are identical
only if the flow rate is exactly high enough to form a jet. Since the sieve tray
serves to distribute the toluene equally over the cross section of the column,
only the curve for nN>22 cm s1 is of interest (Fig. 9.3a). According to this
curve, the liquid flow rate, and thus the velocity through the hole nN should
372 Blass
Fig. 9.3 Sauter mean diameter d32 calculated from drop size measurements at single
nozzles of dN = 1.5 mm for the liquid systems (a) toluene (dispersed phase d ) water
(continuous phase c); and (b) butanol (d ) water (c), is dependent on the mean velocity vN
of the dispersed phase in the nozzle. (From Ref. 5.)
be chosen such that the Sauter diameter reaches its lowest value. This gives
the greatest possible drop surface as per unit volume and, thereby, the largest
mass transfer area:

A 6"d
as ¼ (9.2)
V s d32
when ed represents the holdup of the drops in the total volume.
From experiments, equations have been derived that enable calculation
of the minimum velocity in the nozzle, the nozzle velocity, and the Sauter
diameter at the drop size minimum. They provide the basis for the correct
design of a sieve tray [3,4]. Figure 9.4a shows the geometric design of sieve
trays and their arrangement in an extraction column. Let us again consider
toluene–phenol–water as the liquid system. The water continuous phase flows
across the tray and down to the lower tray through a downcomer. The
toluene must coalesce into a continuous layer below each tray and reaches
Fig. 9.4 Sieve trays with downcomers for liquid systems with (a) high interfacial tension
and dual-flow trays for liquid systems; and (b) with low interfacial tension. In the case of
downcomers, only the phase to be dispersed flows through the holes. The droplets are
formed by jet disintegration. In dual-flow trays, both liquids flow through the same holes
alternately. The larger drops split because of collision with the tray.
374 Blass
Table 9.1 Calculation of Drop Diameter for Dual-Flow Plates

sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

c "d 0:3
Sauter mean diameter d32 ¼ 1:2
g d

g 0:25
Single-drop velocity e ¼ 1:41
c
gdN2 1
Limiting criteria
3
a height such that the static pressure caused by the buoyancy presses the
liquid through the holes at the required velocity. Figure 9.3b shows the
results of another liquid system (i.e., butanol–water). Compared with the sys-
tem toluene–water, its interface tension s is much lower [5]. Therefore, the
drops are much smaller, and very low velocities at the hole cause the for-
mation of a jet. Clearly, surface tension is a very important material property
for extraction systems.
The extremely low liquid velocities at the hole require only very small
stationary layers, which cannot be implemented in technical columns without
difficulties. Therefore, sieve trays without downcomers are used (Fig. 9.4b).
Here, both liquids must flow countercurrently through the same holes.
Neither stationary layers nor liquid jets of the dispersed phase are formed.
According to experimental observations, the bigger drops split up because
they collide with the sieve trays. The drops flow stepwise through the holes in
bigger collectives and alternate with the continuous phase flowing through
the holes. Table 9.1 shows reliable, experimentally proved equations for
medium drop sizes in such dual-flow trays and also a limiting criterion
for sieve trays with or without downcomers according to Hirschmann
and Blass [5]. If the hole diameter dN obtained from the limiting criterion in
Table 9.1 is smaller than 2 mm, dual-flow plates should be selected.
Sieve trays of this type define the possible size of drops, once the sieve
tray has been designed for the intended flow rate of the dispersed phase. The
same can be applied if tower packings are placed in a column to break up the
dispersed phase by colliding with the packing. The drop size can be changed
in an extractor during operation by supplying mechanical energy. The three
most important operations for an additional energy input are mixing, puls-
ing, and centrifuging. Extractors with these systems are more flexible than
spray and simple plate columns in that they can be adapted to changing
liquid systems and operating conditions, although they are more expensive in
initial costs and in maintenance.
9.3.2 Extractors with Liquid Pulsing

Pulsing means that either the whole liquid content of a sieve tray column is
continually pushed up and down by a piston that moves to and fro, or the
whole plate package is moved up and down [3]. Figure 9.5 illustrates the two
extractor constructions schematically. They show about the same efficiency
Fig. 9.5 Sieve tray columns with pulsing of the whole liquid content (left side) or of the
whole plate package (right side, Karr column). The trays are meant to equalize the phase
flows across the column cross section and to disperse the droplets as uniformly as possible.
376 Blass
if the plate geometry is identical. It also shows the principal interface in the
upper part of the column so that the lighter phase is dispersed.
Figure 9.6 demonstrates what happens at the sieve trays when the
whole liquid is pushed up and down. Three different situations can be
observed with the aid of high-speed photographs. During the downward
motion of the liquid pulse, larger drops do not follow the motion of the
pulsing, but accumulate below the plate owing to their relatively high
buoyancy. Thus, once the upward motion starts, the relatively large drops of
this layer, together with the continuous liquid, are pushed through the holes.
Fig. 9.6 Behavior of droplets near the sieve tray in a liquid-pulsed column during
downstroke and upstroke of the liquid content.
The smaller drops follow. Both liquids accelerate in the holes, because the
sum of the cross section of all the holes is less than half the column cross
section. However, this motion is retarded within a short distance, whereby a
zone of drop compaction results above the trays. These phenomena are
modeled based on a balance of maximum and minimum kinetic energy and
the cohesive energy of the droplets [1]. After that, the resulting equation for
the maximum stable drop diameter in the field of pulsing is:
4
dP;max ¼ (9.3)
d (12 22 )
where n1 is the highest nozzle velocity, n2 the velocity at a short distance
above the sieve tray with an approximate value of zero.
New modeling was performed [6] taking into consideration first the
influence of drop viscosity that produced a better agreement with the exper-
imental results than Eq. (9.3). The authors presume that bigger drops have to
be deformed in the sieve tray holes and therefore the circulation inside the
drops is accelerated. Taking into account that the kinetic energy of drops is
changing into surface energy and circulation energy, the authors derive
Eq. (9.4) by balancing the energies for the stable drop size ds at a sieve tray
hole.
dN2
12 þ 20:5 d jcN þ A j
ds2
ds ¼ (9.4)
d (cN þ A )2
where
ds = stable drop diameter
dN = hole diameter
ncN = velocity of continuous phase in the hole
nA = rising velocity of a drop, Eq. (9.15)
nd, rd = dynamic viscosity and mass density of dispersed phase
Figure 9.7a shows values of stable drop diameters that are calculated
with Eqs. (9.3) and (9.4) for special conditions for the liquid system toluene–
water as a function of the product of pulse amplitude, a, and frequency, f,
the so-called pulse intensity. Figure 9.7b is a parity plot and shows that
Eq. 9.4 describes the measured drop sizes for a wide range of velocities and
parameters of the sieve tray geometry quite well.
The mean Sauter diameter, d32, is clearly smaller than the maximum
stable diameter. As experimentally proved by many authors, the ratio of
these is between 0.4–0.6.
There are various further attempts to model the complex flow and
drop-disintegration phenomena for drop size calculation. However, it has
378 Blass
Fig. 9.7 (a) Diameters of stable drops calculated from Eqs. (9.3) and (9.4) for the liquid
system toluene-water as a function of the pulse amplitude a and frequency f. (b) Parity plot
of measured and calculated crop sizes. (From Ref. 6.)
not been possible so far to include mass transfer influences and the inter-
action of dispersion and coalescence. Only empirical equations given in Ref.
[1, Chapter 17] give the best fit to experimental data and have the broadest
range of validity.
0:10
d32 h af 1=4
rffiffiffiffiffiffiffiffiffi ¼ C1 e0:74 exp 3:00 1=4 1=4
0:05 meter g
g

af 1=4
þ exp 28:65 1=4 1=4 (9.5)
g
The best values of the parameter C1 are 1.51, 1.36, and 2.01 for no mass
transfer, c ? d and d ? c direction of transfer respectively. The product af is
considered as the agitation variable in the equation, since the fit could not be
improved if a and f were treated separately. The average absolute value of
the relative deviation in the predicted values of d32 from the experimental
points is 16.3%. Even in packed columns, the separation can be substan-
tially improved by pulsing of the continuous phase resulting from greater
shear forces that reduce the drop size and increase the interfacial area [1,
Chapter 8].
9.3.3 Agitated Extractors

Extraction technique uses mixing tools mainly for adding mechanical energy.
Combinations of stirred tanks and settlers (so-called mixer-settlers) have
been used for many years, and with them the extraction efficiency of one
theoretical stage of separation could be approximated. If higher separa-
tion efficiency is required, several of these stages are connected in series to
form mixer-settler batteries. If the simple stage can be easily calculated, the
evaluation of a battery provides no additional problems as described in
Chapter 8, where the use of such batteries is extensively discussed. However,
the initial costs, the cost for solvents, the need for energy and space, and the
problem of interface control in each stage are considerably higher than for
countercurrent columns [4]. Figure 9.8 shows sections of some frequently
used stirred countercurrent columns [4]. All contain horizontal stator baffles
that divide the column into successive sections, and each section has a mixer,
fixed to a central shaft. Stirred extractors differ according to the types of
mixer and stator baffles. The mixers are meant to disperse one phase into
droplets and to mix the two phases in the section as efficiently as possible.
The stator baffles must support the separation of the two liquids in each stage
and provide the countercurrent flow of the two liquids from stage to stage
and, at the same time, provide a free area as large as possible for the liquid
flows passing through. Section 9.3 discusses this subject.
The problem of liquid dispersion by mixers is now discussed. In
mixing, dispersion takes place in the turbulent shear field of the mixer. The
intensity of the shear field is largely influenced by the geometry of the mixer.
The power consumption P can be calculated by the equation:
P
Ne ¼ (9.6)
c n3R dR5
If the Newton number, Ne, in relation to the Reynolds number, ReR, of the
mixer is known from experiments and the rotor speed, nR, the mixer diam-
eter, dR and the mass density rc of the continuous liquid are given.
380 Blass
nR dR2
ReR ¼ (9.7)
c
Dispersion stirring devices always work in the turbulent flow regime;

therefore, the statistic theory of turbulence [7] is important for understand-
ing the processes during dispersion. To calculate the drop sizes in the tur-
bulent shear field, an energy balance for a single drop is again used, as in the
preceding section. According to Kolmogoroff [7], assuming local isotropic
turbulence, the maximum drop diameter in the local isotropic shear field
follows, which results in:
0:6

dP;max 0:4 (9.8)
c
when C is the energy per unit mass. The same proportionality applies for
the Sauter diameter in stirred tanks. By using the energy per unit mass for
mixers,
n3R dR3 (9.9)
and the Weber number
c n2R dR3
We ¼ (9.10)

it is possible to use:
d32
We0:6 (9.11)
dR
instead of Eq. 9.8.
The most important options for the validity of Kolmogoroff’s theory
are:
Local isotropic turbulence

Equal viscosities and equal densities of the two phases
Negligible influence of the drops on the surrounding phase
Negligible holdup of the drops
The first two options are fulfilled quite well in stirred extractors for a
normal rotor speed and for many liquid systems. This is why empirical
equations for the mean drop size in stirred extractors usually originate from
3
Fig. 9.8 Sections of several stirred countercurrent columns. Horizontal stator baffles
divide the column into successive sections, each fitted with a mixer, all fixed to a central
shaft. The mixers disperse one phase into droplets and mix the two phases in the section as
efficiently as possible. The stator baffles must support the phase separation and the
countercurrent flow of the liquids from state to stage.
382 Blass
Eq. 9.11 and differ in the proportionality constant. For details, the reader is
referred to references 3 and 4. For drop size calculations, the empirical
equations given in Ref. [1, chapter 17] that take into account all the influ-
encing parameters are preferable.
Rotating Disc Contactor (RDC):
0:12 0:16 2 0:59
d32 C1 c d DR c g
¼ pffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi
DR 0:07 þ FrR c DR c
0:25 0:46
h DC
ð9:12Þ
Dc DR
The optimized values of C1 are 0.63, 0.53, and 0.74 for no mass
transfer, c ? d and d ? c, respectively. The value of the holdup is ignored
due to lack of data. Equation (9.12) predicts the drop size with an average
absolute value of the relative deviation of 23%.
Kuehni columns:
d32
= C1 e0:37 n0:11
s [0:14 þ exp ( 18:73FrR )]
DR
0:20 2 0:24
c DR c g
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð9:13Þ
c DR
The values of the constant C1 are 9.81 102 for no mass transfer
and c ? d transfer, and 0.31 for d ? c transfer. The stage number ns, which
varies from 2–17 in the present set of data, shows a rather weak effect on
drop size. Equation (9.13) predicts the drop diameter with an average
absolute value and relative deviation of 17.6%.
9.3.4 Centrifugal Extractors

A centrifugal field is a further method of providing mechanical energy for the
generation and coalescence of droplets. Centrifugal extractors, which have
cylindrical sieve trays arranged concentrically round a shaft, match best with
the gravity extractors described earlier. They are sometimes called rotating
columns. The most popular industrial version is the Podbielniak extractor,
which was used first for penicillin extraction [8], that uses a horizontal shaft.
Figure 9.9 shows a simple experimental centrifuge with a vertical shaft used
to demonstrate the mode of operation. The rotor is covered with a thick glass
disk [9], which makes it possible to record the processes in the rotor with a
high-speed camera. In this design, the two phases must be fed into and taken
out from the rotor with the aid of rotating sealing devices at the shaft. The
heavy phase is fed at the rotor centre and conveyed to the outside by the
centrifugal field force. The light phase is fed at the periphery of the rotor and
Fig. 9.9 Experimental centrifuge with cylindrical dual-flow perforated sheets arranged
concentrically round a vertical shaft. The heavy phase is conveyed to the outside by cen-
trifugal field force; the light phase is displaced to the inside by the heavy phase. With a
control valve in the discharge pipe of the light phase, the degree of filling of the rotor with the
two phases can be defined (i.e., the location of the principal interface). (From Ref. 9.)
displaced to the inside by the heavy phase. This causes a countercurrent flow
of the two phases. With a control value in the discharge pipe of the light
phase, the degree of filling the rotor with heavy and light phase can be defined
(i.e., the location of the principal interface). This causes a high exit pressure
on the light phase, which just compensates the higher centrifugal force of the
heavy phase. The principal interface in Fig. 9.9 can be found at a medium
radius of the rotor. The perforated sheets in Fig. 9.9 have no downcomers for
the continuous liquid, which is common for technical centrifugal extractors.
Then, both liquids must flow through the same holes, just as they do when
dual-flow plates are used in the gravity field. However, other processes take
place here. The heavy phase accumulates at the inside of the sheets as a very
thin layer that then flows through a partial cross section of the holes coun-
tercurrently to the light phase and finally breaks from the hole as a single
384 Blass
drop. Only when the layers are thicker, which can be achieved by down-
comers in the continuous phase, do jets that disintegrate into drops originate
from the holes. The drops, influenced by the centrifugal field, move outward
and form another thin layer in front of the following perforated sheet by
coalescence. The same processes take place at the periphery of the rotor,
where the light phase is dispersed.
Influenced by interfacial tension and centrifugal forces, spherical drops of
various diameters originate at the holes. If we again assume the Sauter
diameter, according to Eq. (9.1), as the mean diameter of the spectrum of
particles, the following equation for heavy and light phases results from
theoretical and experimental results [10]:
rffiffiffiffiffiffiffiffiffiffiffiffiffiffi
c
d32 ¼ 3:22 (9.14)
r!2 d
where o is the angular velocity of the rotor that influences the size of the
drops, and r is the radius of the perforated sheet. Hence, at each stage of the
perforated sheet there is a different mean diameter.
If the distance between two perforated sheets, which is the settling
distance of the drops, is higher than about 40 mm, the drops of the heavy
phase are deformed by the accelerated motion, so that they again disin-
tegrate. Technical distances of the perforated sheets, however, should be
lower than 40 mm so that as many stages as possible can be built into the
rotor to improve the mass transfer performance. Sections 9.4 and 9.5.2 deal
further with this topic. Drop swarms of the light phase, however, do not
change their mean diameter during their retarded motion to the rotor shaft,
no matter how far is the distance to the next sheet.
In addition to the rotating columns previously described, there are a
number of other designs for centrifugal extractors, many originally developed
for the separation of radioactive wastes in nuclear processes (see Chapter 12).
They are both of the mixer-settler type, as discussed in section 9.3.3, and of
the rotating column types.
In the simplest case, a liquid phase in a static mixer is dispersed into
another phase and then fed into a centrifugal separator for phase separation
(external mixer–centrifugal separator) [11,12]. The advantage of this design
is that mixer and settler are separately optimized, and the centrifugal unit
is relatively simple. At best, for each mixer-settler stage, a separation per-
formance corresponding to a theoretical stage can be achieved; for a more
efficient separation, several of this type of mixer-settler may be connected in
series. Furthermore, the mixing can be integrated with phase separation in
the centrifuge into an ‘‘internal centrifugal mixer-separator,’’ as in the
Robatel design. Figure 9.10a shows four such interconnected stages. These
centrifuges are used, for example, by the French nuclear reprocessing
Fig. 9.10 (a) Robatel four-stage centrifugal extractor. The heavy and the light liquids
are introduced near the bottom and top, respectively. They flow countercurrently and leave
the extractor at opposite ends. The extractor consists of a rotating bowl divided by baffles
into horizontal compartments on top of one another. The stationary central shaft carries
mixing disks (A) that run through each compartment and serve to mix and pump the two
phases to the settling chamber of the stage. This chamber comprises a set of two overflow
chutes (B for the heavy phase; C for the light phase), that stabilize the separating area
independently from the outputs. Each compartment has connections to take the settled
phases to the previous and following stages. The standard design has three to eight stages.
(Courtesy of Robatel SLPI and Eries, Genas, France.) (b, see next page) Westfalia coun-
tercurrent centrifugal extraction decanter. The heavy liquid phase containing solids such as
microorganisms and the solvent are fed into the rotating bowl through separate inlets at
different parts of the bowl. Phase contact is effected in countercurrent flow in the feed area
and in the spirals of the scroll. The separation zone is adjusted for variable liquid densities
by means of the ring dam at the cylindrical bowl end. After mass transfer in the contact
zone, the enriched solvent is discharged at the cylindrical bowl end under pressure by a
centripetal pump. The raffinate phase discharges at the conical end of the bowl. All
particles that sediment in the bowl are conveyed by the scroll to the conical end of the
bowl and discharged together with the raffinate phase. (From Ref. 13.)
Fig. 9.10 (continued ).
industry. In the United States, internal centrifugal mixer-separators are also

used, although the design principle is different based on the Sharples type of
centrifuges. Figure 9.10b shows the countercurrent centrifugal extraction
decanter by Westfalia [13], used in the biochemical industry when the raf-
finate phase contains microorganisms.
9.4 FLOW OF PHASES AND LIMITS

OF THROUGHPUT
In a nonpulsed, nonstirred, noncentrifuged flow, drops move through the
continuous liquid driven by the density difference. Their stationary velocity
of motion results from the interaction of buoyancy and drag forces. Figure
9.11 shows the behavior of individual drops in a static continuous liquid.
The stationary relative velocity of the drops np is plotted as a function of
their diameters dp. Unless their interfaces are scrupulously clean, only very
small drops behave like rigid spheres. When the drops are bigger, the shear
forces of the continuous liquid acting on the drop surface cause a circulation
inside the drop. Thus, the drops rotate in their surroundings and move faster
than rigid drops. If the drops are even bigger, they are deformed elliptically,
which makes their motion slower. Finally, when the drops reach a certain
diameter, their shape is no longer stable, and the drop oscillates along and
across its major axis, because periodic whirls form and separate from the
drop. These oscillating drops move along coiled paths. Their axial velocity
component, plotted in Fig. 9.11, remains nearly constant. The drop size can
be increased only up to a maximum value dp,max. If the value is higher, the
drop will break owing to drag forces.
The measured relation between individual drop velocity nP, and drop dia-
meter dP is well represented by the following equation [2]:
1. For circulating drops:
ReP ¼ KL0:15 ðAr0:523 KL0:1435 0:75Þ

if Ar > 1:83KL0:275 (9.15a)
2. For oscillating drops:
ReP ¼ KL0:15 ð4:18Ar0:281 KL0:00773 0:75Þ (9.15b)
The transition point of the two equations is determined by:

Arc0 ¼ 372:9KL0:275 (9.15c)
The basic terms nP and dP and the liquid properties are made non-
dimensional in the equations according to the similarity theory:
388 Blass
Fig. 9.11 Stationary relative velocity, vp, of pentachloroethane drops in motionless

water is dependent on drop diameter, dp. Very small drops behave like rigid spheres, as
shown. Larger drops have an internal circulation and are finally deformed elliptically.
When they have reached a certain diameter, the drops in the end oscillate along and across
their major axis. Their axial velocity is nearly independent of the diameter. Once the drop
size is higher than a maximum value dp,max, the drop will break, owing to the drag forces.
(From Ref. 2.)
P dP
Particle Reynolds number: ReP ¼ (9.16)
d
dP3 gc
Archimedes number: Ar ¼ (9.17)
c2
2c 3
Characteristic number of the continuous liquid: KL ¼ (9.18)
g c4
Now the two-phase flow in a nonpulsed countercurrent column will be

discussed. If the continuous liquid flows toward the drops, the drop velocity
decreases, whereas the holdup of the drops in the flow system increases if
the flow rate of the dispersed phase is maintained. Thus, the remaining
flow area for the flow of the continuous liquid narrows. This interaction
between flows of the two phases leads to a limit of operation when the
buoyancy is not sufficient to overcome the flow resistance. This is a complex
phenomenon on which the deformability of the drops is superimposed, so
that modeling the real processes mathematically would not be particularly
promising.
However, a simple model does work: the liquids should occupy a share of
the cross section according to their holdup and should stream toward each
other as layers. If their superficial velocity, that is, their volumetric flow
rate, related to the complete column cross section, S, is:
V_ c V_ d
_ c ¼ ; _ d ¼ (9.19)
S S
they have an effective velocity of _ c ð1 "d Þ or _ d ="d . Now the relative
velocity nr, between the two liquids is:
_ d _ c
r ¼ þ (9.20)
"d 1 "d
Either nr or ed can be calculated from Eq. (9.20), if there is one more
relationship between the two terms. In Fig. 9.12, nr related to the single-
drop velocity nP according to Eq. (9.15) is plotted as a function of the drop
holdup for droplet swarms with the Archimedes number as a dimensionless
term for the drop diameter for the measured values. It can be seen that
the relative velocity constantly decreases, as the holdup of the drops, ed,
increases and the size of the drops in the swarm decreases.
Many experiments have been published that attempt to predetermine
the dependency of Fig. 9.12. To illustrate the main point, we use the simple
equation:
r
¼ ð1 "d Þk (9.21)
P
The velocity of single drops nP can be calculated by Eq. (9.15). The
exponent k is defined by the empirical correlation:
k ¼ 4:6 0:13 ln Ar (9.22)
which describes well-published experiments in the range 2.5<k<4.6 [14].
The premises have now been provided for calculating the operation
limits (i.e., the highest possible flow rates of a countercurrent column). This
is reached as soon as no further increase of the flow rate can be performed by
the density difference. The consequence is that the light phase leaves the
column at the bottom instead of at the top and the opposite situation for the
heavy phase (see Fig. 9.1). This load limit is also called the flooding point.
The mathematical formulation of this condition makes use of the fact
that an increase of the flow rate cannot increase the drop holdup:
390 Blass
Fig. 9.12 Stationary relative velocity vr, of the drop swarms related to the single drop
velocity vp is dependent on the drop hold-up, ed, and Archimedes number Ar. (From Ref. 2.)
d"d d"d
¼ ¼1 (9.23)
d _ d d _ c
or
d _ d
¼ 0 when _ c ¼ constant (9.24a)
d"d
d _ c
¼ 0 when _ d ¼ constant (9.24b)
d"d
If Eq. (9.20) for the relative velocity nr is used to calculate the superficial
velocity, one obtains the equations for calculating the phase flow velocities
at the flooding point:

dr
_ c ¼ ð1 "d Þ2 r þ "d (9.25a)
d"d

dr
_ d ¼ "2d r ð1 "d Þ (9.25b)
d"d
or, using the dimensionless form [2]:

1=3
c 1=3 2 "d dRer
_ c ¼ Ar ð1 "d Þ Rer þ (9.26a)
c g d"d
1=3
c dRer
_ d ¼ Ar1=3 "2d Rer ð1 "d Þ (9.26b)
c g d"d
where the Reynolds number Rer is calculated with the velocity nr.
Equations (9.26a) and (9.26b) together with Eq. (9.21) for nr and Eq. (9.15)
for nI are the basis for flooding point diagrams shown in Fig. 9.13, where the
ordinate and abscissa contain the phase flow rates in dimensionless form. The
diagram shows the calculated flooding curve of circulating drops for six
different Archimedes values for a liquid system with KL = 1011 and Zc/
Zd = 0.5. In addition, it provides straight lines of constant holdup ed and of
constant flow rate ratios _ c =_ d . If two of these five parameters are given, the
other three can be taken from the diagram. The diagram can also be used for
other values of KL, since their influence on the flooding point is weak for
circulating drops. The effect of other viscosity ratios is demonstrated on the
flooding point curve for Ar = 5 105 and can be transferred to other Ar
values to estimate the results.
The diagram can also be used for oscillating drops, since according to
Fig. 9.11, the rising velocities of oscillating drops hardly depend anymore on
the diameter. The drop diameter attainable depends only on the liquid
system, as stated in Eq. (9.15c). Figure 9.13 lists the Ar numbers according
to Eq. (8.15c) for KL, values of 108, 109, 1010, and 1011, and for these values
the Ar lines are also plotted. These are the flooding curves for oscillating
drops.
To use the flooding point diagram, first it is necessary to decide whether
the drops produced in the extractor are circulating or oscillating. The mean
diameter d32 (see Eq. 9.1) is used for the characteristic drop size. If the flow
rate ratio is known from the thermodynamic design, the superficial velocities
of both phases can be determined at the flooding point. The minimum col-
umn cross-sectional area and diameter necessarily follows directly from the
superficial velocity at the flooding point with Eq. 9.19.
The method of calculation introduced in this chapter not only allows an
exact determination of the column diameter for nonpulsed sieve tray col-
umns, but also allows a good estimation of the diameters of pulsed and
stirred extractors. For the latter, however, more exact specific equations exist
for the flooding point, see for example [1,4].
392 Blass
Fig. 9.13 Flooding point diagram of countercurrent extractors. The drawn lines for six
different Ar values are valid for circulating drops. The dotted lines for three different KL
values are valid for oscillating drops. (From Ref. 2.)
At a maximum flow rate of the two phases, centrifugal extractors with

cylindrical perforated plates also show a flooding point. In contrast to the
processes in a gravity field, however, this is not caused by the countercurrent
flow of the two phases between two perforated plates, since the holdup of
the dispersed phase in the centrifugal field is very low owing to the high drop
velocity. In this case, the two phases in the holes finally block each other so
that the layer of the dispersed phase suddenly grows strongly and the dis-
persed phase also leaves at the continuous phase outlet.
The flooding diagram of Fig. 9.14 shows the flooding point caused
by a flow rate of the two phases that is too high. There are two more flood-
ing points that depend on the difference of the exit pressures of the two
phases. How can they be explained? Section 9.3.4 mentions that the principal
interface in the rotor can optionally be shifted by varying the exit pressure of
the dispersed phase (i.e., the pressure difference of the two phases at the
exit). If the two exit pressures are the same, the rotor must be filled with
heavy phase. The heavy continuous phase then leaves at the exit of the
dispersed light phase. On the other hand, at a specific excess pressure at the
exit of the light phase, the rotor will be filled with light phase. Then, the light
phase will leave at the exit of the heavy one. In both cases, the throughputs
of both phases flooding points are reached. It is possible to predetermine
them [15].
9.5 MIXING OF PHASES

The two liquid phases go different ways during their flow through the ex-
tractor and, therefore, have different residence times. Various reasons for this
can be involved; for example, in a real droplet swarm, the various drop
diameters lead to different rising velocities and motions, or the wake behind
each drop drags the continuous phase in a direction opposite its original
motion. Because of such phenomena, regions of different compositions are
mixed, which reduces the possible mass transfer efficiency. These mixing
processes can be observed only by a close analysis of the residence times of
the phases in the extractor [16]. According to Fig. 9.15, this can be done by
feeding a suitable sensor, for example, a colored tracer into the entering
heavy feed. Then, at two cross sections, a and o, the concentration of the
color intensity, ca or co, respectively, is measured downstream as a function
of time t. It can be noticed that the curve for the sensor widens downstream,
as a result of the various mixing phenomena. The light phase can be traced in
a similar way by measuring its residence time. It is essential that the tracer has
the same flow behavior as the respective phase and that its amount is not
altered during its residence time (e.g., no adsorption at solid surfaces, no
extraction by the other phase and no reaction). Therefore, the functions ca(t)
or co(t) are valid for both the tracer and the main phase.
Fig. 9.14 Flooding point diagram of centrifugal extractors (see Fig. 9.9). Curve 1 marks
the lower limit, when the rotor is filled with heavy phase; curve 2 marks an upper limit,
when the rotor is filled with light phase; curve 3 marks the limit of the total throughput of
both phases.
394 Blass
Fig. 9.15 Principle of residence time measurement of the continuous heavy phase in a
countercurrent extractor. After feeding a short test signal, for example, a colored tracer, the
color intensity ca or co, respectively, is measured downstream at two cross sections a and
o as a function of time. The test signal curve widens downstream, owing to the various
mixing phenomena.
Consider now a mathematical model for the mixing process that allows
the calculation of the distribution function co out of the function ca measured
at the cross section a, where co(t) is called the response curve. This model
must provide one or more parameters that define the degree of mixing in
addition to operation and construction parameters. If the calculated and
measured response curves are almost identical for the defined parameter
values, not only is the model proved, but it will also have yielded a set of
connected operation, construction, and mixing parameters. From a sufficient
number of these measurements, one can work out a functional dependence
for all of these parameters to predetermine the mixing processes in the var-
ious extractors.
9.5.1 Gravitational Extractors

The dispersion model is one of the frequently used models. It describes the
‘‘dispersion of the residence time’’ of the phases according to Fig. 9.16, for
example, in one-dimensional flows by superimposing the plug profile of the
basic flow with a stochastic dispersion process in axial direction, which is
constructed by analogy to Fick’s first law of molecular diffusion:
dc
N_ D ¼ Dax S (9.27)
dh
Equation (9.27) defines the so-called axial dispersion coefficient Dax as a
model parameter of mixing. N_ D is the dispersion flow rate, c the concen-
tration of the tracer mentioned earlier, and S the cross-sectional area of the
column. The complete mole flow rate of the tracer consists of an axial
convection flow and the axial dispersion flow. The balance of the tracer
amount at a cross section of the extractor leads to second-order partial
differential equations for both phase flows at steady state. For example, for
continuous liquids:
@c @2c _ c @c
¼ Dac;c 2 (9.28)
@t @h 1 "d @h
The solution of this partial differential equation is known for special
boundary conditions and measuring conditions at the cross sections a and o
Fig. 9.16 Basic assumptions of the one-dimensional dispersion model. The dispersion
of the residence time of the phases is modeled by superimposing the plug profile of the
basic flow with a stochastic dispersion process in axial direction.
396 Blass
of the measuring zone [17]. The experimenter can then calculate the response
curve in the cross section o, co(t), and can change the dispersion coefficient
Dax,c as the fitting parameter so that the calculated response curve co(t)
agrees best with the measured one. Figure 9.17 for liquid pulsed sieve tray
columns shows that the dispersion model gives a fine reflection of the process
of mixing of the continuous phase. The measured and calculated response
curves for a special value of Dax,c agree very well. However, there are large
deviations for the dispersed phase, in spite of the best fit. This leads to the
conclusion that the dispersion model provides an inadequate description of
this process. Relevant literature offers correlations for dispersion coefficients
Fig. 9.17 Residence times of (a) continuous; and (b) drop phase: in a liquid pulsed sieve
tray extractor according to measurements (squares and crosses) and according to calcu-
lations by the dispersion model (drawn lines). Measured and calculated response curves
agree well for the continuous phase, but not for the dispersed phase. (From Ref. 14.)
in various extractors [1, Chapter 17]. For the continuous phase, they are
usually between 1 and 5 cm2 s1 (in some cases values as high as 20 cm2 s1
are observed); for the dispersed phase they increase by a factor of 2–3.
The design engineer can use the dispersion coefficients determined in
this way for the calculation of the real course of concentrations, c, of any
component in the dispersed (d ) and continuous (c) phases along the coun-
tercurrent column. If the mass transfer between the two phases, the actual
task of an extractor, is included in the balance, the balance equations for an
element of height dh of the extractor for stationary conditions is:
dcc d 2 cc
_ c þ ð1 "d ÞDax;c 2 ¼ kod aðcd cd Þ (9.29)
dh dh
dcd d 2 cd
_ d þ "d Dax;d 2 ¼ kod aðcd cd Þ (9.30)
dh dh
The term on the right side of the equations quantitatively defines mass
transfer from the dispersed to the continuous liquid, which is explained
more fully in section 9.6. The coupled differential equation system can be
analytically or numerically integrated for the appropriate practical bound-
ary conditions.
Figure 9.18 shows schematically the influence of the flow dispersion on
the concentration profile and the possible exit concentrations. Notice that
dispersion changes the possible exit concentrations of the two phases
negatively.
9.5.2 Centrifugal Extractors

For a specific operational condition, Fig. 9.19 shows a curve for the residence
times of the dispersed heavy phase in a centrifugal extractor with three
perforated sheets. These are plotted schematically one after the other from
bottom to top on the right-hand side of the figure [9]. These sheets are fitted
with downcomers and show the formation of jets of the dispersed phase in
the centrifugal field. Extremely short residence times can be seen during
jet formation and motion of the drops in the continuous phase. On the
other hand, there is relatively long residence of the drops within the sta-
tionary layer of this phase in front of the following sheet, i.e., both before and
after coalescence in the stationary layer. On the whole, the liquid molecules
do not remain longer than about 40 s in a three-stage centrifuge, and usually
not longer than about 2 min in a multistage industrial centrifugal extractor.
This short residence time is one main attribute of centrifugal extractors.
Measurements of residence time show another special feature in that all
layers of the two phases between the perforated plates are fully mixed, which
means they behave like ideally mixed agitator tanks. This can be explained by
398 Blass
Fig. 9.18 The influence of flow dispersion on the concentration profiles of the two
phases (mass transfer from the continuous phase to the dispersed phase).
the various peripheral flow velocities at different radii of the rotor and by the
influence of the drops. If sufficient ideally mixed cells are connected in series,
plug flow residence time behavior results. As mentioned at the beginning of
section 9.5, there is no negative influence on the mass transfer by mixing and
the motion of the droplet swarm approaches that of plug flow, since neither
the drop sizes nor their velocity of motion are very different.
The dispersion model introduced in section 9.5.1 cannot be used for
the demonstration of the mixing process in a multistage centrifugal ex-
tractor, because it can describe only the behavior of the extractor overall.
There is no means of distinguishing between the various contributions to
mixing from the single steps. Models, however, that consist of terms for plug
flow patterns and ideal stirring tanks describe the real mixing behavior
exactly [10].
9.6 MASS TRANSFER UNDER PHYSICAL

CONDITIONS IN BINARY SYSTEMS
9.6.1 Countercurrent Flow in a Gravitational Field
We have now learned that the calculation of the flooding point provides
the necessary diameter of the extractor, and that the calculation of the
concentration profiles provides the possible output concentration for a given

length of the extractor, or, conversely, the length of the extractor to obtain
a desired output concentration of the phases. Equations (9.29) and (9.30)
contain the mass transfer terms on the right-hand side of the equations.
The individual, or phase, mass transfer coefficients (i.e., kd for the
dispersed phase and kc for the continuous phase) are defined as:
n_ c ¼ kc cc (9.31a)
n_ d ¼ kd cd (9.31b)
where n_ is the mole flux of the transitional component and Dc the con-
centration differences between the bulk phase and interface [18]. At steady
state, the mole flux of the component from one phase to the interface must
just be able to flow into the other phase. For this transitional state, an overall
mass transfer coefficient ko can be defined. To define ko it must be assumed
that the transitional resistance of the two phases is concentrated in one phase,
Fig. 9.19 The residence times of heavy droplets in a centrifugal extractor (see Fig. 9.9)
with three concentric perforated sheets. Extremely short residence times are a main at-
tribute of centrifugal extractors.
400 Blass
so that the bulk concentration of the resistanceless phase reaches as far as the
interface, as in Fig. 9.20. An overall mass transfer coefficient can then be
defined for the dispersed phase as well as the continuous phase:
n_ d ¼ kod (cd Deq cc ) (9.32a)

cd
n_ c ¼ kod cc (9.32b)
Deq
where Deq ¼ (cd =cc )equilibrium is the distribution ratio. Since both equations
must provide the same transitional rate, it follows that, if interfacial resis-
tances are neglected, the phase and overall mass transfer coefficients are
connected by:
1
kod ¼ (9.33a)
(1=kd ) þ (1=Deq kc )
1
koc ¼ (9.33b)
(1=kc ) þ (1=Deq kd )
It is generally agreed that mass transfer coefficients are only correlated for
negligibly small convectional motion of the transitional component, which is
vertical to the interface. However, when the mass transfer is mutual and
equimolar, no such convections normal to the interface result; otherwise the
transfer coefficient and the driving force must be corrected with the aid of
theories of mass transfer [18]. The transitional rates and, accordingly, con-
vectional flow rates normal to the interface are only low for the extraction
process, so that the uncorrected Eq. (9.31) may be used.
The intensity of mass transfer shown by the mass transfer coefficient
depends on the flow processes inside the drop or in its surroundings and,
thereby, on the various life stages of the drops. During the drop formation,
new interfaces and high concentration gradients are produced near the
interface. The contact times between liquid elements of the drop and the
surroundings that are near the surface are then extremely short. According
to Fick’s second law for unsteady diffusion, it follows that for the phase
mass transfer coefficient [19]:
rffiffiffiffiffiffiffi
4D
k¼ (9.34)

t
where D is the diffusion coefficient of the transitional component in the

respective phase and t is the contact time. The shorter the contact time,
the higher the mass transfer coefficient. Thus, a repeated coalescence and
disintegration of drops is an advantage. In all tray columns, pulsed or
Fig. 9.20 Definition of individual and overall mass transfer coefficients.
nonpulsed sieve tray columns, and stirred columns, the disintegration zone
is relatively small compared with the whole distance that the drops cover in
the extractor. Thus, the influence of the mass transfer kinetics on the integral
mass transfer during the drop formation can be neglected during the lifetime
of the drops.
Next, we are interested in mass transfer during drop motion, which
according to Fig. 9.11 is strongly dependent on drop size. When drops are
very small, and therefore rigid, diffusion in the interior of the drops deter-
mines the mass transfer velocity. When the internal drops circulation occurs,
however, mass transfer is considerably enhanced by the constant transport
of liquid elements to the drop surface. The concentration gradient driving
force is constantly kept at a high level. For oscillating drops, a further rise in
the mass transfer velocity can be expected, since the interior of the drop is
turbulently mixed.
402 Blass
The literature offers numerous calculation models for mass transfer in

single liquid particles. However, they provide only a rough approximation
to reality in industrial columns, since the processes in droplet swarms are
much more complicated, especially when pulsing and rotating motion are
superimposed. For estimation, the following relationships are sufficient:
1. For the mass transfer on the inside of the drop: the equation by Handlos
and Baron [20] for oscillating drops:
r
kd ¼ C (9.35)
1 þ ð d = c Þ
with c = 0.002 [2], and
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
gdP 4
r ¼ þ (9.36)
4 c c dP
The mean drop diameter should be fixed in the range of the transition
region of circulating to oscillating drop, since this gives the most
favorable compromise between the possible transfer area per unit
volume of the column and mass transfer intensity.
2. For the mass transfer coefficient on the outside of the drop kc, Eq. (9.34),
according to the penetration theory by Highbie [19], obtains the contact
time t as the quotient between the rising distance between two stages and
the rising velocity.
After the motion of the drops, their separation by coalescence follows.
During this process, the mass transfer is negligibly small compared with that
where the drops are in motion, if the continuous liquid is not fed directly
into the coalescence region. However, such a technique should be avoided so
as not to disturb the settling process.
For simplicity, this section discusses only the mass transfer of one
component in a liquid–liquid system with negligible miscibility of both
liquids and with one transitional component. On the other hand, calcula-
tions must consider mass transfer rates of several components and more or
less strong variation in the mass flows along the column, where both com-
plicate the equation considerably [21–23]. Chemical reactions may cause
further complications. Their kinetics can enhance the mass transfer coeffi-
cients and, therefore, the reaction equations have to be part of the mathe-
matical model of the extractor [24,25].
9.6.2 Countercurrent Flow in a Centrifugal Field

The general definition of mass transfer coefficients according to Eqs. (9.31)–
(9.33) is also valid for the mass transfer coefficient in a centrifugal field. The
very short residence times in the various steps of flow through a stage indicate
the use of the mass transfer coefficients according to Eq. (9.34) given by
Highbie [19]. Otillinger and Blass [9] have proved that Eq. (9.34) can actually
be used for the two mass transfer coefficients kc and kd in drop formation and
motion, as well as in the coalescence phase, and for the mass transfer between
liquid layers in a stage [1, Chapter 14]. Figure 9.21 shows this for a special
situation, with four actual stages formed of concentric sieve trays. The
number of theoretical stages is used as a measure for the performance of mass
transfer plotted radially in the centrifugal extractor. The increase of the
theoretical number of stages during drop formation and motion within the
radial path of each stage can be clearly seen. The number of stages,
depending on the coalescence and the contact of the liquid layers, can be
observed at specific radii. The straight lines have been calculated using the
mass transfer coefficient from to Eq. (8.34) using the residence times of the
respective flow steps as contact times. The final points of the calculation
Fig. 9.21 The mass transfer performance of a centrifugal extractor (see Fig. 9.9) with
four concentric sheets (stages). This is dependent on the radial path with the field force as
parameter: ns, number of stages; hs, height of a stage; rm, radial drop path in a stage; ri,
inactive radial distance (thickness of perforated sheet and of coalesced layer in a stage).
(From Ref. 9.)
404 Blass
agree well with the measured results for the whole extractor (i.e., that for the
whole radius of the rotor).
9.7 INFLUENCE OF INTERFACIAL ENERGIES

ON FLOW AND MASS TRANSFER
Flow and mass transfer are influenced in many ways by the energetic con-
ditions at the interface [26]. These conditions are described by the interfacial
tension defined by the change of the free enthalpy with interfacial area. In
general, interfacial tensions between liquids, and especially between liquid
mixtures, cannot be calculated, but must be measured.
The presence of a transitional component reduces the interfacial ten-
sion for the majority of extraction liquid systems, so that ds/dx<0. Ten-
sides, surface-active agents, drastically reduce the interfacial tension, even if
their concentration is very small.
Interfacial tension is the parameter in equations influencing the drop
size, as discussed in preceding sections. The smaller the value of s, the
smaller are the resulting drops, if all the other conditions are the same, and
the larger is the transfer area per unit volume. On the other hand, small
drops may show little or no internal circulation, which implies equivalent
consequences for the mass transfer coefficient and a lower rising velocity
and, accordingly, a lower flow rate at the flooding point.
These influences of s, however, can be covered by the equations
introduced previously. In addition to these, there are further energetic
interfacial influences. During the process of mass transfer, local non-
homogeneity of composition and perhaps temperature must be expected
along the interface. As an example, Fig. 9.22 shows the layer of continuous
liquid between two drops. If the mass transfer is directed from the drops into
the liquid layer, the transitional component is relatively more enriched in the
thin region of the layer than in the thick one. This must correspond to a
lower interfacial tension at ds/dx<0 in the thin region of the layer. Since an
interface seeks to have the lowest interfacial energy, a flow of liquid elements
near the interface out of the region of lower tension toward the ranges of
higher tension results. Thus, the thin layer may be thinned until it breaks
and the drops can coalesce. If such drop–drop coalescence is to be pre-
vented, the receiving phase for liquid systems with ds/dx<0 must be dis-
persed; otherwise, the raffinate phase must be dispersed.
Flow near the interface that is influenced by gradients of interface
tension is called Marangoni convection. It may have further modes [27].
Thus, a low Marangoni convection in an interface, which results from small
concentration differences, may be increased to a strong flow in the shape of
rolling cells by mass transfer. These rolling cells transport liquid out of the
Fig. 9.22 Marangoni convection between two drops. If the mass transfer is directed out
of the drops, the transfer component is relatively more enriched in the thin region of the
layer. This corresponds to a lower interfacial tension for most of the extraction liquid
systems (ds/dx<0). Thus, a flow of liquid near the interface out of the region of lower
tension toward the ranges of higher tension results is called Marangoni convection. The
coalesence of the two droplets is promoted.
deeper layers of the phase to the interface and thereby increase mass
transfer. This is desired. In the past, criteria have been developed that allow
the prediction of such Marangoni instabilities [26]; however, the number of
correct and incorrect predictions is equal. Consequently, experiments should
be carried out for a safe prediction of the interfacial instabilities of a par-
ticular liquid system under various conditions. The ‘‘kicking drop’’ method
proves to be very useful. Drops are produced on a capillary in the con-
tinuous phase and, if Marangoni convection appears in the chosen direction
of the mass transfer, a strong vibration of the drop can be observed.
However, strong Marangoni convection can cause interfacial turbulence and
the formation of very small drops. Emulsions like these cannot be separated
in normal settlers and should be avoided a priori.
Finally, it must be noted that tensides that are adsorbed at the inter-
face cause a stiffening of the interface. They hinder or even stop the inner
circulation and oscillation of drops, and reduce the mass transfer intensity.
Moreover, they form a barrier against the mass transfer, so that a further
resistance term should be considered in the overall mass transfer process [28]
in Eq. (9.33). Since the nature and concentration of tensides in industrial
processes cannot be predicted, such phenomena cannot be taken into con-
sideration during equipment calculations.
The wetting of solid surfaces by liquids is also due to interface ener-
gies. As a measure of the wettability, the wetting angle is used, which occurs
at the contact line of the three phases between the solid and the wetting
liquid. It can be measured with suitable optical systems by placing a drop on
a plain, horizontal surface of the material used in the industrial apparatus.
For total wetting, the wetting angle is 08, rising to 1808 if a wetting is not
406 Blass
possible. All dispersion devices in industrial extractors, sieve trays, tower

packings, or mixers should not be easily wetted by the dispersed phase, since
this would result in large drops shaped like doughnuts.
These phenomena of surface energy all are time dependent. The shorter
the contact time of the phases, the more difficult it is for them to develop.
Therefore, flow and mass transfer in centrifugal extractors are hardly affec-
ted.
9.8 SEPARATION OF PHASES

Once mass transfer is completed, the drop phase must be separated from the
continuous liquid. The basic event of the separation process is the coales-
cence of droplets producing a homophase. This can take place in a part of a
countercurrent column especially provided for this purpose (see Figs. 9.1
and 9.5) or in a special settler (Fig. 9.23). If we wish to predetermine the
separation process, the physical course of the droplet coalescence must be
known. Figure 9.24 schematically illustrates the coalescence of a single drop
Fig. 9.23 Settlers for separation of liquid–liquid dispersions. Vh volume flow of heavy
phase; Vl, volume flow of light phase.
Fig. 9.24 Basic phenomena of drop coalescence at a horizontal interface. The drop has
to reach the interface. A thin layer of the continuous phase remains between the drop
and the interface. The thin layer has to drain until it breaks up. Then the drop can flow
into its homophase. Mostly, the drainage process is the time-determining step of this
process.
of lighter phase at a horizontal interface. The drop rises to the interface,

where it is retarded, and deforms both itself and the interface. A thin layer
of the continuous phase remains between the drop and the interface. This
thin layer has to drain, forced by the contact pressure of the drop, until it
reaches its critical thickness and breaks up. Then the drop can flow into its
homophase. In most cases, this drainage process is the time-determining step
of coalescence. If the drainage time can be calculated, the minimum value of
the residence time t of the drop phase in the settler is calculable from the
volumetric flow rate of the dispersed phase, V_ d :
V
¼ (9.37)
V_ d
This is the reference value for the settler volume V. The literature contains
many proposals for modeling the coalescence at horizontal interfaces and the
settling time [see, e.g., Ref. 1, Chapter 13]. However, predetermination of the
coalescence time proves to be very complicated, as different effects hinder the
drainage of the thin layer. Especially, the influence of surfactants, which
occur in every industrial extractor, cannot yet be evaluated using mathe-
matical models. Surfactants produce surface tension gradients during the
drainage process and, therefore, Marangoni convections back into the layer,
which may lead to a drastic increase of the coalescence time [29,30].
For a practical settler design, the height H of the dense dispersion,
which is below the principal interface and has to find space in the settler, is
evaluated by a simple estimation formula:
408 Blass
_
Vd
H¼C (9.38)
S
The factor C and the exponent y must be determined by experiments with

pilot settlers. These terms combine all the unknown parameters. The cross-
sectional area, S, of the settler should be chosen in such a way that the
superficial velocity, _ d , of the dispersed phase does not exceed 5–9 mm s1.
Often aids for coalescing (e.g., packages of inclined plates or filter cartridges
made of metal or plastic) (Fig. 9.23) are used. There are several ways of
predetermining coalescence [31–33]. Furthermore, in special cases electrical
fields and centrifuges are used to support coalescence [34].
9.9 DIMENSIONING OF LARGE-SCALE

EXTRACTORS
9.9.1 The Problem of Scale-Up
The previous sections have pointed out that mathematical models of the
processes must be proved by experiments, or adapted to experimental results
with the aid of pilot extractors. For economic reasons, pilot extractors are
usually much smaller than large-scale industrial apparatus. Pulsed pilot
columns, for example, have a diameter between 30 and 250 mm, whereas
industrial-size columns are up to 2500 mm and more in size. Thus, the
question arises of whether or not the calculations or pilot experiments may be
used for the dimensions of large-scale apparatus. This is a general problem
for engineers.
Only a careful analysis of the differences that appear in construction,
in the processes of flow and mass transfer and measuring and observing
the behavior and the performance of large-scale extractors, can answer
these questions. This provides formula of experience for the factors of
scale-up that have been listed [4] for the countercurrent extractors descri-
bed here.
Thus, the height of the column may need to be scaled up, but the
column diameter may emerge from calculation of the flooding load. Large-
scale circulations that are produced in large columns seem to be the reason
for the decrease of the mass transfer performance as column diameter
increases. This causes the axial mixing to be higher in large columns than
in small ones. Such circulations are not prevented by inserts in the column.
On the other hand, centrifugal extractors have flow and mass transfer
behavior that agrees sufficiently between laboratory and large-scale
equipment.
9.9.2 Problems of Industrial Liquid Systems

Clear test mixtures are used for laboratory experiments to be able to use
fully defined, reproducible liquid systems. Such mixtures, according to
agreement of the European Federation of Chemical Engineering [35], are
listed in Table 9.2. They differ mainly in the interface tension. Compared
with these, the variety of technical systems is very high [36]. To guarantee
reliable dimensions for equipment, engineers must supplement their calcu-
lations by experiments that include the following:
1. The liquid system:

Determination of the binodal solubility curve in a triangular diagram,
the equilibria and their dependences on pH, temperature, salt
content of the raffinate, etc.
Determination of the ratio of solvent to raffinate flow rate
Measurement of the separation time (e.g., in a shaking vessel) to eval-
uate the coalescence behavior
Measurement of the physical properties of feed and product phases (e.g.,
mass densities, viscosities, surface tensions, and wettability) and
their dependence on temperature on typical equipment materials
Qualitative examination of the mass transfer intensity of a single droplet
(kicking drop)
2. Pilot plants:
Selection of the dispersed phase
Preselection of a suitable extractor type on the basis of available
information and testing at pilot scale
Establishment of suitable operating conditions as, for example, the
pulse intensity of pulsed extractors or the speed of rotation of
centrifugal extractors
Determination of the flooding point
Observation of emulsion and dirty layers at the principal interface and
their behavior
Optimization of the operating conditions of the chosen apparatus
Associated corrosion experiments in the pilot plant
3. Evaluation and storage of the results:
Comparison of experimental and theoretical results
Considerations on scale-up
Final decision on extractor type
Generalizing and storing of the experimental results for future appli-
cation
410 Blass
Table 9.2 Important Properties of the Three EFCE Test Systems

1 2 3
Phase I Water
Phase II Toluene n-Butanol Butyl acetate
Formula C7H8 C4H10O C6H12O2
Boiling point (1 bar) 110.48C 117.58C 126.098C
Transferring component Acetone Succinic acid Acetone
Formula C3H6O C4H6O4 C3H6O
wt s 0 35.4 0 1.75 0 14.1

——— ———— 3.13 27.0 3.87 1.0 3.81 11.7
mass-% mN m1 7.67 19.3 6.58 0.7 7.86 9.6
wt rI 0 997.8 0 985.6 0 997.0
——— ———— 3.13 993.7 3.56 995.6 3.03 993.3
mass-% kg m3 7.67 987.8 6.25 1003.1 8.11 986.4
wt rII 0 866.5 0 846.0 0 882.1
——— ———— 3.13 864.5 3.56 866.5 3.03 879.4
mass-% kg m3 7.67 862.6 6.25 881.4 8.11 873.6
wt ZI 0 1.006 0 1.426 0 1.0237
——— ———— 3.13 1.078 3.56 1.536 3.03 1.11
mass-% cP 7.67 1.2 5.29 1.61 9.11 1.28
wt ZII 0 0.586 0 3.364 0.7345
——— ———— 3.13 0.575 3.56 3.749 03.03 0.72
mass-% cP 7.67 0.560 5.29 3.925 8.11 0.6827
0%
wt
8% 0%
wt
- 0%
wt
8%
Deq for (Deq = yt/xt)
6.5%
—————— —————— ——————
0.61–0.83 1.3–1.12 0.9–0.98
wt DI 0.59 1.14 0.69 0.57 0.03 1.093

——— ———— 3.45 1.07 3.66 0.52 31.86 0.598
mass-% 103 mm2 s1
5.96 1.01 5.21 0.47 74.15 1.68
wt DII 2.7 0.43 0.24 0.25 2.2
——— ———— 3.260.72 2.66 3.71 0.23 41.92 2.196
mass-% 10 mm2 s1
3
4.37 2.51 5.56 0.21 79.76 2.506
Source: Ref. 32.
9.10 NOMENCLATURE
Abbreviation Unit Physical quantity

A m2 Area, interfacial area
a m2 m3 Interfacial area per unit volume
a m Pulsation amplitude
c kmol m3 Concentration
D m2 s1 Binary diffusion coefficient
Dax m2 s1 Axial dispersion coefficient
Deq Distribution ratio
d m Diameter
d32 m Sauter mean diameter
E Ws Energy
f 1 s1 Frequency
g m s2 Acceleration from gravity
H, h m Height
k m s1 Mass transfer coefficient
M_ kg s1 Mass flow rate
N_ kmol s1 Mole flow rate
n_ kmol m2 s1 Mole flux
nR 1 s1 Rotor speed
nP Number of drops
nS Number of stages
P kW Power consumption
S m2 Cross-sectional area
t s Time, residence time
V m3 Volume
V_ m3 s1 Volumetric flow rate
9.10.1 Greek Letters

D Difference
e Holdup
Z kg m1 s1 Dynamic viscosity
n m2 s1 Kinematic viscosity
p 3.1415. . .
r kg M3 Mass density
S Sum
s N m1 Surface or interfacial tension
t s Mean residence time
c kW kg1 Specific energy dissipation
412 Blass
Nomenclature (continued )
Abbreviation Unit Physical quantity

o 1 s1 Angular velocity
_ m3 m2 s1 Superficial velocity
n m s1 Velocity
w kmol kmol1 Mole fraction
9.10.2 Dimensionless Numbers

All dimensionless numbers in this chapter are frequently used in chemical
engineering, especially in heat and mass transfer and in multiphase flow [35]:
dp3 gc
Archimedes number Ar ¼
c2
Re2
Ar ¼ here encodes the drop
Fr
diameter
2
Froude number Fr ¼
dg
Fr is the relation between inertia and
gravity forces (not used in Chap. 9)
2 3
Liquid number KL ¼ c 4
g c
Re4 Fr
KL ¼ combines important physical
We3
properties of the phases
P
Newton number Ne ¼
c n3R dR5
Ne relates the power consumption with
important parameters of mixing
P d P
Reynolds number of particle Re ¼
d
Re is the relation of inertia and viscosity
forces
c n2R dR3
Weber number of mixer We ¼

We is the relation of inertia and interfacial
forces
9.10.3 Subscripts
c Continuous phase
d Dispersed phase
D Flow dispersion
i Phase i
i Class of drop sizes
j Jet formation
I Interface
K Column
L Liquid phase
max Maximum
min Minimum
N Nozzle hole
o Overall
P Particle
R Stirrer
r Relative
s Swarm of droplets
t Transfer component
z Cell
a Inlet
o Outlet
* Equilibrium
REFERENCES
1. Godfrey, J. C.; Slater, M. J. Liquid-Liquid Extraction Equipment; John Wiley
and Sons: Chichester, 1994; 630.
2. Philhofer, T; Mewes, D. Siebboden-Extraktionskolonnen; Verlag Chemie:
Weinheim, 1979; 10.
3. Blass, E.; Goldmann, G.; Hirschmann, K.; Mihailowitsch, P.; Pietzseh, P. Ger.
Chem. Eng. 1986, 9, 222.
4. Lo., T. C.; Baird, M. H. I.; Hanson, C. Handbook of Solvent Extraction; John
Wiley and Sons: New York, 1983.
5. Hirschmann, K.; Blass, E. Ger. Chem. Eng., 1984, 7, 280.
6. Wagner, G.; Blass, E. Chem. Eng. Technol., 1999, 21, 475.
7. Kolmogoroff, A. N. Sammelband zur Theorie der statistischen Turbulenz;
Akademieverlag: Berlin, 1958.
8. Barson, N.; Beyer, G. H. Chem. Eng. Progr., 1953, 49, 243.
9. Otillinger, E.; Blass, E. Chem. Eng. Technol., 1988, 11, 312.
10. Otillinger, F. Fluiddynamik und Stoffiibergang in Zentrifugalextraktoren
[doctoral thesis]; TU Munchen, 1988.
414 Blass
11. Reinhardt, H. Chem. Ind., 1972, 363.

12. Rydberg, J.; Persson, H.; Aronsson, P. O.; Selme, A.; Skarnemark, G.
Hydrometallurgy, 1980, 5, 273.
13. Brunner, K. H.; Hemfort, H. Adv. Biotechnol. Processes, 1988, 8, 1.
14. Niebuhr, D. Untersuchungen zur Fluiddynamik in pulsierten Siebboden-Ex-
traktionskolonnen [doctoral thesis], TU Clausthal, 1982.
15. Schilp, R.; Blass, E. Chem. Eng. Commun., 1984, 28, 85.
16. Wen, C. Y.; Fan, L. T. Models for Flow Systems and Chemical Reactors,
Chemical Processes and Engineering; Vol. 3, Marcel Dekker: New York, 1975.
17. Krizan, P. Analyse des dynamischen Verhaltens dispenser Mehrphasensysteme
bei ausgeprsgter Grof3raumstrbmung [doctoral thesis]; TU Munchen, 1987.
18. Bird, R. B.; Stewart, W. E.; Lightfoot, E. N. Transport Phenomena; John
Wiley and Sons: New York, 1960; part III.
19. Highbie, R. Trans. Amer. Inst. Chem. Eng., 1935, 31, 365.
20. Handlos, A. E.; Baron, T. A.I.Ch.E.J., 1957, 3, 127.
21. Steiner, L. Rechnerische Erfassung der Arbeitsweisen von Flüssig-flüssig-
Extraktionskolonnen, VDI-Verlag, Dusseldorf, Reihe 3 Nr. 154, 1988.
22. Pertler, M., Die Mehrkomponentendiffusion in nicht vollstaendig mischbaren
Flussigkeiten [doctoral thesis]; TU Munchen, 1996.
23. Haeberl, M.; Blass, E. Chem. Eng. Res. & Des., 1998, 77, 647.
24. Moertens, M.; Bart, H.-J. J. Chem. Eng. Data, 2000, 23, 353.
25. Czapla, C.; Bart, H.-J. Chem. Eng. Technol. 2000, 23, 1058.
26. Sawistowski, H. Recent Advances in Liquid-Liquid Extraction, Hanson, C.,
Ed., Pergamon Press: Oxford, 1971.
27. Tourneau, M.; Wolf, S.; Stichlmair, J. Phase Boundary Correction in Liquid-
Liquid Systems; DECHEMA monograph 136, 2000.
28. Ollenik, R.; Nitsch, W. Ber. Bunsenges Phys. Chem., 1981, 85, 900.
29. Burrill, K. A.; Woods, D. R. J. Colloid Interface Sci., 1969, 30, 511.
30. Burrill, K. A.; Woods, D. R. J. Colloid Interface Sci., 1973, 42, 15.
31. Rommel, W.; Blass, E.; Meon, W. Chem. Eng. Sci., 1992, 47, 555.
32. Meon, W.; Rommel, W.; Blass, E. Chem. Eng. Sci., 1993, 48, 157.
33. Magiera, R.; Blass, E. Filtr. Sep. 1997, 34, 369.
34. Kriechbaumer, A. Entwicklung eines elektrischen Emulsionsspalters und Un-
tersuchung zu den Koaleszenzvorgiingen und Stabilitdtsparametern im clek-
trischen Feld [doctoral thesis]; Graz, 1984.
35. Misek, T.; Berger, R.; Schroeter, J. Standard Test Systems for Liquid Extrac-
tion; Institution of Chemical Engineers: Rugby, 1985.
36. Hampe, M. J. Ger. Chem. Eng., 1986, 9, 251.
10
Extraction of Organic Compounds
RONALD WENNERSTEN Royal Institute of Technology,
Stockholm, Sweden
10.1 INTRODUCTION
The separation of organic mixtures into groups of components of similar
chemical type was one of the earliest applications of solvent extraction. In
this chapter the term solvent is used to define the extractant phase that may
contain either an extractant in a diluent or an organic compound that can
itself act as an extractant. Using this technique, a solvent that preferentially
dissolves aromatic compounds can be used to remove aromatics from ker-
osene to produce a better quality fuel. In the same way, solvent extraction
can be used to produce high-purity aromatic extracts from catalytic refor-
mates, aromatics that are essentially raw materials in the production of
products such as polystyrene, nylon, and Terylene. These features have
made solvent extraction a standard technique in the oil-refining and pet-
rochemical industries. The extraction of organic compounds, however, is
not confined to these industries. Other examples in this chapter include the
production of pharmaceuticals and environmental processes.
In these applications, solvent extraction constitutes an extraction stage
during which an organic phase is in contact with an aqueous phase or
another immiscible organic phase. The extract is then recovered by dis-
tillation or washing with an aqueous or organic phase.
Using new solvents and having a better understanding of the chemistry
involved in more specific interactions, solvent extraction has become a very
interesting separation technique for high-value organic chemicals (e.g.,
amino acids). Furthermore, liquid extraction using two phases with very high
water concentration has found applications for the separations of proteins.
415
416 Wennersten
Supercritical extraction with carbon dioxide as a solvent is also a most

promising technique for a wide area of applications.
The development of better thermodynamic models has made it pos-
sible to simulate industrial processes from only a minimum of experimental
data. This facilitates process integration and optimization, with a minimum
of pilot plant tests.
10.2 DESIGN CRITERIA

10.2.1 General: Choosing an Organic Solvent
Although the solvent extraction process might seem simple from an engi-
neering point of view and has relatively low equipment costs, the design of an
economically attractive process requires a complex strategy that takes
account of many dependent variables. There are numerous instances in which
an improper comparison between solvent extraction and other separation
techniques has been made because of a quite unsystematic investigation of
the solvent selection step. The substance to be extracted, the solute, and the
extracting solvent, made up of extractant and a diluent, interact in such a
complicated way that the extraction step cannot be isolated from the
recovery steps in the selection of the solvent. There are at least three inte-
grated stages to be considered:
1. Liquid–liquid contact
2. Solvent recovery
3. Raffinate cleanup
In the extraction of organic components, as will be shown, these three
steps can all be equally important for costs of operating and capital costs.
Design calculations cannot be made based on distribution coefficients alone.
Thus, several cases have shown that the optimum economic approach was
neither obvious nor anticipated in preliminary studies.
The initial requirement in the development of a solvent extraction
process for the recovery or separation of organic substances is knowledge of
the feed composition and flow rate and of the restrictions on the product
quality and raffinate composition. Starting with these conditions fixed, a
preliminary solvent selection can be made from solvents representing different
process configurations such as low-boiling solvents, high-boiling solvents, or
solvents from which the solute is stripped back into an aqueous phase.
There is no ‘‘universal solvent,’’ and solvent selection must be made
individually for each separation problem. The specific choice must be based
on a general knowledge of the different interactions in both phases. Among
the desirable features for the extracting solvent are the following:
Extraction of Organic Compounds 417
1. A high capacity for the species being extracted

2. Selectivity, dissolving to a large extent one of the key components
without dissolving other components to any large extent
3. Low mutual solubility with water
4. Easy regeneration
5. Suitable physical properties, such as density, viscosity, and surface
tension
6. Relatively inexpensive, nontoxic, and noncorrosive
Obviously, no solvent satisfies all these requirements, and the selection
of a desirable solvent involves a compromise between these and other fac-
tors. When a preliminary selection has been made, a secondary screening
can be performed based on simplified calculations of minimum energy
requirement, since the capital costs for similar process configurations will
not vary too much.
In the final selection, rigorous calculations must be carried out together
with laboratory tests on phase separation properties, chemical stability,
verification of equilibrium data used in the calculations, and so on. At this
stage, the data obtained have to be examined critically to ensure that there is
sufficient potential for the project to be carried to the next phase. If sufficient
information has been collected for a conceptual flow sheet, a decision can
then be made on whether or not to run a pilot plant. From pilot plant data,
the process can be scaled up according to principles given in the literature.
This outline of the principles of designing a solvent extraction process is
general, but some of the individual steps require experience and effective aids,
such as computer programs and experimental equipment. Some of the dif-
ficult steps are discussed in greater detail in the following examples.
10.2.2 Defining the Separation Problem

The industrial stream to be treated, the feed, will not be an analytical
grade solute dissolved in water, but often contains several known and
unknown substances, both organic and inorganic. To be able to make an
initial selection of possible solvents, it is necessary to make a classification of
the individual substances present in the feed and of the groups of substances
with chemical similarities, for instance, paraffins, aromatics, salts, or others.
The following questions have to be answered:
What is the pH of the solution?
Which acids or bases are present?
Are there additives, such as surfactants, that will affect the phase separation
properties?
Are solids present?
418 Wennersten
What is their nature?

Are there restrictions concerning product quality or raffinate compositions?
Are there any toxic components?
What is the stability of the substances?
When the foregoing criteria have been determined, one can proceed to the
next step.
10.2.3 Primary Stage in Solvent Selection

It is advisable to begin with a literature search, preferably on a computer-
based system or in a literature source book [1], to answer the question: what
is the state of art concerning the extraction of the substance of interest or of
similar materials? A preliminary selection of solvents must be based on an
estimate of chemical interactions that could be important. These interac-
tions are mainly ones of solute–solvent and solvent–solvent (see Chapter 2).
Solute–solvent interactions are dependent on the natures of the solute and
the solvent. The interaction is least when both are nonpolar; it becomes
stronger as the polarity of the molecules increases; and it is strongest
when there are interactions caused by formation of hydrogen bonds,
charge transfer complexes, or coordination complexes. In reactive solvent
extraction, it is desirable to find a solvent that can form strong, specific
interactions with the solute to be extracted. The well-known rule of
thumb that ‘‘like dissolves like’’ does not always apply. The solubilities of
nonpolar substances are sometimes larger in polar than in nonpolar sol-
vents.
The explanation for this could be that, although the solvent–solvent
interactions may be extensive in a polar solvent, the gain in energy from the
nonpolar solute interaction with a polar solvent can be still more so. In such
a case, this larger gain outweighs the disadvantage of the elimination of the
solvent–solvent interaction, and a higher solubility can be expected. The
classification of the solvents according to their chemical structure and
functional groups has been described in Chapters 2, 4, and 11 and is not
treated here. If specific chemical interactions are present, as when an acid is
extracted with a basic solvent, the complex formed may exhibit a low
solubility in the solvent. Here, the solvent should comprise both an
extractant (in this case a base) and a diluent that has a high solubility for the
complex. The optimal diluent has the following characteristics:
It improves the distribution of the solute to be extracted.

It improves the selectivity.
It increases the density difference between the phases and lowers the visc-
osity of the solvent.
At this stage, it may not be possible to distinguish any promising

process configuration for the recovery of the solvent. From the general
criteria just given, a selection should be made of several solvents repre-
senting the following classes:
Solvents having a lower boiling point than the solute.
Solvents having a higher boiling point than the solute.
Solvents from which the solute could be stripped back into an aqueous
phase, perhaps utilizing a temperature shift or a chemical reaction.
The recovery process must also be kept in mind, and fundamental
vapor–liquid data, such as the formation of azeotropes, should be examined.
Azeotropic data can be found in the literature [1], but are sometimes con-
tradictory. Finally, solvents that are unstable, toxic, expensive, and high
grade should be avoided, unless the product price is high and the feed flow
rate is low.
10.2.4 Secondary Stage in Solvent Selection

As mentioned earlier, quantitative optimization of all solvent properties is a
difficult and time-consuming procedure, since many process design alter-
natives must be considered for each solvent. Consequently, the solvent-
screening strategy proposed here is to identify, as a primary step, a group of
solvents that appears to have most promise as extractants for the solute. In
the second step, computer simulations that use process simulation packages
must be applied to quantify the effects of various process design alternatives.
This involves collecting thermodynamic data through literature or experi-
ments and fitting the data to a mathematical model.
10.3 PHASE EQUILIBRIA

Design of extraction processes and equipment is based on mass transfer and
thermodynamic data. Among such thermodynamic data, phase equilibrium
data for mixtures, that is, the distribution of components between different
phases, are among the most important. Equations for the calculations of
phase equilibria can be used in process simulation programs like PROCESS
and ASPEN.
There are mainly four possibilities of acquiring the necessary phase
equilibrium data for a certain system:
1. Data at the actual state (P,T-area of interest) are available from the
literature.
2. Data are available at other states requiring different thermodynamic
correlations for interpolation or extrapolation.
420 Wennersten
3. Data are available for similar systems at actual or other states-again

requiring different thermodynamic correlations for interpolation or
extrapolation.
4. Data are totally missing-experimental work is necessary to obtain
data.
There are thousands of references and several source books on liquid–
liquid equilibrium (LLE) data for organic systems in the literature [1].
However, the available data may be incomplete or unreliable, which makes
experimental work necessary. Different thermodynamic models can then be
used for interpolation and extrapolation.
10.3.1 Theory
Two main approaches can be taken when developing thermodynamic
equations for the correlation and prediction of equilibrium data. Both are
semiempirical in that they are based on simplifications of rigorous ther-
modynamic expressions, but include parameters that have to be fitted to
experimental data.
The first method, which is the more flexible, is to use an activity
coefficient model, which is common at moderate or low pressures where the
liquid phase is incompressible. At high pressures or when any component is
close to or above the critical point (above which the liquid and gas phases
become indistinguishable), one can use an equation of state that takes into
account the effect of pressure. Two phases, denoted a and b, are in equi-
librium when the fugacity f (for an ideal gas the fungacity is equal to the
pressure) is the same for each component i in both phases:
fi ¼ fi for i ¼ 1; 2; 3; . . . n (10.1)
(Reference states for both phases are chosen at the same temperature.)
From classic thermodynamics, the following relation can be derived
[2]:
Z P
fi 1 @V RT
ln ¼ ln i ¼ T; P; nj dP (10.2)
xi P RT 0 @ni P
Z 1
fi 1 @P RT P
ln ¼ ln i ¼ T; V; nj dV ln (10.3)
xi P RT V @ni V RT
where xi is the mole fraction of component i, and fi is the fugacity coeffi-

cient of component i in the mixture.
These equations are general and exact and express the fugacity of
component i in a mixture as a function of the measurable quantities pressure
(P), temperature (T ), volume (V ), and concentration (x).
The equilibrium condition is thus:
i xi ¼ i xi for i ¼ 1; 2; . . . n (10.4)
To solve the integrals in Eqs. (10.2) and (10.3), it is necessary to have
volumetric data; for example, an equation of state (EOS) of type:
V ¼ f1 ðT; P; n1 ; n2 . . .Þ (10.5)
or
P ¼ f2 ðT; P; n1 ; n2 . . .Þ (10.6)
Most of the equations of state are pressure explicit, and Eq. (10.6) can be
used for equilibrium calculations. As the integration is made from V to ?
the EOS has to be valid in the density range from zero to the actual density.
10.3.2 Equations of State

The EOSs are mainly used at higher pressures or when some of the com-
ponents are near or above their critical point. The most commonly used
cubic equations of state are the Peng-Robinson [3] and the Soave-Redlich-
Kwong [4] equations.
The Peng-Robinson equation has the following appearance:
RT aðTÞ
P¼ (10.7)
b ð þ bÞ þ bð bÞ
where n is the molar volume, a accounts for interactions between species in
the mixture, and b accounts for differences in size between the species of the
mixture.
For pure components:
b ¼ 0:07780ðRTc =Pc Þ (10.8)
aðTÞ ¼ aðTc ÞðTR Þ (10.9)
aðTc Þ ¼ 0:45724ðR2 T 2 =Pc Þ (10.10)

1=2
1=2 ðTR Þ ¼ 1 þ mð1 þ TR Þ (10.11)
m ¼ 0:37464 þ 1:54226! 0:26992!2 (10.12)
where Tc and Pc are the critical temperature and pressure, TR is the reduced
temperature (TR=T/Tc), and o is the acentric factor for component i.
422 Wennersten
For mixtures, one has to apply a mixing rule; for example, for a:
X
N X
N
a¼ xi xj aij (10.13)
i j
and
pffiffiffiffiffiffiffiffi
aij ¼ ð1 ij Þ ai aj (10.14)
where dij is an adjustable interaction parameter obtained from experimental

data, and xi and xj are mole fractions of component i and j in the liquid
phase.
For b the following mixing rule can be used:
N X
X N
b¼ xi xj bij (10.15)
i j
and
bij ¼ ð1 ij Þbðbi þ bj Þ=2c (10.16)
where Zij is an adjustable parameter determined along with dij by fitting the
equation of state to equilibrium data. The cubic equations of state are
further considered by Marr and Gamse [5].
10.3.3 Activity Coefficient Models

At low or moderate pressure, when the liquid phase is incompressible, an
activity coefficient model (g model) is more flexible to use than an equation of
state. This method often works, even for strongly nonideal systems invol-
ving polar and associating components.
The fugacity of a component i in a liquid mixture is given by:
fiL ¼ i xi fi0 (10.17)
where fi is the fugacity of component i at an arbitrary chosen reference state
and gi is the activity coefficient that indicates the deviation from ideality.
The reference state is often chosen as pure component i as a liquid at
temperature T and pressure P, which gives:
"Z #
P
fi ¼ Pi i exp
0 0 0
ði =RTÞdP (10.18)
P0i
where Pi is the vapor pressure for pure component i, fi is the fugacity

coefficient for component i at saturation, and ni is the molar volume for
component i as a liquid.
For components at supercritical conditions, the reference fugacity, fi,

is usually taken as Henry’s constant:
fi0 ¼ Hi ¼ lim ð fiL =xi Þ (10.19)
xi !0
where Hi is a function of pressure and temperature. The activity coefficient gi

is obtained from Gibbs ‘‘excess’’ energy as:

@nT GE
RT ln i ¼ P; T; nj (10.20)
@ni
To calculate gi, an expression is required such as:
GE ¼ f3 ðT; P; n1 ; n2 . . .Þ (10.21)
For most liquids at moderate pressures, the pressure dependence can be
neglected and the temperature dependence is small. The most important
parameter is the composition.
Many empirical and semiempirical equations for GE have been published in
the literature. The equations have to fulfill the boundary condition:
i ! 1 as xi ! 1
for the reference state fio chosen as a pure component as a liquid and:
i ! 1 as xi ! 0
for components in supercritical state, with reference state fio=Hi.
The earliest equations for Gibbs excess energy, like Margules and
van Laar, were largely empirical. More recent equations and NRTL and
UNIQUAC are based on a semiempirical physical model, called the two-
liquid theory, based on local composition. The molecules do not mix in a
random way, but because of different bonding effects, the molecules prefer
a certain surrounding. This results in a composition at the molecular level,
the local composition, which differs from the macroscopic composition.
The local mole fractions cannot be measured easily, but must be related
to the overall composition. A simple correlation can be obtained from sta-
tistical thermodynamics for which the quotient of the local mole fractions
equals the quotient of the total mole fractions times a Boltzmann’s factor
according to:
x21 x2
¼ exp½ð 21 11 Þ=RT (10.22)
x11 x1
x12 x1
¼ exp½ð 12 22 Þ=RT (10.23)
x22 x2
where (m21m11) and (m12m22) are two adjustable parameters that can be
obtained by correlation to binary experimental data.
424 Wennersten
Activity coefficient methods work fairly well at temperatures well below the
critical, at which the liquid phase is largely incompressible, and up to
moderate pressures.
10.3.4 Liquid–Liquid Equilibria Using

Activity Coefficient Models
Binary data can be represented with a Tx diagram that shows the mutual
solubility as function of temperature. Most of the binary systems belong to
one of the classes in Fig. 10.1. For ternary systems, experimental data are
usually obtained at constant temperature and given in ternary diagrams.
There are many types of systems, but more than 95% belong to one of the
two classes shown in Fig. 10.1.
The fugacity for two liquid phases in equilibrium can be calculated
with Eq. (10.17). At equilibrium, the temperature in both phases is the same,
Fig. 10.1 Different types of liquid–liquid systems. (a), (b) Solubility as function of
temperature for binary systems; (c), (d) ternary systems. (Dashed lines are examples of tie
lines, which connect the two phases in equilibrium located at the binodal.)
which gives the same reference state for component i in both phases, and the
equilibrium condition then reduces to
i xi ¼ i xi (10.24)
where a and b denote the two phases.
The activity coefficients can be calculated using any of the existing
models if the binary parameters for all combinations of binary pairs are
known. These parameters are obtained by fitting to experimental data. For
ternary systems, one can either simultaneously fit all six parameters or
first determine the parameters using binary data for those binary systems
that have a phase separation and the rest of the parameters from ternary
data.
In principle, as the g models do not have an internal temperature depen-
dence, the extrapolations in temperature must be carried out with great
care. When extrapolating in temperature, the parameters in the g models
should be made temperature dependent by fitting to experimental data at
different temperatures. Parameter estimation is obtained by minimizing an
object function such as:
X
N X
n
Q¼ ½ðXik Xik Þ2 þ ðXik Xik Þ2 (10.25)
k¼1 i¼1
As an example, the system ethyl acetate–acetic acid–water can be used and

equilibrium data using the UNIQUAC equation calculated [6]. Using the
NRTL equation would have given similar results.
In principle, there are three situations:
1. Experimental liquid–liquid equilibrium (LLE) data are available.
2. Only experimental vapor–liquid equilibrium (VLE) data are available.
3. Experimental data are lacking.
In the first case, the six parameters in the UNIQUAC equation are
obtained by fitting to experimental LLE data, shown in Fig. 10.2 as tie lines.
The binodal curve has been calculated and is shown in Fig. 10.2 as a
solid line. Furthermore, the calculated distribution of acetic acid between
both phases is shown in Fig. 10.3. From these figures, it can be seen that the
fitting to experimental data is good.
In the second case, where LLE are lacking, VLE data are used to fit
the parameters in the UNIQUAC model. These parameters, for the three
binaries, were obtained from the literature in which VLE data were given at
the following temperatures:
Ethyl acetate–acetic acid, 315 K
Ethyl acetate–water, P=20 kPa (303–333 K)
Acetic acid–water, 313 K
426 Wennersten
Fig. 10.2 Liquid–liquid equilibrium for the system ethyl acetate–acetic acid–water at
303 K.
As can be seen in Figs. 10.2 and 10.3, the agreement with experimental LLE
is not very good.
Generally, one can say that parameters obtained from VLE should
be used only for qualitative calculations. The result is considerably im-
proved if the parameters for the binary ethyl acetate–water are calculated
from solubility data.
When both VLE and LLE data are missing, it is not possible to
assume ideal solutions. The separation into two liquid phases indicates a
strong deviation from ideality. Here, the remaining possibility is to use
some of the existing group contribution methods, such as UNIFAC, for
which the parameters are based on LLE data.
10.4 INDUSTRIAL ORGANIC APPLICATIONS

OF LIQUID EXTRACTION
10.4.1 Petroleum and Petrochemical Processing
Oil refineries and the petrochemical industry constitute one of the largest
application areas for liquid extraction. The largest applications in this field
are the treatment of lubricating oils and the separation of aromatic and
aliphatic hydrocarbons. Other examples are the production of high-purity
fiber-grade caprolactam, recovery of acrylic acid, and production of anhy-
drous acetic acid. The success of solvent extraction in the petroleum and
petrochemical industry is due to careful process integration and also the
development of large-scale column extractors.
The development of different processes and solvents for the separation
of aromatics from aliphatics has reached a rather stable state. A number of
different processes, some of them with capacities of several hundred thou-
sand tons of aromatics per year, are in operation. The more important ones
are listed in Table 10.1.
The Union Carbide Process
The process equipment consists mainly of two extraction columns with
pulsating trays and four distillation columns according to Fig. 10.4 [7]. The
feed, with a high content of aromatics, is pumped to the middle of the first
extraction column where the aromatics are extracted with the solvent S1
(tetraethylene glycol). In the lower part, the extracted aromatics are washed
with S2 (dodecane). The outgoing raffinate phase R1 (containing aliphatics,
S2, and a small amount of S1) is distilled in the fourth distillation column.
The distillate, D4 consists of pure aliphatics, whereas the bottom stream
Fig. 10.3 Distribution coefficients for acetic acid.

428 Wennersten
Table 10.1 Different Processes for Separation of Aromatics from Aliphatics

Extraction
Process Extraction solvent temp. (K) Equipment
UDEX Polyethylene glycols + 423 Sieve plate column
water
Union Carbide Tetraethylene glycol 373 Pulsed plate column
Shell Sulfolane 393 RDC
DMSO Dimethylsulfoxide 298 RDC
Arosolvan N-Methylpyrrolidone 308 Vertical mixer-
settler column
Formex N-Formylmorpholine 313 Sieve plate column
Carmex Methyl carbonate 298 RDC
S2 + DS1 is recirculated as reflux to the first extraction column. The leaving

extract phase E1, containing aromatics and S1, is fed to the second
extraction column where it is washed with more S2. The washed organic
phase Sl is recirculated to the first column, while the extract phase E2,
containing aromatics and S2 + DS1, goes to distillation.
The distillates from the distillation columns constitute the pure
aromatics, while the bottom stream from the third column, containing
S2 + DS1, is recirculated to the second extraction column. The operating
conditions for this process are given in Table 10.2.
Re-refining Process
Solvent extraction is used in a technology for upgrading of low-value,
contaminated hydrocarbons into high-value products such as base lubri-
cating oil stock and clean-burning industrial fuels (Fig. 10.5). Thus, the re-
refining process may be used to extract gasoline and diesel from refinery
bottoms.
To produce fuel, propane is mixed with the refinery bottoms to
separate heavy hydrocarbons from middle distillates remaining in the
bottoms. These middle distillates can be processed through vacuum dis-
tillation to produce gasoline, diesel, and marine diesel [8].
10.4.2 Pharmaceutical Industry

Liquid–liquid extraction is extensively used in the pharmaceutical industry
for the production of drugs and isolation of natural products [9]. These
products are often heat sensitive and cannot be recovered by methods such
as atmospheric distillation or evaporation. Owing to competition, little
detailed information is available on current commercial operations. A clas-
sic example and the best documented, which has encountered problems
general for pharmaceutical compounds, is the purification and concentra-

tion of penicillin [10].
Production of Penicillin
The fermentation broth containing penicillin is first filtered to remove
mycelium and adjusted to pH 2–2.5 to convert it to the largely undissociated
penicillinic acid before it is fed to the first extraction stage where it is con-
tacted with, for example, butyl acetate (Fig. 10.6). The precise pH used is a
compromise between the stability of the penicillin and the partition coeffi-
cient. The partition coefficient increases with pH, but unfortunately, it is
accompanied by a rapid decrease in product stability [11]. To minimize
the time the product is in the free-acid form, the feed is contacted with
the extractant before acidification and the broth/extractant/acid efficiently
contacted using an in-line mixer. It is also necessary to carry out the acid-
ification and extraction at low temperature to minimize decomposi-
tion. Extraction takes place in countercurrent centrifugal contactors, e.g., a
Podbeilniak extractor to ensure rapid throughput. Once in the organic phase
the product is more stable, being removed from the aqueous acidic phase.
However, because of entrained acid, it is still important to convert the
penicillinic acid to a stable salt as soon as possible. This is achieved by
mixing the extract phase with a 2% potassium phosphate solution at pH 6.
The potassium penicillin is recovered by filtration and the solvent recircu-
lated. Recovery of the organic phase from the aqueous raffinate is also very
important to minimize costs and environmental impact. Butyl acetate, being
a low-boiling solvent, can be recovered easily by distillation.
Fig. 10.4 The Union Carbide process.

430 Wennersten
Table 10.2 Some Operating Conditions for the Union Carbide Process
First extraction Second extraction
column column
Equipment type Pulsed plate Pulsed plate
column column
Number of ideal stages 10 7
Extraction temperature 373 K 373 K
Solvent Tetraethylene glycol Dodecane
Reflux Dodecane -
Sl/S2 9 1.43
Sl/F 7 -
Aromatics in feed - 87 vol%
Recovery of benzene and - >99.5%
toluene (D1, D2) - -
Recovery of C8-aromatics (D3) - >98.5%
Purity of benzene - 99.9%
Problems that can arise in the process include the formation of

emulsions during extraction from the presence of surface-active impurities
in the filtered broth. The effect of these can be minimized by introducing
appropriate surfactants that can also reduce the accumulation of solids in
the extraction equipment. In addition, other organic impurities are present
that can be coextracted with the penicillin. It has been found that a number
of these can be removed by adsorption onto active carbon.
Most of the penicillin is used as intermediates in the production of,
for example, cephalosporin, but if it is necessary to produce pure penicillin
for pharmaceutical use it can be purified by reextraction at pH 2–2.5 and
further stripping with a phosphate solution at pH 6.
This extraction of penicillin is an example of the direct use of a polar
organic compound to partition a solute. An alternative process that has been
considered is to use organic reagents that form ionpairs with penicillin [12]:
R4 Nþ þ
aq þ Paq $ R4 N Porg
Here the authors found that the penicillin anion (P) could be
extracted efficiently with a secondary amine (Amberlite LA-2) in the pH
range 5–7 where the product is most stable. This type of process is used
extensively in hydrometallurgy (Chapter 11) and can be used to extract both
anionic species using cations as shown earlier, or cationic species using
organic acid anions. In hydrometallurgy, the system normally uses a hydro-
carbon diluent, but for pharmaceutical applications more polar diluents are
generally required.
Although the use of such chemically assisted extraction procedures is

unlikely to displace the established extraction processes for commercial ex-
traction of penicillin, there are a number of other systems in biotechnology
where ion pair formation is used (section 10.5).
10.4.3 Separation of Isomers

Pure isomers are often used as starting products for fine chemicals (e.g.,
different drugs). The separation of isomers entails great difficulties because
they often have boiling points differing by only a fraction of a degree and
they have closely similar solubilities in many solvents. Solvent extraction
processes for the separation of isomers, therefore, have to rely more on
chemical reactions than on nonspecific physical interactions between the
solute and the solvent.
Separation of m-Xylene from o- and p-Xylene
Xylenes react reversibly to form complexes with boron trifluoride (BF3) in
the presence of liquid hydrogen fluoride (HF). This fact can be utilized in
the separation of m-xylene from its isomers, because the m-xylene complex
is much more stable than the others and also more soluble in hydrogen
Fig. 10.5 The re-refining process.

432 Wennersten
active carbon in
Solvent solvent
Water Solvent
acid
Penicillin
G or V
containing
broth
Broth filtration Solvent extraction Carbon filtration

to distillation to distillation
Carbon treatment
Mycelium Extracted broth Carbon
filtrate (+ solvent) (+ solvent + impurities)
Cycle 15h Solvent Potassium or sodium

acetate in solvent
Solvent
Hot air To distillation
solvent
Penicillin (+ impurities)
G or V
potassium or Crystal
sodium salt filtration
Crystal filtration
and drying Crystal washing Crystallization
to distillation
solvent + impurities
Fig. 10.6 Purification of penicillin.
fluoride. After extraction with the HF-BF3 mixture, m-xylene is recovered

by vaporizing and recycling the HF and BF3 [13].
Separation of Alkylnaphthalene Isomers Using Cyclodextrins
Dimethylnaphthalenes are important starting materials for surfactants and
polymers. Thus 2,6-dimethylnaphthalene is used to produce 2,6-dimethyl-
carboxylic acid used in the manufacture of polyesters. The separation of the
2,6- and 2,7-dimethylisomers is difficult as their physical properties are very
similar, so other approaches have to be found. Cyclodextrins have a structure
where six a-1,4-D(+)-glucopyranose molecules are linked in a doughnut
shape (Fig. 10.7). They can accommodate various organic molecules in the
central apolar cavity forming inclusion complexes. Selectivity depends mainly
on the size and shape of the guest molecule [14]. The mixture of dimethyl-
naphthalenes in 1-methylnaphthalene is mixed with an aqueous solution of
substituted cyclodextrins and the system allowed to settle into two phases.
The upper phase consists of the oil sample and the lower phase contains the
cyclodextrin/2,6-dimethyl isomer inclusion complex. After separating the
phases, the aqueous phase is mixed with an organic solvent that dissociates
the inclusion complex and extracts the dimethylnaphthalene product. Analysis

of the product showed 96% isomeric purity of the 2,6-dimethyl isomer from a
50:50 mixture of 2,6- and 2,7-isomers [15].
The use of cyclodextrins is still in the development stage and such
systems have not yet been used commercially.
Dissociation Extraction
This is a separation technique for the separation of mixtures of organic
acids or bases that exploits the differences in the dissociation constants of
the components of the mixture. Application of dissociation extraction has
been considered chiefly for the separation of organic acids and bases
occurring in coal tar (e.g., m- and p-cresols) [16]. These isomers have very
close boiling points, but they have appreciable differences in acid strength,
as shown by their relative dissociation constants of 9.8 1011 for m-cresol
and 6.7 1011 for p-cresol. If a mixture of these bases is partially neu-
tralized by a strong base, there is a competition between the two acids for
reaction with the base. m-Cresol, the stronger acid, reacts preferentially
with the base to form an ionized salt, insoluble in organic solvents. The
weaker acid, p-cresol, remains predominantly in its undissociated form and,
accordingly, is more soluble in an organic solvent. When this process is
applied with countercurrent flow of organic solvent and an aqueous base,
the mixture of cresols may be separated. The m-cresol may be generated
from its salt in the aqueous phase by treatment with a strong mineral acid.
10.4.4 Environmental Applications

Solvent extraction has many features that make this separation technique
applicable for the removal of organic pollutants from wastewater. The
organic solutes can be recovered, and there is thus a potential for economic
credit to the operation. In comparison with biological treatment, it is not
subject to toxicity instabilities. The disadvantage is that, even with solvents
having low solubilities in water, solvent losses can be substantial owing to
Fig. 10.7 Structure of cyclodextrin.

434 Wennersten
Fig. 10.8 The Phenosolvan process.
high flow rates of wastewaters. In practice, solvent extraction, therefore,

has to be combined with other separation methods, such as distillation, to
fulfill the environmental restrictions. Complex process considerations have
then to be taken into account to evaluate the economics of different alter-
natives.
Phenosolvan Process
In this process developed by Lurgi [17], the phenolic effluent is contacted
with the solvent in a multistage mixer-settler countercurrent extractor
(Fig. 10.8). The extract, containing phenol, is separated into phenol and
solvent by distillation and solvent is recycled to the extractor. The aque-
ous raffinate phase is stripped from solvent with gas, and the solvent
is recovered from the stripping gas by washing with crude phenol and
passed to the extract distillation column.
Initially the process utilized butyl acetate as a solvent, but more
recently isopropyl ether has been used, although the latter has a much lower
partition coefficient for phenol. The reason for this choice of solvent is that
the separation of solvent and phenol by distillation is easier and less costly.
In addition, isopropyl ether is not subject to hydrolysis, which can be a
problem when using esters as solvents.
The Phenosolvan process is used in the treatment of wastewaters

from plants involving phenol synthesis, coke ovens, coal gasification, low-
pressure carbonization, and plastic manufacturing. The residual phenol
content after dephenolization is usually in the range 5–20 mg dm3 and
plants treating 500 m3 water per day are in operation.
10.5 APPLICATIONS OF LIQUID EXTRACTION IN

BIOTECHNOLOGY
The recovery of products from biotechnological processes has traditionally
been focused on bench-scale separation approaches, such as electrophoresis
or column liquid chromatography. These methods are difficult to scale up to
production levels and often become prohibitively expensive for medium-
and low-value products.
Recently however, it has been recognized that liquid extraction is a
potential method in the primary recovery of fermentation cell culture prod-
ucts, such as proteins and amino acids. The separation problem, however, is
difficult because the product mixtures are often complex, including cell
debris and enzymes. Proteins are not suitable for conventional solvent
extraction because of incompatibility with organic solvents, but can be
handled in aqueous two-phase systems or by extraction in reverse micellar
systems (Chapter 15).
Although biotechnological processes often have been stated to be
energy efficient in that the reaction temperature is low, it is important to
realize that the product concentrations are low and that the product recovery
step is often the most energy consuming.
10.5.1 Aqueous– Organic Systems

Solvent extraction has been proposed as an alternative for the separation of
organic acids and amino acids from fermentation broth [18]. Thus, citric
acid is currently produced in bulk by fermentation followed by recovery of
the acid by precipitation as the calcium salt. Purification of the impure
product is achieved by dissolving in sulphuric acid to precipitate calcium
sulphate, leaving the citric acid in solution from which it is recovered by
evaporation. An alternative process has been proposed using solvent
extraction [18]. Citric acid can be extracted by both high molecular weight
amines and organophosphate solvating reagents. Important criteria have
been established for the processing of citric acid by solvent extraction: a
good distribution coefficient, but <10 to allow for easy stripping with water,
and a low tendency to form emulsions. Stripping with water is important
because, although the citric acid could be recovered with a base, the
436 Wennersten
formation of a citrate salt would require further processing as in the con-

ventional flow sheet, thus removing the advantages of solvent extraction.
Wennersten found that the extractant system tri-n-butylphosphate
(TBP) in Shellsol A diluent was temperature dependent with extraction
decreasing with increasing temperature. Therefore, extraction from the
broth at room temperature (228C) and water stripping at 608C provided an
efficient process (Fig. 10.9). The extraction with long-chain amines was
hampered by the formation of emulsions and thus poor separation. Emul-
sion formation depended on the diluent with hydrocarbons, e.g., Alamine
336/Shellsol A, given the best separation.
10.5.1.1 Carboxylic Acids
Acetic acid is an important intermediate organic tonnage chemical that may
be produced by the petroleum industry and fermentation. The latter process
requires the recovery of acetic acid from water solutions, and several tech-
niques have been applied to this separation, including solvent extraction,
azeotropic distillation, and extractive distillation. A comparison of eco-
nomics between azeotropic distillation and solvent extraction combined
with azeotropic distillation (Table 10.3) shows that the introduction of
40
10ºC
22ºC
30 60ºC
Extract phase conc., g\l
80ºC
20
10
10 20 30 40 50
Raffinate phase conc., g\l
Fig. 10.9 Isotherms for citric acid extraction with TBP.

Table 10.3 Economics of Different Processes for Dehydration of Acetic Acid

Plant capital Steam usage Operating cost
(45,000 tons p.a. £m) (lb/lb acetic acid) (£/ton acetic acid)
Azeotropic distillation 0.5 6 4–5
Solvent extraction and 0.2–0.25 1.2–2.5 2–2.5
azeotropic distillation
solvent extraction successfully introduces savings in plant investment and

operating costs [19]. This process flow sheet is outlined in Fig. 10.10.
A number of studies have been carried out on the extraction of other
important carboxylic acids from fermentation liquors [20,21].
10.5.1.2 Amino Acids
Amino acids, because they contain both carboxyl (-COOH) and amino
(-NH2) groups, can behave as cations at low pH and anions at high pH
and have a zwitterion character at intermediate pH values. Although they
have no net charge in this intermediate pH range, they are hydrophilic and
so their solubility in nonpolar diluents is low. In addition, their extraction
with carbon-bonded oxygen donor extractants is poor. Therefore, the most
common method for extraction is to convert the amino acid into one of
the ionic forms and use an appropriate ion-pair extractant. Thus extrac-
tion can be effectively carried out by a cationic extractant (HA), for
instance, either a phosphoric acid extractant such as H(DEHPA), [22], or an
organosulphonic acid such as H(DNNSA), where the extracted species is
(RNH3COOH)+ (A) [23]. The extent of extraction depends on the relative
pKa values of the two acids, with extraction favored when the extractant has
a lower pKa than the extractant. Thus glycine (pKa=2.34) is extracted by
H(DNNSA) (pKa=0.68) over the pH range 1–5, but H(DEHPA) has a
higher pKa value (3.50), and little glycine extraction is found. However,
some amino acids with two amino groups, e.g., histidine, form both singly
and doubly charged cations and here the singly charged cation can be
extracted by H(DEHPA) or even Versatic acid (pKa * 4.8).
In high-pH media, the amino acid carries a net negative charge and so
can be extracted by an anionic extractant, usually an alkylammonium salt,
e.g., tri-n-octylamine hydrochloride (R3NHCl). Here the extracted species is
(R3NH)+ (RNH2COO) [24,25]. In both cases, stripping is easy using either
dilute alkaline or acidic solutions respectively to regenerate the extracting
species.
Of the two extracting systems, the use of an anionic extractant is more
common to minimize coextraction of inorganic cations by the organic acids.
438 Wennersten
Fig. 10.10 Acetic acid flow sheet.
However, similar problems can also occur with the alkylammonium ex-
tractants, especially as the pH of the solution must be sufficiently high to
produce an adequate concentration of the amino acid anions, i.e., more than
two units above the pK of the amino acid. Thus, coextraction of hydroxyl
ions competes with the amino acid anions and in addition lowers the pH of
the aqueous phase and reduces the concentration of amino acid anions. To
avoid such problems, a large buffer capacity is required in the aqueous phase
with the buffer chosen to minimize coextraction of anions.
10.5.1.3 Ethanol
One of the problems associated with the production of ethanol by fermen-
tation is the inhibition of the yeast activity with ethanol concentrations
greater than about 12 wt%, thus defining an upper practical limit for
ethanol production. Removing ethanol during fermentation by solvent
extraction would allow a greater total product yield from a given batch of
feed. The application of in-line extraction of ethanol from the fermentation
broth has attracted interest. Problems do arise with direct extraction from
broths with the effect of entrained organic solvents on microbial growth and
the reduction in mass transfer of the ethanol by the presence of cell debris
and other biological material. These problems can be minimized by suitable
choice of the extracting solvent, e.g., long-chain aliphatic alcohols such as
butanol, hexanol, etc., and removal of cell debris by filtration [26]. The
process operates as a closed loop with the filtered broth circulating through
a suitable column extractor before returning to the fermenter via a carbon

column to remove residual organic solvent (Fig. 10.11).
10.5.2 Aqueous Two-Phase Systems

Under certain conditions, polymer incompatibility in aqueous solutions can
lead to the formation of two phases with high water content. With such a
system, it is possible to separate sensitive biological molecules, such as
proteins, without denaturation, which would be the case for an ordinary
aqueous–organic solvent system.
Intracellular enzymes and other microbial proteins are of increasing
interest. Processes for their purification may be divided into five stages:
1. Disruption of the cells
2. Removal of cell debris
3. Concentration and enrichment
4. High-resolution purification
5. Concentration and finishing
An overall goal for these steps is to obtain a high yield while retaining the
biological activity of the proteins. The required purity of the protein is
determined by its end use. Enzymes that are to be used as technical catalysts
require a lower purity than if they are to be used for analytic purposes or as
pharmaceuticals.
Fig. 10.11 Extraction of alcohol.

440 Wennersten
The phenomenon of aqueous two-phase polymer systems has been

known since the 19th century. However, the utilization of aqueous two-
phase systems as a separation method for biological materials was explored
much later by Albertsson in the mid-1950s. Albertsson performed an
extensive investigation on the ability of different water-soluble polymers to
form aqueous two-phase systems and how biological macromolecules and
cell particles partitioned in these phase systems [27].
The system polyethylene glycol (PEG)–dextran–water is still the most
used and best-studied aqueous polymer two-phase system. A phase diagram
for a typical two-phase system is shown in Fig. 10.12 for the PEG-dextran
system. Both polymers are separately miscible with water in all proportions.
As the polymer concentration increases, phase separation occurs, with the
Fig. 10.12 Phase diagram and phase compositions of the dextran–polyethylene glycol
system D 48-PEG 4000 at 293 K. All values are % w/w.
formation of an upper phase, rich in PEG, and a lower phase, rich in

dextran, each phase consisting of more than 80% water. Within the two-
phase region, any mixture of the three components split into two phases,
with compositions dictated by the intersections of the tie line passing
through the mixture point with the binodal.
The formation of two aqueous phases can be exploited in the recovery
of proteins using liquid–liquid extraction techniques. Many factors con-
tribute to the distribution of a protein between the two phases. Smaller
solutes, such as amino acids, partition almost equally between the two
phases, whereas larger proteins are more unevenly distributed. This effect
becomes more pronounced as protein size increases. Increasing the polymer
molecular weight in one phase decreases partitioning of the protein to that
phase. The variation in surface properties between different proteins can be
exploited to improve selectivity and yield. The use of more hydrophobic
polymer systems, such as fatty acid esters of PEG added to the PEG phase,
favors the distribution of more hydrophobic proteins to this phase. In
Fig. 10.13, partition coefficients for several proteins in a dextran–PEG sys-
tem are given [27].
Cells and cell particles generally partition almost entirely to one of the
two phases or to the interface and one phase. Thus, it is possible to partition
proteins away from the cell debris and, thereby, reduce the need for
downstream processing.
The phase behavior of the polymers is also dependent on the type and
concentration of salt present. Many times a sufficiently high concentration
of salt in a single polymer solution can induce phase separation to form one
salt-rich and one polymer-rich phase. Sodium and potassium phosphate are
commonly used salts.
The efficiency of protein partitioning in aqueous two-phase systems
can be considerably enhanced by the introduction of specific ligands into the
phase system. The technique is called affinity partitioning, which is analo-
gous to the well-known affinity chromatography concept widely used in
analytical and preparative protein separations. In this technique, a small
fraction of one of the phase-forming polymers is substituted by the same
polymer with the specific ligand covalently attached to it. Since ligands
interact strongly with specific proteins, the effect on the partition coefficient
can be very dramatic, and there are examples when distribution coefficients
have been changed by a factor of 10,000. The ligands that are used for
affinity partition are mostly different dyes linked to a PEG molecule.
Recent developments in aqueous two-phase partitioning have been
published by Raghavarao et al. [28].
Biotechnological processes, in general, are run at low concentrations
of reactants and products. One reason for this is that most biocatalysts are
442 Wennersten
Fig. 10.13 Distribution coefficients of several proteins in the dextran–polyethylene

glycol system in 0.0005 M KH2PO4, 0.0005 M K2HPO4, and with increasing concentra-
tion of NaCl; the latter is expressed as moles NaCl added per kilogram standard phase
system.
sensitive to product inhibition. The reaction may stop when only a small
fraction of the substrate has been converted into product. In such cases,
there is a strong need for continuous extractive procedures during which the
product is continuously removed. The integration of bioconversion and
extraction is called extractive bioconversion.
The large differences in surface tension and dielectric constants
between water and common organic solvents cause enzymes to unfold when
they are exposed to the interface between the two solvents. This problem can
be overcome through extractive bioconversions in aqueous two-phase sys-

tems.
A collection of cutting-edge methods intended to provide practical
guidelines for those who are new to the area of separations in two-phase
systems has been published by Hatti-Kaul [29].
10.5.3 Process Considerations

An important issue to be addressed in industrial applications of two-phase
polymer systems for enzyme recovery is the economic or product quality
requirement that the phase-forming polymers and salts be recycled.
In some rare cases, the polymer is accepted in the product (e.g., in a
process for purification of enzymes for detergent purposes). This process
utilizes the system PEG–Na2SO4–water, and more than 95% of the enzyme
is extracted in a two-stage countercurrent process. No recycling of PEG is
necessary, since the polymer is a part of the product. Usually, however,
recycling of phase constituents is necessary, and several methods can be
applied. As in the process described earlier, enzymes can be recovered from
salt solutions by ultrafiltration, which also can be used to separate PEG
from lower molecular weight products. If the product is charged, an ion-
exchange chromatographic step could be sufficient for isolation. Several
methods to precipitate the polymer are also available.
Because of the small density difference between the two aqueous
phases, the phase separation properties of the system are poor. Solid-free
systems can be separated in a few hours or overnight in simple cylindrical
vessels. Centrifugal separators are usually necessary for phase separation
when cell debris is present because of the relatively high viscosity of the
bottom phase. The performance of centrifugal separators is greatly reduced
by a high solid concentration in the top phase. Thus, when extracting
proteins from disintegrated cells, conditions should be such that the protein
is partitioned to the top phase and the cell material to the bottom phase.
10.5.4 Economic Considerations

In a study by Hustedt et al. [30], comparisons were made between the costs
of liquid extraction recovery of formate dehydrogenase and conventional
processing using centrifugation, precipitation, and column chromatography.
The comparison was favorable for liquid extraction, and the major cost
saving was the reduction in process time, compared with the conventional
process. The dominating cost for the extraction process was that of phase
polymers, in particular the dextran, which can contribute up to 90% of the
materials cost in the two-phase system. This fact stresses the importance of
developing methods for recovery of the phase constituents.
444 Wennersten
10.6 SUPERCRITICAL FLUID EXTRACTION

Solvent extraction processes usually run at ambient pressures and tem-
peratures. If higher pressures are applied, it is mostly because a higher
extraction temperature is required when equilibrium or mass transfer con-
ditions are more favorable at an elevated temperature. Distillation, on the
other hand, is usually carried out at higher temperatures and ambient
pressures. To avoid thermal degradation, the pressure sometimes has to be
lowered below ambient pressure. Distillation is based on the differences in
vapor pressures of the components to be separated, whereas solvent
extraction utilizes the differences in intermolecular interactions in the liquid
phase.
Supercritical fluid extraction (SFE) is a separation technique that, to a
certain extent, unites the principles of distillation and solvent extraction.
SFE utilizes the very special properties of fluids at supercritical conditions,
as is discussed below.
Figure 10.14 is a P-T diagram that illustrates the difference between
conventional solvent extraction, supercritical extraction, and distillation.
The choice of pressure and temperature area depends on the separation that
is required.
Fig. 10.14 Pressure-temperature (P-T) diagram for a pure component.

Fig. 10.15 Solubility of naphthalene in ethene at 298 K and 318 K.
10.6.1 Physical Principles

Figure 10.15 illustrates the solubility of naphthalene in ethene as a function
of pressure at 298 and 318 K. The solubility of naphthalene increases dra-
matically when passing the critical point of ethene (Tc=282 K, Pc=5 MPa).
The increasing solubility is due to the rapid increase of the density of ethene
after passing the critical point.
Figure 10.16 illustrates the solubility of naphthalene in supercritical
ethene as a function of temperature at different pressures. In Fig. 10.16 the
temperature dependence of solubility is different in different pressure areas.
At high pressure, an increase in temperature is followed by an increase in
solubility, whereas at lower pressures the opposite effect occurs.
An increase in temperature at constant pressure, on one hand, leads to
a decrease in solvent density, which would lower the solubility. On the other
hand, an increase in temperature results in an increase in vapor pressure of
naphthalene. At high pressures, the density dependence on temperature is
small compared with the effect of vapor pressure, which results in an
increased solubility. At lower pressures, the density effect dominates when
increasing the temperatures, resulting in a decrease in solubility.
Figure 10.16 is characteristic for most mixtures in which the
mutual solubility of the components is limited. From Fig. 10.16 it is
446 Wennersten
Fig. 10.16 Solubility of naphthalene in supercritical ethene as function of temperature at

different pressures.
possible to derive simplified, general principles for extraction processes

in the supercritical area. In Fig. 10.17, three different principles are
shown schematically. Figure 10.17a illustrates the principle for extraction
and solvent recovery through a change in pressure. In Fig. 10.16, the
point E1 represents the state in the extractor (30 MPa). The extract phase
is passed through an expansion valve at which the pressure is reduced
to 8 MPa (point S1 in Fig. 10.16). Naphthalene now precipitates in the
separator. The solvent is recompressed and returned to the extractor. In
Fig. 10.17b, naphthalene is extracted at 6 MPa and 285 K (point E2
in Fig. 10.16). The extract phase is passed through a heat exchanger in
which the temperature is raised to 315 K (point S2 in Fig. 10.16). Also,
in this case, naphthalene precipitates in the separator. The solvent passes
through another heat exchanger to lower the temperature and is then
returned to the extractor. The extracted substances can also be recovered
through adsorption (e.g., on activated carbon), which is shown in Fig.
10.17c.
The potential to vary the density of the solvent without passing any
phase borders make the supercritical area very interesting from the point of
view of separation. Through a gradual decrease in pressure it is possible, for
example, to fractionate a mixture of substances.
10.6.2 Advantages and Disadvantages of

Supercritical Extraction
The main disadvantage of SFE is the elevated pressure, requiring more
expensive process equipment. The critical pressures, however, are below the
pressures used in many high-pressure processes in the petrochemical
industry today.
Table 10.4 gives critical data for the most common solvents used in
high-pressure extraction. Table 10.5 illustrates the favorable mass transport
properties that can be achieved in the supercritical area owing to a low
viscosity and a high diffusivity, compared with the liquid phase.
Fig. 10.17 (a) Extraction and separation through reducing the pressure; (b) extraction
and separation through a change in temperature; and (c) extraction and separation through
adsorption.
448 Wennersten
Table 10.4 Critical Properties for Some Solvents

Critical pressure
Solvent Critical Temp. (K) (MPa)
Ethene 283.8 5.05
Trifluoromethane 299.8 4.69
Carbon dioxide 304.2 7.29
Ethane 305.4 4.82
Dinitrogen monoxide 309.7 7.17
Sulfur hexafluoride 318.8 3.77
Propene 365.8 4.54
Propane 370.0 4.24
Table 10.5 Physical Properties of Fluids in Different States

Supercritical
Property Symbol Liquid Units Gas fluid phase
Density r g cm3 103 1 0.3
Diffusivity D cm2 s1 101 5106 103
Dynamic viscosity Z g cm1 s1 104 102 104
Table 10.6 Solubilities of Different Substances in Liquid Carbon Dioxide

Fraction dissolving
Completely miscible (% wt) Insoluble
Stannous chloride Water 0.1 Urea
Benzine Iodine 0.2 Glycine
Benzene Naphthalene 2.0 Phenylacetic acid
Pyridine Aniline 3.0 Oxalic acid
Acetic acid o-Nitroanisole 2.0 Succinic acid
Caprylic acid Oleic acid 2.0 Malic acid
Ethyl lactate Lactic acid 0.5 Tartaric acid
Amyl acetate Butyl stearate 3.0 Citric acid
Glyceryl triacetate Ethyl anthranilate 6.0 Ascorbic acid
Ethanol Glyceryl monoacetate 1.0 Dextrose
Hexanol Glycerol 0.05 Sucrose
Benzaldehyde n-Decanol 1.0 -
Camphor - - -
Limonene - - -
Thiophene - - -
The separation properties in SFE are dependent on the choice of

solvents, as well as on the solutes. The most popular solvent, carbon
dioxide, is a rather nonpolar solvent, which dissolves mainly nonpolar
solutes. Solubilities of selected compounds in liquid carbon dioxide
are given in Table 10.6. The solubility and selectivity can be altered by
adding small amounts of polar solvents, called entrainers (e.g., water or
ethanol).
10.6.3 Applications
Much of the interest in SFE has been focused on using carbon dioxide to
extract different natural products from solid materials. Examples of large
industrial processes in this area are decaffeinating coffee beans and hop
extraction.
In the area of extracting solutes from aqueous solutions, many systems
have been screened in feasibility tests that have used carbon dioxide as a
solvent. A partial list of the solutes includes ethanol, acetic acid, dioxane,
acetone, and ethylene glycol. The reason for these efforts has been potential
low energy costs compared with distillation and the environmental advan-
tages of using carbon dioxide.
Other systems that have been investigated are the separation of bio-
cides from edible oils and fractionation of different components in vegetable
oils. An example of the latter is given next. For an extensive literature survey
of work done in the area of SFE, see references [31–34].
Separation of Mono-from Triglycerides
In Fig. 10.18, the quasi-ternary system carbon dioxide–acetone–glycerides
at 13 MPa is shown [34]. As shown in Fig. 10.18, acetone increases the
solubility of glycerides in the carbon dioxide phase. The entrainer has
the function of increasing the solubility of the solute and enhancing the
selectivity.
In Fig. 10.19, a simplified process scheme is given to illustrate the
separation of a nonvolatile compound from a mixture, using an entrainer.
The process consists of two columns. In the first column, the nonvolatile
monoglycerides are separated from the mixture, and in the second column,
they are separated from the solvent.
The mixture to be separated is fed, together with the entrainer, to the
middle of the first column. Here, the solvent, carbon dioxide and acetone, is
supercritical to provide high solubility of the monoglycerides. The super-
critical phase leaves the top of column I and goes to the lower part of
column II. In column II, the binary solvent entrainer is subcritical and in
the bottom of this column, the monoglyceride leaves, together with the
entrainer. Part of it is returned as reflux to column I, whereas the rest goes
to distillation for the separation of acetone. With a bottom temperature of
450 Wennersten
Fig. 10.18 Phase equilibrium for the quasi-ternary system CO2–acetone–glycerides at

13 MPa and 343, 373, and 374 K.
353 K and a pressure of 13.5 MPa in column I and 383 K at the top of
column II, 95% of the monoglycerides could be separated selectively from
the mixture.
Supercritical fluid extractions with solid feed stocks are industrially
carried out batchwise because of lack of equipment for feeding solid
materials to a pressurized extractor.
With liquid feed solutions, however, it is possible to work in a
manner analogous to traditional solvent extraction. Pressurized columns
can be of the packed-bed type or agitated by magnetic stirrers. Because
of the efforts of pilot plant tests, much of the scale-up work has to
be carried out in laboratory extractors. From solubility measurements, it
is possible to determine parameters in thermodynamic models (e.g.,
equations of state), which can be used for the simulation of large-scale
applications.
Fig. 10.19 Process for separation of a nonvolatile component from a mixture.
10.7 NOMENCLATURE
The symbols used follow standard textbooks for chemical engineering [see,
Ref. 35].
fi Fugacity of component i
xi Mole fraction of component i in liquid phase
P Pressure
V Volume
v Molar volume
T Temperature
fi Fugacity coefficient of component i
ni Number of moles of component i
a Adjustable parameter in equations of state that accounts for
interaction between species in a mixture
b Adjustable parameter in equations of state that accounts for differences
in size between species in a mixture
oi Accentric factor for component i
Hi Henry’s constant for component i
GE Gibbs excess energy
gi Activity coefficient for component i
452 Wennersten
aI Parameter in the Peng-Robinson EOS Equation 10.11 or phase [e.g.,

Eq. (10.1)]
b Phase [e.g., Eq. (10.1)]
BIBLIOGRAPHY
Kennedy, J. F.; Cabral, J. M. S. Recovery Processes for Biological Molecules; Wiley:
1993.
Hatti-Kauk, R. Ed., Aqueous Two-Phase Systems; Humana Press, 2000.
Schugerl, K. Solvent Extraction in Biotechnology; Springer-Verlag: 1994.
Thornton, J. D. Ed., Science and Practice of Liquid-Liquid Extraction; Oxford Sci-
ence Publications: Vol. 2, chapters 3–5, 1992.
REFERENCES
1. Dortmunder Datenbank DDB (DETHERM database) [online]. Available from
STN International, DECHEMA e.V. server, 2001.
2. Smith, J. M.; Van Ness, H. C. Introduction to Chemical Engineering Thermo-
dynamics; McGraw-Hill: New York, 1975.
3. Peng, D. Y.; Robinson, D. B. Ind. Eng. Chem. Fundam., 1976, 15, 59.
4. Soave, G. Chem. Eng. Sci., 1972, 27, 1197.
5. Marr, R.; Gamse, T. Chem. Eng. Process., 2000, 39, 19.
6. Abrams, D. S.; Prausnitz, J. M. A.I.Ch.E., 1975, 21(1), 116.
7. Somekh, G. S.; Proc. ISEC 71, Lyon; Society of Chemical Industry: London
1971, 1, 323.
8. Kane, L.; Romanow-Garcia, S.; Nakamura, D. Hydrocarb. Process., 1995, 74,
36.
9. Verrall, M.S. in Science and Practice in Solvent Extraction, J. D. Thornton, Ed.,
Oxford Science Publications, Vol. 2, Chapter 3, 1992.
10. Kirk-Othmer Encyclopedia of Chemical Technology, 4th Edition 1993; Vol. 10,
pp. 125ff.
11. Reuben, B. G.; Sjoberg, K. Chemtechnol., 1981, 11, 315.
12. Reschke, M.; Schurgel, K. Chem. Eng. J. 1994, 28, B1-29.
13. Bailes, P. J. Chem. Ind., 1977, 69.
14. Bender, M. L.; Komiyama, M. Cyclodextrin Chemistry; Springer-Verlag: Ber-
lin, 1978.
15. Uemasu, I. Value Adding Through Solvent Extraction (Proc ISEC’96) edited
Shallcross, D. C., Paimin, R.; and Prvcic, L. M. Ed., Univ. Melbourne Press,
1966; Vol. 2, 1635p.
16. Pratt, M. W. T. Handbook of Solvent Extraction. Lo, T. C., Baird, M. H. I., and
Hanson, C. John Wiley and Sons: New York, 1983, p. 605.
17. Wohler, E. Removal and Recovery of Phenol and Ammonia from Gas Liquor;
Lurgi Kohle Mineraloltechnik GmbH, 1978.
18. Wennersten, R. J. Chem. Tech. Biotechnol., 1983, 33, 85.
19. Lloyd-Jones, E. Chem Ind., 1967, 1590.

20. Kertes, A. S.; King, C. J. Biotechnol. Bioeng,. 1986, 28, 269–282.
21. Canari, R.; Hazan, B.; Bloch, R.; Eyal, A. M. Value Adding Through Solvent
Extraction, (Proc ISEC’96) Shallcross, D. C., Paimin, R., and Prvcic, L. M.,
Univ. Melbourne Press, 1996, Vol. 2, 1517p.
22. Yagodin, G. A.; Yurtov, E.; Golubkov, A. S. Proc ISEC’86, Munich; DE-
CHEMA: 1986, 3, 677.
23. Lukhezo, M.; Kelly, N. A.; Reuben, B. G.; Dunne, L. J.; Verrall, M. S. Value
Adding Through Solvent Extraction, (Proc ISEC’96) Shallcross, D. C., Paimin,
R., and Prvcic, L. M. Eds., Univ. Melbourne Press, 1, 87 (1996).
24. Hano, T.; Matsumoto, M.; Ohtake, T.; Sasaki, K.; Hori, F.; Kawano, Y. J.
Chem. Eng. Japan, 1990, 23(6): 734.
25. Schugerl, K. Solvent Extraction in Biotechnology, Springer-Verlag, Berlin, 1994,
p. 113.
26. Wang, H. Y.; Robinson, E. M.; Lee, S. S. Biotechnol. Bioeng. Symp., 1981, 11,
555–565.
27. Albertsson, P-A. Partition of Cell Particles and Macromolecules, Second Edi-
tion; John Wiley and Sons: New York, 1971.
28. Raghavarao, K. S. M. S.; Guinn, M. R.; Todd, P. Sep. Purif. Methods, 1988,
27(1), 1.
29. Hatti-Kaul, R. Aqueous Two-Phase Systems: Methods and Protocols, Humana
Press: Totowa, NJ 2000.
30. Hustedt, H.; Kroner, K.H.; Kula, M-R. Proc. Eur. Congr. Biotechnol., 1984, 1,
597.
31. Stahl, E.; Quirin, K.-W.; Gerard, D. Dense Gases for Extraction and Refining;
Springer-Verlag: Berlin, 1987.
32. Bungert, B.; Sadowski, G.; Arlt, W. Ind. Eng. Chem. Res., 1998, 37(8), 3208.
33. Teja, A. S.; Eckert, C. A. Ind. Eng. Chem. Res., 2000, 39(12), 4442.
34. Peter, S.; Brunner, G. Ger. Chem. Eng., 1978, 1, 26.
35. Reid, R. C.; Prausnitz, J. M.; Sherwood, T. K. The Properties of Gases and
Liquids; McGraw-Hill: New York, 1977.
11
Solvent Extraction in Hydrometallurgy
MICHAEL COX* University of Hertfordshire, Hatfield, Hertfordshire,
United Kingdom
11.1 INTRODUCTION
The use of solvent extraction as a unit operation in hydrometallurgy now
extends to a wide range of metals from a variety of feed materials including
low-grade ores, scrap and waste, and dilute aqueous solutions. The tech-
nology was pioneered in the 1940s for the extraction of uranium from its ores
and, later, for the treatment of wastes from spent reactor fuel, still an
important use of the technique today (see Chapter 12). The knowledge
gained led to processes for the recovery of other high-value metals and the
separation of elements such as the rare earths, zirconium–hafnium, and
niobium–tantalum that, before the introduction of this technology, could be
separated only by lengthy batch techniques with many recycle steps to obtain
the desired purity. Gradually, solvent extraction was seen to have applica-
tions for the recovery of other less valuable, but important, metals such as
cobalt and nickel. At the time, the process was confined to rather small
operations. It was only after the development of selective chelating acidic
reagents in the 1960s that liquid–liquid extraction was seen to be a com-
mercially viable addition to the unit operations of hydrometallurgy and was
able to compete with alternative processes like cementation.{ Liquid–liquid
*Retired.
{
Cementation is the process of recovery of metals from dilute aqueous solution by reductive
precipitation using another metal with a more negative electrode potential, e.g., Cu2++
Feo?Cuo + Fe2+. The product, in this case ‘‘cement’’ copper, is relatively impure because of
iron contamination. However, cementation can be used in conjunction with a solvent extraction
flow sheet to remove small amounts of a metallic impurity, for example, removal of copper from
a nickel solution by cementation with nickel powder. Here the dissolved nickel conveniently
augments the nickel already in solution.
455
456 Cox
extraction is now an economic alternative to pyrometallurgy for metal

extraction, especially when physical beneficiation of the ore to provide a
suitable concentrate for smelting is difficult. The alternative hydrometal-
lurgical route (Fig. 11.1) involves leaching of the ore to provide a leachate
that, after any necessary solid–liquid separation, can be fed to the extraction
circuit. Control of relative liquid flows allows the concentration of the
desired metal and, thereby, economic processes for the recovery of metals
from dilute aqueous wastes such as mine waters and wash solutions from the
metal-plating industries.
Liquid–liquid extraction of metals has been used for many years as a
concentration technique in analytical chemistry; consequently, it was logical
to use the types of organic compounds employed in these systems for the
development of industrial reagents for the same metals. However, because
the overall requirements for these two end uses are different, modifications
to the basic structure were necessary to take account of the particular
requirements of hydrometallurgy, summarized by the following criteria:
1. The ability to transfer the desired metal selectively across the aqueous–
organic interface in both directions
2. The ability of the reagent–diluent mixture to function efficiently with
the proposed feed and strip solutions in terms of rates of operation and
stability toward degradation
3. The ability of the reagent to perform with maximum safety to plant,
personnel, and environment at minimum cost
4. The ability of the process to interface with other unit operations both
upstream (leaching) and downstream (winning) in the overall extraction
flow sheet
It can be seen that these criteria differ from those appropriate to
analytical applications. Thus, in the latter, it is not essential to strip the
metal from the organic phase, especially if a colorimetric method is being
used. Also the chemistry of the feed solution is under the control of the
analyst, who can change pH and add buffers and masking agents at will.
Finally, because so little reagent is used, toxicity and cost are not prime
considerations. Thus, although the general class of organic molecule will
find application in both analytical and hydrometallurgical processes, the
detailed structures of the compounds will be different. Examples of such
modifications can be found in a number of texts [1–3], and factors associated
with commercial reagent design have been discussed [3–6].
Solvent Extraction in Hydrometallurgy 457
Fig. 11.1 Hydrometallurgical flow sheet using solvent extraction.
11.2 THEORY OF METAL EXTRACTION

Although a few metal compounds are sufficiently covalent to be extractable
into an inert organic phase (e.g., halides of arsenic, antimony, gallium, and
mercury), for a number of reasons, such as their ease of hydrolysis, this type
of compound is unlikely to feature in a commercial extraction process.
Normally, metal salts exist in aqueous phases as hydrated species and, as
such, are incompatible with those nonpolar organic diluents normally
used in hydrometallurgy. Therefore, to achieve extraction, species must be
modified to make them more hydrophobic by such procedures as (1) reac-
tion of the metal cation with suitable anions to produce a neutral complex,
(2) formation of an ion pair, and (3) replacement of hydrated water mole-
cules by an organic solvating reagent.
Reagents that are capable of such reactions are termed acidic, basic or
ion pair, or solvating, and their modes of action are typified by the following
equations:
Mnþ þ nHA ! MAn þ nHþ (11.1)
MXðnpÞ
p þ ðn pÞR3 NHþ X ! R3 NHðnpÞ MXp þ ðn pÞX
(11.2)
458 Cox
MYq xH2 O þ rB ! MYq Br þ xH2 O (11.3)
Discussion of the theory and application of such reagent systems is outside

the scope of this chapter, but details can be found in Chapters 4 and 6, and
in other texts [1,3,7,8].
A survey of current commercial extractants is given in Tables 11.1–
11.4. The situation on availability of such extractants is continually chang-
ing, as modifications are always being made, and reagents may be added or
deleted as the markets alter. Similarly, in the commercial world names and
business interests of companies change. Thus, manufacturers are not in-
cluded in this table. Commercial solvent extraction processes have expanded
considerably since the 1960s, and there is an increasing use of solvent
extraction technology in hydrometallurgy. Flow sheets have been devised for
most of the transition elements and for a large number of other metals and
metalloids, especially if these are available naturally only at low concentra-
tions or are required at high purities. Of these elements, a few have been
chosen for more detailed investigation on the basis of their commercial
importance or because they demonstrate the interesting use of chemical
properties to achieve a commercial process. Other texts and review articles
are available that cover different elements, and these are included in the
bibliography.
11.3 COBALT– NICKEL

Cobalt and nickel are both important elements for metallurgists, and their
adjacent positions in the first-row transition series results in very similar
chemical behavior, which causes problems in their separation. Any separa-
tion based on solvent extraction will be difficult unless some differences in
chemical behavior can be exploited. Such differences include a higher rate of
water exchange of the hexa-hydrated divalent cation for cobalt than for
nickel, which might result in the development of a kinetic, rather than
thermodynamic, separation. In addition, although the redox potentials of
cobalt(II)–cobalt(III) and nickel(II)–nickel(III) couples are similar for the
hexa-aquoion, on complexation, cobalt(II) is much more readily oxidized
than nickel(II). Also, once formed, the cobalt(III) complexes are rela-
tively inert to substitution. Finally, in the divalent state, cobalt has a
strong tendency to form tetrahedral complexes in concentrated electrolytes
(e.g., CoCl42), whereas in such aqueous solutions nickel(II) retains hexa-
coordination (e.g., Ni(H2O)6Cl2).
These general differences in chemical behavior have been exploited to
provide the solvent extraction processes currently used or proposed for
cobalt–nickel separation. All these processes remove cobalt from nickel
Table 11.1 Commercial Acidic Extraction Reagents
Reagent class Structure Extractants
Carboxylic
acids
Naphthenic acids
Versatic acids
R1 = R2 = C5 = Versatic 10
R1 = R2 = C4–C5Versatic 911
Phosphorus
acids
Phosphoric R1 = OH; R2 = C4H9CH(C2H5)CH2O-

mono-2-ethylhexylphos-
acids phoric acid
R1 = R2 = C4H9CH(C2H5)CH2O- di-2-ethylhexylphosphoric
acid (DEHPA)
R1 = R2 = p-CH3(CH2)6CH2C6H5O- di-p-octylphenylphosphoric
acid (OPPA)
Phosphonic R1 = C4H9CH(C2H5)CH2-; 2-ethylhexylphosphonic
acids R2 = C4H9CH(C2H5)CH2O- acid mono-2-ethylhexyl
ester (PC88A), (P-507)
Phosphinic R1 = R2 = C4H9CH(C2H5)CH2- di-2-ethylhexylphosphinic
acids acid (P-229)
R1 = R2 = CH3(CH2)3CH2CH(CH3)CH2- di-2,4,4-trimethylpentyl-
phosphinic acid
(CYANEX 272)
Thiophosphorus R1R2P(S)OH
acids
monothiophosphinic
acid (CYANEX 302)
R1R2P(S)SH
dithiophosphinic
acid (CYANEX 301)
Sulfonic acid
5,8-dinonylnaphthanenesul-
phonic acid DNNSA
460 Cox
Table 11.2 Commercial Chalating Acidic Extractants
a-Hydroxyoximes R1C(NOH)CH(OH)R2 5,8-diethyl-7-hydroxydodecan-

6-oxime (LIX63)
b-Hydroxyaryloximes
R1 = phenyl R2 = C9H19 LIX65N

LIX65N + LIX63 (1%) LIX64N
R1 = CH3 R2 = C9H19 LIX84
R1 = H R2 = C9H19 P1
P1 + nonyl phenol P5000 series
P1 + tridecanol PT5050
P1 + ester modifier M5640
R2 = C12H25 LIX860
LIX84 + LIX860 LIX984
LIX980 + tridecanol LIX622
8-Quinolinol
R = 5,5,7,7-tetramethyl-1-octenyl- (pre-1976)
= 4-ethyl-1-methyloctyl- (post 1976) Kelex 100
R = unknown side chain Kelex 108
R = unknown saturated alkyl LIX26
b-Diketones R1COCH2COR2
R1 = R-C6H5 R2 = CH3(CH2)5 LIX54
solution, as no commercial flow sheet has yet been devised to remove nickel
selectively from cobalt. Two different solvent extraction systems are used
commercially involving (1) anion exchangers and (2) acidic chelating ex-
tractants, and these will be considered in turn.
11.3.1 Anion Exchangers

As noted earlier, cobalt(II) in strong electrolyte solutions will readily form
tetrahedral anionic complexes that can be extracted into an immiscible
Table 11.3 Commercial Basic Extractants

Primary amines RNH2
R = (CH3)3C(CH2C(CH3)2)4 Primene JMT
Secondary amines R1R2NH
R1 = C9H19CH = CHCH2
R2 = CH3C(CH3)2(CH2C(CH3)2)2 Amberlite LA-1
R1 = CH3(CH2)11
R2 = CH3C(CH3)2(CH2C(CH3)2)2 Amberlite LA-2
R1,R2 = CH3(CH2)12 Adogen 283
R1,R2 = CH3CH(CH3)CH2(CH2)6 HOE F2562
1 2 3
Tertiary amines RRRN
R1,R2,R3 = CH3(CH2)7 Trioctylamine
Alamine 300
R1,R2,R3 = C8–C10 mixture Alamine 336
Adogen 364
Hostarex A 327
R1,R2,R3 = (CH3)2CH(CH2)5 Tri-isooctylamine
Adogen 381
Alamine 308
Hostarex A 324
R1,R2,R3 = (CH3)2CH(CH2)7 Tri-isodecylamine
Alamine 310
R1,R2,R3 = CH3(CH2)11 Adogen 363
Alamine 304
R1,R2,R3 = CH3(CH2)12 Adogen 383
R1,R2,R3 = C28H57 Amberlite XE204
R1 = CH3(CH2)7, R2 = CH3(CH2)9
R3 = CH3(CH2)11 Adogen 368
Quaternary amines R1R2R2N(CH3) + Cl
R1,R2,R3 = C8–C10 mixture Aliquat 336
Adogen 464
HOE S 2706
462 Cox
Table 11.4 Commercial Solvating Extractants

Carbon-oxygen extractants
Amides R1CONR22
R1 = CH3, N503
R2 = CH3(CH2)5CH(CH3)
R1 = R2 unknown A101
Ethers CH3(CH2)3OCH2CH2OCH2- Dibutylcarbitol
CH2O(CH2)3CH3 (Butex)
Ketones (CH3)2CHCH2COCH3 Methyl isobutylketone
(MIBK)
Phosphorus-oxygen extractants
Phosphorus esters R1R2R3PO
R1,R2,R3 = CH3(CH2)2CH2O Tri-n-butylphosphate
(TBP)
R1,R2 = CH3(CH2)2CH2O,
R3 = CH3(CH2)2CH2 Dibutylbutylphosphonate
(DBBP)
R1,R2 = CH3O, R3 = CH3 Dimethylmethylphosphonate
Phosphine oxides R1R2R3PO

R1,R2,R3 = CH3(CH2)6CH2 Tri-n-octylphosphine oxide
(TOPO) CYANEX 921
Phosphorus-sulfur extractants
Phosphine sulfides R1R2R3PS
R1,R2,R3 = CH3(CH2)6CH2 CYANEX 471X
1 2
Carbon-sulfur extractants R R S
R1,R2 = C6H13 Dihexylsulfide
R1,R2 = C7H15 Diheptylsulfide
organic phase by amines or quaternary ammonium compounds. The most

commercially important aqueous-phase ligand is chloride, although thio-
cyanate has also been studied by INCO Canada. Nickel/cobalt ratios of
2000:1 in the nickel raffinate and a cobalt product with a 1000:1 ratio over
nickel are readily achievable. In chloride medium, the degree of formation of
CoCl24 and, hence, the level of extraction, depend on the chloride concen-
tration. This is shown in Fig. 11.2, which also shows the level of separation
possible from nickel (>104) at a hydrochloric acid concentration of 8–
10 mol dm3. This extraction process can be represented by Eq. (11.4), and the
reaction can be easily reversed by water washing of the loaded organic phase.
þ
4 þ 2R3 NH Cl
CoCl2 ! ðR3 NHÞ2 CoCl4 þ 2Cl (11.4)
Figure 11.3 illustrates the effect of hydrochloric acid concentration on

cobalt extraction, with a maximum cobalt extraction observed at approxi-
mately 9 mol dm3. Above this concentration, the HCl2 ion competes with
CoCl42 for the extractant; replacement of hydrochloric acid by a metal
chloride solution, such as LiCl, eliminates this maximum. Fig. 11.3 also
shows that extraction depends on the concentration and type of extractant;
thus, extraction is greater for tertiary>secondary>primary amines. This
anion-exchange process has been used commercially in Norway (Falcon-
bridge Nikkelwerk) and Japan (Sumitomo) for cobalt–nickel separation.
11.3.2 Acid Extractants

For simple acidic extractants, such as carboxylic or sulfonic acids, the
similarity in formation constants does not produce cobalt–nickel separation
Co
factors (SNi *2) sufficiently large for commercial operation (Fig. 11.4).
Data for pH versus extraction for some chelating acid extractants does seem
to offer the possibility of separation [e.g., for the hydroxyoxime Acorga P50,
the pH50 for nickel(II) is 3.5 and for cobalt(II) 5.0]. Normally, this pH
difference would be suitable for a separation process, but this particular
system has hidden complications. The rate of nickel extraction is very slow
compared with cobalt and, in addition, although cobalt is initially extracted
Fig. 11.2 Extraction of metal chlorocomplexes by amines. (From Ref. 2.)

464 Cox
as the divalent species, atmospheric oxidation to cobalt(III) readily occurs,

which prevents normal stripping processes. A similar oxidation of the loa-
ded organic phase has been observed with substituted quinolines and b-
diketones. The complete exclusion of air from the circuit enables one to
obtain high levels of cobalt stripping, but this complicates the operation. It
has also been shown that, if cobalt(III) already exists in the aqueous phase,
then hydroxyoximes and some b-diketones do not extract cobalt at all,
although it has been reported that Kelex 100 will extract cobalt(III). This
behavior suggests a possible cobalt–nickel separation route by the formation
of stable cobalt(III) species, such as cobaltammines, in the aqueous phase
and extraction of nickel by, for example, hydroxyoximes.
The oxidation of cobalt(II) in the organic phase can be minimized
by the addition of donor molecules, such as the solvating extractants,
Fig. 11.3 Extraction of metal chlorocomplexes as a function of chloride concentration.

(From Ref. 2.)
Fig. 11.4 Extraction of cobalt and nickel with naphthenic acid (0.5 mol dm3), dashed
lines, and a mixture of naphthenic acid (0.5 mol dm3) and isotridecanal oxime (1.0 mol
dm3), both systems in toluene. (From Ref. 10.)
tri-n-butylphosphate (TBP), tri-n-octylphosphine oxide (TOPO), and car-

boxylic acids. In addition to stabilizing cobalt(II), various degrees of
synergism can be obtained, enhancing the separation of nickel from cobalt.
Although separation factors of up to 150 have been observed with the
system LIX* 63–carboxylic acid, again, there are problems with the slow
rate of nickel extraction. By using a nonchelating oxime in place of LIX 63,
separation factors of about 200 have been obtained, without any reduction
in the rate of nickel extraction or oxidation of cobalt (see Fig. 11.4). The
structure of the extracted complex for both metals is MR2(oxime)2, and
relative ligand field stabilizing energies for nickel and cobalt have been
proposed as the mechanism for synergism [9,10]. One problem with the use
of this system in continuous operation is that the acid strip solution tends to
hydrolyze the oxime.
Earlier it was noted that nickel and cobalt could be extracted by
carboxylic and sulfonic acids, with nickel being extracted at the lower pH.
However, with alkylphosphorus acids, a selectivity reversal is observed, with
cobalt being favored under acid conditions. The cobalt–nickel separation
factor has been shown to depend upon metal concentration, reagent struc-
ture, diluent, temperature, and the presence of a diluent modifier. Thus, with
increasing cobalt concentration the color of the extractant phase changes
*LIX is a trademark of Cognis, previously Henkel.

466 Cox
from pink to blue, indicating a change from octahedral coordinated species

(e.g., [CoR2(HR)2]) to an unsolvated polymeric tetrahedral species [CoR2]n.
Similar behavior was shown by changing the other variables listed. In
contrast, nickel remains octahedrally coordinated, with a structure [Ni(R2)-
HRx(H2O)4–x], maintaining a green color in the organic phase. This beha-
vior was first noticed for di-2-ethylhexylphosphoric acid (DEHPA), and the
chemistry of separation was elucidated by Barnes et al. [11]. Since the
introduction of a phosphonic acid ester (PC88A, Daihachi) and a phos-
phinic acid (CYANEX* 272, Cytec) further studies [12] have shown a
reagent dependence on the cobalt–nickel separation factor, increasing from
14 (DEHPA) to 280 (PC88A) and 7000 (CYANEX 272), at pH 4 and 258C.
Other factors that increased the separation include a change from aliphatic
to aromatic diluents and an increase in temperature. Introduction of diluent
modifiers, such as the solvating extractants TBP or isodecanol, reduces the
separation factor by depolymerizing the system and partially solvating the
extracted complex to form octahedral species. Di-2-ethylhexylphosphoric
acid has been proposed commercially to separate cobalt–nickel with either
column contractors or mixer-settlers [13], and PC88A has been incorporated
into a flow sheet [14]. A problem with these circuits arises from oxidative
degradation of the diluent by cobalt, which depends on the aromatic content
of the diluent, on the temperature, and on cobalt loading [15]. This degra-
dation can be eliminated by introduction of antioxidants such as nonyl
phenol. It is surprising that such oxidative degradation of diluents has only
recently been observed, when it is well known that cobalt compounds can
act as oxygen carriers and oxidation catalysts. In addition to the diluents,
oxidation of hydroxyoximes and 8-quinolinols has also been reported.
11.3.2.1 Nickel Laterite Processing
This discussion has centered on the use of solvent extraction to separate
cobalt from nickel, as until recently the use of solvent extraction in nickel
flow sheets has been minimal. However, with the increasing interest over the
past decade in the development of hydrometallurgical processes for nickel
lateritic ores, the use of solvent extraction to produce high-purity nickel
products has become important. The flow sheets that have been developed
are interesting as they indicate a number of different ways of using solvent
extraction. Lateritic ores are formed by the chemical weathering of surface
rocks and two main types are found, limonitic ores containing large amounts
of iron (40%) and typically 1.4% nickel and 0.15% cobalt, and saprolitic
ores that contain less iron (*15%) and cobalt (0.05%) but more nickel
(2.4%) and silica. The saprolitic ores are amenable to pyrometallurgical
*CYANEX is a trademark of Cytec Inc., previously Cyanamid.

treatment usually to produce ferronickel, but the lower content of nickel and
higher levels of cobalt in the limonitic ores makes them more suitable to
hydrometallurgical processing. As the metal values are widely distributed
through the limonitic ore, preconcentration is not feasible so that large
amounts of the ore have to be leached, giving large volumes of a dilute leach
liquor. Two reagents are used for leaching, ammonia in the Caron process,
which dissolves the metal as ammoniates e.g., [Ni(NH3)6]2+, and acid, which
gives hydrated cations. Details of the complete hydrometallurgical flow
sheets are outside the scope of this chapter and can be found elsewhere [13].
Here discussion will concentrate on the solvent extraction flow sheets to
illustrate how the technology can separate the desired values from a range of
other metals present in the leach liquor.
11.3.2.1a Ammonia Leachate
The ammonia leach leaves the iron as an insoluble residue that is removed
by filtration. The filtrate is contacted with a ketoxime reagent (e.g., LIX84I,
Cognis) to extract the nickel [14].
þ
NiðNH3 Þ2þ
x þ 2HðoximeÞ ! NiðoximeÞ2 þ 2NH4 þ ðx 2ÞNH3
(11.5)
However, this will also extract cobalt(II) under the same conditions and once
extracted the cobalt is oxidized to cobalt(III) and cannot be stripped.
Therefore, to avoid this problem the cobalt is oxidized prior to extraction. In
this way, nickel can be selectively extracted from a feed solution (9 g dm3 Ni
and 0.3 g dm3 Co) leaving about 0.01 g dm3 Ni in the raffinate. Nickel is
subsequently stripped from the loaded organic phase with strong ammonia.
One of the main disadvantages of this flow sheet is that the cobalt(III) causes
some oxidation of the reagent and a process of reoximation is required [14].
Also, ammonia transfers to the organic phase, which again represents an
operating loss. However the introduction of solvent extraction raises the
nickel content of the products to >99.5% nickel as compared to 85–90%
nickel with the earlier flow sheet that did not use solvent extraction [13].
An alternative route to cobalt/nickel separation following ammonia
leaching in the Caron process is to produce a mixed Co/Ni carbonate
product. This can then be dissolved in sulfate media and processed using one
of the acidic extractants described later [13].
11.3.2.1b Acid Leaching
Acid leaching is usually carried out using sulfuric acid under pressure to
dissolve the majority of the iron minerals and to release the cobalt and
nickel. If this is carried out at 150–2508C, then the iron(III) is precipitated as
haematite or jarosite, reducing the amount of iron in the leachate. The
468 Cox
leachate produced from acid leaching contains higher concentrations of

other elements that can cause problems in the solvent extraction circuit,
especially calcium, magnesium, and silica. Silica is a particular problem
because of the tendency to promote crud formation (see Chapter 8). Various
flow sheets involving different reagent have been proposed to treat the
leachate. The three main flow sheets will be considered in the following
sections.
11.3.2.1c Bulong Process
The Bulong operation is situated near Kalgoorlie in Western Australia and
in the year 2000 produced 9000 t/a nickel and 729 t/a cobalt. Following the
pressure leach, the leachate is neutralized with lime to precipitate residual
iron, chromium, and aluminum. The clarified filtrate is contacted with
CYANEX 272, containing TBP to inhibit third phase formation, at pH 5.5–
6.0 and 508C to extract the cobalt, zinc, and manganese and maximize the
rejection of nickel (Fig. 11.5). Nickel is then extracted from the raffinate
with a carboxylic acid (e.g., Versatic 10) at pH 6.2–7.0. Coextracted calcium
and magnesium are scrubbed using the loaded strip solution and report to
the raffinate. The nickel is finally stripped from the organic phase with
recycled electrolyte from the nickel electrowinning* circuit. Coextracted
impurities are removed from the cobalt circuit initially by sulfide pre-
cipitation of cobalt, copper, and zinc from the loaded strip liquor. The
sulfide cake is then redissolved in acid and the zinc extracted with DEHPA,
the copper removed by resin ion exchange and the cobalt electrowon.
Problems associated with this flow sheet include cross contamination
of the organic phases and aqueous solubility losses of the Versatic acid.
11.3.2.1d Cawse Process
Purification of the pressure acid leachate from treatment of the Cawse
limonitic ore involves the precipitation of cobalt and nickel hydroxides with
magnesium oxide following prior neutralization with limestone to remove
iron and other impurities (Fig. 11.6). The hydroxide cake is selectively
leached with ammonia, rejecting any iron, manganese, or magnesium but
redissolving any copper and zinc in the precipitate. The cobalt is then oxi-
dized to cobalt(III) prior to extraction, to prevent coextraction with nickel
with the hydroxyoxime LIX 84I (Cognis). The nickel is stripped with dilute
sulfuric acid and electrowon. Copper and some cobalt build up in the
organic phase and these levels are controlled by including a bleed stream,
*Electrowinning is an electrolytic process where cathodic reduction is used to recover metal from
an electrolyte using an inert anode. The process differs from electro-refining where the anodic
reaction is dissolution of the metal, that is then reversed at the cathode.
Fig. 11.5 Bulong process flow sheet (simplified).

470 Cox
Fig. 11.6 Cawse process flow sheet (simplified).

from which the copper is removed with a more concentrated acid strip and
cobalt by stripping in the presence of a reducing agent. One other major
problem is the oxidative degradation of the oxime extractant by the
cobalt(III) requiring the addition of a reoximation step to maintain ex-
tractant capacity. The cobalt in the raffinate is recovered by sulfide pre-
cipitation. The Cawse mine produced 8500 t/a nickel and 1900 t/a cobalt in
2000 at a cost in 1999 of $0.11 kg1 after cobalt credits, the lowest cost
production in the world.
11.3.2.1e Murrin-Murrin Process
The Murrin-Murrin mine is also near Kalgoorlie in Western Australia.
This is the largest of the three projects, with a first-phase production of
45,000 t/a nickel and 3000 t/a cobalt. Future expansion has been suggested
to 115,000 t/a nickel and 9000 t/a cobalt. The pressure acid leachate, fol-
lowing removal of the insoluble residue, is treated with hydrogen sulfide
under pressure to precipitate nickel and cobalt and reject impurity ele-
ments. The sulfides are then dissolved in a pressure oxygen leach to pro-
duce a sulfate liquor that is treated with CYANEX 272. Initially zinc is
removed at low pH, followed by a pH adjustment and extraction of cobalt
(Fig. 11.7). Any coextracted nickel is scrubbed from the loaded organic
phase with a dilute cobalt sulfate solution and the cobalt stripped with
dilute sulfuric acid. Both the cobalt and nickel are recovered from solution
as powders by hydrogen reduction that also produces ammonium sulfate
as a by-product. The main problem in this flow sheet has been the crys-
tallization of nickel ammonium sulfate because of the high nickel con-
centration (100 g dm3) in the feed to extraction. This severely reduced the
production from the plant and the problem was solved by preloading the
extractant with nickel prior to extraction by contact with the nickel sulfate
raffinate.
11.3.3 Summary
Cobalt and nickel can be separated by several solvent extraction processes,
the choice depending on the total flow sheet requirements, on the ratio of the
metals in the feed solution, and on the required purity of the products.
The various systems can be summarized as follows:
1. Extraction of cobalt from chloride media by amines
2. Extraction of cobalt by organophosphorus acids, especially CYANEX
272, from acidic solutions
3. Extraction of nickel after prior oxidation to cobalt(III) using hydro-
xyoximes
4. Extraction of nickel by mixtures of simple oximes and carboxylic acids
472 Cox
Fig. 11.7 Murrin-Murrin process flow sheet (simplified).

11.4 COPPER
There has been a long history of the extraction of copper by a hydro-
metallurgical route, with the Rio Tinto process being operated in Spain
from the mid-18th century. In the 19th century, the use of acid leaching of
low-grade oxidized ores was practiced in Russia and the United States, with
the resulting copper solution being reduced to metal by scrap iron. This
process of leaching–cementation requires an adequate supply of acid and
iron, and it produces only a low-grade copper powder that requires further
treatment by smelting before marketing. The introduction of electrowinn-
ing of copper from acid solution provided a number of advantages over
cementation [2] in that the product, cathode copper, had a much higher
value than the cement copper and also, during electrowinning some of the
sulfuric acid was regenerated for leaching. However, the demands placed on
the feed liquor to electrowinning were severe enough to limit the intro-
duction of the leach-electrowin route. Since high concentrations of other
metals such as iron and magnesium have effects on the current efficiency
and nature of the copper deposit, concentrations of these and several other
elements should be as low as possible. This is difficult for iron and mag-
nesium, in particular, because of their wide occurrence in mineral deposits.
Solvent extraction provides a possible process whereby a pure feed to
electrowinning could be obtained if a suitable extractant for copper was
available.
Until recently, the favored leaching acid had been sulfuric acid,
in which the metals exist as cationic species in solution. Examination
of selectivity series for the common extractants available in the 1950s
(Fig. 11.8) indicates that separation of copper from magnesium, for
example, would be easily achieved by carboxylic or phosphoric acids, but
with both these extractants, iron(III) would also be extracted with the
copper. Although it would be possible to precipitate iron as the hydroxide
by addition of alkali, this would complicate the process. However, as
copper extraction does not occur below pH 3, the addition of alkali is
necessary to achieve optimum extraction. This increases the overall costs
of the extraction circuit and prevents the direct return of the extraction
raffinate to leaching. Thus, it was obvious that to take advantage of the
benefits of the process, new extractants that could selectively extract
copper from iron at the pH of the leach liquor would be required. The
search for such reagents was based on extractants used in analyti-
cal chemistry, such as benzoin oximes, 8-quinolinol, and b-diketones (see
Table 11.3). The use of these reagents in commercial operations is
described individually.
474 Cox
11.4.1 Hydroxyoxime Extractants

The first commercial reagents designed specifically for copper extraction
were an aliphatic a-hydroxyoxime (LIX63) and a b-hydroxyoxime based on
benzophenone (LIX64). LIX63, with pH50 values of 3–4, was unable to
extract copper from commercial leach liquors without alkali addition. Also,
this reagent extracted iron(III) at lower pH values than copper. These dis-
advantages were largely overcome by LIX64, which was able to extract
copper selectively from iron in the pH range 1.5–2.5, but with the dis-
advantage of a slow rate of extraction. This was solved by the addition of a
small amount of LIX63. The mixture, LIX64N, for many years was the
extractant of choice for commercial copper extraction from acid solution.
The extracted complex had a 2:1 oxime/metal stoichiometry with the two
oxime molecules linked by hydrogen bonds to give an overall planar con-
figuration (Fig. 11.9). The copper could be stripped with sulfuric acid (150–
200 g dm3), thus interfacing with spent electrolyte from the electrowinning
Fig. 11.8 Extraction of metals from sulfate media by DEHPA. (From Ref. 4.)
Fig. 11.9 Structure of bis(5-methylsalicyldoximato)copper(II).
circuit. Because of the commercial success of these reagents, other manu-

facturers produced modifications of the basic b-hydroxyoxime structure with
slightly different characteristics. Accordingly, chlorination of the benzene
ring gave a reagent (LIX70) capable of extracting copper at pH 1, but re-
quired concentrated acid (>300 g dm3) for stripping. Other developments
centered on modification of the oximic side chain from phenyl (LIX64,
LIX65N) to methyl (SME529 Shell) and hydrogen (P50 ICI/Acorga),
which varied both equilibrium and kinetic performance. The importance of
these reagents in the development of large-scale commercial operations is
described in several texts [1–3]. These references also contain detailed
descriptions of some of the commercial operations.
More recently, a series of second-generation commercial reagents have
become available that have been designed to produce a solvent phase to
match the requirement of particular feedstocks or equipment. It is now
generally accepted that the aldoximes, based on salicylaldehyde oxime, have
a number of advantages in terms of fast extraction kinetics and high
extractant strength, allowing the extraction of copper from highly acidic
leachate with good selectivity over iron(III). However, increasing the ex-
tractant strength increases the acid strength required for stripping, which
reduces the overall copper transfer when interfaced with a normal tank-
house spent electrolyte. Two approaches have been proposed to overcome
these difficulties. Avecia with their Acorga* reagents have showed that
addition to the hydroxyoxime of nonyl phenol (P5000 series) or tridecanol
(PT5050) improves the stripping performance, with little effect on extrac-
tion. More recently, other reagent modifiers, such as long-chain alkyl esters,
have been added to P50 to produce the M series (Acorga), which are claimed
to show improved commercial performance by reduction of entrainment
and crud formation [16,17]. An alternative to this approach developed
by Cognis involves a 1:1 mixture of an aldoxime and ketoxime, thereby
*Acorga is a trademark of Avecia, previously ICI, now owned by Cytec (2003).

476 Cox
combining the extractive strength, selectivity, and fast kinetics of the

aldoxime with the proved hydrolytic stability and stripping performance of
the ketoximes. Both reagents seem to largely retain their individual char-
acteristics in the mixture. The acetophenone oxime LIX84 is used in these
mixtures (see Table 11.3). The aldoxime chosen has a 5-dodecyl ring sub-
stituent that is reported to be more stable than the corresponding 5-nonyl
derivative [18]. An added commercial benefit of these second-generation
reagents has been the ability to use them in the latest flow sheet designs with a
general change of circuit configuration from the three extraction–two strip of
the 1970s, to either a two extraction–two strip for copper feed concentrations
>3 g dm3 or two extraction–one strip for feeds 0.5–3 g dm3. This change
has gradually occurred as operators have gained experience in running large
extraction circuits and have become more concerned over capital expenditure
than running costs. Table 11.5 shows the 2001 worldwide production capa-
city of copper solvent extraction plants, and approximately 30% of copper
production now uses the leach–solvent extraction–electrowinning process. It
is interesting to note the number of plants with a capacity >100,000 tons that
have been constructed in Chile since 1990. The total production capacity in
Chile of 1,430,000 tons represents 56% of the world’s capacity for leach–
solvent extraction–electrowon copper.
An alternative circuit for copper extraction involves ammoniacal
leaching, and hydroxyoximes have been studied for this application [2].
However, because of their tendency to extract ammonia, these reagents have
been superceded by b-diketones.
11.4.2 Quinoline-Based Extractants

A 7-alkyl-substituted derivative of 8-quinolinol (Kelex 100, Schering; see
Table 11.3) was also designed for the selective extraction of copper from
iron(III) in acid media. Here, the separation was based on the slow
extraction of iron(III) relative to copper, presumably related to steric hin-
drance induced by substitution in the 7-position. With the short contact
times and low pH values found in copper extraction circuits, separation
factors >1000 were obtained. If the extraction was run to equilibrium, then
iron(III) was preferred to copper. The reagent had an advantage over
hydroxyoximes in its ability to extract copper from high-concentration,
high-acid feeds, and it did not require high acid concentrations for stripping.
However, unfortunately the quinolinic nitrogen atom is readily protonated
and thus takes up acid on stripping, which either would be transferred to the
extraction circuit or would have to be removed by a water wash, thereby
complicating the circuit. Therefore, it is unlikely that Kelex 100 would be
used for commercial copper extraction.
Table 11.5 Production Capacity of Acid Leach Copper SX-EW Plants (2001)a
Capacity
Name / location Leaching method (tons / year) Startup
USA
Nord Copper Johnson Camp Heap (oxide) 3,000 1990
Arimetco, NVb Heap (oxide) 2,500 1989
Asarco Ray, AZ Dump/Heap (oxide/sulf ) 41,000 1979
Asarco Silver Bell Heap/Dump (oxide/sulf ) 20,000 1997
BHP, Miami, AZ In situ/Agit. (sulf/tails) 12,500 1976
BHP, Pinto Valley, AZ Dump (oxide/sulf ) 8,750 1981
BHP, San Manuel Heap/In situ (oxide) 18,000 1986
Burro Chief, Tyrone, NM Heap/Dump (oxide/sulf ) 71,500 1984
Cyprus Bagdad. AZ Dump (oxide/sulf ) 14,000 1970
Cyprus Miami, AZ Heap (oxide/sulf ) 76,000 1979
Cyprus Sierrita, AZ Dump/Heap (oxide/sulf ) 26,000 1987
Cyprus Tohono, AZb In situ (oxide) 21,900 1981
Phelps Dodge, Chino, NM Heap (oxide/sulf ) 59,700 1988
Phelps Dodge, Morenci, AZ Heap/Dump (oxide/sulf ) 355,000 1987
Equitorial Mineral Park, AZ Dump (oxide/sulf ) 2,500 1992
Equatorial Tonopahb Heap (sulfide) 26,000 2000
Peru
Cyprus Cerro Verde Heap (oxide/sulf ) 73,000 1977
Southern Peru (Toquepala) Dump/Heap (oxide/sulf ) 56,000 1995
Centromin Mine Water (oxide/sulf ) 5,000 1980
Mexico
Cananea I and II, Mexico Dump (sulfide) 40,000 1980
La Caridad, Mexico Dump (oxide/sulf ) 24,000 1994
Canada
Gibraltar, Canadab Dump (oxide/sulf ) 5,475 1986
Africa
ZCCM, Zambia Agitation (oxide/tails) 90,000 1974
Sanyati Project (Reunion) Heap (oxide) 3,500 1995
Bwana Mkubwa Agitation (tails) 10,000 1998
Australia
BHAS Agitation (sulfide) 4,380 1984
Girilambone Heap (sulfide) 20,000 1993
Gunpowderc Heap (sulfide) 1990
Mt. Gordon Pressure (sulfide) 45,000 1998
Olympic Dam Agitation (oxide) 21,900 1989
Nifty Heap 16,500 1993
Cloncurryb Heap (oxide/sulf ) 1996
Murchison United (Mt.Cuthbert) Heap (oxide/sulf ) 5,500 1996
(continued)
478 Cox
Table 11.5 (continued)

Capacity
Name / location Leaching method (tons / year) Startup
Chile
Andacollo, Chile Heap (oxide/sulf ) 20,000 1996
Punta Del Cobre (Biocobre) Heap (oxide/sulf ) 8,000 1993
Cerro Colorado Heap (oxide/sulf ) 100,000 1994
Cerro Dominador Heap (oxide) 1,000 1999
Chuquicamata Ripios Plant Dump/Vat (oxide) 171,000 1988
Chuquicamata LGS Dump (sulfide) 20,000 1993
Collahuasi Heap (oxide) 50,000 1998
Dos Amigos, Chile Heap (sulfide) 10,200 1997
El Abra, Chile Heap (oxide) 220,000 1996
El Soldado Heap (oxide) 7,500 1992
El Salvador I and II Heap (oxide/sulf ) 25,000 1994
El Teniente In situ (sulfide) 5,000 1986
Escondida Oxide Heap (oxide) 125,000 1998
Ivan Zar Heap (oxide/sulf ) 10,000 1995
Lomas Bayas Heap (oxide) 60,000 1998
Los Bronces Heap (oxide) 5,000 1998
Mantoverde Heap (oxide) 42,000 1995
Michilla Lince I and II Heap (oxide) 50,000 1992
Quebrada Blanca Heap (sulfide) 75,000 1994
Radomiro Tomic Heap (oxide) 260,000 1997
Santa Barbara, Chile Vat (oxide) 42,000 1995
Tocopilla Heap (oxide) 3,500 1984
Zaldivar Heap (oxide/sulf ) 120,000 1995
a
Data kindly supplied by Dr. G. A. Kordosky (Cognis).
b
Shutdown.
c
Partially incorporated into Mt. Gordon.
Another commercial reagent based on quinoline is LIX34, which is a

derivative of 8-alkylarylsulfonamidoquinoline. It has good extractant and
stripping performance and, unlike Kelex 100, does not protonate easily.
However, the cost of production outweighed these advantages.
11.4.3 b-Diketone Extractants

Although b-diketones cannot compete with hydroxyoximes for the extrac-
tion of copper from acid media, they do have advantages for extraction from
ammoniacal leach solutions because, unlike hydroxyoximes, they do not
transfer ammonia. The extraction follows Eq. (11.6), and the extent of
extraction depends on the concentration of ammonia and ammonium salts.
CuðNH3 Þ2þ
4 þ 2HA ! CuA2 þ 2NHþ
4 þ 2NH3 (11.6)
Kinetics of extraction are fast and stripping is possible with solutions of low
acidity and high copper concentration, as found with spent tankhouse
liquors. A process using the reagent LIX54 for the recovery of copper from
printed circuit board etch liquors has been very successful commercially (see
Chapter 14). With these reagents, a higher copper transfer is achieved than
with the hydroxyoximes, as illustrated by a tenfold reduction in solvent
extraction plant size when changing from LIX64 to LIX54 for a leachate
from a lead dross process. Some problems have been found with the long-
term use of LIX54 in ammoniacal circuits due to the formation of a ketimine
[Eq. (11.7)] that reduces the capacity of the LIX54 and could not be ade-
quately removed by a normal clay cleanup treatment. Only one compound
was identified, suggesting that the reaction could be inhibited by steric
hindrance about the carbonyl group. Synthesis of compounds where the a-
hydrogen atoms on the C7H15 group were substituted by alkyl groups
removed this problem.
C6 H5 COCH2 COC7 H15 þ NH3 !C6 H5 COCH2 CðNHÞC7 H15 (11.7)
11.4.4 Nitrogen-Based Extractants

Recent interest in chloride hydrometallurgy in the processing of complex
sulfide ores has resulted in the development of a reagent to remove copper
from such leachates. Here, there is a requirement for an extractant to
operate without pH control and with a high selectivity over a wide range of
metals and metalloids. Ion-pair extractants, the normal choice for chloride
media [see Eq. (11.2)], are not suitable for this application, as copper has a
lower tendency to form chloroanions than other elements in the leachate,
such as iron(III), zinc, and lead. A reagent based on a pyridine carboxylic
ester derivative operates as a solvating reagent for CuCl2, according to Eq.
(11.8).
CuCl2 þ 2B ! CuCl2 B2 (11.8)
Unlike the amine extractants, this compound does not protonate easily and
provides excellent selectivity over those metals and metalloids that exist as
chloroanions in this medium. Extraction of the latter species becomes sig-
nificant only above 8 mol dm3 chloride, as protonation becomes significant.
This extractant [Acorga CLX 50], which contains 50% of the active ingre-
dient, has been proposed for an integrated leach, extraction, electrowinning
circuit (Cuprex process) [19].
480 Cox
11.5 PRECIOUS METAL EXTRACTION

The traditional route for the extraction and separation of precious metals
consists of a series of selective dissolution–conditioning and precipitation
steps that are generally inefficient in terms of the degree of separation that
can be achieved. Even when conditions allow the complete precipitation of
the desired element from solution, the precipitate will still be contaminated
by entrained mother liquor, thereby requiring extensive washing, which
generates wash streams containing small amounts of these valuable ele-
ments. Consequently, the traditional route required multiple precipitation
and redissolution steps to provide the desired product purity, which, with
the consequent wash streams, involved increased recycling and retreatment.
This is not only very labor-intensive, but it also locks up considerable
amounts of these elements, which, in turn, places an economic constraint on
the whole operation. In some circumstances, a year can elapse between
dissolving a sample of scrap at the start of the flow sheet and the emergence
of the pure elements for marketing. During this time, the value of the ele-
ments is locked up in the plant and, hence, becomes a financial liability on
the operation.
To overcome some of these problems and to produce a flow sheet that
could accept a wide range of feed materials, the major refining companies
embarked on research programs that could replace the traditional routes.
These have led to the production of three overall flow sheets that reflect the
chemical complexity of these elements and the difference in the raw material
feed to the different refineries. To understand these schemes, it is necessary
to outline some of the fundamental chemistry of the precious metals.
These elements are noble metals and, as such, can be dissolved only
with great difficulty. The usual leaching agent is hydrochloric acid, with the
addition of chlorine to increase the solution oxidation potential. This strong
chloride medium results in the almost exclusive formation of aqueous
chloroanions, with, under certain circumstances, the presence of some
neutral species. Very seldom are cationic species formed in a chloride
medium. However, these elements do possess a range of easily accessible
oxidation states and, with the possibility of a number of different anionic
complexes that are dependent on the total chloride concentration, this
provides a very complicated chemistry. A summary of the most important
chloro complexes found in these leach solutions is given in Table 11.6, from
which the mixed aquochloro and polynuclear species have been omitted.
The latter are found especially with the heavier elements.
The chloro species also differ in their rates of substitution, which, in
general terms, are slower than those of the base metals and follow the order
Au(I), Ag(I)Pd(II)>Au(III)>Pt(II)>Ru(III)Rh(III)>Ir(III)>Os(III)
Table 11.6 Common Chloro Species of the Precious Metals
Oxidation Major
Metal state chloro species Comments
Silver (Ag) I AgCl Insoluble
AgCl2 High HCl concentration
Gold (Au) I AuCl2 Very stable
III AuCl4
Platinum (Pt) II PtCl42 Conversion IV to II slow
IV PtCl62 Most common species, kinetically inert
Palladium (Pd) II PdCl42 Most common
IV PdCl62 Conversion II to IV difficult
Iridium (Ir) III IrCl63 Both species stable and
IV IrCl62 conversion IV to III easy
Rhodium (Rh) III RhCl63
Ruthenium (Ru) III RuCl63 Complex equilibria between III and IV
IV RuCl62 depends on redox potential and
chloride concentration
Osmium (Os) III OsCl63
IV OsCl62 Os(IV) more stable than Ru(IV)
Ir(IV), Pt(IV), with the states from Rh(III) being termed inert. Thus,
kinetic factors tend to be more important, and reactions that should be
possible from thermodynamic considerations are less successful as a result.
On the other hand, the presence of small amounts of a kinetically labile
complex in the solution can completely alter the situation. This is made even
more confusing in that the basic chemistry of some of the elements has not
been fully investigated under the conditions in the leach solutions. Conse-
quently, a solvent extraction process to separate the precious metals must
cope with a wide range of complexes in different oxidation states, which
vary, often in a poorly known fashion, both in kinetic and thermodynamic
stability. Therefore, different approaches have been tried and different flow
sheets produced.
It can be seen that the three commercial flow sheets (Figs. 11.10–11.12)
do have common features. Hence, because gold and silver can be easily
separated from the remaining precious metals, they tend to be removed from
the flow sheet at an early stage. In addition, significant quantities of gold
and silver exist, in either primary ores or waste materials that are not
associated with the platinum group metals. Silver often occurs as a residue
from the chloride leaching operation and so is removed first, followed by
gold, which can be extracted as the tetrachloroauric(III) acid, HAuCl4, by a
solvating reagent.
482 Cox
Fig. 11.10 INCO flow sheet for extraction of precious metals.

Fig. 11.11 Matthey Rustenberg flow sheet for extraction of precious metals.
484 Cox
Fig. 11.12 Lonrho flow sheet for extraction of precious metals.

Notice that none of the flow sheets uses solvent extraction exclusively.
Because the aqueous chemistry of osmium and ruthenium is very complex,
most operators remove these elements by distillation of the tetraoxides,
MO4. Also, it has been advantageous to use ion exchange to separate and
concentrate rhodium. The various extraction routes for individual elements
are discussed in the following sections.
11.5.1 Gold
Gold exists in the leach solution as AuCl 4 , and all the processes rely on the
extraction of this ion or the parent acid by ion-pair or solvating extractants,
respectively. The early studies by Morris and Khan [20] led to the adoption
of dibutylcarbitol (diethyleneglycol dibutylether) by INCO in a gold refinery
several years before the complete solvent extraction refining flow sheet was
produced. The extractant has a high selectivity for gold (Fig. 11.13) and
coextracted elements can be easily scrubbed with dilute hydrochloric acid
[22]. Because of the high distribution coefficient, recovery of gold is best
achieved by chemical reduction. Thus, oxalic acid at 908C produces easily
filtered gold grain. However, the slow kinetics necessitate a batch process.
An alternative solvating extractant, 4-methyl-2-pentanone (methyl iso-
butylketone; MIBK) is used by Matthey Rustenberg Refiners (MRR). This
has a lower selectivity for gold (Fig. 11.14), so a simple acid scrub is not
sufficient to remove impurities [22]. However, there is an advantage to the
overall flow sheet in that some of the base elements, such as tellurium,
arsenic, and iron, are removed at an early stage. Again, gold is recovered by
chemical reduction with, for example, iron to give an impure product for
subsequent refining. A third type of solvating extractant, long-chain alco-
hols have also been proposed, the most effective being 2-ethylhexanol,
which can be easily stripped at ambient temperature with water [24]. Gold,
like silver, being classified as a soft acid, should prefer interaction with soft
donor atoms like sulfur. Therefore, it is no surprise to find that extraction
with the trialkylphosphine sulfide, CYANEX 471X, is very efficient and
provides excellent separation from elements such as copper, lead, bismuth,
and other metalloids found in anode slimes [22]. Stripping can be achieved
using sodium thiosulfate, which allows the separation of recovery and the
reduction to the metal.
Amine salts can also be used to extract anionic species, but, in general,
these are less selective for gold. However, Baroncelli and coworkers [25]
have used trioctylamine (TOA) to extract gold from chloroanions of, for
example, copper, iron, tin, and zinc, in aqua regia solution. The selectivity is
attributed to the presence of TOA HNO3 in the organic phase. Coextracted
metals are scrubbed with dilute nitric acid, and gold is recovered by
486 Cox
Fig. 11.13 Extraction of elements from a chloride medium by dibutylcarbitol. (From

Ref. 21.)
a hydrochloric acid-thiourea solution. The same authors also used N,N0 -di-
n-butyloctanamide as a solvating extractant with aqua regia solutions. The
protonated forms of Kelex 100 and LIX26, both acting as ion pair extrac-
tants, have been proposed by Demopoulos et al. [26] for gold extraction,
with recovery as gold grain by hydrolytic stripping.
Alkaline cyanide solutions are another common lixiviant for gold,
giving complex cyano species in solution. These anionic complexes may be
extracted from aqueous solution by amines as ion pairs [27,28]. Recently,

another extractant has been developed specifically to extract gold and silver
from these media. The reagent, LIX79, is a trialkylated guanidine (Fig.
11.15) that can extract both metals rapidly from solutions within the
approximate pH range 9–11.5 (gold) and 9–10 (silver). At higher pH values
the metals are stripped from the organic phase as the guanidine becomes
deprotonated [29,30].
Environmental concern over the use of cyanide leaching has stimulated
the development of several alternative leaching processes. Of those tested,
thiosulfate is the most promising, although the cost of this reagent is a dis-
advantage. Once again the gold is present in the leachate as an anionic
Fig. 11.14 Extraction of elements from chloride medium by MIBK. (From Ref. 23.)
488 Cox
Fig. 11.15 Structure of LIX79.
complex, [Au(S2O3)2]3, so extraction is possible with either amines, at

pH <8.0, or solvating reagents, e.g., TBP (pH >10) [31]. It was also found
that a mixture of these reagents gave some synergism over the pH range 9–11.
11.5.2 Silver
Although silver is not treated by solvent extraction in any of the flow sheets,
silver is recovered from aqueous solution in several other situations. For
these processes, Cytec developed reagents with donor sulfur atoms to extract
this ‘‘soft’’ element. For example, tri-isobutylphosphine sulfide (CYANEX
471X) extracts silver from chloride, nitrate, or sulfate media selectively from
copper, lead, and zinc [32]. The silver is recovered from the loaded organic
phase by stripping with sodium thiosulfate, and the metal recovered by
cementation or electrolysis. Silver can also be extracted from chloride
solution by a dithiophosphinic acid (CYANEX 301) [33].
11.5.3 Palladium and Platinum

In addition to the presence of these elements in ores, they are also available
from recycled feeds, such as catalyst wastes, and as an intermediate bulk
palladium–platinum product from some refineries. The processes that have
been devised to separate these elements rely on two general routes: selective
extraction with different reagents or coextraction of the elements followed
by selective stripping. To understand these alternatives, it is necessary to
consider the basic solution chemistry of these elements. The two common
oxidation states and stereochemistries are square planar palladium(II) and
octahedral platinum(IV). Of these, palladium(II) has the faster substitution
kinetics, with platinum(IV) virtually inert. However even for palladium,
substitution is much slower than for the base metals so long as contact times
are required to achieve extraction equilibrium.
Two commercial flow sheets (Figs. 11.10 and 11.11) use either di-n-
octyl sulfide or a hydroxyoxime as selective extractants to remove the pal-
ladium, followed by tri-n-butylphosphate (TBP; INCO) and an amine
(MRR) to recover platinum. The extraction of palladium by the sulfide is

slow with 2–3 hr to achieve equilibrium, which inhibits flow sheet design.
However, it has been shown that tertiary amines can act as a kinetic accel-
erator, reducing equilibrium times to minutes, but no information has been
published on the effect of these amines on palladium–platinum separation.
High loading of the sulfides is possible, and stripping is easy with aqueous
ammonia. The overall extraction and stripping equations are as follows:
4 þ 2R2 S
PdCl2 ! ðR2 SÞ2 PdCl2 þ 2Cl (11.9)

ðR2 SÞ2 PdCl2 þ 4NH3 ! PdðNH3 Þ2þ
4 þ 2R2 S þ 2Cl (11.10)
Neutralization of the strip solution with hydrochloric acid gives Pd(NH3)2-
Cl2 as product. One of the problems that has emerged is the formation of di-
n-hexylsulfoxide [34] by oxidation of the sulfide. This may cause several
problems including extraction of iron(III) that is strongly dependent on the
HCl concentration. The iron can easily be stripped by water. There have also
been indications of a buildup of rhodium in the extract phase that again can
be explained by the extraction of anionic rhodium species by the sulfoxide.
One benefit from the presence of the sulfoxide is that the rate of palladium
extraction is increased by the presence of the protonated sulfoxide at high
acidities; however, this kinetic enhancement is less that found with
TOA HCl, which remains protonated even at low acidities.
In this flow sheet, the aqueous raffinate from extraction is acidified to
5–6 mol dm3 with hydrochloric acid to optimize platinum extraction by the
solvating extractant TBP. The coextraction of iridium is prevented by
reduction with sulfur dioxide, which converts the iridium(IV) to the (III)
species, which is not extractable. Once again, kinetics are a factor in this
reduction step because, although the redox potentials are quite similar,
[Ir(IV)/(III) 0.87 V; Pt(IV)/(II) 0.77 V], iridium(IV) has a relatively labile
d 5 configuration, whereas platinum(IV) has the inert d 6 arrangement. The
species H2PtCl6 is extracted by TBP, from which platinum can be stripped
by water and recovered by precipitation as (NH3)2PtCl2.
The second selective extractant route (that of MRR) uses a hydroxy-
oxime to remove palladium. The actual compound used has not been spe-
cified, but publications refer to both an aliphatic a-hydroxyoxime and an
aromatic b-hydroxyoxime. The a-hydroxyoxime LIX63 has the faster
extraction kinetics, but suffers from problems with stripping. For the b-
hydroxyoxime, a kinetic accelerator in the form of an amine (Primene
JMT*) has been proposed. The precise mode of operation of this accelerator
is unknown, but it may be a similar process to that proposed for the sulfide
*Trademark of Rohm and Hass.

490 Cox
system [i.e., a form of proton-transfer catalysis where the protonated amine

extracts PdCl42, which then reacts with the hydroxyoxime in the organic
phase: Eq. (11.11)].
4 þ R3 NHCl
PdCl2 ! ðR3 NHÞ2 PdCl4 þ 2HA
! PdA2 þ 2R3 NHCl þ 2Cl (11.11)
Good loading of palladium and palladium–platinum separation fac-
tors of 103–104 are found. Stripping of the palladium is easy with hydro-
chloric acid, and (NH3)2PdCl4 is obtained by precipitation. The raffinate
from palladium extraction (see Fig. 11.10) is treated to remove osmium and
ruthenium and, again, sparged with sulfur dioxide to reduce iridium(IV),
following PtCl62 species removed by an amine. Stripping of the platinum is
difficult because of the high-distribution coefficient, but techniques that
have been successful include the use of strong acids, such as nitric or per-
chloric, to break the ion pair, or alkali to deprotonate the amine salt. The
choice depends very much on the amine chosen; for example, if a quaternary
amine is used, the deprotonation reaction is not available, and a strong acid
has to be used. On the other hand, with other amines, the possibility of
+4(6x)
forming inner sphere complexes such as Pt(RNH2)xCl(6x) is an
important consideration, as these are nonextractible and cause a lockup of
platinum in the organic phase. The formation of such complexes is enhanced
by a deprotonation step that releases the free amine into the organic phase.
As the ease of formation of platinum–amine complexes decreases in the
order primary>secondary>tertiary, the reason for the choice of the latter
for platinum extraction can be understood.
The phosphine sulfide CYANEX 471X has been proposed as an
alternative to the foregoing reagents for palladium–platinum separation.
Here, the faster extraction of palladium affords separation factors of 10:1,
with a 10 min contact time. The chemistry of the process is similar to that of
the alkyl sulfides. Various other separations have been published; for
example, TBP will extract platinum(IV) chloroanions preferentially to pal-
ladium(II).
An alternative general process involves coextraction of palladium and
platinum, followed by selective stripping (see Fig. 11.12). A novel amino acid
extractant, made by the reaction between chloracetic acid and Amberlite*
LA-2, a secondary amine, is used to extract the two elements from the leach
liquor. However, in the given flow sheet, no selective stripping is employed,
both elements being stripped by hydrochloric acid. The resulting chlor-
oanions are then separated by using di-n-hexylsulfide to extract PdCl42.
*Trademark of Rohm and Hass.

Other coextraction-selective strip routes that have been proposed

include the following:
1. Extraction with a secondary amine, followed by stripping of the
palladium by hydrazine, and bicarbonate to strip platinum by deprotona-
tion of the amine ion pair:
ðR3 NH2 Þ2 PdCl4 þ 2N2 H4 ! PdðN2 H4 Þ2 Cl2 þ 2R3 NH2 Cl
(11.12)
ðR3 NH2 Þ2 PtCl6 þ 2HCO

3 ! PtCl2
6 þ 2R3 NH þ 2H2 O þ 2CO2
(11.13)
2. Extraction with a tertiary amine, Alamine* 336, followed by
palladium stripping with thiourea (Tu), and platinum with thiocyanate:
ðR3 NH2 Þ2 PdCl4 þ 4Tu ! PdðTu2 ÞCl2 þ 2R3 NHCl (11.14)
ðR3 NHÞ2 PtCl6 þ 2NCS ! PtCl2

6 þ 2R3 NHNCS (11.15)
However, both of these above systems suffer from the possibility of lockup
of metal in the organic phase, with the formation of unstrippable complexes,
problems with stripping and metal recovery from the strip solutions.
3. Extraction with 8-quinolinol derivatives, which can act as either a
chelating acid or, after protonation, as an anion extractant. It was observed
[35] that palladium extraction was favored at low acidity, but platinum was
extracted best at high acidity (Fig. 11.16). At about 2 mol dm3 acid, the two
metals were extracted equally. This suggested the following extraction
equations:
4 þ 2H2 ACl
PdCl2 ! PdA2 þ 6Cl þ 4Hþ (11.16)
6 þ 2H2 ACl
PtCl2 ! ðH2 AÞ2 PtCl6 þ 2Cl (11.17)
The reaction to form the palladium complex is similar to that reported for
amine salts, although here, because a bidentate chelating ligand is used, no
chlorine atoms are retained in the complex, and the system is easy to strip.
Also, as both reactions involve initial ion pair extraction, fast kinetics are
observed with 3–5 min contact time to reach equilibrium at ambient tem-
perature. The extraction conditions can be easily adjusted in terms of acidity
to suit any relative metal concentrations and, because the reagent is used in the
protonated form, good selectivity over base metals, such as iron and copper,
*Trademark of Cognis.
492 Cox
Fig. 11.16 Extraction of platinum(IV) and palladium(II) by LIX26 (5%), isodecanol

(5%) in Solvesso (90%). (From Ref. 35.)
which must be extracted by the chelate mechanism, is found. Differential

stripping is possible from the loaded organic phase by variation of the acidity
of the strip solution; that is, dilute solutions to release platinum and more
concentrated for palladium, and separation factors >200 can be obtained.
4. Finally, another development is the use of N-alkylamides that
selectively extract platinum from strong acid media, presumably by ion pair
formation between the protonated amide and the chloroanions:
RCONHR0 þ Hþ ! RCONH2 R0þ (11.18)
2RCONH2 R0þ þ PtCl2

6 ! ðRCONH2 R0 Þ2 PtCl6 (11.19)
Stripping is easy with dilute acid or water, which deprotonates the amide
and releases the metal anion. The process favors metals in high oxidation
states [e.g., platinum(IV) or iridium(IV)], so that reduction of any iridium
must precede the extraction of platinum and palladium, as with other sys-
tems. The platinum–palladium separation factors depend on the amide
structure, but are rather small, about 25–55, so that scrubbing of the loaded
organic phase is required to give a good separation [36].
11.5.4 Indium and Rhodium

In the commercial flow sheets, these elements are left in the aqueous raffi-
nate after platinum and palladium extraction. Indium can be extracted in
the +IV oxidation state by amines (see Fig. 11.11), or TBP (see Figs. 11.10
and 11.12). However, although the separation from rhodium is easy, the
recovery of iridium may not be quantitative because of the presence of
nonextractable iridium halocomplexes in the feed solution. Dhara [37] has
proposed coextraction of iridium, platinum, and palladium by a tertiary
amine and the selective recovery of the iridium by reduction to Ir(III).
Iridium can also be separated from rhodium by substituted amides [S(Ir/
Rh): 5 105).
Rhodium is recovered in all the examples by precipitation from the
raffinate after ion exchange, but solvent extraction with an amine followed by
thiourea stripping has been suggested [37]. The problem in these systems is
that extraction is favored by low acidity and low chloride concentrations so
under the conditions found in the commercial flow sheets rhodium extraction
would be low [38,39]. In addition, stripping from the organic phase is difficult
with the possibility that the rhodium remains in the organic phase as the
dimer, Rh2Cl93. Extraction with 8-hydroxyquinoline extractants has also
been studied. Here extraction increased with acidity and chloride con-
centration to a maximum of 40%. This is explained by the formation of
the RhCl63 species that is then extracted as an ion pair by the protonated
8-hydroxyquinoline in a similar way to palladium discussed earlier [40].
However, once again stripping was a problem. A critical review explaining
these problems in more detail has been published [41].
11.6 LANTHANIDE EXTRACTION

The lanthanide group of elements (Table 11.7) is very difficult to separate by
traditional methods because of their similar chemical properties. The tech-
niques originally used, like the precious metals, included laborious multiple
fractional recrystallizations and fractional precipitation, both of which
required many recycle streams to achieve reasonably pure products. Such
techniques were unable to cope with the demands for significant quantities of
certain pure compounds required by the electronics industry; hence, other
separation methods were developed. Resin ion exchange was the first of these
494 Cox
Table 11.7 Properties of Lanthanide Elements

Ionic radius
Element Atomic number Atomic mass (Ln3 + ) (Å)
Lanthanum (La) 57 138.9 1.061
Cerium (Ce) 58 140.1 1.034
Praesodymium (Pr) 59 140.9 1.013
Neodymium (Nd) 60 144.2 0.995 Light
(Promethium (Pm) 61 - 0.979 earths
Samarium (Sm) 62 150.4 0.964
Europium (Eu) 63 152.0 0.950
Gadolinium (Gd) 64 157.3 0.938 Middle
Terbium (Tb) 65 158.9 0.923 earths
Dysprosium (Dy) 66 162.5 0.908
Holmium (Ho) 67 164.9 0.894
Erbium (Er) 68 167.3 0.881 Heavy
Thulium (Tm) 69 168.9 0.869 earths
Ytterbium (Yb) 70 173.0 0.858
Lutetium (Lu) 71 175.0 0.848
(Yttrium (Y) 39 88.9 0.88
used commercially, but it suffered from the restrictions of low resin capacity
that required large columns, and also the system had to use a batch opera-
tion. This process has now been largely superceded by solvent extraction.
Both ion exchange and solvent extraction separations are based on the steady
decrease in size across the lanthanide elements, which result in an increase in
acidity or decrease in basicity with increasing atomic number. This causes a
variation in formation coefficient of any metal–extractant complex, allowing
preferential binding to an ion exchange resin, or extraction of the complex
into an organic phase (see Chapters 3 and 4). The variations in formation
coefficients and hence, separation factors, between adjacent elements, are
small; so many equilibria and thus a large number of extraction units are
required. Even then, separation is possible only by recycling both the exit
streams. This configuration, rarely used outside lanthanide production, is
essential to the preparation of pure products or the separation of adjacent
elements. The operation is shown in Fig. 11.17. The bank of mixer-settlers is
fed close to the midpoint, with a mixture of components, and by returning
the exit streams to the process, 100% reflux is obtained. The components are
force-fed to build up their concentration, and the reflux is continued until
the desired degree of separation is achieved. Then, the operation is stopped
and the individual mixer-settlers emptied to give the required products.
Figure 11.18 gives a typical separation of lanthanum, neodymium, and

praseodymium from nitrate media, using TBP as extractant. The number of
extraction units can be varied according to the products required; for
example, lanthanum can be concentrated in the aqueous raffinate and neo-
dymium in the organic extract in holding tanks external to the mixer-settlers,
thereby decreasing the number of units required. It should be noted that
yttrium, although not a member of the lanthanide series, often accompanies
them in ores. The charge/size ratio of yttrium places it between dysprosium
and holmium, although, depending on the extraction system, it can behave as
either a light or heavy element.
The lanthanides occur in a number of ores, but as the relative con-
centration of individual elements differ according to the source, several
extraction schemes have been developed to take advantage of this natural
variability. To avoid confusion caused by such variations, this section will
consider only the more general extraction schemes. The first operation in the
flow sheets involves leaching of the ore and, in most cases, the leaching agent
is one of the common mineral acids. The choice of extractant depends on the
acid used; for example, the formation of nitrate complexes allows the use of
solvating extractants (e.g., TBP), whereas the carboxylic or organophos-
phorus acids may be used with either hydrochloric, nitric, or sulfuric acid
media. Occasionally, mixtures of two extractants will improve the separa-
tion of individual elements.
Lanthanides are also found as minor components in other ores, par-
ticularly in association with uranium or in phosphate rock. These are often
coextracted with the major product and can be economically recovered
from the waste streams resulting from the uranium or phosphoric acid
extraction.
Fig. 11.17 Mixer-settler bank under reflux configuration.

496 Cox
Fig. 11.18 Concentration profiles for mixer-settler bank under total reflux: system TBP
(50%) from nitrate medium. (From Ref. 42.)
11.6.1 Solvating Extractants

The extraction of the lanthanides from nitrate media with TBP [Eq. (11.20)]
depends on several factors including pH, nitric acid concentration, ionic
strength, and concentration of the elements in the aqueous phase.
LnðNO3 Þ3 xH2 O þ 3TBP ! LnðNO3 Þ3 3TBP þ xH2 O (11.20)

The situation is complicated by the ability of TBP to extract nitric acid as
HNO3 xTBP, which then competes with the lanthanides, so the process is
optimized in terms of high pH, but low nitric acid concentration. Increased
ionic strength has the advantage of salting out the aqueous phase complexes
into the organic phase, whereby neutralization of excess nitric acid increases
the distribution coefficient. The total concentration of lanthanides in the feed
affects the separation factors between adjacent elements; the separation
factor decreasing as the concentration rises, with values in the range S = 0.7–
2 (concentration of rare earth oxides 60– 460 g dm3) (see Table 11.8). In
contrast, the separation factors between yttrium and the lanthanides increase
with dilution, so that a process with a dilute feed can be used to produce
purified yttrium. A commercial flow sheet has been designed to produce pure
lanthanum using 50% TBP, the lanthanum remaining in the raffinate after
extraction. The organic phase is stripped with water, and the aqueous
solution concentrated and extracted with the same organic phase to separate
praseodymium and neodymium. A typical profile of the mixer-settler bank is
shown in Fig. 11.18. Other solvating organophosphorus extractants have
been studied to a small extent and these show some processing advantages
with, for example, an organophosphonate showing higher separation factors
than TBP.
11.6.2 Amine Extractants

Amines provide an alternative method of extraction from nitric acid media,
extracting the nitrato-complexes of the lanthanides as ion pairs. Because of
competition from nitric acid, the process has to operate above pH 2, usually
with metal nitrates as salting out agents. The extraction also depends on the
amine structure, with asymmetric compounds giving the highest distribution
coefficients. Again, separation from yttrium is favored, with separation
factors in the range of 5–20. An interesting process to produce pure yttrium
from a mixed lanthanide feed depends on this element behaving as a heavy
lanthanide in nitrate media and as a light lanthanide in thiocyanate solution.
Thus, using a nitrate solution and an amine extractant, the light elements
can be extracted from yttrium and the heavy lanthanides; then, using
thiocyanate media, yttrium can be extracted from the remaining elements in
the raffinate (Fig. 11.19).
11.6.3 Acidic Extractants

Both carboxylic (e.g., Versatic acid) and organophosphorus acids have been
used commercially to extract the lanthanides. The extraction follows the
formation of the metal–extractant complex [Eq. (11.21)] and so depends on
the pH of the feed.
Ln3þ þ 3H2 A2 ! LnA3 ðHA3 Þ3 þ 3Hþ (11.21)
Table 11.8 Separation Factors for Adjacent Lanthanides Using TBP from Nitrate
Medium
Separation factors
Aqueous phase
conc. (Ln2O3 g dm3) Sm/Nd Gd/Sm Dy/Gd Ho/Dy Er/Ho Yb/Er
460 2.26 1.01 1.45 0.92 0.96 0.81
430 - - 1.20 0.96 0.65 -
310 2.04 1.07 1.17 0.94 0.82 -
220 1.55 0.99 1.08 0.89 0.78 -
125 1.58 0.82 0.92 0.83 0.72 -
60 1.40 0.78 0.89 0.77 0.70 0.63
Source: Ref. 42.

498 Cox
Fig. 11.19 Distribution coefficients for lanthanide extraction with Aliquat 336 from
nitrate (solid squares) and thiocyanate (solid circles) media. (From Ref. 1.)
Organophosphorus acids, of which DEHPA has received the most atten-

tion, are more powerful extractants than Versatic acid and extract the heavy
lanthanides even at low pH values. This does give problems with stripping,
and the heaviest elements, ytterbium and lutetium, require 20% hydrofluoric
or 50% sulfuric acid as strip solutions. Variation of pH thus allows the
separation of heavy, including yttrium, from light elements. Separation
factors between adjacent elements follow the order phosphinic phos-
phonic>phosphoric>carboxylic acids, although few data are available for
the first two extractants (Table 11.9). In addition, with organophosphorus
acids, higher separation factors are found from chloride than from sulfate or
nitrate media. Thus, parameters that can be used to design a flow sheet
include type and concentration of extractant, pH of the feed solution, nature
of anion (organophosphorus acids), and nature and concentration of the
strip solution.
Table 11.9 Separation Factors for Adjacent Lanthanides Using Acidic Extractants
Ce/La Pr/Ce Nd/Pr Sm/Nd Eu/Sm Gd/Eu Tb/Gd Dy/Tb Ho/Dy Er/Ho Tm/Er Yb/Tm Lu/Yb
a
Versatic 911 3.00 1.60 1.32 2.20 - 1.96 - - 1.17 1.28 - - -
Solvent Extraction in Hydrometallurgy
DEHPAa - - 1.40 4.00 2.30 1.50 5.70 2.00 1.80 2.40 3.40 - -
DEHPAb 5.24 1.86 1.32 9.77 2.14 1.99 - - - - - - -
Phosphonic acidb 13.8 3.47 1.51 13.80 2.45 1.86 6.92 3.23 2.17 2.82 2.63 3.71 1.74
Phosphinic acidb 16.98 1.86 1.32 13.80 2.14 1.32 6.02 2.82 1.74 2.63 3.47 2.63 1.51
a
Source: Ref. 42.
b
Source: Ref. 43. (All organophosphorus extractants used in this work had 2-ethylhexyl substituents.)
499
500 Cox
11.6.3.1 Versatic Acid

Extraction takes place by a cation-exchange mechanism with maximum
extraction occurring close to the pH of hydrolysis (i.e., about pH 6). No
preference is shown for a particular anion in the feed solution, and separation
factors are less dependent on the concentration of elements in the feed than
for TBP. The high cerium–lanthanum separation factor indicates that Ver-
satic acid would be a better extractant than TBP to produce pure lanthanum.
The heavy elements are strongly extracted with yttrium lying between dys-
prosium and gadolinium. A process has been developed to produce pure
yttrium by first using Versatic acid to remove the light elements in the raf-
finate. The extract is then stripped and the concentration of the strip phase
adjusted so that, on reextraction with TBP, yttrium behaves as a light ele-
ment, and separation from the remaining heavy elements is achieved.
However, the small separation factors still require both extractions to be
performed under total reflux in a large (56-stage) mixer-settler train.
11.6.3.2 Organophosphorus Acids
Di-2-ethylhexylphosphoric acid has received most attention, but both the
organophosphonic and organophosphinic acids potentially offer some
processing advantages. Organophosphorus acids are more powerful
extractants than Versatic acid and, thus, extract at lower pH values. Con-
sequently, a 0.1 mol dm3 DEHPA solution completely extracts all the ele-
ments above europium at pH 2 and, as noted earlier, this makes the heavy
elements difficult to strip. Although higher separation factors are found
from chloride than from nitrate or sulfate media, increasing the acidity of
the chloride solution decreases these values. A similar, but smaller, depen-
dency of separation factors on acidity has also been found with sulfate
systems.
Several different flow sheets have been developed around DEHPA,
which, as discussed earlier, vary according to the precise composition of the
feed material. A typical process using a nitric acid leach of a xenotime ore
follows as an example.
The leach liquor is first treated with a DEHPA solution to extract the
heavy lanthanides, leaving the light elements in the raffinate. The loaded
reagent is then stripped first with 1.5 mol dm3 nitric acid to remove the
elements from neodymium to terbium, followed by 6 mol dm3 acid to
separate yttrium and remaining heavy elements. Ytterbium and lutetium are
only partially removed; hence, a final strip with stronger acid, as mentioned
earlier, or with 10% alkali is required before organic phase recycle. The
main product from this flow sheet was yttrium, and the yttrium nitrate
product was further extracted with a quaternary amine to produce a
99.999% product.
Fig. 11.20 Effect of DEHPA concentration on extraction of lanthanides. (From Ref. 2.)
Changing the concentration of extractant will give a different split of

the lanthanides (Fig. 11.20); accordingly, the flow sheet can be modified to
take advantage of variations in feed composition or to produce a particular
element as product.
The rate of extraction of the lanthanides by the 2-ethylhexyl ester of
2-ethylhexylphosphonic acid (e.g., PC88A, P507) is slower than for DEHPA
but the higher distribution coefficients allow extraction from more dilute
solutions. Also, as acidic stripping is easier than with DEHPA, the reagent
has been used commercially to separate lutetium from terbium and ytter-
bium.
Recently emphasis has shifted to the use of organophosphinic acids for
the extraction of lanthanides [44] with CYANEX 272, providing significant
advantages over other organophosphorus acid extractants for the heavy
lanthanides with greater average separation factors and easier stripping.
Furthermore, a 1:1 mixture of CYANEX 272 and a carboxylic acid gives
significant synergism for the heavy lanthanides with higher loadings
502 Cox
(0.2 mol dm3) than either CYANEX 272 (0.08 mol dm3), DEHPA, or the
phosphonic acid ester (P507) (0.17 mol dm3) [45]. In addition, stripping from
the synergistic mixture with either hydrochloric or nitric acid is easy. The
latter enables the production of high-purity products free from chloride ions.
11.6.4 Summary
Liquid–liquid extraction provides one of the easiest methods for the
separation of these commercially important elements. It is more convenient
than ion exchange, allowing both semicontinuous operation and the use of
more concentrated feed solutions. However, the large number of mixer-
settlers required imposes a considerable capital investment in the plant, and
their use in the reflux mode entails that some elements are locked into the
process for long periods, which also has economic implications. Therefore,
flow sheets are developed to provide elements with a ready market at the
earliest opportunity, leaving the remaining elements as intermediates for
separation at a later stage or for sale as mixed lanthanides.
The choice of reagent depends very much on the nature of the feed, but
the introduction of organophosphinic acids either alone or in combination
with carboxylic acids in a synergistic mixture offer great potential for
commercial development.
Further discussion on the extraction of lanthanides, especially con-
cerning their separation from actinides, can be found in Chapter 12.
11.7 CONCLUSION
Solvent extraction is now a proven technology for the commercial extrac-
tion, separation, and concentration of a wide range of metals both from
primary and secondary sources (see Chapter 14). In recent years, there has
been a reduction in the development, production, and marketing of new
commercial extractants as the overall costs of such activities increases.
However, the use of established reagents in new hydrometallurgical appli-
cations continues to expand.
REFERENCES
1. Lo, T. C.; Baird, M. H. I.; Hanson, C. Handbook of Solvent Extraction; Wiley-
Interscience: New York, 1983.
2. Ritcey, G.; Ashbrook, A. W. Solvent Extraction, Part 2; Elsevier; Amsterdam,
1979.
3. Ritcey, G.; Ashbrook, A. W. Solvent Extraction, Part 1; Elsevier; Amsterdam,
1984.
4. Cox, M.; Flett, D. S. Chem Ind., 1987, p. 188.
5. Dalton, R. F.; Price, R. Proceedings of Oslo Symposium 1982; Society of

Chemical Industry: London, paper 11–23, 1982.
6. Hudson, M. Hydrometallurgy, 1982, 9, 149.
7. Marcus, Y.; Kertes, A. S. Ion Exchange and Solvent Extraction of Metal
Complexes; Wiley-Interscience: New York, 1969.
8. Sekine, T.; Hasegawa, Y. Solvent Extraction Chemistry; Marcel Dekker: New
York, 1977.
9. Preston, J. S. Hydrometallurgy, 1983, 10, 187.
10. Preston, J. S. Hydrometallurgy, 1983, 11, 105.
11. Barnes, J. E.; Stetchfield, J. H.; Williams, G. O. R. J. Inorg. Nucl. Chem., 1976,
38, 1065.
12. Nishimura, S. Extraction Metallurgy ’81; Institute of Mining and Metallurgy:
London, p. 404, 1981.
13. Sole, K. C.; Cole, P. M. Ion Exchange and Solvent Extraction; Marcus, Y. and
Sengupta, A. K. Eds.; Marcel Dekker: New York, 2002; Vol. 15, 143.
14. Mackenzie, J. M. W.; Virnig, M. J.; Boley, B. D.; Wolfe, G. A. ALTA 1998
Nickel Cobalt Pressure Leaching and Hydrometallurgy Forum, Melbourne;
ALTA Metallurgical Services, 1988.
15. Flett, D. S.; J. Chem. Tech. Biotechnol. 1999, 74, 99.
16. Dalton, R. F.; Seward, G. W. Proceedings of Reagents in the Minerals Industry,
Jones, M. J. and Oblatt, R., Eds., Institute of Mining and Metallurgy: London,
1984, 106.
17. Dalton, R. F.; Seward, G. W. Proceedings ISEC ’86; Munich; DECHEMA:
Frankfurt-am-Main, 2: 11, 1986.
18. Kordosky, G. A.; Olafson, S. M.; Lewis, R. G.; Deffner, V. L.; House, J. E.
Sep. Sci. Technol., 1987, 22, 215.
19. Dalton, R. F.; Price, R.; Henmana, E.; Hoffmann, B. Separation Process in
Hydrometallurgy, Davies, G. A. Ed.; Ellis-Horwood: Chichester; 1987, 466.
20. Morris, D. F. C.; Khan, M. A. Talanta, 1968, 75, 1301.
21. Rimmer, B. Chem. Ind. 1974, 63.
22. Soldenhoff, K.; Wilkins, D. Solvent Extraction for the 21st Century, Proc.
ISEC’99; Society of Chemical Industry: London, 2001, 1, 777.
23. Cleare, M. J.; Charlesworth, P.; Bryson, D. J. J. Chem. Tech. Biotechnol., 1979,
29, 210.
24. Grant, R. A.; Drake, V. A. Proc ISEC2002, Cape Town; 2002, 940.
25. Baroncelli, F.; Carlini, D.; Gasparini, G. M.; Simonetti, E. Proceedings ISEC
’88, Moscow; Vernadsky Inst.; Academy of Sciences USSR: Moscow, 1988;
Vol. 4, 236.
26. Demopoulos, G. P.; Pouskouleli, G.; Ritcey, G. M. Proceedings ISEC ’86,
Munich; DECHEMA: Frankfurt-am-Main, 1986; Vol. 2, 581 pp.
27. Villaescusa, I.; Miralles, N.; de Pablo, J.; Salvado, V.; Sastre, A. M.; Solv. Extn
Ion Exch. 1993, 11, 613.
28. Argiropoulos, G.; Cattrall, R. W.; Hamilton, I. C.; Paimin, R. Value Adding
Through Solvent Extraction, Proc. ISEC’96; University of Melbourne, 1996;
Vol. 1, 123 pp.
504 Cox
29. Kordosky, G. A.; Sierokoski, J. M.; Virnig, M. J.; Mattison, P. L. Hydro-

metallurgy; 1992, 30, 291.
30. Virnig, M. J.; Kordosky, G. A. Value Adding Through Solvent Extraction, Proc.
ISEC’96; University of Melbourne, 1996; Vol. 1, 311 pp.
31. Zhao, Jin; Wu, Zhichun; Chen, Jiayong. Value Adding Through Solvent Ex-
traction, Proc. ISEC’96; University of Melbourne, 1996; Vol. 1, 629 pp.
32. Capela, R. S.; Paiva, A. P. Proc. ISEC2002; Cape Town, 2002; 335 pp.
33. Facon, S.; Rodriguez, M. Avila; Cote, G.; Bauer, D. Solvent Extraction in the
Process Industries, Proc ISEC93; Elsevier Applied Science, 1993, Vol. 1, 557 pp.
34. Du Preez, A.; Preston, J. S. Proc. ISEC2002; Cape Town, 2002, 896 pp.
35. Pouskouleli, G.; Kelebek, S.; Demopoulos, G. P. Separation Processes in Hy-
drometallurgy; Davies, G. A. Ed.; Ellis-Horwood: Chichester, 1987, 175 pp.
36. Grant, R. A.; Murrer, B. A.; European Patent No. 0210004, 1986.
37. Dhara, S. C. Proceedings of Precious Metals Mining, Extraction, and Proces-
sing; Kudryk, V. Corrigan, D. A. Laing, W. W. Eds.; Metallurgical Society
AIME, 1984, 199 pp.
38. Dolgikh, V. I.; Borikov, P. I.; Borbat, V. F.; Kouba, E. F.; Gindin, L. M. Tsvet
Metal. 1967, 40, 30.
39. Fedorenko, N. V.; Ivanova, T. I.; Russ. J. Inorg. Chem., 1965, 10, 387.
40. Benguerel, E.; Demopoulos, G. P.; Cote, G.; Bauer, D. Solv. Extn. Ion Exch.
1994, 12, 497.
41. Benguerel, E.; Demopoulos, G. P.; Harris, G. B. Hydrometallurgy, 1996, 40,
135.
42. Brown, C. G.; Sherrington, L. G. J. Chem. Technol. Biotechnol., 1979, 29, 193.
43. Preston, J. S.; du Preez, A. C. Mintek Report M378; Council for Mineral
Technology: Randburg, South Africa, 1988.
44. Li, D. Solvent Extraction for the 21st Century, Proc ISEC’99; 2000; Vol. 2,
1081 pp.
45. Yuan, C.; Ma, H.; Pan, C.; Rickelton, W. A.; Value Adding Through Solvent
Extraction, Proc ISEC’96; 1996, Vol. 1, 733 pp.
BIBLIOGRAPHY
In addition to the list at the end of Chapter 1, texts concerned with appli-
cations of solvent extraction in hydrometallurgy include the following:
Burkin, R. (ed.). Critical Reviews in Applied Chemistry, Vol. 17; Extractive

Metallurgy of Nickel, John Wiley and Sons: Chichester, 1987.
Chapman, T.W. Extraction-metals processing. In Handbook of Separation
Process Technology, Chap. 8 Rousseau, R.W. Ed.; John Wiley and
Sons: New York, 1987, 467.
Murthy, T. K. S.; Koppiker, K. S.; Gupta, C. K. Solvent extraction in
extractive metallurgy. In Recent Developments in Separation Science;
Li, N.N. and Navratil, J.D. Eds.; CRC Press: Boca Raton, Florida,
1986, Vol. 8, 1 p.
Osseo-Asare, K.; Miller, J. D. Eds. Hydrometallurgy; Research, Development
and Plant Practice; Metallurgical Society AIME; New York, 1982.
Proceedings of Extraction Metallurgy Conferences, published by Institution
of Mining and Metallurgy, London, 1981, 1985, 1989.
Schulz, W. W.; Navratil, J. D. Eds. Science and Technology of Tributyl
Phosphate, Vols. 1 and 2; CRC Press: Boca Raton, Florida, 1987.
Szymanowski, J. Hydroximes and Copper Hydrometallurgy; CRC Press:
Boca Raton, Florida, 1993.
Yamada, H.; Tanaka, M. Solvent extraction of metal carboxylates. Adv.
Inorg. Chem. Radiochem. 1985, 29, 143.
12
Solvent Extraction in Nuclear Science
and Technology
CLAUDE MUSIKAS* Commissariat à l’Energie Atomique, Paris, France
WALLACE W. SCHULZ Consultant, Albuquerque, New Mexico, U.S.A.
JAN-OLOV LILJENZIN* Chalmers University of Technology, Göteborg,
Sweden
12.1 INTRODUCTION
In the year 2000, 15% of the world’s electric power was produced by
433 nuclear power reactors: 169 located in Europe, 120 in the United States,
and 90 in the Far East. These reactors consumed 6,400 tons of fresh enri-
ched uranium that was obtained through the production of 35,000 tons of
pure natural uranium in 23 different nations; the main purification step was
solvent extraction. In the reactors, the nuclear transmutation process yielded
fission products and actinides (about 1000 tons of Pu) equivalent to the
amount of uranium consumed, and heat that powered steam-driven turbines
to produce 2,400 TWh of electricity in 2000.
Different countries have adapted various policies for handling the
highly radioactive spent fuel elements:
1. in the once through cycle, the fuel elements are stored in temporary
facilities, usually with the intention to finally deposit them in under-
ground vaults; and
2. in the recycling concept, the spent fuel pins are dissolved in acid and
‘‘reprocessed’’ to recover energy values (i.e., unused uranium and other
fissionable isotopes, notably plutonium), while the waste is solidified
and stored in underground vaults.
In some countries, the main purpose of reprocessing is to recover
plutonium for weapons use. The main separation process in all known
reprocessing plants is solvent extraction.
*Retired.
507
508 Musikas et al.
As all uranium-burning reactors produce plutonium, which may be

used for nuclear weapons, nuclear power for electricity production has
become a political issue. Thus, some countries have decided to abstain from
using nuclear energy, while others efficiently pursue it. This decision is partly
affected by fear of the effects of nuclear radiation. That and connected prob-
lems are widely discussed among the public and decision makers. However,
in this chapter we only provide a technical/scientific overview of current
industrial-scale applications of solvent extraction chemistry and technology
in the nuclear industry. A principal focus is on solvent extraction processes
developed to partition and concentrate long-lived actinides and certain fis-
sion products from nuclear waste solutions, with the purpose of reducing
handling difficulties and risks from the nuclear waste.
12.2 BACKGROUND
12.2.1 Historical Aspects
Nuclear science and technology have played key roles in development of
liquid–liquid solvent extraction for the industrial-scale separation of metals.
The high degree of purity of materials needed for nuclear applications,
together with the high value of these materials, favors solvent extraction as
the separation method of choice when compared with other methods. The
need for a continuous, multistage separation system is unique to reproces-
sing of spent nuclear reactor fuels because of the requirement for nearly
quantitative recovery and separation of fissile uranium and plutonium from
some 40 or more fission products belonging to all groups of the periodic
table. Only a brief account of the historical development and application of
solvent extraction processes and technology in the nuclear field is included
here. Numerous other authors have reviewed various aspects of this history;
readers interested in historical details should consult references [1–8].
In 1942, the Mallinckrodt Chemical Company adapted a diethylether
extraction process to purify tons of uranium for the U.S. Manhattan Project
[2]; later, after an explosion, the process was switched to less volatile ex-
tractants. For simultaneous large-scale recovery of the plutonium in the
spent fuel elements from the production reactors at Hanford, United States,
methyl isobutyl ketone (MIBK) was originally chosen as extractant/solvent
in the so-called Redox solvent extraction process. In the British Windscale
plant, now Sellafield, another extractant/solvent, dibutylcarbitol (DBC or
Butex), was preferred for reprocessing spent nuclear reactor fuels. These
early extractants have now been replaced by tributylphosphate [TBP], diluted
in an aliphatic hydrocarbon or mixture of such hydrocarbons, following the
discovery of Warf [9] in 1945 that TBP separates tetravalent cerium from
Solvent Extraction in Nuclear Science 509
trivalent rare earths. Tributylphosphate has several major advantages over

ethers and ketones, especially in its stability toward nitric acid, the preferred
acid for nuclear fuel reprocessing [10]. Tributylphosphate allows easy and
quantitative recovery of U(VI) and Pu(IV) from nitric acid solutions because
of the high distribution ratios of these elements. Salting-out agents, such as
Al(NO3)3, used in early extraction processes are no longer necessary. Fur-
thermore, TBP has a low toxicity.
The solvent extraction process that uses TBP solutions to recover
plutonium and uranium from irradiated nuclear fuels is called Purex (plu-
tonium uranium extraction). The Purex process provides recovery of more
than 99% of both uranium and plutonium with excellent decontamination of
both elements from fission products. The Purex process is used worldwide to
reprocess spent reactor fuel. During the last several decades, many variations
of the Purex process have been developed and demonstrated on a plant scale.
In 1948, high-molecular-weight alkylamines were introduced as ex-
tracting agents by Smith and Page [11]. Industrial-scale processes for re-
covery of uranium from ore leach liquors were developed at the U.S. Oak
Ridge National Laboratory [12]. Plutonium purification processes that were
based on extraction by trilaurylamine [TLA] were developed in France [13]
and were used for a time in the French UP2 Purex plant at the La Hague site.
Solvent extraction research has been a major program in many national
institutions involved in nuclear technology. Research in U.S. laboratories
during the 1960s led to the synthesis of new organophosphorus extractants
and to detailed investigations of their extraction chemistry. Monofunctional
extractants such as di-(2-ethylhexyl)phosphoric, -phosphonic, or -phosphinic
acids were first investigated [14] at the Argonne National Laboratory
(ANL, Chicago). The limited affinity of these monofunctional extractants for
actinide ions in acidic media was an incentive to investigate the behavior of
bifunctional extractants for the separation of the tri-, tetra-, and hexavalent
actinides. Efficient new organophosphorus molecules were later synthetized
in the U.S. [15] and in the former Soviet Union [16]. However, because
poor back extraction of some metal ions [e.g., U(VI), Pu(IV), and Am(III)]
by these polyfunctional organophosphorus molecules made them unsuitable
for process applications, dihexyldiethylcarbamoylmethylenephosphonate
(DHDECMP) and octylphenyldiethylcarbamoylphosphine oxide (CMPO),
which have one phosphoryl and one amide functional group, showed pro-
mise for use and are still under study for actinide cation separations. As the
best process is always a compromise between several parameters, changes in
the preference of the engineers occurred over time due to the technical pro-
gress made [17].
In the European Community (EC) countries, two additional criteria
for extractant selection are thought to be important. The first is related to
510 Musikas et al.
the amount of incombustible waste produced in the process, which can be

reduced if the organic solvents are completely incinerable: i.e., they must only
contain C, H, O, and N atoms. The second is related to the degradation prod-
ucts of the extractant. To avoid full solvent regeneration before each solvent
recycle, the degradation products must not interfere strongly with the extrac-
tion and back extraction of the major actinides: dialkylphosphoric acids
are formed as degradation products of trialkylphosphates, -phosphonates,
or -phosphine oxides, and they prevent the back extraction of metals at the
low acidities needed in some parts of the process. Thus, the degradation
products must be carefully eliminated before each solvent recycle, leading to
generation of additional waste volumes.
12.2.2 Chemical Aspects

A primary goal of chemical separation processes in the nuclear industry is to
recover actinide isotopes contained in mixtures of fission products. To
separate the actinide cations, advantage can be taken of their general chem-
ical properties [18]. The different oxidation states of the actinide ions lead to
ions of charges from +1 (e.g., NpOþ 4+
2 ) to +4 (e.g., Pu ) (see Fig. 12.1),
which allows the design of processes based on oxidation reduction reactions.
In the Purex process, for example, uranium is separated from plutonium by
reducing extractable Pu(IV) to nonextractable Pu(III). Under these condi-
tions, U(VI) (as UO2þ 4+
2 ) and also U(IV) (as U ), if present, remain in the
TBP phase.
In waste management programs, it might be desirable to isolate the
minor actinides (Np, Am, Cm) and recycle them in nuclear reactors (see later
flow sheets). These actinides appear together with the trivalent lanthanides,
which have undesirable nuclear properties (e.g., high neutron reaction cross
sections), and thus it becomes necessary to remove the trivalent lanthanides.
Unfortunately, the trivalent lanthanides behave chemically like the trivalent
actinides, because their similar ionic radii decrease with the atomic number in
each series and overlap, resulting in complexes of very similar strength,
leading to very small separation factors of the 4f (lanthanide) and 5f (acti-
nide) M3+–X complexes.
The actinide elements are ‘‘hard acid cations’’ and their bonds are
principally ionic; thus, ionic interactions are the main driving force for bond
formation. Consequently, ‘‘type A ligands’’ such as oxygen donors form
complexes with the actinide ions much better than the ‘‘type B ligands’’ (see
Chapter 3). To produce different stabilities in the complexes of the ions of the
4f and 5f series, the bonds must have some covalent character. As the degree
of covalency increases with the decrease of the Pauling electronegativity of
the donor atoms in the ligand, ligands with donor atoms less electronegative
Fig. 12.1 Oxidation states of the actinide elements: . most stable ions in aqueous
solutions; ++ oxidation states observed in aqueous solutions; + , unstable ions observed
only as transient species. *In solids precipitated from alkaline solutions.
than oxygen have more covalency. In addition, the trivalent actinides have
slightly more covalency than the lanthanides [19]. This has led to the search
for new extractants containing more covalent donor atoms, such as sulfur or
nitrogen. Such extractants have been shown to provide better separation of
the trivalent ions of the 4f and 5f transition series (see section 12.9.1).
After the separation of the actinides from the high-level waste, it is
desirable to remove certain other fission products from the nuclear wastes.
Some Cs+ and Sr2+ are low-charged cations that react well with macro-
cyclic ligands (e.g., crown ethers, calixarenes). Research to synthesize and
investigate the properties of macrocyclic ligands for application in nuclear
waste treatment has been an active effort internationally. Some of the results
obtained are discussed in section 12.7.
Extractants that lead to the formation, in the organic phase, of com-
plexes in which the metal is completely surrounded by oxygen atoms are
often quite selective for the actinides. This is true with TBP, which selec-
tively extracts actinides from nitrate solutions, forming complexes such as
Th(NO3)4(TBP)2 and UO2(NO3)2(TBP)2 (see Fig. 12.2). Nitric acid is pre-
ferred over other strong inorganic acids because it is the least corrosive acid
for the stainless steel equipment used in large-scale reprocessing of spent
nuclear fuels. Its low complexing affinity for the d transition ions and its
ability to fill two coordination sites favors binding with large actinide ions.
12.3 URANIUM PRODUCTION

The process used for recovery of uranium from its ores depends on the
nature of the ore. All the processes include a leaching step that solubilizes
the metal. Solvent extraction is used most frequently for the recovery and
512 Musikas et al.
Fig. 12.2 Distribution ratios of actinide ions and some selected metals between un-
diluted TBP and HNO3 or HCl solutions.
purification of uranium from the leaching liquors [20–22]. The production

processes and their relative importance are listed in Table 12.1. The ex-
traction processes are subject to variation in different mining areas, so the
data in Table 12.1 should be considered only as illustrative of uranium
solvent extraction processes that have been used. Today, the demand for
mining new uranium for use in nuclear reactors is not as high as before.
Orders for new nuclear reactors are few and, in some countries, uranium is
available from dismantled nuclear weapons. Presently the price of uranium
is low and the production from mines is constant, or in many countries
decreasing.
12.3.1 Recovery from Sulfuric Acid Leach Liquors

Two extractants are used to recover uranium from sulfuric acid leach
liquors: di-(2-ethylhexyl)phosphoric acid (HDEHP) in the Dapex process
and commercial mixtures of trioctyl to tridecylamines (e.g., Adogen 364 or
Alamine 336) in the Amex process. The chemical principles on which these
processes are based are given in Table 12.2. Details, such as the nature of the
diluent, extractant concentration, and washing and stripping conditions,
vary from plant to plant. The advantages and disadvantages of the Dapex
and Amex processes are compared in Table 12.3. The Amex process is more
Table 12.1 Commercial Processes for Recovery of Uranium from Ores
Relative importance in the
Type of ores/main locations Leachant Purification process production of U
CaCO3, basic, U present Na2CO3 Precipitation Na2U2O7 (no 11
mainly as U(VI) Grants (New solvent extraction steps)
Mexico), Rio Algon (Utah)
Weakly basic to acidic ores H2SO4 + oxidant Dialkylphosphoric acids or 80

Canada, Australia, U.S.A. trialkylamine extraction of U(VI)
(contain U(IV)) from H2SO4 liquors
High-grade pitchblende ores HNO3 TBP extraction of U(VI) Few%
Zaire, Canada,
Czechoslovakia, South Africa
Phosphates Florida H2SO4 leads to a Synergistic extraction of U(VI) Few % will vary with U price.
phosphoric acid solution by HDEHP-TOPO or extraction The phosphate rocks are the
that contains 50–200 ppm of U(IV) by dialkylphosphoric second largest potential
of U acids mixtures source of U after sea water
513
514
Table 12.2 Chemical Basis for the Amex and the Dapex Processes for U(VI) Recovery from H2SO4 Leach Liquors
Amex Dapex
Extraction UO22 + + SO42 + 2(R3NH)2SO4 ! (R3NH)4UO2(SO4)3 UO22 + + 2(HDEHP)2 ! UO2(HDEHPDEHP)2 + 2H +

R3N = Alamine 336 or Adogen 364
(trialkylamines with alkylgroups with 8
to 10 carbons)
Stripping Acidic stripping Alkaline stripping UO2(HDEHP
(R3NH)4UO2(SO4)3 + 4HX ! DEHP)2 + 4Na2CO3 ! UO2(CO3)34 +
4R3NHX + UO22 + + 3HSO4 + H + 4NaDEHP + H2O + CO2 : + 4Na+
HX = HCl or HNO3 (TBP must be added to avoid the third-
Neutral stripping phase formation owing to low organic
(R3NH4)4UO2(SO4)3 + (NH4)2SO4 + 4NH3 ! phase solubility of NaDEHP.)
4R3N + UO2(SO4)34 + + 6NH4+ + SO42 Acidic stripping
Alkaline stripping UO2(HDEHP)2 (DEHP)2 + 2H + ! 2(HDEHP)2 + UO22 +
(R3NH)4UO2(SO4)3 + 7Na2CO3 !
4R3N + UO2(CO3)34 + 4HCO3 +
3SO42 + 14Na +
Musikas et al.
Table 12.3 Comparison of the Amex and Dapex Processes
Point of
comparison Amex process Dapex process
Coextracted Mo(VI), Zr(IV) Fe(III), Th(IV), V(IV),

Ti(IV), Mo(VI), Rare Earths
Stripping Simple, many possibilities With Na2CO3 which
necessitates
addition of TBP to
avoid the separation
of NaDEHP in a third phase
Extractant stability Rather low stability More stable than amines
Diluent Tendency to form third phases Many diluents can be used
with aliphatic hydrocarbons.
Long-chain alcohols as
phase modifiers or aromatic
diluents are necessary
Extraction kinetics Rapid Slow
Phase disengagement Slow (sensitive to Rapid
suspended solids)
widely used than the Dapex process because of the greater selectivity of
trialkylamines than HDEHP for uranium in H2SO4 solution.
12.3.2 Recovery from Nitric Acid Leach Liquors

High-grade pitchblende ores are leached with nitric acid to recover uranium.
Extraction of uranium from nitrate solutions is usually performed with
TBP. TBP-based solvents are used in several areas of the nuclear industry,
especially for reprocessing of spent nuclear fuels and for refining the ura-
nium product of the Amex and Dapex processes. Extraction of uranium by
TBP solvents is described in sections 12.3.4 and 12.5.
12.3.3 Recovery from Phosphoric Acid Solutions

The concentration of uranium contained in phosphate rocks (50–200 ppm) is
higher than that in seawater (see section 12.3.5). Even though economic
recovery of uranium from phosphate rock is difficult, several phosphoric
acid plants include operation of uranium recovery facilities.
Various processes have been used for uranium extraction from
phosphoric acid solution; their main features are listed in Table 12.4. The
HDEHP-TOPO process is increasingly preferred over others because of the
stability of the extractant and the well-understood chemistry of the process.
516
Table 12.4 Main Industrial Processes for the Recovery of U from Wet Phosphoric Acid
Extractant Mechanism of extraction Stripping Remark
Octylpyrophosphoric acid U4 + +2H2OPPA ! In 25% H2SO4 + 10–20% Cheap but unstable

(H2OPPA) U(OPPA)2 + 4H + (high HF: U(OPPA)4 + 4HF ! extractant. Difficulties
distribution ratios) 2H2OPPA + UF4 in stripping.
Octylphenyl U4 + + (HDOPA)2 + H2MOPA ! Oxidation to U(VI): Cheap extractant.
orthophosphoric (DOPA)2 MOPA + 4H + U(DOPA)2 MOPA + ½O2 + Difficulty of
acids (mono plus (high distribution ratios) 3H2O ! UO22 + + H2MOPA stripping. Phase
di) H2MOPA + + (HDOPPA)2 + 2OH disengagement
HDOPA problems.
HDEHP–TOPO UO22 + + 2(HDEHP)2 + TOPO ! Reduction to U(IV) in Rather expensive
UO2 (DEHPHDEHP)2 + 2H + H3PO4 or use of (NH4)2CO3: extractant. Must
(rather low distribution ratios) UO2(HDEHP DEHP)2TOPO + oxidize U(IV) to
3(NH4)2CO3 + 2NH3 ! U(VI) before
4NH4DEHP + TOPO + extraction.
UO2(CO3)34 + 4NH4+
Musikas et al.
Today, phosphoric acid for use in the pharmaceutical or food industry must
be free from uranium and other harmful metals. In the future, the processing
of phosphate rock for production of fertilizers may be an important source
of uranium because uranium removal may become compulsory to avoid its
dissemination into the environment from the use of phosphate-based fertil-
izers.
12.3.4 Recovery as a By-Product from Production

of Base Metals
Large-scale winning of copper by acidic leaching of copper ores sometimes
results in waste solutions containing appreciable amounts of uranium. The
uranium bearing aqueous raffinate from copper extraction is usually a dilute
sulfuric acid solution. Uranium can be recovered using the same technique
as described in section 12.3.1. A typical example is uranium production at
the Olympic Dam mine in Australia, where the copper ore bodies are esti-
mated to contain a total of over a million metric tons of uranium.
12.3.5 Recovery from Seawater

The chemical properties of U, such as the stability of the U(VI) ions as
UO2þ (2x2) ,
2 and its water soluble anionic carbonato complexes, UO2 (CO3 )x
where x = 2 or 3, make sea water the largest single source of U on earth. Its
low concentration, 3.3 mg per m3, is the main problem for its recovery.
Estimates of the price of U from seawater have shown that the cost is several
orders of magnitude higher than for recovering U from high-grade ores.
However, recent research on calixarenes [23] has showed that this type of
macrocycle can achieve very high separation factors between U(VI) and
other cations if the number of functional groups on the calixarenes corre-
sponds to the number of ligands needed to saturate the coordination sphere
of the UO2þ2 ion in its equatorial plan. This is achieved with calixarene(5)
or (6). High separation factors are needed for the recovery of U as the
seawater contains many other elements. Figure 12.3 shows the structure
of two lipophilic p-tert-butylcalixarenes(6), bearing carboxylate or hydro-
xamate groups on the lower rim, which have been considered as extractants;
some separation factors are reported in Table 12.5. It can be seen that
the best functional group for the calixarene(6) is the hydroxamic group,
CONHOH, probably because, at the pH of seawater, neutral U(VI) species,
UO2(calix(6))H4, are formed in the organic phase; with carboxylate groups
the lower pKa of R-COOH favors anionic species. It appears that with the
carboxylate extractant Ni2+ , Zn2+, and Fe3+ have an important negative
effect upon the U(VI) extraction, but with the hydroxamate only Fe3+ in
excess affects the U(VI) extraction.
518 Musikas et al.
Fig. 12.3 Developed formula for calixarene(6) selective extractants for U(VI) extraction
from sea water. (From Ref. 23.) (1) Carboxylate functions; (2) hydroxamate functions.
12.4 RECOVERY OF THORIUM FROM ITS ORES

Thorium, which consists mainly of the isotope 232Th, has often been pro-
posed as a secondary source of nuclear energy [24,25]. Irradiation of 232Th
with neutrons produces 233Th (t½ = 22.4 min), a beta-emitter that decays to
fissionable 233U. The world reserves of 232Th are about three times larger
than those of U. However, because the U nuclear fuel cycle was developed
first and is easier to use for nuclear energy production, the thorium cycle has
not received as much attention. However, thorium must be considered as a
future energy source. Fission of thorium produces less hazardous waste,
because the amount of long-lived transuranium isotopes is about 100 times
less than in the uranium cycle.
Accelerator-driven reactors projects based on the Th cycle have been
proposed [26].
Table 12.6 contains, in a simplified way, the composition, location,
and treatment of main thorium ores. The purification of thorium by TBP
extraction is illustrated in Fig. 12.4. This purification takes place after the
dissolution of Th in nitric acid, generally from a hydroxide cake. When the
Th is dissolved in sulfuric acid, purification is achieved by extraction with
long-chain alkylamines.
12.5 REPROCESSING OF IRRADIATED

NUCLEAR FUELS
12.5.1 Irradiated Uranium
The initial objective for reprocessing irradiated nuclear fuels [27,28] was
to recover and purify 239Pu for military applications. However, with the
introduction of commercial nuclear power reactors, reprocessing of the

irradiated fuels was considered as the best use of the available uranium re-
sources. Reprocessing also allowed recovery of valuable fissionable nuclides
that could be recycled in fast breeder reactors.
12.5.1.1 Purex Process Flow Sheet
Irradiated UO2 is dissolved in nitric acid, resulting in a dissolver solution
with the approximate composition listed in Table 12.7. This is treated by the
Purex process. The main steps in the conventional Purex process are shown
schematically in Fig. 12.5. All existing plants listed in Table 12.8 use some
variation of the Purex process. Typically, the extractant composition (per-
centage TBP, diluent) and the extraction equipment (i.e., pulse columns,
mixer-settlers, etc.), vary from plant to plant. However, the upper con-
centration limit is 30% TBP to prevent a phase reversal due to the increased
density of the fully loaded solvent phase.
Figure 12.5 illustrates the basic components of the Purex process; three
purification cycles for both uranium and plutonium are shown. High levels
of beta and gamma radioactivity are present only in the first cycle, in which
99.9% of the fission products are separated. The other two cycles, based
upon the same chemical reactions as the first cycle, obtain additional decon-
tamination and overall purity of the uranium and plutonium products.
Distribution ratios of the actinides in largest concentration and
valency state, and some important fission products, which form the basis for
the design of the Purex process flow sheet, are shown in Table 12.9.
A typical flow sheet for the first cycle of the Purex process is shown in
Fig. 12.6. The main steps are as follows:
Table 12.5 Comparison of the Extraction Selectivity of Two p-

tert-butylcalix(6)arene, Uranophiles, Bearing 6 Carboxylates 1(6)
or 6 Hydroxamates 2(6) Groups on the Lower Rim
% UO22+ extr. % UO22 + extr.
x+
Metal ion M /UO22+ by 1(6) by 2(6)
None 0 100 100
Mg2+ 1000 100 100
Ni2+ 10 77 98
Zn2+ 12 51 96
Fe2+ 10 0 66
Fe3+ 1 0 100
Note: The Organic Phase is 104 mol dm3. Calixarene in CHCl3. Aqueous
Phase is 2 105 mol dm3 UO2(CH3COO)2 Buffered to pH 5.9 by
102 mol dm3 acetate. Phase Volume Ratio Org/Aq = 0.2.
520 Musikas et al.
Table 12.6 Commercial Processes for the Recovery of Th from Its Main Ores
Types of ores Ore
(main locations) composition Leaching Purification
Monazites (Brazil, (La, Ce, Th) Hot concent- Dissolution of oxides

India, Australia, PO4 rated NaOH into HNO3 and then TBP;
South Africa, or H2SO4 alkylamines for Th
United States) in sulfate solutions
Thorianite ThO2 HNO3 or TBP for nitrates
Uranothorianite (U,Th)O2 H2SO4 Alkylamines for sulfates
Thorite ThSiO4 Hot H2SO4 Alkylamines
Uranothorite (U,Th)SiO4 Hot H2SO4 Alkylamines
(United States, Canada)
1. Coextraction of 99.9% of Pu and U and some 90% of 237Np from 3–

4 mol dm3 HNO3 solution into 30 vol% TBP, diluted with a mixture of
aliphatic hydrocarbons or, occasionally, with pure dodecane. Fission
products and trivalent actinides (Am, Cm) remain in the aqueous raf-
finate.
2. U-Pu partitioning: U(VI) remains in the TBP phase when Pu(IV) is
reduced to Pu(III) and stripped (back extracted ). Several reducing
agents [e.g., U(IV), Fe(II), and hydroxylamine nitrate (NH2OH
HNO3)] can be used; all have their advantages and disadvantages.
Reduction of Pu(IV) to Pu(III) in the U-Pu partition step is a process
that has to be carefully controlled, because inadequate reduction can
result in accumulation of plutonium in the partitioning equipment and
increase hazards from nuclear criticality:
(a) U(IV) solutions have to be prepared at the plant site and usually
have to be stabilized against oxidants by addition of a nitrite
scavenger such as hydrazine (N2H4), which can lead to the for-
mation of highly explosive hydrazoic acid (HN3). Although the use
of U(IV) as reducing agent has the advantage of adding no new
salts to the high-level waste, it does increase the solvent load and
can lead to third phase formation. As it follows the uranium
stream in the Purex process, it is not suitable in the second and
third cycles, where Pu is purified from traces of U.
(b) Instead other reducing agents, such as ferrous sulfamate or hydro-
xylamine, are preferred [29]. Ferrous sulfamate, Fe(SO3NH2)2,
rapidly reduces Pu(IV) to Pu(III) and the sulfamate ion reacts with
HNO2 to prevent the HNO2-catalyzed oxidation of Pu(III) by
NO 3 . However, ferrous sulfamate adds some undesirable inor-

ganic salts to the aqueous high-level waste.
(c) Hydroxylamine is a kinetically slow reducing agent and has to be
used in conjunction with other reagents, such as U(IV) and
hydrazine.
(d) Electrolytic reduction of Pu(IV) to Pu(III) appears to be a very
useful method of reduction. However, there is no large-scale plant
experience, and a nitrate scavenger, such as hydrazine, is often
necessary.
3. Uranium stripping: Dilute HNO3 solutions at 45–508C are used to
remove uranium from the TBP phase. Traces of the fission products
ruthenium and zirconium are eliminated in the second and third cycles
of the Purex process. Also, in the second and third cycles, neptunium
and the last traces of plutonium are removed from the uranium product.
12.5.1.2 Solvent Degradation and Regeneration
Routine regeneration of the TBP solvent is an important step in the Purex
process. The TBP, as well as the diluents, are degraded by hydrolysis and
Fig. 12.4 Simplified flow sheet used in the recovery of thorium from its ores.
522 Musikas et al.
Table 12.7 Composition of Dissolver Solution from Several Reactor Fuel Types
Graphite Pressurized Fast breeder
Fuel type characteristics gaseous reactor water reactor reactor
Irradiation MW day per metric ton 4,000 33,000 80,000
Activity b,g after 150 days of cooling 1,400 4,500 18,000
(Ci per kg of fuel)
U(VI), mol dm3 1.05 1.05 0.85
Pu(IV), mol dm3 0.00275 0.01 0.22
HNO3 mol dm3 3 3 3
Zr, mol dm3 0.0013 0.0095 0.0184
Ru, mol dm3 0.0005 0.0058 0.02
Fission products (kg per metric ton) 4.16 35 87
Transplutonium elements, mol dm3 0.0002 0.0028
Note: 1 Ci (Curie) corresponds to 3.7 1011 Bq (Becquerel, or dis./s).

Source: Ref. 82.
radiolysis. Degradation of TBP results in products such as dibutylphos-

phoric acid (HDBP), monobutylphosphoric acid (H2MBP), and phosphoric
acid. Depending upon the type of diluent, hydrolytic and radiolytic attack
produces nitrato esters, nitro compounds, carboxylic acids, ketones, and
others. The degradation products of TBP are stronger complexants for ca-
tions than TBP itself, and their complexes are not very soluble in the organic
phase. Elimination of the degradation products, particularly the organo-
phosphorus species, prior to solvent reuse is an important factor in the
success of the Purex separations.
Degraded TBP process solvent is typically cleaned by washing with
sodium carbonate or sodium hydroxide solutions, or both. Such washes
eliminate retained uranium and plutonium as well as HDBP and H2MBP.
Part of the low-molecular-weight neutral molecules such as butanol and
nitrobutane, entrained in the aqueous phase, and 90–95% of the fission
products ruthenium and zirconium are also removed by the alkaline washes.
Alkaline washing is not sufficient, however, to completely restore the inter-
facial properties of the TBP solvent, because some surfactants still remain in
the organic phase.
Various additional (secondary solvent cleaning) procedures have been
proposed, including vacuum distillation [30], treatment with macroreticular
anion-exchange resins [31], and treatment with activated alumina [32,33].
These secondary cleanup operations increase solvent quality, and continue
to be studied to better evaluate their process and economic advantages. In
the French UP3 Purex plant, vacuum distillation is used to regulate TBP
concentration and solvent quality with beneficial effects. The latter include
523
Fig. 12.5 Schematic representation of the reprocessing of light-water reactor spent fuels. The solvent extraction steps are shown in the dotted
frames (HA, MA, LA, for high, medium, and low activity wastes, respectively).
524 Musikas et al.
Table 12.8 Principal Operating PUREX Plants

Site and plant Capacity Year
Country name (ton per year) Solvent commissioned
France Marcoule (UPI) 400 30% TBP–TPHa 1958
La Hague (UP2) 400 30% TBP–TPH 1976
La Hague (UP3) 800 30% TBP–TPH 1990
La Hague 800 30% TBP–TPH 1992
(UP2–800)
India Tarapur 100 30% TBP–Shellsol 1982
Japan Tokia Mura 210 30% TBP–dodecane 1977
United Kingdom Sellafield 1200 30% TBP–OKb 1992
Windscale 1000 20% TBP–OK 1971
United States Savannah River 1800 7.5% TBP–dodecane 1954
or
30% TBP–dodecane
a
TPH, acronym for hydrogenated tetrapropylene, a branched dodecane.
b
OK, odorless kerosene.
retention of metals and use of less concentrated Na2CO3 in the solvent wash,
which helps to limit the amount of sodium salts that must be vitrified.
12.5.1.3 Purex Process Improvements
Reprocessing of spent reactor fuels has been the subject of much interna-
tional debate [34]. Strategies ranging from no reprocessing at all to complete
reprocessing with, in some cases, recovery and transmutation of long half-
life nuclides into stable or shorter half-life nuclides, have been proposed.
Several countries, such as the United Kingdom, France, and Japan, have
adopted an intermediate policy that consists of recovery of fissile material
from spent fuels and storage of the highly radioactive wastes after vitrifi-
cation in deep geological repositories.
To increase the economic benefits of reprocessing, some improvements
in the Purex process appear to be highly desirable. Two approaches can be
followed:
1. make changes within the Purex process, or
2. develop new processes that are more advantageous.
The chemistry of the Purex process has been extensively investigated,
and it seems very difficult to find new major improvements. However, some
improvements in some process details may contribute to a decrease in re-
processing costs, especially if several changes are made simultaneously.
Examples of such changes are presented in next sections.
12.5.1.3a Decreased Number of Cycles

A decrease in the number of uranium and plutonium purification cycles
from three to two, or even one, would be highly advantageous. First-cycle
decontamination factors of uranium from neptunium and from the fission
products ruthenium and zirconium must be significantly improved to realize
such a decrease.
12.5.1.3b Improved and Simplified Separation
of Neptunium-237
Two routes are available to decontaminate uranium and plutonium from
neptunium [35]:
1. Route all neptunium as Np(V) to the high-level waste in the first
extraction cycle
2. Allow most of the neptunium to coextract with U(VI) as Np(IV) or
Np(VI) and separate it in downstream uranium purification cycles
The latter alternative is currently followed in Purex plants in France
and the United Kingdom and was used for years in the Purex plant at the
U.S. Hanford site. Drake [36] has discussed laboratory and plant-scale
experience in coextracting neptunium and uranium and in separating nep-
tunium from uranium. Simplified Purex process operations would be achieved
by diverting neptunium to the high-level waste in the first extraction con-
tactor. Results of bench-scale experiments to find effective reagents for
establishing and maintaining nonextractable Np(V) in Purex process ex-
traction feed have been discussed by Drake. Much more work is necessary to
Table 12.9 Distribution of Various Metallic Species Between 30%

TBP–Dodecane, 80% Saturated with U(VI), and HNO3 Solutions at 258C
3 mol dm3 1 mol dm3 0.01 mol dm3

a
Metal HNO3 HNO3 HNO3
U(IV) 3.3 1.5 0.8
Np(VI) 0.8 0.4 -
Np(V) Negligible Negligible Negligible
Np(IV) 0.12 0.04 Hydrolyzes
Pu(IV) 0.7 0.18 Hydrolyzes
Zr(IV) 0.01 0.0015 Hydrolyzes
Ru 0.0013 0.006 -
Ce(III) 0.0004 0.004
Sr(II) Negligible Negligible Negligible
Cs(I) Negligible Negligible Negligible
a
Tracer level concentrations except for uranium.
526
Fig. 12.6 Flow sheet for the first purification cycle in the Purex process (TPH is hydrogenated tetrapropylene, a commercial branched dodecane).
develop reductants that can be used successfully and routinely in operating

Purex plants. However, removal of all Np from the high-level raffinate before
vitrification may be required in the future. In that case, it would be desirable
to extract Np as fully as possible in the first extraction cycle.
12.5.1.3c Partitioning of Technetium-99
Long-lived (t½ = 2.1 105 years) 99Tc, present as TcO 4 in Purex process
HNO3 feed solutions, is partially coextracted with uranium and plutonium in
the first cycle. Unless separated in the Purex process, 99Tc contaminates the
uranium product; subsequent processing of the UO2(NO3)2 solution to UO2
can release some of the technetium to the environment. The presence of
technetium in the purification steps as well as in the uranium product causes
several other complications. Thus it is desirable to route all Tc into the high-
level waste. Efforts in this direction have been described in some recent flow
sheets [37].
12.5.1.3d Improved Waste Management
Techniques for improving Purex process waste management could have
a substantial financial impact on the costs of fuel reprocessing. Figure 12.7
Fig. 12.7 Volumes of wastes containing long half life isotopes generated by the re-
processing of spent nuclear fuels as a function of time. (From Ref. 38.) Comparison with
the nonreprocessing option.
528 Musikas et al.
shows the volumes of deep underground wastes generated for 1 ton of

U reprocessed in the UP3 plant from its start in 1980 [38]. The volumes
of each kind of waste are given and the volume necessary to store unrepro-
cessed spent fuel is shown for comparison. Several improvements have been
made since 1980, but more volume reduction is desirable, particularly in the
amount of sodium in spent solvent washes and in solutions used to decon-
taminate process equipment. Further, it is desirable from an environmental
point of view to remove all alpha emitters and long-lived 137Cs and 90Sr from
Purex process acid waste solutions, so-called waste partitioning; discussed in
sections 12.6 and 12.7, respectively.
12.5.1.3e Replacement of Tributylphosphate with
an Alternative Extractant
From the time the Purex process was first developed, various groups in
many countries have suggested candidate extractants to replace TBP in the
Purex process. For reasons discussed by McKay [10], all of these attempts to
replace TBP failed, in spite of advantages claimed for the replacement
reagents. French [39] and Italian [40] workers have investigated the prop-
erties of several N,N-alkylamides with a view to their use in a Purex-type
process. The following properties of N,N-dialkylamides are worthy of
note: (1) they are completely incinerable; (2) radiolytic and hydrolytic
degradation products are not deleterious to process performance; (3)
reducing agents are not necessary to separate plutonium from uranium; and
(4) physical properties of N,N-dialkylamide-diluent solutions are suitable
for use in existing Purex plant equipment. Figure 12.8 shows a chemical
flow sheet using an N,N-dialkylamide extractant for reprocessing spent
light-water reactor fuels, but pilot-plant scale tests have not yet been per-
formed.
12.5.2 Reprocessing of Irradiated Thorium Fuels

Natural thorium spiked with 233U, 235U, or 239Pu has been used as fuel in
several large test reactors. After irradiation, thorium fuels contain Pa, 233U,
Np, and fission products. The Thorex solvent extraction process is generally
used to reprocess spent Th-based fuels. As in the Purex process, the solvent is
TBP diluted in an appropriate mixture of aliphatic hydrocarbons. Figure 12.9
shows the Thorex process flow sheet used by Küchler et al. [41] for repro-
cessing high-burn-up thorium fuel.
Several difficulties arise from the use of TBP in the Thorex pro-
cess.Thorium (IV) is the major element present, and the loading capacity of
a 30% TBP–dodecane solution is considerably lower for the tetravalent
actinides than it is for U(VI); thus, a larger organic/aqueous phase ratio
Fig. 12.8 Flow sheet of the first purification cycle of U and Pu in a process using an
N,N-dialkylamide as extractant. (DOBA is C3H7CON(CH2CHC2H5C4H9)2; DOiBA,
(CH3)2CHCON(CH2CHC2H5C4H9)2; diluent TPH (branched dodecane).
must be used to achieve suitable recovery of the thorium. Distribution ratios

of Th(IV), Pa(V), and Zr(IV) are such that a good separation of Th(IV) is
possible only at low aqueous feed acidity. Two extraction-strip cycles are
necessary with fuels containing high quantities of hydrolyzable fission
products. In the first cycle, carried out with acidic aqueous feed, thorium
and uranium are separated from the bulk of the fission products. In the
second cycle, performed with an acid-deficient feed solution, the final puri-
fication of uranium and thorium is achieved. The performance of this two-
stage process is shown in Table 12.10.
If reprocessing of spent Th fuels is needed in the future, many draw-
backs of the Thorex process can be avoided by using N,N-dialkylamides
instead of TBP. It is known from studies of Pu(IV)–U(VI) separation with
amides that U(VI)–Th(IV) separation in acidic medium is easier than with
TBP. The advantages of amides over phosphates could probably be con-
sidered more freely for Thorex than for Purex plants because no Thorex
plant has been built yet. The extraction behavior of Pa(V) has not been
yet investigated but in view of the general chemical properties of the two
extractants, smaller distribution ratios are expected with the N,N-dialkyl-
amides.
530
Fig. 12.9 Two-stage acid Thorex process for highly irradiated fuels. Numbers in the frames indicate stage number, whereas numbers on
Musikas et al.
the lines indicate flow volumes relative to the feed volume (DOD and FP are dodecane and fission products).
12.6 SOLVENT EXTRACTION OF ACTINIDES FROM

WASTE SOLUTIONS
Several kinds of radioactive wastes are produced by the nuclear industry,
including those classified as high-level (HLW), low-level (LLW), and trans-
uranic (TRU) wastes. Typically, the latter are those solid and liquid wastes
that contain 3.7 106 Bq kg1 (100 nCi g1) or more of transuranic nuclides,
with a half-life (t½) greater than 20 years. In several countries, TRU wastes
must be eventually disposed of in deep geological repositories. Conversely,
low-level wastes, which contain <3.7 106 Bq kg1 (<100 nCi g1) of transura-
nium elements can qualify for disposal in less expensive near-surface facil-
ities. For economic reasons, therefore, there continues to be great interest in
several countries to develop efficient and practicable solvent extraction
processes for nearly quantitative extraction of all actinides, including, in
particular, trivalent americium and curium, from any and all HNO3 waste
solutions. In contrast with tetravalent and hexavalent actinides, which are
well extracted by the TBP and other monofunctional organophosphorus
reagents, Am3+ and Cm3+ are poorly extracted from strongly acid and
unsalted solutions. It is desirable to find reagents that will effectively extract
the trivalent actinides, as well as any residual IV and VI actinides that may
be present.
Several classes of reagents that have been or are currently being
investigated for their ability to extract actinides from both HNO3 and HCl
waste solutions are discussed in the next sections.
Table 12.10 Performance of Two-Stage Thorex Process and a Uranium

Purification Cycle
Decontamination factor
Thorex two-stage
Contaminant Thorium Uranium Third uranium cycle

Total gamma 6 104
10 5
170
Total beta 3 105 2 106 3 103
144
Ce 6 106 3 107 2 105
106
Ru 5 104 5 104 170
95
Zr 6 105 6 106 2 104
233
Pa 103 2 104
Source: Ref. 32.

532 Musikas et al.
12.6.1 Bifunctional Organophosphorus Reagents

Bifunctional organophosphorus compounds of the generic formula
(R0 Op )(ROm )PO-(CH2 )n -PO(ROm )(R0 Op )
are known to form complexes with metal ions [42]. These compounds include
diphosphonates (m = 1, p = 1), diphosphinates (m + p = 1), or diphosphine
oxides (m = p = 0). Higher distribution ratios for +3 actinides from strong
HNO3 solutions are obtained when n = 1, which corresponds to the ideal size
of a molecule for chelation of metal ions.
Bifunctional organophosphorus reagents in which one of the PO
groups is replaced by an amide (NCO) group have the generic formula:
(R0 Op )(ROm )PO-(CH2 )n -CONR1 R2 ;
with m and p = 0 or 1 and 0 < n < 5 (see Appendix D, 13 and 14). These
carbamoyl-organophosphorus compounds are much more effective than
diphosphonates, diphosphinates, and diphosphine oxides in extracting Am3+
and Cm3+ as well as (IV) and (VI) actinides from strong HNO3 media.
Replacement of one PO group with an amide group also allows much
greater flexibility in selecting hydrocarbon diluents and reagents for stripping
of actinides.
Two carbamoyl-organophosphorus extractants have been intensively
tested for their usefulness in plant-scale removal of any and all actinides
from HNO3 waste solutions: dihexyl-N,N-diethylcarbamoylmethylpho-
sphonate [(C6H13O)2POCH2CON(C2H5)2] and octyl(phenyl)-N,N-di-isobutyl-
carbamoylmethylphosphine oxide [CMPO; C8H7(C6H5)POCH2CON(CH2
CH(CH3)2)2]. The latter extractant is used in the Truex process [43]. Figure
12.10 illustrates a reference Truex process flow sheet for use with an HNO3
waste solution. In the Truex process, CMPO is diluted with TBP in either an
aliphatic hydrocarbon or a halogenated hydrocarbon (e.g., tetrachloro-
ethylene) diluent. The TBP serves as a modifier, which greatly increases the
organic phase solubility of CMPO–metal complexes.
Nash et al. [44] have identified the principal hydrolytic and radiolytic
degradation product of CMPO to be C8H17(C6H5)POOH. Small con-
centrations of this acidic degradation product substantially increase the
difficulty of stripping actinides from the Truex process solvent. The washing
of degraded Truex process solvents with dilute Na2CO3 solutions restores
the quality of the solvent to near-pristine condition.
Of the many radioactive and inert constituents in waste solutions, only
99
Tc and lanthanides coextract with actinides into the Truex process solvent.
Trivalent lanthanides follow Am3+ and Cm3+ when the latter are stripped
with dilute HNO3; depending on the disposition of the americium and
Fig. 12.10 Generic Truex process flow sheet for use with nitric acid wastes (NPH is a
mixture of aliphatic hydrocarbons).
curium, separation of the associated lanthanides may be necessary. The 99Tc

follows uranium in the Truex process and co-strips with U(VI) into dilute
Na2CO3 solutions. Anion-exchange resin and solvent extraction procedures
have been used for separating uranium and technetium.
The Truex process has been extensively and successfully tested in the
United States with several different actual nitric and hydrochloric acid waste
solutions, and overall decontamination factors (Truex process feed to raf-
finate) of better than 105 have been achieved. This is encouraging for the
pursuit of investigations for HLW declassification to avoid the storage of
large volumes of wastes in deep underground repositories. Several runs
of the Truex process have also been made in Japan to study the partitioning
of the actinides contained in HLW [45], but with less success (e.g., coex-
traction of Ru). Many bifunctional organophosphorus extractants have
been tested in Russia [46], but for practical applications they have chosen an
534 Musikas et al.
extractant close to CMPO, diphenyldibutyl-CMPO. With this derivative the

solubility and third phase problems are worse than with CMPO when ali-
phatic hydrocarbons are used as diluent. Hence, the Russian workers
propose to use a fluoroether ‘‘fluoropol-732’’ (CF3 C6H4NO2) [47] to
overcome these problems [48]. A review and comparison of the existing
processes has been published by U.S. authors [49].
12.6.2 1,3-Propanediamides (DIAMEX Process)

Bifunctional diamides, particularly the 1,3-propanediamides, are good ex-
tractants for actinides, including trivalent americium and curium, from
HNO3 waste solutions [50,51]. Figure 12.11 shows distribution ratios of
selected ions between 0.5 mole dm3 (C4H9CH3NCO)2CHC2H5OC6H13
[DMDB-(2-(3,6-OD),1,3-DA)P] in tert-butylbenzene and various aqueous
HNO3 solutions. Figure 12.12 indicates the effect of radiation upon some of
the distribution ratios. The perceived advantages of bifunctional diamides as
extractants for actinides in waste solutions include their ease of inciner-
ability and the fact that radiolytic and hydrolytic degradation products are
not deleterious to the process. These early results were confirmed by
extensive work in the last decade [52,53] in an EU collaboration program to
develop a process (DIAMEX) for the extraction of actinides. Thirty new
extractant molecules were synthesized and rated in order of ability to extract
trivalent actinides; selectivity toward the fission and corrosion products;
third phase boundaries as a function of the aqueous nitric acid concentra-
tion; and the loading by lanthanides [52,53].
The development of the Diamex process started with N,N0 -dimethyl-
dibutyltetradecylmalonamide [(CH3C4H9NCO)2CHC14H25, DMDBTDMA],
but an alternative molecule, N,N 0 -dimethyldioctylhexyloxyethylmalonamide
[(CH3C8H17NCO)2CHC2H4OC6H13, DMDOHEMA] was found to present
some advantages (and also disadvantages). Among the advantages is a better
decontamination from Mo(VI) (99.7% removal compared to 95% for
DMDBTDMA), Fe(III) (98.1% removal compared to 27%), and Ru (79%
compared to 50%) in countercurrent bench tests. The disadvantages are
mostly the formation of more degradation products, which were more
soluble in the organic phase, due to the replacement of the CH3C4H9N group
by CH3C8H17N.
Several tests using countercurrent separation in mixer-settler or cen-
trifugal extractors with simulated and genuine high-level waste showed that
the recovery of An(III, IV, VI) is quantitative. The back extraction of
An(III) was complete. These early flow sheets were not designed to strip
U(VI) and Pu(IV). The distribution ratios of these ions at low acidities are
lower than those measured with CMPO and one can guess that the stripping
Fig. 12.11 Distribution ratios of various metallic species between 0.5 mol dm3
(C4H9CH3NCO)2CHC2H4OC2H4OC6H13 [DMDB(2-3,6-OD,1,3-DA)P] t-butylbenzene solu-
tions and HNO3 solutions.
of Pu(IV) and U(VI) from Diamex will not be a difficult task. In addition,
the degradation products of the solvent do not retain the actinides in the
organic phase, as happens with the organophosphorus extractants.
Inactive metals in the wastes, such as Fe(III), Mo(VI), and Zr(IV), are
retained in the aqueous feed by addition of suitable quantities of oxalic acid.
As with CMPO, some Ru is extracted. Currently, the extraction of Pd, Tc,
Np and their poor stripping remain a problem for which process mod-
ifications are necessary. The flow sheet in Fig. 12.13 and Table 12.11 has
536 Musikas et al.
Fig. 12.12 Extraction of actinide ions by irradiated 1 mol dm3 [DMDB-(2-3,6-OD,1,3-

DA)P] (C4H9CH3NCO)2CHC2H4OC2H4OC6H13 into t-butylbenzene solutions: solid line,
irradiated in presence of 5 mol dm3 HNO3; broken line, irradiated in presence of
0.5 mol dm3 HNO3.
been tested in a battery of 16 centrifugal extractors with a genuine raffinate

from a small-scale Purex process at the European Institute for Transur-
anium Elements (ITU), Karlsruhe.
12.6.3 Monofunctional Organophosphorus Reagents

During development of the Truex process, various monofunctional organo-
phosphorus reagents [e.g., TBP; DBBP (dibutylbutylphosphonate); HDEHP
(di-(2-ethylhexyl)phosphoric acid)] were considered for removal of alpha-
emitters from HNO3 waste solutions. Two typical flow sheets for the use of
HDEHP and TBP are shown in Fig. 12.14. Because of the poor performance
of these two extractants, the processes were complicated. Distribution ratios
of trivalent actinides (e.g., 241Am) into TBP and DBBP are only high
enough for practical applications from either highly salted or low-acid
solutions, or very concentrated (i.e., 15–16 mol dm3 HNO3) solutions.
Actual plant-scale experience with DBBP was very disappointing [56]
because of the inability to adequately control the feed pH values in the range
required for effective americium extraction.
The use of HDEHP for extraction of actinides from waste solutions
also has several drawbacks, including extraction of trivalent ions at a pH
at which tetravalent ions such as Zr(IV) and Pu(IV) are hydrolyzed,
and difficulties in stripping tetravalent and hexavalent actinides. All in all,
monofunctional organophosphorus reagents are vastly inferior to their
bifunctional counterparts for extracting Am(III) and other actinides from
strong HNO3 media.
Many of these drawbacks were circumvented in a process developed at
Chalmers University, which was successfully demonstrated in continuous
operation using the old high-level raffinate concentrate from Purex pro-
cessing of low burn-up fuel. Although operation was very easy and stable,
the process flow sheet was complicated using bromoacetic acid, HDEHP,
TBP, and lactic acid [57–59].
12.7 EXTRACTION OF Sr-90 AND Cs-134, -135, -137

FROM ACIDIC WASTES
To facilitate near-surface disposal of nuclear wastes, efficient removal of
the t½ = 30 year fission product 90Sr is desirable. The SREX process [60]
Table 12.11 Decontamination Factors for Extraction (Mass in the Feed to Mass in
Raffinate) and for the Stripping [(Mass in the Feed Minus Mass in the Raffinate)/Mass in
the Organic Out] in DIAMEX Hot Test with Flow Sheet in Fig. 12.13
Element Extraction Stripping Element Extraction Stripping
a
Zr(IV) 1 - Mo(VI) 1 -a
b
Tc(VII) 120 30 Ru(II) 1.3 61
Pd(II) 157 96 La(III) 167 >46000
Ce(III) 211 >22500 Pr(III) 216 >44000
Nd(III) 157 >50000 Sm(III) 110 -a
Eu(III) 113 >300000 Gd(III) 49 >300000
Np(V, VI) 29 29b Am(III) 258 >25100
Cm(III) 301 3150
a
Below ICP-MS determination limit.
b
Accumulation in the back extraction section.
538 Musikas et al.
Fig. 12.13 Schematic drawing for the hot test of the DIAMEX process in a 16-stage
centrifugal extractor battery. (From Ref. 53.)
provides such removal from highly acidic solutions. The process employs an
n-octanol solution of the commercially available macrocyclic ether (cf.
Appendix D, Fig. 21), di-tert-butyldicyclohexyl-18-crown-6 (DtBuCH18C6)
as the extractant for 90Sr. Batch contacts with simulated waste solutions
show that the SREX process solvent is highly selective for strontium; of the
many contaminants only barium and technetium coextract with strontium to
any extent. Irradiation tests confirm that the SREX process solvent is
satisfactorily resistant to radiolysis and hydrolytic attack is also minimal.
The SREX process has been successfully tested in both batch contacts and
continuous countercurrent tests with simulated waste solutions. An overall
90
Sr decontamination factor (SREX process feed to SREX process raffinate)
of 4600 was achieved from Truex wastes.
In the past, the extraction of 90Sr and 137Cs was investigated for some
practical applications (heat and gamma ray sources) but the main interest
today is for decreasing the thermal power and the potential hazard of
nuclear waste in underground repositories. Results of extractions with some
Fig. 12.14 Schemes for partitioning Purex high-level waste. (a, data from Ref. 54; b, data from Ref. 55.)
539
540 Musikas et al.
other macrocyclic extractants are collected in Table 12.12, and the formulae
of some of the extractants are shown in Fig. 12.15.
12.8 OTHER NUCLEAR SOLVENT EXTRACTION

APPLICATIONS
12.8.1 Production and Purification of Transplutonium
Elements
During the period 1960–1980, work was carried out in the United States [64]
and in France [65] to isolate macro quantities of transplutonium elements
obtained from irradiated plutonium. Particular emphasis was placed upon
separation and purification of gram to kilogram amounts of 241Am for use
in smoke detectors and other applications [66]. The solvent extraction
processes used varied over time; two typical chemical separation flow sheets
are illustrated in Fig. 12.16.
12.8.2 Characterization of New Radioisotopes

Solvent extraction techniques are frequently used to identify and char-
acterize products of nuclear reactions, because the elemental (M) distribu-
tion ratios (DM) that provide the scientific information can be measured at
extremely low metal concentrations, as described in Chapter 4, providing
information about the chemical properties of the metal ion. Figure 12.17
illustrates an example of such studies. Sophisticated radiometric analytical
techniques greatly facilitate measurement of distribution ratios of radio-
nuclides at very low concentrations.
Often the products of nuclear reactions have very short half-lives. This
is especially true for the heaviest elements obtained by bombardment of
heavy targets with heavy ions. To identify and characterize such short-lived
nuclides, fast separations are required; solvent extraction techniques are well
suited to provide the required fast separations. For example, the SISAK
method [68] has been successfully used in conjunction with in-line gas jet
separators at heavy ion accelerators to identify short half-life actinide iso-
topes produced by collision of heavy atoms. The Sisak method involves use
of centrifugal contactors, with phase residence times as low as tenths of a
second, in conjunction with in-line radiometric detection equipment.
12.8.3 Measurement of Radioactivity by Liquid

Scintillation in Organic Solutions
Organic media are suitable for detecting and counting radioactivity
by scintillation methods, because the quenching of scintillation in organic
Table 12.12 Distribution Ratios of Cs + and Sr2+ Between Aqueous Nitric Acid Solutions and Various Organic Solvents
Solvent and extraction reaction Aqueous phase DM Refs.
0.01 mol dm3 dicarbollide (CoDC) in nitrobenzene 1 mol dm3 HNO3 + 0.03 mol dm3 Sr2 + 15.3 [61]
polyethylene glycol
Mn + + nHCoDC ! M(CoDC)n + nH + 0.1 mol dm3 HNO3 + 0.03 mol dm3 Sr2 + 569 [61]
polyethylene glycol
0.01 mol dm3 dicarbollide (CoDC) 0.5 mol dm3 HNO3 Cs + 22 [61]
in nitrobenzene 0.1 mol dm3 HNO3 Cs + 102.6
1 mol dm3 TBP + 0.1 mol dm3 HDNNS + 0.05M crown 1(C1) 3 mol dm3 HNO3 Sr2 + 0.06 [62]
into kerosene: Mn + + m(HNO3)C1 + nHDNNS ! 0.5 mol dm3 HNO3 Cs + 1.59
M(C1)m(DNNS)n + nH + + mHNO3 Cs + 1
1 mol dm3TBP + 0.1 mol dm3 HDNNS + 0.05 mol dm3 3 mol dm3 HNO3 Sr2 + 1.98 [62]
crown 2(C2) in kerosene: Mn + + m(HNO3)C2 + nHDNNS ! Cs + *0
M(C2)m(DNNS)n + nH + + mHNO3 0.5 mol dm3 HNO3 Sr2 + 2.57
2
10 mol dm3 di n-octyloxy calix[4]arene crown 6 1 mol dm3 HNO3 Cs + 33 [63]
(CA1) in orthonitrophenyl hexyl ether:
Csþ þ NO 3 þ CA1 ! CsNO3 CA1
102 mol dm3 di n-octyloxycalix[4]arene crown 6 1 mol dm3 HNO3+ Cs + 25 [63]
(CA1) in orthonitrophenyl hexyl ether: 4 mol dm3 NaNO3
Csþ þ NO 3 þ CA1 ) ! CsNO3 CA1
102 mol dm3 benzyloxycalix[8]arene N-diethylamide 1 mol dm3 HNO3 Sr2 + 20 [63]
(CA2) into orthonitrophenyl hexyl ether: Cs+ <.001
Sr2þ þ 2NO 3 þ CA2 ! SrðNO3 Þ2 CA2
102 mol dm3 benzyloxycalix[8]arene N-diethylamide 1 mol dm3 HNO3+ Sr2 + 5 [63]
(CA2) in orthonitrophenyl hexyl ether: 4 mol dm3 NaNO3 Cs + 0.1
Sr2þ þ 2NO 3 þ CA2 ! SrðNO3 Þ2 CA2
541
Fig. 12.15 Formulae of the ligands used for Cs(I) or Sr(II) extraction from acidic media
and cited in Table 12.12.
542
Fig. 12.16 Flow sheets used for separation of transplutonium elements from irradiated Pu: (a) France (Fontenay-aux-Roses); (b) United States
(Oak Ridge National Laboratory). (Data from (a) Ref. 64; (b) Ref. 65.)
543
544 Musikas et al.
Fig. 12.17 Determination of the oxidation state of submicro traces of 256Lr (open cir-
cles) and 255No (solid circles). Solvent: 0.2 mol dm3 TTA–MIBK; aqueous phases: buf-
fers at various pHs. (From Ref. 67.)
environments is much less than in water. This is used in the PERALS

method [69], in which radionuclides are transferred by solvent extraction
into an organic phase of aromatic molecules. In the Perals system an elec-
tronic device based on the delay of counting after detection allows the
discrimination of alpha and beta particles. This extends the range for the
detection of alpha-emitting nuclides in solution. The good yield of scintil-
lation measured by the number of photons detected by energy absorbed also
allows the consideration of low-resolution alpha spectrometry, and this also
improves the interest in radioactivity analysis by scintillation. The absorp-
tion of neutrinos in organic media followed by the counting by scintillation
of the radionuclides formed is the basis of several projects in experimental
physics dealing with the fundamental properties of neutrinos [70].
12.9 NEW DEVELOPMENTS

12.9.1 Partitioning of Wastes
The fission products contain some 50 elements located in the middle part of
the periodic system. Some of the fission products are produced in con-
siderable amounts, though their radioactivity is rather low [71], and they
were therefore thought to have a potential value on the mineral market.
Some other fission products were considered as suitable for radiation

sources (heat sources, radiography, etc.). Thus, processes were designed to
isolate these from the rest of the fission products. Today, there is little
interest in such applications.
However, there is renewed interest in separating some key nuclides
(e.g., 90Sr, 99Tc, 129I, 134,135,137Cs), some of which have important medical
applications. However, the prime interest comes from the desire to generate
wastes that could be more easily and safely stored in deep geological
repositories. In the most favorable situation, they could be safely stored in
relatively inexpensive near-surface engineered facilities. The separation of
the remaining actinides (traces of Pu, Np, Am, Cm, etc.) and of the most
hazardous long-lived fission products is referred to as partitioning. The
partitioned radionuclides would either be transmuted (e.g., actinides, 99Tc,
129
I) in nuclear reactors or accelerators to form stable or short-lived
nuclides, or stored (90Sr and 137Cs) in hardened engineered near-surface
facilities for 300–500 years to allow them to decay to negligible levels.
Solvent extraction is thought to be a key technique for partitioning and such
partitioning could improve the economics of final waste handling.
The current conventional approach (Fig. 12.18a) to partition radio-
nuclides from Purex process acidic high-level waste involves a sequential
series of separation processes to remove selected nuclides or groups of
radionuclides. An example of this approach is the United States ‘‘clean use
of reactor energy’’ (CURE) concept [72] that utilizes the Truex process to
remove actinides, 99Tc, and lanthanides; the Srex process to remove 90Sr
from the raffinate from the Truex process; and precipitation of cesium
phosphotungstate from the Srex process raffinate by addition of phospho-
tungstic acid. Results of batch contacts to test the Truex and Srex process
portions of the Cure concept were discussed previously. The cesium phos-
photungstate precipitation process has been used extensively on a plant scale
at the U.S. Hanford site [73].
The Cure concept and the Japanese OMEGA concept [74] are quite
similar, although they will likely employ different reagents. For example, for
removal of actinides from Purex process high-level waste, Japanese scientists
are investigating the properties of di-isodecylphosphoric acid. The Japanese
Omega concept also involves removal of valuable rhodium and palladium
from the high-level waste.
Even though the Cure and Omega concepts are year-2000-state-of-the-
art, they have a number of potential disadvantages including the need to
handle the highly radioactive waste in three or more extraction cycles.
Improved partitioning processes that will accomplish the separations in
fewer steps are highly desirable. For example, a solvent extraction process
that provides for quantitative recovery of both 90Sr and 137Cs in a single
546
Fig. 12.18 Concepts for partitioning radionuclides from Purex process high-level waste: (a) today’s technology; (b) tomorrow’s technology.
Note: Numbers in parentheses indicate separations that must be performed remotely.
extraction cycle is very much needed. In addition, a desirable goal is to find,

if possible, an extractant that will extract (III), (IV), and (VI) actinides, plus
TcO 4 , and reject the lanthanides to the aqueous raffinate. One such pro-
gram is the SPIN (solvent partition–incineration) program [75] for separa-
tion of the trivalent actinides from the trivalent lanthanides. This separation
is necessary because nuclear incineration of the actinides requires that the
neutron-consuming lanthanides be removed.
Recent investigations [76,77] indicate that certain soft donor ligands
containing nitrogen and sulfur atoms extract trivalent actinides better than
trivalent lanthanides from nitric acid solutions. However, the particular
extractants studied then were not considered suitable for plant-scale use
because of poor chemical stability [e.g., di(2-ethylhexyl)dithiophosphoric
acid] against radiation or relatively high aqueous phase solubility (tris-
pyridyl-triazine from dinonylnaphthalene sulfonate TPTZ mixtures into
CCl4). These results, however, encourage further investigations of soft
donor ligands for use in preferential extraction of (III) actinides, as well as
(IV) and (VI) actinides, from HNO3 solutions. The search for good extrac-
tants remains a challenge for the future, because ligands bearing soft donor
atoms will bind better with the d transition elements contained in the fission
and corrosion products.
12.9.2 Trivalent Actinide-Lanthanide Group

Separations
Considerable research has been done within the EU through the SPIN
program [51,52,62,75] to improve methods for the separation of the minor
actinides and some of the fission products from HLW. Some of the
results have been already discussed. The goal was to optimize the formula of
the soft donors (N and S) extractants in order to use them in large-scale
separations of trivalent actinides from trivalent lanthanides. Both types of
donors were studied. For the nitrogen donors the research started from tris-
pyridyl-triazine. Substituting three tert-butyl groups on the pyridine rings
[tri-(tert-butylpyridine)triazine] increased the lipophilicity of the extractants
and allowed two drawbacks to be overcome, i.e., the solubility of the
nitrogen donor in the aqueous phase as an acidic species and the rather high
pH for extraction of trivalent ions. The solubility of the nitrogen ligand in
aqueous solution became negligible and the influence of the branched alkyl
substituents on the Am(III), Eu(III) distribution ratios can be seen in
Fig. 12.19. The results shown indicate that the design of a flow sheet can be
considered if an alternative to a-bromocapric acid can be found. This acid is
undesirable because its bromine atom can lead to the production of cor-
rosive HBr by radiolysis or hydrolysis. a-cyanocarboxylic acids have been
548 Musikas et al.
Fig. 12.19 Distribution ratios of Am(III) and Eu(III) between 1 mol dm3 a-bromocapric
acid + 0.02 mol dm3 nitrogen ligand into TPH as a function of aqueous HNO3 con-
centration. (From Ref. 52.)
tested, but their low solubility in TPH is another drawback [53]. Altogether
78 nitrogen donor molecules were synthetized. Among them are the tri-
dentate nitrogen donors from the family of the 2,6-bis-(5,6-dialkyl-1,2,4-
triazine-3-yl)-pyridine prepared by Kolarik et al. [78].
Sulfur donors belonging to the dialkyl-thiophosphorous acids were
also investigated, and nine molecules have been considered. Separation
factors between Am(III) and Eu(III) as high as 17,000 have been found and
confirm the ability of sulfur donors to promote intergroup 5f-4f separations
of trivalent ions. As indicated by the Pauling electronegativities of N and S,
the S donors should bind better to the 5f trivalent ions than the N donors.
With the individual sulfur donors or synergistic mixtures investigated, the
extraction reactions were:
M3þ þ 3R2 PSSH ! M(R2 PSS)3 þ 3Hþ
M3þ þ 3R2 PSSH þ nR03 POM(R2 PSS)3 (R03 PO)n þ 3Hþ
The many laboratories involved in this work allowed a large experimental

program to be set up to test the behavior of these soft donor extractants in
countercurrent separations. The trivalent actinide-lanthanide separation
process was named SANEX and four chemical systems have been explored
(Fig. 12.20):
Fig. 12.20 Extractants, diluents, and aqueous feeds for the trivalent 5f–4f ion group
separations used in the SANEX processes.
550 Musikas et al.
1. SANEX 1 was based on the preferential extraction of trivalent actinides

over lanthanides using as solvent 0.02 mol dm3 tris-pyridine (Tpy) +
1 mol dm3 a-bromocapric acid in tert-butylbenzene. Flow sheet calcu-
lations showed that this process is operable but is very sensitive to the
flow ratio (organic/aqueous) and the aqueous acidity, which must be
maintained around 0.01 mol dm3.
2. SANEX 2 used a solvent composed of 0.1 mol dm3 4,6-di-(pyridin-2-yl)-2-
(3,5,5-trimethylhexanoylamino)-1,3,5-triazine (BADPTZ) + 1 mol dm3
a-cyanodecanoic acid + 0.5 mol dm3 DMDBTDMA with TPH as
diluent. The aqueous feed was a mixture of actinides and lanthanides,
mimicking the genuine output of the Diamex process in 0.2 mol dm3
HNO3.
3. SANEX 3 used the 2,6-bis-(5,6-dialkyl-1,2,4-triazine-3-yl)pyridine fam-
ily of extractants containing the propyl group, which are able to selec-
tively extract trivalent actinides from 1 mol dm3 HNO3 [53,78]. The
extractant was 0.04 mol dm3 in a mixture of 70%/30% TPH/octanol. A
flow sheet based on this solvent has been tried and is shown schemati-
cally in Fig. 12.21; the results are given in Table 12.13. The feed was an
Fig. 12.21 Flow sheet for the hot test of SANEX 3 process using n-Pr-BTP.
Table 12.13 Decontamination Factors, DF, for the Extraction (Mass in

the Feed/Mass in the Actinide Back-Extraction) in the SANEX 3 Hot Test
Depicted in Fig. 12.21
% in actinide DF
Element % in raffinate strip % in solvent feed/strip
Am(III) 0.11 98.34 1.71 1.017
Cm(III) 0.15 93.89 8.07 1.065
Y(III) 114a 1.8 <0.06 54
La(III) 119a <0.05 <0.05 >2000
Ce(III) 118a <0.03 <0.03 >3800
Pr(III) 120a <0.05 <0.05 >1800
Nd(III) 119a <0.02 <0.02 >6300
Sm(III) 114a 0.4 <0.08 246
Eu(III) 112a 2.3 <0.37 43
Gd(III) 108a 3.1 <0.28 32
RuNO(III) 109a <1 0.66 >100
Pd(II) 0.09 0.97 75.5 103
Fe(II) 6.8 <2 73.4 >50
a
Uncertainties of analysis are responsible for mass balances >100%.
active solution simulating the aqueous output of the Diamex process. It

can be seen that Pd(II) and Fe(III) accumulate in the solvent, while Ru
is not extracted. The decontamination factors for Eu and Gd are rather
low [78], and kinetic or phase disengagement problems are probably
responsible for the poor separation.
4. SANEX 4: The general properties of the sulfur donors, dithio-
phosphoric acids, published earlier [79] were confirmed [53]. More
chemically stable molecules bis-(2,4,4-trimethylpentyl)dithiophos-
phinic acid, CYANEX 301 [80], or bis-(chlorophenyl)dithiophos-
phonic acid [81] were investigated. Flow sheets based on synergistic
mixtures of 0.5 mol dm3 bis-(chlorophenyl)-dithiophosphonic acid and
0.25 mol dm3 TOPO, which permit rather acidic feeds (0.5 mol dm3
HNO3) to be used, were also tested. The first experiments showed
several problems such as formation of gels, decomposition of the
extractant at the high acidities needed for stripping, and extraction and
accumulation of Fe, Pd, Tc, and Np in the solvent. Low extraction and
stripping kinetics prevent these experiments reaching the separation
performance expected from the test-tube distribution ratios. However,
very high Am(III)-Eu(III) separation factors, e.g., 17,500, were mea-
sured, indicating that the sulfur donors have a high potential for
552 Musikas et al.
trivalent actinide lanthanide group separation and must be considered

further.
12.9.3 Uranium Mining– Milling Operations

As noted earlier (see section 12.3.1), the Amex process, which uses long-chain
amines, is preferred over the Dapex process, which uses HDEHP, for solvent
extraction of uranium from H2SO4 leach solutions. Because the surfactant
properties of amine sulfates are conducive to formation of objectionable
emulsions, the Amex process is very sensitive to the presence of solids in the
H2SO4 leachate. For acceptable phase coalescence in the Amex process, feeds
should contain no more than 20 ppm solids. The Dapex process can tolerate
feeds containing as much as 100 ppm solids.
Thus there is considerable incentive to find extractants that could tol-
erate higher quantities of solids in H2SO4 leach liquors. Stripping of uranium
from the Amex process extractant and subsequent regeneration of the amine
solvent also consume considerable quantities of acid and base. Recovery
of uranium from H2SO4 solutions would be simplified if a convenient neu-
tral extractant could be found. An extractant with better selectivity for
vanadium and molybdenum than HDEHP and long-chain amines is also
desirable.
Radium, thorium, and other radionuclides accumulate in uranium mill
tailings. The potential environmental effects of these radionuclides has
become of increasing concern to the public. In the future, it may be necessary
to modify existing uranium recovery processes to accommodate removal of
radium and perhaps other radioactive decay products of uranium.
Thorium is produced in smaller amounts than uranium, but if its
production increases in the future the tailings problem will be very similar.
The rare earth industry also produces comparable radioactive effluents
because many minerals that contain rare earths (e.g., monazites) also in-
clude Th and U.
12.10 SUMMARY
Solvent extraction technology has an unchallenged role in the nuclear fuel
cycle and such technology has been aggressively developed and implemented
over the last five decades to a fully mature and complete status. Incentives,
either technical or economic, to make any short-term changes or perceived
improvements to present advanced engineering-scale fuel cycle practice are
not apparent and likely do not exist. Long-term changes in fuel processing
strategies, e.g., implementation and transmutation of actinides and selected
fission products, have been suggested by some. But even in this case, workable
and efficient solvent extraction systems employing either organophosphorus

or organonitrogen-based reagents for removing and separating actinides
from high-level waste solutions are known and await only pilot-scale test-
ing and demonstration in countercurrent equipment with fully radioactive
feed solutions, should a decision be made to implement a transmutation
approach.
Of course, even without engineering-scale economic and technical
drivers, studies of new nuclear solvent extraction technology can and will
proceed at academic and national laboratory institutions. Areas for which
new technology could be beneficial include, among others, development of
extractants that can be readily incinerated; detailed information concern-
ing the kinetics of extraction of various solutes; and perhaps, development
of contactors with very short residence times. Extraction kinetics must be
more carefully investigated in the future to be able to take advantage of
kinetic differences, especially between the actinides and the d-transition ele-
ments.
REFERENCES
1. Schulz, W.; Burger, L. L.; Navratil, J. D. Eds. Science and Technology of
Tributyl Phosphate, Vol. 3, Applications of Tributyl Phosphate in Nuclear Fuel
Processing; CRC Press: Boca Raton, Florida, 1990.
2. Hewlett, G.; Anderson, O. E. The New World 1939/1946; Pennsylvania State
University Press: State College, Pennsylvania, 1962.
3. Long, T. Engineering for Nuclear Fuel Reprocessing; American Nuclear Society:
La Grange Park, Illinois, 1978.
4. Wymer, R. G.; Vondra, B. L. Eds. Light Water Nuclear Reactor Fuel Cycle;
CRC Press: Boca Raton, Florida, 1981.
5. Stoller, S. M.; Richards, R. B. Eds. Reactor Handbook, 2nd Ed.; Interscience
Publishers: New York, 1961.
6. Lo, T. C.; Baird, M. H. I.; Hanson, C. Eds. Handbook of Solvent Extraction;
John Wiley and Sons: New York, 1983.
7. McKay, H. A. C.; Miles, J. H.; Swanson, J. L. Science and Technology of
Tributyl Phosphate; Schulz, W. W., Burger, L. L., Navratil, J. D., Eds.; CRC
Press: Boca Raton, Florida, Vol. 3, Chap. 1, 1990.
8. Danesi, P. R. Development in Solvent Extraction Alegret, S. Ed.; Ellis Horwood:
Chichester, U.K., 1988; Chap. 12.
9. Warf, J. C. J. Am. Chem. Soc., 1949, 71, 3257.
10. McKay, H. A. C. Science and Technology of Tributyl Phosphate; Schulz, W. W.
Navratil, J. D. Eds.; CRC Press: Boca Raton, Florida, 1990; Vol. 1, Chap. 1.
11. Smith, L. E.; Page, J. E. J. Chem. Soc., 1948, 67.
12. Coleman, C. F.; Brown, K. B.; Moore, J. G.; Crouse, D. J. Ind. Eng. Chem.,
1958, 50, 1756.
13. Chesne, A.; Koehly, G.; Bathellier, A. Nucl. Sci. Eng., 1963, 17, 557.
554 Musikas et al.
14. Peppard, D. F. Liquid-liquid extraction of metal ions. In Advances in Inorganic

Chemistry and Radiochemistry; Academic Press: New York and London, 1966;
Vol. 9, p. 1.
15. Schulz, W. W.; Navratil, J. D. Solvent extraction with neutral bidentate organic
phosphorus reagents. In Recent Developments in Separation Science, Li, N. N.
Ed.; CRC Press: Boca Raton, Florida, 1982; Vol. VII, pp. 31–72.
16. Myaesoedov, B. F.; Chmutova, M. K. New methods of transplutonium ele-
ments isolation, purification and separation. In Separation of f Elements, Nash,
K. L. Choppin, G. R. Eds.; Plenum Press: New York, 1995; p. 11.
17. Saran, M. S. Solv. Ext. Ion Exch., 1989, 7, 783.
18. Arhland, S.; Liljenzin, J. O.; Rydberg, J. Comprehensive Inorganic Chemistry;
Trotman-Dickenson A. E. Ed.; Pergamon Press: New York, Vol. 5, 1973; p. 465.
19. Musikas, C.; Le Marois, G.; Fitoussi, R.; Cuillerdier, C. Actinide Separations;
Navratil, J. D.; Schultz, W. W. Eds.; ACS Symposium Series 117, 1980; p.131.
20. Lloyd, P. J. D. Handbook of Solvent Extraction; Lo, T. C. Baird, M. H. I.
Hanson, C. Eds.; John Wiley and Sons: New York, 1983; p. 763.
21. Ritcey, G. M.; Ashbrook, A. W. Solvent Extraction, Principles and Applica-
tions to Process Metallurgy; Part 2; Elsevier Scientific: New York, 1979; Chap.
4.18.
22. Benedict, M.; Pigford, T. H.; Levi, H. W. Nuclear Chemical Engineering; 2nd
Ed.; McGraw-Hill: New York, 1981; Chap. 5.8.
23. Nagasaki, T.; Shinkai, S. J. Chem. Soc. Perkin Trans. 1991, 2, 1063.
24. Sherrington, L. Handbook of Solvent Extraction; Lo, T. C. Baird, M. H. I.
Hanson, C. Eds.; John Wiley and Sons: New York, 1983; Chap. 25.6.
25. Benedict, M.; Pigford, T. H.; Levi, H. W. Nuclear Chemical Engineering; 2nd
ed.; McGraw-Hill: New York, 1981; Chap. 6.8.
26. Carminati, F.; Klapisch, R.; Revol, J. P.; Roche, Ch.; Rubio, J. A.; Rubbia, C.
Report CERN/AT/93-47(ET) November 1993.
27. Naylor, A.; Wilson, P. Handbook of Solvent Extraction; Lo, T. C. Baird, M. H. I.
Hanson, C. Eds.; John Wiley and Sons: New York, 1982; p. 784.
28. Benedict, M.; Pigford, T. H.; Levi, H. W. Nuclear Chemical Engineering;
McGraw-Hill: New York, 1981; p. 466.
29. Miles, J. H. Science and Technology of Tributyl Phosphate; Schulz, W. W.
Burger, L. L. Navratil, J. D. Eds.; CRC Press: Boca Raton, Florida, 1990;
Vol. 3, Chap. 1, Part 2.
30. Guillaume, B.; Germain, M.; Puyou, M.; Rouger, H. Proceedings of Interna-
tional Conference on Nuclear Fuel Reprocessing, RECORD 87; Soc. Francaise
d’Energie Nucleaire: Paris, 1987; Vol. 1, p. 433.
31. Schulz, W. W. Macroreticular Ion Exchange Resin Cleanup of Purex Process
Solvent; U.S. Atomic Energy Commission Report ARH-SA-58, Atlantic
Richfield Hanford Co.; Richland, Washington 1970.
32. Mailen, J. C. Nucl. Technol., 1998, 83, 182.
33. Reif, D. J. Nucl. Technol., 1988, 83, 190.
34. Proceedings of the International Conference on Nuclear Reprocessing, RECORD
87; Soc. Francaise d’Energie Nucleaire: Paris, Vol. 1, 1987.
35. Kolarik, Z.; Shuler, R. Proceedings of Extraction’84; Inst. Chem. Eng. Sym-
posium Series No. 88, Pergamon Press: London, 1984; p. 83.
36. Drake, V. A. Science and Technology of Tributyl Phosphate; Schulz, W. W.;
Burger, L. L.; Navratil, J. D. Eds.; CRC Press: Boca Raton, Florida, 1990;
Vol. 3, Chap. 3.
37. Baron, P.; Boullis, B.; Germain, M.; Gué, J. P.; Miquel, P.; Poncelet, F. J.;
Dormant J. M.; Dutertre, F. Extraction cycles design for La Hague plants,
Proceedings of the International Conference on Future Nuclear Systems Emerg-
ing Fuel Cycles and Waste Disposal Options, Global ’93; Sept.12–17, Seattle
Washington, 1993.
38. Hugelmann, D.; Pradel, Ph. Revue Générale Nucléaire, 1995, 1, 30.
39. Musikas, C. Sep. Sci. Technol., 1998, 23, 1211.
40. Gasparini, G. M.; Grossi, G. Solv. Extr. Ion Exch., 1986, 4, 1233.
41. Küchler, L.; Schäfer, L.; Wojtech, B. Kerntechnik, 1970, 12, 327.
42. Schulz, W. W.; Navratil, J. D.; Recent Developments in Separations Science;
Li, N. N. Ed.; CRC Press: Boca Raton, Florida, 1980; Vol. 7, p. 31.
43. Schulz, W. W.; Horwitz, E. P. Sep. Sci. Technol., 1988, 23, 1191.
44. Nash, K. L.; Rickert, P. G.; Horwitz, E. P. Solv. Extr. Ion Exch., 1989, 7,
655.
45. Ozawa, M.; Nemoto, S.; Toghashi, A.; Kawata, T.; Onishi, A. Solv. Extr. Ion
Exch., 1992, 10, 829.
46. Prybylova, G. A.; Chmutova, M. K.; Nesterova, N. P.; Myaesoedov, B. F.;
Kabachnik, M. I. Radiokhimiya 1991, 33, 70.
47. Myaesoedov, B. F.; Chmutova, M. K.; Smirnov, I. V.; Shadrin, A. U. Global
’93: Future Nuclear Systems: Emerging Fuels Cycles and Waste Disposal Op-
tions, American Nuclear Society: La Grange Park, Illinois, 1993; p. 581.
48. Babain, V. V.; Shadrin, A. Yu. New Extraction Technologies for Management of
Radioactive Wastes in Chemical Separation Technologies and related Methods of
Nuclear Waste Management, Choppin, G. R.; Khankhasayev M. Kh. Eds.;
NATO Science Series, 2- Environmental Security; Kluwer Academic Publishers
1999; Vol. 53.
49. Horwitz, E. P.; Shulz, W. W. Metal Ion Separation and Preconcentration Pro-
gress and Opportunities Bond, A. H. Dietz, M. L. Rogers, R. D. Eds., American
Chemical Society, 1998.
50. Musikas, C.; Hubert, H. Solv. Extr. Ion Exch., 1987, 5, 877.
51. Nigond, L.; Musikas, C; Cuillerdier, C. Solv. Extr. Ion Exch., 1994, 12, 297.
52. Madic, C.; Hudson, M. J. High Level Liquid Waste Partitioning by Means of
Completely Incinerable Extractants; European Commission, Nuclear Science
and Technology, EUR18038 EN 1998; p. 208.
53. Madic, C.; Hudson, M. J.; Liljenzin, J. O.; Glatz, J. P.; Nannicini, R.;
Facchini, A.; Kolarik, Z.; Odoj, R. New Partitioning Techniques for Minor
Actinides; European Commission, Nuclear Science and Technology, EUR19149
EN 2000; p. 286.
54. Liljenzin, J. O.; Rydberg, J.; Skarnemark, G. Sep. Sci. Technol., 1980, 15,
799.
556 Musikas et al.
55. Cecille, L.; Dworschak, H.; Girardi, E.; Hunt, B. A.; Mannone, F.; Mousty, E.
Actinides Separations; Navratil, J. D. Schulz, W. W. Eds.; American Chemical
Society: Washington D.C., 1980; p. 427.
56. Schulz, W. W. The Chemistry of Americium; U.S. Department of Energy
Technical Information Center; Oak Ridge, Tennessee, 1976; p. 203.
57. Liljenzin, J. O.; Persson, G.; Svantesson, I.; Wingefors, S. The CTH-process for
HLLW treatment, Part I–General description and process design. Radiochim.
Acta. 1984, 35, 155.
58. Persson, G.; Wingefors, S.; Liljenzin, J. O.; Svantesson, I. The CTH-Process for
HLLW treatment, Part II-Hot test. Radiochim. Acta. 1984, 35, 163.
59. Persson, G.; Svantesson, I.; Wingefors, S.; Liljenzin, J. O. Hot test of a TAL-
SPEAK procedure for separation of actinides and lanthanides using
recirculating DTPA-Lactic acid solution. Solv. Extr. Ion Exch., 1984, 2(1), 89.
60. Horwitz, E. P.; Dietz, M. L.; Fisher, D. E. Solv. Extr. Ion Exch., 1990, 8, 557.
61. Rais, J.; Selucky, P.; Kyrs, M. J. Inorg. Nucl. Chem., 1976, 38, 1376, 1742.
62. Shuler, R. G.; Bowers, C. B. Jr.; Smith, J. E. Jr.; Van Brunt, V.; Davis, M. W.
Jr. Solv. Extr. Ion Exch., 1985, 3, 567.
63. Dozol, J. F.; Scwing-Weill, M. J.; Arnaud-Neu, F.; Böhmer, V.; Ungaro, R.;
van Veggel, F. C. J. M.; Wipff, G.; Costero, A.; Desreux, J. F.; de Mendoza, J.
Extraction and Selective Separation of Long-Lived Nuclides by Functionalized
Macrocycles; European Commission, Nuclear Science and Technology series,
EUR19605 EN 2000; p. 198.
64. King, L. J.; Bigelow, J. E.; Collins, E. D. Transplutonium Elements-Production
and Recovery; Navratil, J. D. Schulz, W. W. Eds.; ACS Symposium Series 161,
American Chemical Society: Washington, D.C. 1981; p. 133.
65. Berger, R.; Koehly, G.; Musikas, C.; Pottier, R.; Sontag, R. Nucl. Appl.
Technol., 1970, 8, 371.
66. Schulz, W. W. The Chemistry of Americium; U.S. Department of Energy
Technical Information Center: Oak Ridge, Tennessee, 1976, p. 30.
67. Silva, R.; Sikkeland, T.; Nurmia, M.; Ghiorso, A. Inorg. Nucl. Chem. Lett.,
1970, 6, 733.
68. Skarnemark, G.; Skälberg, M.; Alstadt, J.; and Björnstad, T. Nuclear Studies
with the Fast In-Line Chemical Separation System SISAK, Physica Scripta,
1986, 34, 597.
69. McDowell, W. J. Radioactivity Radiochem., 1992, 3(2), 26.
70. Raghavan, R. S. Phys. Rev. Lett. 1997, 78, 3618.
71. Choppin, G.; Liljenzin, J. O.; Rydberg, J. Radiochemistry and Nuclear Chem-
istry; 3rd Ed.; Butterworth-Heinemann, 2002; p. 593.
72. CURE: Clean Use of Reactor Energy, U.S. Department of Energy Report
WHC-EP-0268; Westinghouse Hanford Co.: Richland, Washington, 1990.
73. Schulz, W. W.; Bray, L. A. Sep. Sci. Technol., 1987, 22, 191.
74. Mukaiyama, T.; Kubota, M.; Takizuka, T.; Ogawa, T.; Mizumoto, M.; and
Yoshida, Y. Partitioning and Transmutation Program OMEGA at JAERI,
Proceedings of the GLOBAL’95 conference; Versailles, France, 1995, 1, 110.
75. Barré, J. Y.; Bouchard, J. French R&D strategy for the back-end of the fuel
cycle. Proceedings of the Global ’93; Seattle, Washington, September 1993; 12–
17.
76. Musikas, C.; Vitart, X.; Pasquiou, J. Y.; Hoel, P. Chemical Separations; King
C. J.; Navratil, J. D. Eds.; Litarvan Literature: Denver, 1986; Vol. 2, p. 359.
77. Musikas, C.; Vitorge, P.; Pattee, D. Proceedings International Solvent Extrac-
tion Conference ISEC 83; American Institute of Chemical Engineers: Denver,
1983; p. 6.
78. Kolarik, Z.; Müllich, U.; Gassner, F. Solv. Ext. Ion Exch., 1999, 17, 23, 1155.
79. Musikas, C. Proceedings of the International Symposium on Actinides/Lantha-
nides Separations; Honolulu, Hawaii, 1984; and Actinide-Lanthanide Separa-
tion; World Scientific: Singapore, 1985 p. 19.
80. Zhu, Y. Radiochim. Acta. 1995, 68, 95–98.
81. Modolo, G.; Odoj, R. J. Radioanal. Nucl. Chem. 1998, 228(1,2), 83.
82. Miquel, P. Bull. Inf. Sci. Tech., CEA 1973, 57, 184.
13
Analytical Applications of Solvent Extraction
MANUEL AGUILAR, JOSÉ LUIS CORTINA, and ANA MARÍA
SASTRE Universitat Politècnica de Catalunya, Barcelona, Spain
13.1 INTRODUCTION
The role of solvent extraction in analytical chemistry has steadily increased
since the mid-1950s as a powerful separation technique applicable both to
trace and macro levels of materials. Work in this field has provided the
basis for a rich store of analytical methodology characterized by high sen-
sitivity and selectivity as is described in Chapters 2–4. Developments of new
extractants and their application to separation of a growing variety of
compounds are recognized as an important area of analytical chemistry.
Advances in this field over the last 50 years have been reported in a large
number of publications, among them several monographs and reviews [1–5].
Because of the great range of concentrations (from weightless trace levels of
carrier-free radioisotopes to macro levels of several weight percent of metal
ions) for which quantitative separations by solvent extraction are applicable,
this technique is equally useful in analytical, preparative, and process
chemistry. Solvent extraction has been also used as a separation step in
many analytical techniques and methods in response to the new problems
posed in many other fields such as medicine, biology, ecology, engineering,
etc. Among these, automatic methods of analysis have gained a notable
momentum and have motivated the development of a large number of
commercial instruments.
Solvent extraction has also played a major role in sample pre- or
posttreatment to improve selectivity and sensitivity. Initially simple schemes
of liquid–liquid extraction were used as separation methods for the clean-
up and preconcentration of samples, mainly because of its simplicity,
559
560 Aguilar et al.
reproducibility, and versatility. Later, the distribution between two liquid

phases was used as an efficient tool in chromatographic separation processes.
More recently, the same principles have been implemented using supercritical
fluids in preconcentration, cleanup, and column separation schemes.
In the past decades, the principles and the properties of solvent
extraction reagents were used to develop extraction chromatography tech-
niques, to design different types of membrane electrodes, and to prepare and
develop impregnated materials. In such materials where the extractant is
placed in a solid matrix, commonly an organic polymer or an inorganic
adsorbent, advantage is taken of their improved properties when high
volumes of aqueous samples are to be treated.
13.2 ROLES OF SOLVENT EXTRACTION IN

ANALYTICAL CHEMISTRY
The analytical process can be defined as a set of operations separating the
untreated, unmeasured sample from the results expressed as required in
accordance with the ‘‘analytical black box’’ concept. However, following
modern schemes of analysis, the total analytical process can be defined by
a set of operations as shown in Fig. 13.1 [6]. According to these ideas, the
so-called preliminary operations comprise a series of steps such as sam-
pling, sample preservation and treatment (e.g., dissolution, disagregga-
tion), separation techniques, development of analytical reactions, and
transfer of an adequate portion of the treated sample to the detector. The
second stage of the analytical process requires the use of one or several
instruments to generate pertinent information. The resulting analytical
signal (optical, electrochemical, thermal, etc.) should be unequivocally
related to the presence, amount, or chemical structure of one or several
analytes. Solvent extraction has been extensively used in two of these steps
in the analytical process: step 4, sample preparation (pretreatment, sep-
aration) and step 5, which is more concerned with measurement. Details
of procedures and methodologies are described in the following para-
graphs.
13.3 SOLVENT EXTRACTION IN

SAMPLE PREPARATION AND
PRETREATMENT STEPS
In sample preparation or sample pretreatment steps there are a number of
important operations that may include: dissolution of the sample, trans-
formation of the elements into specific inorganic forms, conversion of the
Analytical Applications of Solvent Extraction 561
Fig. 13.1 General scheme of the analytical process.
analytes into alternative chemical species, separation of the analyte from

other chemical species present, and preconcentration. The significance of
these preliminary operations is generally very critical because they could be
the source of major errors that may hinder analyte preconcentration and
elimination of matrix effects. On the other hand, these procedures can be
rather complex and time consuming and thus require ample dedication.
Then, the first stage of the analytical process decisively influences the pre-
cision, sensitivity, selectivity, rapidity, and cost. Sample pre- and post-
treatments in analytical schemes are intended to:
1. Improve selectivity, by removal of interfering species from the sample
matrix
2. Improve sensitivity by means of preconcentration
3. Prevent the deterioration of the analytical system by a sample cleanup
step
13.3.1 Operation Modes

Liquid–liquid extraction is by far the most popular separation method for
the cleanup and preconcentration of samples because it is simple, repro-
ducible, and versatile. There are several ways to achieve these objectives,
from the original discontinuous (‘‘batch’’) and nonautomatic techniques to
continuous separation techniques incorporated with automated methods of
analysis. The methodologies can be classified into two general types:
562 Aguilar et al.
13.3.1.1 Discontinuous Methods (Batch)

Discontinuous methods are performed in conventional separating funnels in
one or more steps. In ultratrace analyses, tapered or specially profiled quartz
tubes are recommended because of their easier cleaning (more compact size),
the introduction of less contaminating material, and easier centrifugation in
the case of difficulties with phase separation. Shaking must be continued
until equilibrium is reached, which may last seconds, minutes, or (rarely)
hours, depending on the physicochemical properties of the system; more
than 2–5 min requires a mechanical shaker. Microscale extraction carried
out in autosampler tubes, followed by direct automatic introduction of the
organic phase into the atomizer, is recommended.
13.3.1.2 Continuous Methods
Continuous methods can be performed by different procedures, such as
solvent recirculation, extraction chromatographic techniques, or counter-
current chromatography. The introduction of extractors in continuous
methods such as segmented flow analysis (SFA), flow injection analysis
(FIA), and completely continuous flow analysis (CCFA) that are used as pre-
or postcolumn devices in chromatographic separation techniques (liquid
chromatography) have been used widely in the last decade. Continuous-flow
extraction involves segmenting of an aqueous stream with an organic sol-
vent and separation of the phases using a membrane separator [7]. This
option is seldom used in practice because the preconcentration factors
are small and solvent consumption is large. An interesting possibility
is offered by countercurrent extraction, in which gravitational forces
retain the organic phase while the aqueous phase is pumped through it. The
basic components of a continuous liquid–liquid extractor are shown in
Figure 13.2.
As can be seen in this figure, a continuous liquid–liquid extraction
system consists of three main parts and performs the following functions:
Fig. 13.2 General scheme of a continuous solvent extraction contactor.

1. It receives the two streams of immiscible phases and combines them into
a single flow with alternate and regular zones of the two phases (solvent
segmenter).
2. It facilitates the transfer of material through the interfaces of segmented
flow in the extraction coil, the length of which, together with the flow
rate, determines the duration of the actual liquid–liquid extraction.
3. It splits, in a continuous manner, the segmented flow from the extrac-
tion coil into two separated phases (phase separator).
The functions of these three parts are based on the same fundamental
principle, i.e., the selective wetting of internal component surfaces by both
organic and aqueous phases. In general, it is found that organic solvents wet
Teflon surfaces whereas aqueous phases wet glass surfaces.
13.3.2 Solvent Extraction for Separation Steps

Solvent extraction can facilitate the isolation of analyte(s) from the major
component (matrix) and/or the separation of the particular analyte from
concomitant trace or minor elements. Extraction is usually a fast and simple
process, that demands only very simple equipment but having as a dis-
advantage its rather low preconcentration coefficient. In practice, the sep-
aration/preconcentration step consists either of selectively removing the
matrix without affecting the analyte(s) or of isolating the analyte(s) with-
out affecting the matrix. There is a variety of separation/preconcentration
methods available [8]. The most popular include solvent extraction, pre-
cipitation, sorption and chromatographic techniques, volatilization, and
electrodeposition. The choice is dictated by the sample to be analyzed, the
analytes and concentration levels to be determined, and the characteristics of
the determination technique. In any case, the incorporation of a separation/
preconcentration step increases the analysis time, may result in losses of the
analyte(s), and at the same time raises demands both on the purity of
reagents used and the analytical expertise required.
The selectivity of the extraction is expressed by the separation factor S,
which is derived from the individual distribution coefficients (D1, D2) for
two species (1, 2):
S ¼ D1 =D2 (13.1)
This coefficient gives quantitative information for the separation of both

species. Working with mixtures of extractants, additional experimental
information is needed and the synergistic coefficient (SC) defined by [9]:
SC ¼ log D12 =ðD1 þ D2 Þ (13.2)

564 Aguilar et al.
where D12 denotes the distribution coefficient of the species for the mixture
of extractants. In analytical chemistry, for an extraction system based on a
pH-dependent extraction reaction, as in the case of acidic extractants, the
value of SC has been estimated by the following equation:
SC ¼ n pH50 (13.3)
where n is the charge of the metal ion and DpH50 is the difference of pH
corresponding to 50% extraction when the total concentration of the
extraction system is the same for the single system and for the mixtures.
13.3.3 Preconcentration for Trace

Element Determination
Preconcentration is an operation in which the relative ratio of trace com-
ponents vs. the macro component is increased; it is aimed, typically; for
overcoming limited detection characteristics of the instrumental technique.
The efficiency of this operation, the preconcentration factor (PF ), is defined
in terms of recovery as:
PF ¼ ½AT =½AoT (13.4)
where [AT] and [AoT ] are the concentrations of the microcomponent in the
concentrate and in the sample respectively. The need for preconcentra-
tion of trace compounds results from the fact that instrumental analytical
methods often do not have the required selectivity and/or sensitivity. Thus,
the combination of instrumental techniques with concentration techniques
significantly extends the range of application of the instrument. The chem-
ical techniques used in preconcentration can provide, in many cases, analyte
isolation, as well as high enrichment factors. The concept of sample pre-
concentration prior to determination could apply to many situations other
than just concentration enrichment and minimization of matrix effects.
Selection of a preconcentration scheme based on any liquid–liquid extrac-
tion step will depend upon on the type of analyte and/or sample matrix.
Concentration factors reasonably achieved by solvent extraction have
practical limits. However, considering a reasonable practical limit of ex-
tracting a 100 cm3 sample with 1–2 cm3 of organic phase, a concentration
factor of 100 to 50 can be achieved. One of the limitations of the pre-
concentration factor in batch solvent extraction is the difficulty of obtaining
good separation of the small volume of organic phase. Significant portions
of the organic solvent may adhere to the walls of the vessel, requiring
repeated washings, which decreases the concentration factor.
Thus, to conduct successful analyses for many organic and inorganic

compounds at trace concentrations, it is necessary to extract these com-
pounds and use a concentration step prior to analysis. Many of the tech-
niques developed for preconcentration are described in specialized books
[10]. Proper choice of the extracting solvent can often be the critical step in
the procedure.
13.3.3.1 Preconcentration of Inorganic Compounds
There has been a growing demand for highly sensitive analytical methods of
trace components in complicated matrices. Recent advances of new ligands,
chelating reagents, and instrumentation have improved detection and spec-
ificity. However, in the presence of an interfering matrix, or when the con-
centration of the component of interest is far less than the instrumental
detection limit, it is necessary to concentrate the trace component with high
selectivity beforehand. Although many types of reagents can be used to
extract a wide variety of metal ions from different media, chelating agents are
typically widely used. It is interesting to note that a considerable number of
methods for specific extractions with reagents such as 8-hydroxyquinoline
have been developed in the past, basically because the instrumentation at
that time was not sufficiently specific and so prior separation was required.
For example, interference in atomic absorption determinations are common,
and so separations of interfering species are frequently required. Group
preconcentration of trace elements has the objective of isolating the max-
imum number of elements in a single step using the minimum number of
reagents, with the excess reagents being retained in the aqueous phase. On the
other hand, by suitable choice of the appropriate pH and sometimes using
appropriate masking agents, these reagents may become highly selective and
even specific for a particular species. In some cases exchange techniques in
which a less stable metal chelate is the source of the chelating agent may be
very useful. Extraction reagents available for this purpose fall into several
basic categories (solvating, chelating, ion exchangers, etc.) as described in
Chapter 4.
A series of preconcentration techniques based on the use of a solid
phase carrying a liquid metal extractant has been developed [11,12]. These
solid phases substitute the solvent (liquid phase) carried of the liquid–liquid
extractant with a solid phase. Such impregnated materials have been pre-
pared using different types of solid supports: paper, resin beds, gels, poly-
urethane foams, flat and hollow fiber membranes, clays, silica, etc. The
major advantage of most of these solid adsorbents is that the functional
group is immobilized on a solid substrate, therefore providing the possibility
of either batch extraction of the analytes from solution or using the solid
phase in a column.
566 Aguilar et al.
13.3.3.2 Preconcentration and Separation

of Organic Compounds
A complete scheme for trace analysis of organic compounds generally
consists of sampling, extraction, prefractionation, and analysis by gas chro-
matography (GC) or gas chromatography/mass chromatography (GC-MS).
Determination of organics at parts per trillion (ppt) levels can be performed
by combining sensitive and selective detection with sample preconcentra-
tion. Increasing the degree of preconcentration can make the limits of
detection extremely small. Many standard methods of analysis [13] often
include preconcentration as an integral part of the sampling and extraction
procedure. It is especially important for environmental samples where many
toxic and carcinogenic compounds are distributed from a wide variety of
sources. Because of their serious damaging effects at low levels, to assess
their environmental impact it is necessary to achieve the greatest possible
analytical sensitivities. In some cases, detection limits as low as a few ppt are
required. In addition to the polynuclear aromatic hydrocarbons (PAHs),
much attention has been focused on many halogenated pollutants such as
pesticides, polychlorinated biphenyls (PCBs), trihalomethanes (THMs),
polychlorinated dibenzofurans (PCDFs), and dibenzo-p-dioxines. Such sub-
stances can be found in air, water, and solid and biological samples. Most
preconcentration techniques fall into two classes: solvent extraction fol-
lowed by solvent reduction or sorbent trapping with subsequent solvent
elution or thermal desorption. There are many variations of these methods,
and they are frequently used in combination.
13.3.3.2a Solvent Extraction Reduction
Solvent extraction reduction is most frequently performed mainly in con-
nection with the extraction of solid and biological samples by liquid parti-
tion. Extractions are typically accomplished using a Soxhlet apparatus that
provides the benefits of multiple extractions. By repeated distillation and
condensation of the solvent, the apparatus allows multiple extractions using
the same (small) volume of solvent. Soxhlet extraction has been a standard
method for many decades, and it is often the method against which other
extraction methods are measured and verified [14].
Preconcentration will only give precise results for quantitative work if
the initial extraction technique gives high, or at least known and repro-
ducible, recoveries of the desired compounds from the initial sample. As
typically, several cycles are needed, and the solvent containing the extracted
compounds must be concentrated to a small volume. This is normally carried
out by evaporation under reduced pressure. Volatile compounds may be lost
in this procedure. However, for many applications the compounds of interest
are nonvolatile compared to the extracting solvent. In such applications, the

use of a rotary evaporation technique is rapid and straightforward, although
severe losses of even nonvolatiles can be experienced without careful sample
handling. The major problem involves the physical handling of the sample, as
even the very small volume losses to the glass walls of the recovery flask or
the disposable pipettes commonly used for sample transfer may result in
significant and nonreproducible component loss.
13.3.3.2b Accelerated Solvent Extraction (ASETM)
Accelerated solvent extraction is a relatively new extraction technique using
equipment that holds the sample in a sealed high-pressure environment to
allow conventional solvents to be used at higher temperatures [15]. Using a
higher temperature without boiling, ASE allows smaller volumes of solvent
to be used in a single-stage extraction. Extraction kinetics are also faster, so
the entire process is much faster than Soxhlet extraction. After heating, the
cell is allowed to cool to below the normal boiling point of the solvent and
pressure is applied to the cell to force the solvent and extracted materials
through a filter.
13.4 SOLVENT EXTRACTION AS A MEANS

OF ANALYTICAL DETERMINATION
In the previous section, the role of solvent extraction was limited to pre-
paring the analyte for subsequent analysis. A large majority of procedures
that use solvent extraction in chemical analysis are used in this fashion.
However, the extraction itself, or rather the distribution ratio characteriz-
ing it, may provide an appropriate measured signal for analysis. Examples
of this use of solvent extraction are found in spectroscopy, isotope dilution
radiometry, and ion-selective electrodes using liquid membranes. In the
latter case, electrochemical determinations are possible by controlling the
local concentration of specific ions in a solution by extraction.
13.4.1 Analytical Methods Based on

Spectrophotometric Detection
The oldest application of solvent extraction in spectrophotometric deter-
minations uses extraction from the original aqueous solution and sub-
sequent back-extraction into a second aqueous phase. Here the extractant
provides only separation or concentration, as in the case of Np(IV) and
Pu(IV) determination [16]. However, as only a few element species (e.g.,
MnO 2
4 , CrO4 ) are capable of absorbing light in the UV-VIS range, usually
all spectrophotometric methods are based on reactions of analytes with
568 Aguilar et al.
color-forming reagents. The role of the most important organic reagents has
been discussed [17] and a comprehensive dictionary is available [18].
Chelating reagents are the most popular. Although most of these reac-
tions were initially developed for aqueous phase measurement, in many cases
taking into account the low aqueous solubility of the analyte-chromogenic
reagents, extraction into an organic phase was used for direct spectroscopic
measurements. Among the most important reagents used are the follow-
ing [19].
Dithizone (diphenylthiocarbazone) is a weak acid insoluble in water at
pH <7 but readily soluble in CCl4 and CHCl3. It reacts with many metal
ions to form chelates that can be extracted into the organic phase from
which the excess of the green reagent is stripped with dilute NH3. The most
stable dithizonates (Pt, Pd, Au, Ag, Hg, and Cu) are extracted from strongly
acid solutions. Other metals (Bi, Ga, In, Zn) are extracted from weakly acid
media and some (Co, Ni, Pb, Tl, Cd) from neutral or alkaline media. Some
compounds are extracted rapidly (Ag, Hg, Pb, Cd) and others more slowly
(Pd, Cu, Zn), while a few (Rh, Ir, and Ru) require prolonged heating to be
formed.
Azo dyes contain an azo link between two aromatic rings possessing a
orthohydroxy group. The most important reagents include PAN, PAR, and
Arsenazo III and generally they offer high sensitivity for the majority of
transition metals.
Chelating dyes include triphenylmethane reagents (e.g., Pyrocatechol
Violet, Eriochrome Cyanine R, Chrome Azurol S, Xylenol Orange) and
xanthene reagents (fluorones, e.g., Gallein, Pyrogallol Red, phenylfluorone,
and salicylfluorone). They form chelates with most metals. Ionic surfactants
make it easier to dissociate the protons of chelating triphenylmethane
reagents and facilitate reactions with easily hydrolyzable metals (Be, Al, Fe,
Se, Ti, Zr), leading to very sensitive but poorly selective methods.
Nonchelating dyes include basic triphenylmethane dyes (e.g., Brilliant
Green, Malachite Green, Crystal Violet), xanthene dyes (e.g., Rhodamine B,
Rhodamine 6G), azine dyes (e.g., Methylene Blue), and acid dyes (e.g.,
Eosin, Erythrosin). These are intensely colored and when paired with an
oppositely charged analyte ion lead to high sensitivities.
Crown ethers are not chromogenic unless they contain a pendant
chromogen able to dissociate a proton in a basic medium. The resulting
anion interacts strongly with the crown-complexed cation compensating the
electric charge. The formation of a zwitterion leads to a hydrophobic ex-
tractable species with a considerably shifted absorption maximum compared
with the protonated species. This allows the same spectrophotometric
determination to be used for a large number of metal ions, provided the
appropriate crown compound is used in each case. Another method involves
using specific macrocyclic reagents that incorporate a chromophoric group,

even if the latter is not involved in the complexation of the metal ion. This
latter use of crown compounds is a development of an earlier known
method, used when no reagent could be found, which combines selectivity
for a certain metal and spectrophotometric sensitivity. In such cases, the
analyte is first extracted by a highly selective extractant, after which the
reagent providing the photometric determination is added to the extractant
phase [20].
13.4.2 X-Ray Fluorescence

Although this method is generally used on solid samples, it may be beneficial
in some cases to combine the procedure with solvent extraction. In addition
to the usual advantages of separation and preconcentration (before the solid
sample is prepared), it may be easier to prepare the required solid sample by
evaporation or crystallization from an organic extract phase rather than
from an aqueous solution [21]. Of special interest is the technique that uses
an extraction system that is solid at room temperature, but has a relatively
low melting point. Liquid–liquid extraction is performed at a temperature
above the melting point and after separation of the phases, the organic melt
is allowed to solidify, to give a purified concentrate of the analyte [22].
13.4.3 Atomic Absorption Spectroscopy and

Flame Emission Spectroscopy
It may seem that the high selectivity of these techniques would make a
separation process superfluous. However, preconcentration, especially the
removal of the solution medium, is often essential in atomic absorption
determinations. Extraction into an organic phase decreases the detection
limit by increasing the rate of introduction of the solution into the flame,
since most organic solvents have a lower viscosity and surface tension than
aqueous solutions. In addition, the sensitivity of the determination may be
influenced favorably by changes in flame temperature and composition [23].
Using the combination of solvent extraction-flame atomic absorption,
attention must be paid to the problems of vapor toxicity or noncombus-
tibility of the extractant or its diluent. For example, benzene or chlorine
containing solvents should always be avoided. Favored extraction systems
are solutions of a chelating agent such as oxine (8-hydroxyquinoline), a
ketone (methylisobutylketone), or butyl acetate [24].
The advantages of solvent extraction in combination with atomic
absorption apply equally well for flame emission spectroscopy. In addition,
the latter analytical method often requires separation of the analyte from a
large excess of other components. This may be achieved either by extracting
570 Aguilar et al.
the elements to be determined or by carrying out the spectrometric analysis

of an aqueous solution from which the interfering components have been
removed by extraction [25].
13.4.4 Polarography (Voltammetry)

Solvent extraction has also been used to enhance the selectivity of polaro-
graphic determinations. Such measurements are normally carried out in
aqueous solutions, and extraction followed by back-extraction has been
widely used. However, it may be unnecessary to perform a back-extraction if
the organic extractant phase has a sufficiently high dielectric constant to
dissolve sufficient background electrolyte for a voltammetric determination
or if the organic phase can be diluted with suitable polar solvents, such as
methanol or acetonitrile [26].
13.4.5 Activation Analysis

Activation analysis is the main radiometric method used in analytical
chemistry. The sample is irradiated in a neutron flux, and the resulting
specific radioactivity of the analyte determined. Although high-resolution
gamma spectrometers would seem to eliminate the need for highly selective
separation of individual gamma-emitting radioactive elements, group sep-
aration still may be necessary. Solvent extraction can be used in combina-
tion with activation analysis in two ways: either by separating the analyte
from interfering components and preconcentrating before irradiation, or by
performing the neutron activation and separating the analyte before the
radiometric determination. The latter method has the advantage of elim-
inating the need for a blank correction. However, when the interfering
components become highly radioactive and the resulting radioisotopes are
long-lived, it is preferable to remove them before irradiation.
13.4.6 Isotope Dilution Radiometry

In this type of radiometric analysis, a tracer quantity of a radioactive iso-
tope is added to the analyte, which is then partly extracted using a specific
extractant. Since the extractant may be considered as reacting totally with
the analyte, the ratio of radioactivity in both phases provides the con-
centration of analyte in the sample. This method, first developed by Stary
[2], has proved to be useful in several systems.
13.4.7 Liquid Scintillation Counting

Carbon-14 and tritium are radioisotopes with b-emissions of very low energy
that are extremely difficult to detect with any form of window counter, due to
self-absorption of the b-particles and absorption within the counter window.

To reduce the self-absorption losses, it is desirable to mix the active sample
homogeneously with the detecting material. This can be done by counting the
sample in the gaseous phase. The gaseous activity can then be intimately
mixed with the filling gas of any type of gas ionization detector, thus mini-
mizing the effect of b-absorption and resulting in high counting efficiencies.
In this, the radioactive sample and a scintillator material are both dissolved
in a suitable solvent, and the resulting scintillations are detected and counted
[27]. The method is called liquid scintillation counting. If a compound con-
taining an a- or b-emitting isotope is dissolved in a solvent such as toluene,
the radioactive emissions result in the formation of electronically excited
solvent molecules. If the solution also contains a small amount of a suitable
scintillator, the excited solvent molecules rapidly transfer their excitation
energy to the scintillator, forming electronically excited scintillator mole-
cules, which then relax by the emission of photons. All scintillator solutions
contain:
1. a solvent;
2. a primary solute, the scintillator material; and may contain
3. a secondary solute.
The functions of the solvent are to keep the scintillator or solute in solution,
and to absorb the decay energy of the radioisotope for subsequent transfer
to the solute. Solvents fall broadly into three categories:
1. Effective solvents, e.g., aromatic hydrocarbons such as toluene and
xylene.
2. Moderate solvents, e.g., many nonaromatic hydrocarbons.
3. Poor solvents; unfortunately this includes virtually everything else
including the most common laboratory solvents such as alcohols,
ketones, esters, and chlorinated hydrocarbons.
13.4.8 Other Methods

Several other analytical procedures exist in which solvent extraction may be
applied. Thus extraction has been used in a limited number of analyses with
procedures such as: (1) luminescence (fluorimetry), where, for example, the
detection limit of rhodamine complexes of gallium or indium can be in-
creased by extraction [28]; (2) electron spin resonance using a spin-labelled
extractant [29]; and (3) mass spectrometry, where an organic extract of the
analyte is evaporated onto pure Al2O3 before analysis [30].
Several thousand articles have been published on analytical procedures
for chemical elements in which solvent extraction is involved. The scope of
this chapter limits us to only one example for each element (Table 13.1).
572 Aguilar et al.
Table 13.1 Elements Determined by an Analytical Procedure Involving Solvent

Extraction
Element Extractant Analytical technique Ref.
Alkali metals crown ether compounds Spectrophotometry (AAS) 30
Lithium dipivaloylmethane Spectrophotometry (VIS) 31
Potassium crown ether compounds Spectrophotometry (VIS) 5
Rubidium/Caesium nitromethane Spectrophotometry (VIS) 32
Copper dithizone Isotope dilution 33
diethyldithiocarbamate Spectrophotometry (VIS) 17
Silver diethyldithiocarbamate Isotope dilution 34
dithizone Spectrophotometry (VIS) 17
Gold diphenyldipyridylmethane Activation analysis 35
Beryllium b-diketones Spectrophotometry (AAS) 36
Magnesium 8-hydroxyquinoline Spectrophotometry (VIS) 37
Calcium tributylphosphate/CCl4 Spectrophotometry (VIS) 38
Strontium polyethylene glycol Activation analysis 39
Barium crown ether compounds Spectrophotometry (AAS) 40
Zinc dithizone Spectrophotometry (VIS) 41
thioxine Molecular fluorescence 17
Cadmium diethyldithiocarbamate Polarography 42
Mercury dithizone Spectrophotometry (VIS) 43
Aluminium 8-hydroxyquinoline Spectrophotometry (VIS) 44
Gallium diisopropyl ether Activation analysis 45
thioxine Spectrophotometry (VIS) 17
8-hydroxyquinoline Spectrophotometry (VIS) 17
Indium b-mercaptoquinoline Polarography 46
Thallium diethyldithiocarbamate Spectrophotometry (VIS) 47
Scandium mesityl oxide Spectrophotometry (VIS) 48
Yttrium diantipyrylmethane Complexometry 49
Lanthanides diethylhexylphosphoric acid Activation analysis 50
Thorium trioctylphosphine oxide Spectrophotometry (VIS) 51
Uranium quaternary amine Spectrophotometry (VIS) 52
thenoyltrifluoroacetone Spectrophotometry (VIS) 17
Germanium dibutyl ether Spectrophotometry (AAS) 53
Tin diisopropyl ether Activation analysis 54
Table 13.1 (Continued )

Lead diethyldithiocarbamate Polarography 42
Titanium monolaurylphosphoric acid Isotope dilution 55
Zirconium trioctylphosphine oxide Spectrophotometry (VIS) 56
Hafnium thenoyltrifluoroacetone Spectrophotometry (VIS) 57
Phosphorous butanol Spectrophotometry (VIS) 58
Arsenic carbon tetrachloride Spectrophotometry (AAS) 59
thioxine Spectrophotometry (AAS) 17
Antimony dithiocarbamate Spectrophotometry (AAS) 60
thioxine Spectrophotometry (AAS) 17
Bismuth dithiocarbamate Spectrophotometry (VIS) 61
Vanadium Cupferron Spectrophotometry (AAS) 62
thenoyltrifluoroacetone 17
Niobium tetraphenylarsonium salt Spectrophotometry (VIS) 63
Tantalum methylisobutyl ketone Spectrophotometry (AAS) 64
Selenium diethyldithiocarbamate Spectrophotometry (AAS) 65
Tellurium methylisobutyl ketone Spectrophotometry (AAS) 66
Chromium methylisobutyl ketone Activation analysis 67
Molybdenum N-benzoyl-N-phenyl Polarography 68
hydroxylamine
Tungsten methylisobutyl ketone Spectrophotometry (AAS) 69
Manganese tetramethylene-dithiocarbamate Spectrophotometry (AAS) 70
Technetium tetraphenylarsonium chloride Spectrophotometry (VIS) 71
Rhenium ALIQUAT 336 Activation analysis 72
Iron acetylacetone Polarography 33
Cobalt 1-nitroso-2-naphthol Isotope dilution 73
Nickel dimethylglyoxime Polarography 74
thenoyltrifluoroacetone Spectrophotometry (VIS) 17
Platinum diphenyldithourea Spectrophotometry (VIS) 74
(Continued )
574 Aguilar et al.
Table 13.1 (Continued )

Palladium diphenyldithourea Spectrophotometry (VIS) 74
Rhodium diphenyldithourea Spectrophotometry (VIS) 74
thenoyltrifluoroacetone Spectrophotometry (VIS)
Ruthenium 8-hydroxyquinoline Spectrophotometry (VIS) 17
13.5 SOLVENT EXTRACTION–BASED MATERIALS

FOR ANALYTICAL APPLICATIONS
13.5.1 Solvent-Impregnated Resins
In the last decade, the development of new impregnated materials with che-
lating and complexing properties has acquired great importance. These mate-
rials are prepared principally by the simple immobilization of complexing
organic reagents by adsorption onto conventional macroporous polymeric
supports (polar and nonpolar). They provide two important advantages over
conventional ion exchange resins: the possibility of selecting the functional
group to be immobilized and the facility for continuous column operation.
Such macromolecular resins containing an extractant within the lattice
of the polymer were developed to bridge the gap between the two techniques
of solvent extraction and ion exchange and are roughly classified into
extractant-impregnated sorbents and Levextrel resins [12]. The former are
prepared by soaking a polymeric resin in a diluent containing the extractant,
then evaporating the diluent (dry impregnation method) or retaining the
diluent (wet method). Levextrel resins are prepared by adding an extractant
to the mixture of styrene monomers during bead polymerization with divi-
nylbenzene. This method has increased in importance mainly because of its
adaptability to preconcentration, separation, and/or determination of ana-
lytes. Additionally, they could be used in continuous-flow systems and in
solid phase spectrophotometry by using chemical reactions that occur at
interfaces (e.g., solid/liquid or gas/solid). Furthermore, most of the analytes
could be analyzed directly on the solid matrix using X-ray spectrometry,
neutron activation analysis, molecular absorption, fluorescence spectros-
copy, or isotopic dilution methods. Alternatively, analytes may be eluted
from the column and the analysis completed on the solution. In this context,
atomic absorption (AAS) or inductively coupled plasma spectroscopy (ICP)
are mainly used.
The use of impregnated resins in the preconcentration and separation

of trace metal ions provides the following advantages:
The active part of the resin (complexing ligand) can be selected to be
compatible with the nature of the metal ion and the matrix of the
sample, and with the analytical procedure to be applied.
The impregnation procedures of the complexing molecules are simple.
Their structure and composition could be compatible with integrated
detection systems, when used in solid phase spectroscopic measure-
ments.
13.5.1.1 Solid-Phase Spectrophotometry Applications

Solid-phase spectrophotometry (SPS) is a technique based on the precon-
centration of the species of interest onto a solid, aided by complexants or
other reagents, and subsequent measurement of the spectrophotometric
properties of the species in the solid phase [75]. Depending on the spectro-
photometric responses of the analytes or analyte complexes, several proce-
dures based on absorbance and fluorescence measurements have been
developed. In the first case, the absorbance of a resin containing the analyte
fixed as a colored chromogenic species is measured directly. In the second
case, solid phase fluorimetry (SPF), the diffuse reflected fluorescence, is
measured. Most procedures using color or fluorescence measurement are
based on the addition to the sample solution of a resin impregnated with a
highly specific chromogenic agent for the analyte.
Ion exchanger colorimetry has been used as a sensitive and rapid
method for vanadium analysis by immobilization of 2[2-(3-5-dibromopyri-
dyl)azo]-5-dimethylaminobenzoic acid onto an ion exchanger resin AG1X2
[75]. Solid phase fluorimetry can be useful for the analysis of very dilute
solutions in water analysis or trace metal determination; thus a chelating
8-(benzene-sulfonamido)quinoline, immobilized on Amberlite XAD2 sup-
port, has been used for the spectrofluorimetric determination of Zn(II) and
Cd(II) [76].
13.5.1.2 Applications for Fiber Optic Chemical Sensors
The development of chemical sensors based on optical measurements has
grown steadily in importance during the last decade. While a large variety of
devices are possible, they share a common feature in multiple applications,
i.e., an immobilized reagent phase that changes its optical properties upon
interaction with an analyte on either a continuous or reusable basis. Systems
in which chelating liquid extractants are immobilized onto solid polymeric
supports have been used for chemical sensing. Particular attention has been
given to fiber-optic devices for measuring and controlling metal ions and
organic compounds in aqueous media. The applications of such devices
576 Aguilar et al.
have covered areas such as environmental applications, industrial process

control, and biomedical and clinical applications. Small sensors based on
immobilization of acid-base indicators such as Bromothymol Blue, Bro-
mophenol Blue, Bromocresol Purple, Phenolphtalein, Phenol Red, Chloro-
phenol Red Alizarin, on nonionic macroporous supports, Amberlite XAD-2
and Amberlite XAD-4, have been developed and used for pH measurements
[77]. These fiber-optic probes provide advantages over conventional elec-
trodes in safety, reliability, applicability, and cost. A sensor based on per-
ylene dibutyrate adsorbed onto Amberlite XAD4 has been characterized in
some depth [78]. Fiber optic sensors, particularly fluorescent sensors, have
become the object of considerable interest among researchers in recent
years. Their performance is based on the change in fluorescent properties of
organic reagents immobilized on a solid matrix upon contact with solutions
of metals in a continuous system.
13.5.2 Liquid Membranes in Analytical Chemistry

The use of liquid membranes in analytical applications has increased in the
last 20 years. As is described extensively elsewhere (Chapter 15), a liquid
membrane consists of a water-immiscible organic solvent that includes a
solvent extraction extractant, often with a diluent and phase modifier, im-
pregnated in a microporous hydrophobic polymeric support and placed
between two aqueous phases. One of these aqueous phases (donor phase)
contains the analyte to be transported through the membrane to the second
(acceptor) phase. The possibility of incorporating different specific reagents
in the liquid membranes allows the separation of the analyte from the matrix
to be improved and thus to achieve higher selectivity.
Solvents used in liquid membranes should have special characteristics
such as low aqueous solubility, as a thin film of solvent is in contact with
large volumes of aqueous solutions, and low viscosity to provide large dif-
fusion coefficients in the liquid membrane. Furthermore, the analyte should
have large partition coefficients between the donor and the membrane phase
to give good extraction recovery and, at the same time, interfering sub-
stances in the sample should have low partition coefficients for efficient
cleanup.
Two configurations of liquid membranes are mainly used in analytical
applications: flat sheet liquid membranes that give acceptable extraction
efficiencies and enriched sample volumes down to 10–15 mL, and hollow
fiber liquid membranes that allow smaller enriched sample volumes. Flat
sheet liquid membrane devices consist of two identical blocks, rectangular or
circular in shape, made of chemically inert and mechanically rigid material
(PTFE, PVDF, titanium) in which channels are machined so that when
assembled the channels face each other. The channels are about 0.10–
0.25 mm deep, 1.5 mm wide, and differ in length from 15 cm to 250 cm to
provide volumes from 12 mL to 1000 mL. In the smallest devices the channels
can be U-shaped grooves, while in other separators they are arranged in
spirals with the feed inlet on the periphery and the outlet in the center
(Fig. 13.3). The impregnated membrane, which is prepared by soaking the
Fig. 13.3 Schematic diagram of two different membrane units. (a) Membrane separator
unit composed of two machined blocks of PTFE or Ti (A), and PTFE membrane (B),
impregnated with stationary liquid. (b) Membrane unit. The PTFE membrane is placed
between the two blocks made of titanium. The two channels (donar and acceptor) that are
formed have a nominal volume of 12 mL.
578 Aguilar et al.
support in the organic phase, is clamped tightly between the planar surfaces
of the two blocks. Such devices can be coupled to chromatographic col-
umns, but are too large for on-line connection to packed capillary liquid
chromatography or capillary electrophoresis. To decrease the volume while
preserving the enrichment efficiency requires shallower channels with the
possibility of clogging problems. In such cases, porous hollow fiber modules
as supports for the liquid membrane can be used. The hollow fiber module
may have only one fiber of a microporous polymer about 15 mm long with
an internal diameter of 300–500 mm giving a lumen volume for the acceptor
phase of 1.9 mL and annular donor volume of approximately 0.5 mL.
13.5.2.1 Applications of Liquid Membranes in
the Analysis of Selected Samples
The most frequently used pretreatment methods for extraction and enrich-
ment of analytes are liquid–liquid solvent extraction and solid-phase
extraction (SPE). Liquid–liquid solvent extraction often gives a good cleanup
from the matrix. It allows, by incorporating different specific reagents,
improvement of the separation of the analyte from the sample. However,
liquid–liquid solvent extraction has some drawbacks. It is laborious and
difficult to automate and connect on-line to analytical instruments. In
addition, large amounts of organic solvents should be avoided for environ-
mental and health reasons. Supported liquid membranes (SLM) are an
attractive alternative based on the efficient cleanup of liquid–liquid extrac-
tion, using small amounts of organic solvents and avoiding the possibility of
emulsion formation. Using appropriate carriers, the SLM technique offers
very selective extraction of analytes in very complex samples. Furthermore,
preconcentration is often required in trace analysis to improve detection
limits and this can be also achieved using liquid membranes [79–81] with the
possibility of obtaining over 100 times enrichment of heavy metal ions as well
as various organic pollutants.
The most important features of liquid membranes are that they offer
highly selective extraction, efficient enrichment of analytes from the matrix
in only one step, and the possibility of automated interfacing to different
analytical instruments such as liquid chromatography, gas chromatography,
capillary zone electrophoresis, UV spectrophotometry, atomic absorption
spectrometry, and mass spectrometry [82].
13.5.2.1a Application in Cleanup Procedures
Many analyses of organic compounds in liquid samples require selective
cleanup and concentration. Direct on-line coupling of sample preparation to
the analytical instrumentation minimizes sample handling and thereby
the risk for contamination or loss of analyte. Also, on-line coupling makes
automation of the process possible, resulting in more reproducible and

economical analysis. Membrane extraction techniques are then well suited to
automated interfacing with various separation techniques (chromatography,
capillary electrophoresis) as well as with various detection techniques such
as UV-VIS and flow-through sensors.
1. Liquid membrane sample cleanup coupled to gas–liquid chromatog-
raphy. Several types of analysis including environmental [83–85] and bio-
logical samples [86–88] that need both cleanup and concentration prior to
chromatographic determination can be advantageously pretreated by liquid
membranes. Thus on-line combination of supported liquid membrane
extraction in a hollow fiber configuration and column liquid chromatog-
raphy with a phenol oxidase-based biosensor as a selective detector has been
used for the determination of phenols in human plasma. The phenols are
selectively extracted into a porous PTFE membrane impregnated with a
water-immiscible organic solvent [89]. Also a liquid membrane in a minia-
turized hollow fiber module coupled on-line to packed capillary liquid
chromatography enables the use of very small volumes and has been used to
determine amines in plasma using 6-undecanone as extractant.
2. Liquid membrane sample cleanup coupled to capillary zone electro-
phoresis. Capillary zone electrophoresis (CZE) is a technique that can be
used for determination of a great variety of analytes and it is especially
suitable when only small sample volumes are available and high separation
power is needed. Biological samples are difficult to analyze by CZE due to
their complexity and the low concentration of analytes. Thus, before CZE
determination, a cleanup procedure is necessary to eliminate the interfering
matrix. On the other hand, to improve the detection limit of CZE, a pre-
concentration step is needed [90]. The use of a liquid membrane of 6-
undecanone in a PTFE support has been used successfully as a prior step
before CZE determination of bambuterol in human plasma to obtain a high
degree of cleanup from the plasma and thus avoid adsorption problems in
the CZE capillary [91]. The samples were concentrated more than 1000
times, allowing the detection limit to be lowered about 400 times. A hollow
fiber miniaturized supported liquid membrane can be used as sample pre-
treatment for on-line connection to CZE for determining basic drugs in
human plasma [92]. The analyte is extracted from the outside of the hollow
fiber (feed phase) through the liquid membrane containing the organic
solvent into the strip phase, which flows through the fiber lumen and can
then be injected into the CZE capillary.
13.5.2.1b Application to Enrichment Procedures
1. Liquid membrane enrichment coupled to mass spectrometry. Membrane
introduction mass spectrometry (MIMS) is an established method of sample
580 Aguilar et al.
analysis that couples rapid introduction via a semipermeable membrane with

the sensitivity and specificity of a mass spectrometer. Membranes are chosen
to enrich the analyte concentration in the sample stream entering the mass
spectrometer while rejecting the bulk of the mobile phase.
The membranes used are typically composed of cross-linked silicones
and are suitable for on-line monitoring of volatile organic and inorganic
compounds [93–94]. An alternative material is microporous PTFE, which
has more rapid responses as well as lower selectivities and higher fluxes of the
mobile phase compared to nonporous silicone membranes. More recently,
developments in membrane introduction systems include the use of liquid
membranes composed, for example, of a polyphenyl ether diffusion pump
fluid [95–96]. This membrane has the advantage that it can take any desirable
analyte and the selectivity can be modified using appropriate reagents.
2. Liquid membrane enrichment coupled on-line with ion chromatog-
raphy. Low molecular mass carboxylic acids in low concentrations in air or
soil samples can be determined by ion chromatography coupled on-line to a
selective enrichment system consisting of a supported liquid membrane,
impregnated with tri-n-octylphosphine oxide in di-n-hexyl ether [97–98]. The
system allows the determination of these carboxylic acids at micromolar
levels in the presence of interfering ions such as nitrite, chloride, sulfate,
iron, and aluminum.
3. Liquid membrane enrichment coupled on-line with atomic absorption
spectrophotometry. Metal ions can be readily determined by atomic absorp-
tion spectrometry. Nevertheless, preconcentration is sometimes required and
supported liquid membranes can be used as an attractive alternative to other
pretreatment methods such as liquid–liquid extraction [99–101]. The trans-
port of the metal species across the liquid membrane is generally performed
by carriers in the membrane such as alkylphosphorous acids, long-chain
alkylamines, or chelating reagents. These carriers react with metal ions to
give neutral species that are soluble in nonpolar solvents and diffuse across
the membrane. In the case of cationic metal species, cationic exchanger
extractants or chelating extractants are used frequently as carriers. In such
systems, the metal ions are transported from the aqueous donor solution to
the acceptor phase and protons are countertransported from the acceptor
to the donor phase. The driving force for the process is the pH gradient
between the donor and acceptor phases and this allows the metal ions to be
transported against their concentration gradient with the result that the metal
ions concentrate in the acceptor phase.
In the case of metal ions present as anionic complexes in the donor
phase, solvating or ion-pairing extractants can be used as carriers. Here the
metal ions and counterions are cotransported across the membrane from
the donor to the acceptor phase. By using a complexing or reducing agent in
the acceptor phase, it is possible to back-extract the metal ions into the
acceptor phase.
Metal ions such Cu2+, Cd2+, and Pb2+ can be preconcentrated
from water samples using liquid membranes containing 40% w/w of di-2-
ethylhexylphosphoric acid in kerosene diluent in a PTFE support. The
liquid membrane can be coupled on-line to an atomic absorption spectrom-
eter and has been shown to be stable for at least 200 h with extraction
efficiencies over 80%, and enrichment factors of 15 can be obtained. A
liquid membrane has also been used for sample cleanup and enrichment of
lead in urine samples prior to determination by atomic absorption spec-
trometry [100]. The experimental setup for metal enrichment is shown in
Fig. 13.4. Lead was enriched 200 times from urine [80] and several
metals were enriched 200 times from natural waters [88]. Using hollow fiber
Fig. 13.4 Schematic setup for a membrane-based metal enrichment system.

582 Aguilar et al.
geometry and crown ethers in the membrane, enrichment factors up to 3000

have been obtained for lead [102].
Other applications of supported liquid membranes have been related
to metal speciation. For example, recently a system for chromium speciation
has been developed based on the selective extraction and enrichment of
anionic Cr(VI) and cationic Cr(III) species in two SLM units connected in
series. Aliquat 336 and DEHPA were used respectively as carriers for the
two species and graphite furnace atomic absorption spectrometry used for
final metal determination. With this process, it was possible to determine
chromium in its different oxidation states [103].
13.5.3 Micelles in Analytical Chemistry

Another of the new techniques for extractive preconcentration, separation,
and/or purification of metal chelates, biomaterials, and organic compounds
is based on the use of surfactant micellar systems.
Surface-active agents aggregate in aqueous solutions to form micelles
if the concentration in aqueous solutions exceeds the critical micelle con-
centration (CMC). Dilute aqueous solutions of certain surfactant micelles,
when the conditions (i.e., temperature, pressure, and electrolyte concentra-
tion) of the solution are changed, have the ability to separate into two
isotropic liquid phases: a surfactant-rich phase with a small amount of water
(surfactant phase or coacervate phase) and a phase containing an almost
micelle-free dilute aqueous solution. This separation is reversible so that on
changing the conditions, e.g., cooling, the two separated phases merge to
form a clear solution once again.
This phenomenon can be exploited for separation and concentration
of solutes. If one solute has certain affinity for the micellar entity in solution
then, by altering the conditions of the solution to ensure separation of the
micellar solution into two phases, it is possible to separate and concentrate
the solute in the surfactant-rich phase. This technique is known as cloud
point extraction (CPE) or micelle-mediated extraction (ME). The ratio of
the concentrations of the solute in the surfactant-rich phase to that in the
dilute phase can exceed 500 with phase volume ratios exceeding 20, which
indicates the high efficiency of this technique. Moreover, the surfactant-rich
phase is compatible with the micellar and aqueous-organic mobile phases in
liquid chromatography and thus facilitates the determination of chemical
species by different analytical methods [104].
The most common surfactants for analytical applications are nonionic
(polyoxyethylene glycol monoethers, polyoxyethylene methyl-n-alkyl ethers,
t-octylphenoxy polyoxyethylene ethers, and polyoxyethylene sorbitan esters
of fatty acids) that demonstrate cloud point behavior with increasing solution
temperature, and zwitterionic surfactants (ammonioethylsulfates, ammo-
niopropylsulfates, ammoniopropanesulfonates, phosphobetaine, dimethyl-
alkylphosphine oxides) that show cloud point behavior on decreasing
solution temperature. Thus, cloud point temperature depends on the struc-
ture of the surfactant and its concentration and range from 0.5C to 120C
[105,106]. In a homologous series of polyoxyethylated surfactants, the
cloud point temperature increases with the hydrocarbon chain length and
increasing length of the oxyethylene chain. At a constant oxyethylene con-
tent in the surfactant molecule, the cloud point temperature is lowered by
decreasing the molecular mass of the surfactant and by branching of the
hydrophobic group.
The most important advantage of cloud point extraction is that only
small amounts of nonionic or zwitterionic surfactants are required and
consequently the procedure is less costly and more environmentally benign
than other conventional extraction techniques such as liquid–liquid extrac-
tion and solid–liquid extraction [107,108]. Moreover, CPE offers the pos-
sibility of combining extraction and preconcentration in one step.
13.5.3.1 Experimental Protocols
Operation procedures are performed by adding a small volume of a con-
centrated nonionic or zwitterionic surfactant solution to an aqueous sample
(50–100 cm3) containing the analyte, taking into account that the final sur-
factant concentration must be greater than its CMC value so that micelles are
present in solution. The analyte, depending on its affinity to the micelles, is
incorporated into the micellar aggregates in solution. The solution is heated
(in the case of nonionic surfactants) or cooled (with zwitterionic surfactants)
until the cloud point temperature is reached and the solution is allowed to
settle (settling temperature) in a thermostatted bath set at a temperature
above or below that of the cloud point (for nonionic and zwitterionic sur-
factants respectively) until the phases separate. As the density of both phases
in some cases is quite similar, centrifugation is often recommended to facil-
itate the physical separation. The analyte is concentrated in the surfactant-
rich phase in a small volume (50–400 mL) [105,106]. Figure 13.5 shows the
steps involved in CPE prior to analysis.
The surfactant selected for CPE technique should not have too high a
cloud point temperature. In practice, it is possible to obtain almost any
desired temperature by choosing an appropriate mixture of surfactants, as
cloud point temperatures of mixtures of surfactants are intermediate
between those of the two pure surfactants, or by the choice of an appro-
priate additive (i.e., salts, alcohols, organic compounds) [105].
584 Aguilar et al.
Fig. 13.5 Steps involved in cloud point extraction (CPE) prior to HPLC, GC, and CE
analysis.
13.5.3.2 Applications of Cloud Point Extraction to the

Analysis of Selected Samples
Cloud point extraction has been applied to the separation and pre-
concentration of analytes including metal ions, pesticides, fungicides, and
proteins from different matrices prior to the determination of the analyte by
techniques such as atomic absorption, gas chromatography, high perfor-
mance liquid chromatography, capillary zone electrophoresis, etc.
1. Cloud point extraction of metal ions. The use of cloud point
extraction as a separation technique was first introduced by Watanabe for
the extraction of metal ions forming sparingly water soluble complexes
[109]. Since then, the technique has been applied successfully to the ex-
traction of metal chelates for spectrophotometric, atomic absorption, or
flow injection analysis of trace metals in a variety of samples [105–107,110].
Other metal complexes such as AuCl 4 or thiocyanato-metal complexes can
be extracted directly using nonionic surfactants such as polyoxyethylene
nonyl phenyl ether (PONPE) or Triton X prior to their determination by

ICP or visible spectrometry [111,112]. The main advantages of cloud point
extraction of metal ions include the simplicity of the extraction procedure
and the possibility of obtaining high preconcentration factors ranging from
10 to 100 allowing the development of analytical methods for determining
metal ions at very low concentrations. For example, Pd(II) can be deter-
mined in a surfactant phase of Triton X-100 with coproporphyrin III as a
complexing agent in a procedure based on phosphorescence at room tem-
perature with detection limits of 20 nmol dm3. Some parameters for the
cloud point extraction of metals with different surfactant micelles are given
in Table 13.2 [105,107].
2. Cloud point extraction from biological and clinical samples. The most
frequent use of CPE is for the separation and purification of biological
analytes, principally proteins. In this way, the cloud point technique has
been used as an effective tool to isolate and purify proteins when combined
with chromatographic separations. Most of the applications deal with the
separation of hydrophobic from hydrophilic proteins, with the hydrophobic
proteins having more affinity for the surfactant-rich phase, and the hydro-
philic proteins remaining in the dilute aqueous phase. The separation of
biomaterials and clinical analytes by CPE has been described [105,106,113].
3. Cloud point extraction of environmental samples. More recently, CPE
has been used for sample preparation and preconcentration of organic ana-
lytes such as pesticides [114,115], herbicides, polycyclic aromatic hydro-
carbons (PAH) [116], polychlorinated biphenyls (PCBs) [117], and phenols
[118] in environmental samples prior to their determination. The compat-
ibility of the surfactant-rich phase from CPE with micellar or conventional
hydroorganic mobile phases allows subsequent determination of the analyte
by thin layer chromatography, HPLC [119,120], micellar electrokinetic
capillary chromatography, or CZE [121]. Cloud point extraction is a rapid,
simple, sensitive, and efficient sample pretreatment for trace environmental
analysis and offers some advantages over conventional liquid–liquid ex-
traction technology in terms of enhanced detection limits from the large
preconcentration factors, elimination of analyte losses during evaporation of
solvents used in liquid–liquid extraction, and elimination of toxic solvents
[107].
13.5.4 Ion-Selective Electrodes

An important advance in ion-selective electrodes (ISEs) and related systems
was based on the concept of polymeric liquid membranes developed by
Eisenman [122]. The principle of this approach was to incorporate an
organic compound as the ionophore into a polyvinyl chloride membrane
586 Aguilar et al.
Table 13.2 Summary of Cloud Point Extractions of Metals Chelates Using Nonionic
Surfactants Micelles
Metal ion Ligand Nonionic surfactant Experimental conditions
pH; CFa, %Eb
Ni(II) TANc Triton X-100 pH 7.0 (phosphate); CF = 30
PAN Triton X-100 pH 5–6
Zn(II) PAN PONPE-7.5 pH 10 (carbonate); CF = 40
QADI PONPE-7.5 pH 9; NH3; CF = 40
Ni(II), Zn(II), Cd(II) PAMP PONPE-7.5
Au(III) HCl PONPE-7.5 HCl; %E > 95
Ni(II), Cd(II), Cu(II) PAMP PONPE-7.5 pH 5.6
Ni(II) PAN OPd pH 6.0; CF = 15–25
Transition metal ions TAN PONPE-7.5
Fe(III), Ni(II) TAC Triton X-100
U(VI) PAN Triton X-114 pH 9.2; %E = 98
U(VI), Zr(IV) Arsenazo Tween 40e pH 3; %E > 96
Cu(II), Zn(II), Fe(III) Thiocyanate PONPE-7.5 %E = 72.5–96.8
Er(III) CMAP PONPE-7.5 CF = 20
Gd(III) CMAP PONPE-7.5 CF = 3.3; %E = 99.88
Cd(II) PAN Triton X-114 CF = 60
Ni(II), Zn(II) PAN Triton X-114
Ru(III) Thiocyanate Triton X-110 CF = 5–10
Au(III) HCl PONPE-10f %E > 90
Ga(III) HCl PONPE-7.5f %E & 90
Ag(I), Au(III) DDTP Triton X-114f CF = 9–130
Cu(II) LIX54 Igepal CO-630f
a
Concentration factor.
b
Percent extracted.
c
Abbreviations for ligands: CMAP: 2-(3,5-dichloro-2-pyridylazo)-5-dimethylaminophenol; DDTP:
o,o-diethyldithiophosphoric acid; LIX54: dodecylbenzoylacetone; PAN: 1-(2-pyridylazo)-2-naphthol;
QADI: 2-(8-quinolazo)-4,5-dipheylbenzimidazole; PAMP: 2-(2-pyridylazo)-5-methylphenol; PAP: 2-
(2-pyridylazo)-phenol; TAC: 2-(2-thiazoylazo)4-methylphenol; TAN: 1-(2- thiazoylazo)-2-naphthol.
d
OP surfactants refer to (polyethyleneglycol octylphenyl ethers).
e
In this extraction, the concentrated surfactant-rich phase was a solid rather than a liquid.
f
Abbreviations for surfactants: Igepal CO-630: nonylphenoxypoly(ethylenoxy)ethanol; PONPE-7.5:
polyoxyethylene(7.5)nonylphenyl ether; PONPE-10: polyoxyethylene(10)nonylphenyl ether; TRI-
TON X: t-octylphenoxypolyoxyethylene ether.
together with an appropriate plasticizer and additive to provide the mem-

brane with the properties of a liquid phase. Ionophores are lipophilic,
electron-rich complexing agents capable of reversibly binding ions and
transporting them across organic membranes, and the ultimate performance
of such devices depends strongly on the choice of ionophore. The mode of
action of these electrodes is therefore a reversible ion exchange process

accompanied by diffusion and migration effects.
Potentiometric sensors, such as ion-selective electrodes, ion-selective
microelectrodes, and ion-selective field effect transistors (ISFET) are fre-
quently used in analytical systems. Moreover, the recently introduced ion-
selective optodes based on absorbance or fluorescence measurements provide
additional sensors for a broad range of monitoring applications [123,124]. In
these systems, the active constituent of the liquid membrane is an ion-pairing
or complexing agent that is selective for a limited number of specific ions.
This reagent is dissolved in a low-polarity organic solvent, which must also
have a low vapor pressure to minimize losses through evaporation, high
viscosity to prevent rapid loss during flow across the membranes, and low
solubility in the aqueous sample solution with which it comes into contact.
The complexing reagents are subdivided into two groups:
1. Electrically charged ligands, often called liquid ion exchangers, which
are ionizable organic molecules of high molecular weight
2. Electrically neutral complexing agents, ionophores, that are capable of
enveloping metal ions in a ‘‘pocket’’ of oxygen or nitrogen ligands and
can serve as selective extractants for cations
The construction of a liquid membrane electrode is rather similar to
that of the glass electrode, in that it requires an internal reference electrode.
In addition, a solid support is required into which the active liquid reagents
are incorporated. This support may be a porous polyvinyl chloride (PVC)
plug, a porous polymeric film, or a commercial membrane. Many different
designs have been suggested to solve the problem of interposing the liquid
membrane between the aqueous solutions. The support prevents leakage of
the organic solvent into the aqueous solution and the same time maintains
efficient contact with the analyte solution. Two of these designs are illus-
trated in Fig. 13.6.
The electromotive force (emf ) of liquid membrane electrodes depends
on the activity of the ions in solution and their performance is similar in
principle to that of the glass electrode. To characterize the behavior of liquid
membrane electrodes, the linearity of the emf measurements vs. concentra-
tion of a certain ion in solution is checked. Additional performance data are
the Nernstian slope of the linear range and the pH range over which the
potential of the electrode is constant.
Eisenman [122] pioneered the development of this type of electrode
and created the theory on which the measurements are based, and investi-
gated the selectivity of certain membranes. Equally significant were the
contributions of Buek, Durst, Worf [126], Koryta [127], and Simon, which
were compiled into an important monograph [128].
588 Aguilar et al.
Fig. 13.6 Schematic diagrams of two liquid membrane electrodes: (a) a commercial
Onion electrode; (b) an improved version developed by Szczepaniac and Oleksy [125].
The ion selectivity of a membrane can be established by measuring the

potential difference between two identical reference electrodes. One elec-
trode is immersed in the specimen solution, the other in a reference solution,
and the membrane is interposed between them. The composition of solution
2 is constant, and, if both solutions contain monovalent ions, the ion-
exchange process can be described as follows:
J þ ðaqÞ þ I þ ðmembraneÞ () J þ ðmembraneÞ þ I þ ðaqÞ (13.5)
and the potential difference is given by:
E ¼ E0 þ 0:059 log fai þ Kij aj g (13.6)
where E0 is a constant related to the reference electrodes and the reference
solution, ai and aj are the activities of ions i and j in the specimen solution,
and Kij is the selectivity constant of the membrane.
Inspection of Eq. (13.6) shows that the selectivity behavior of a liquid
membrane is specified completely by the membrane selectivity constant, Kij,
which in turn is dependent on the equilibrium constant of Eq. (13.5) and on
the mobility of ions i and j within the membrane. For the case in which the
membrane consists of a neutral carrier [129], the exchange reaction can be
presented as:
J þ ðaqÞ þ ISþ ðmembraneÞ () JSþ ðmembraneÞ þ I þ ðaqÞ (13.7)

If the ligand (neutral carrier) is capable of completely enveloping the

cations, the terms Kij and E0 of the membrane depend only on the ratio of
the complex stability constants Kjs =Kis of the ions with the ionophore.
Liquid membrane electrodes are not capable of being specific for only
one ion in solution. There is always some interference from other ions in
solution with the given analyte. The selectivity coefficient provides an
indication of the ability of an electrode to measure a particular ion in the
presence of another ion. The response of an electrode to an interfering ion
can be included in the Nernstian equation:
0
E ¼ E00 þ 0:059=z0 log faA þ KAB aB z=z g (13.8)
where:
E00 = a constant emf, including that of the reference electrode and the
standard potential of the ion being measured
aA = activity of species A with charge Z0
aB = activity of interfering ion, B, with charge Z
KA/B = selectivity constant of the electrode for A over B
13.5.4.1 Liquid Ion Exchangers

Liquid ion exchange carriers must again be subdivided into cation- and
anion-specific ion exchangers. For cations, extensive use is made of high
molecular mass organic anions, such as sulfonates, carboxylates, thio-
carboxylates, and phosphonates. An important example of this membrane
type is the calcium-selective electrode, in which the liquid membrane consists
of long-chain alkylphosphates [130–131]. These compounds are dissolved in
di-n-octylphenylphosphate, which enhances the selectivity for calcium rela-
tive to magnesium and the other alkaline earth metals. A mercury(II) ion-
selective electrode uses a thiophosphonyl derivative of thiobenzamide as the
complexing agent [125]. For anions, stable ion pairs are formed with large
cationic organic molecules, such as substituted organophenanthroline nickel
or iron compounds dissolved in decanol.
Of special interest is the nitrate electrode, which has found many
applications in quantitative analysis of nitrates in biological fluids and
agricultural products. A nitrate-selective electrode [132] based on a tributyl-
octadecylphosphonium ligand is also selective for perrhenate and per-
chlorate ions.
13.5.4.2 Neutral Carrier-Based Cation-Selective
Electrodes
A number of polyether molecules have been described that form membranes
selective for four monovalent and for six divalent cations [129,133,134].
Inspection of Table 13.3 shows that two classes of ligands can be distinguished:
590 Aguilar et al.
Table 13.3 Some Commercial Ion Exchange Electrodes

Ion Ligand Area of application Refs.
Li+ Trioctylphosphine oxide + Lithium in blood serum 136

neutral carrier and pharmaceuticals
K+ Valinomycin dissolved in Potassium in feldspar, urine, 137
diphenyl ether blood serum, seawater, vegetables
NHþ
4 Nonactin or dinactin Activity of nitrate reductase, 138
ammonium in mineral water,
fruit juice, beer, urine, sewage water
Ca2+ Phosphates and Protein binding studies, calcium in 131
phosphonates sewage water, mineral water, blood
serum, biological fluid
NO
3 Tributyl octadecyl- Nitrate in agricultural products, soils, 132
phosphonium nitrate foods
the first is the macrotetrolides (antibiotics), such as nonactin and dinactin,

which have eight oxygen atoms in a flexible ring and are specific for
ammonium ions. Another antibiotic that has assumed considerable impor-
tance is valinomycin, which has a high selectivity for potassium across the
membrane of living cells. The second class of ligands is composed of synthetic
crown ethers, some acrylic polyethers, and cryptands. Figure 13.7 demon-
strates how different geometric arrangements of these polyethers provide
compounds that systematically bind ions, from Li+ to Pb2+.
Five liquid membrane electrodes (Table 13.3) are now commercially
available and have found wide application in the testing of electrolytes in
biological and technological systems. All five electrodes perform well in the
concentration range over which the Nernstian slope is maintained, i.e., from
101–105 mol dm3. These electrodes to a certain extent have replaced in
both chemical and clinical laboratories the more traditional instrumental
methods of analysis, such as flame photometry and atomic absorption
spectrometry. There are, of course, many more liquid membrane electrodes,
but the availability of satisfactory solid electrodes has greatly restricted their
development and practical application.
13.6 SOLVENT EXTRACTION– BASED TECHNIQUES

The term chromatography now embraces a variety of processes that are
based on the differential distribution of the components in a chemical
mixture between two phases. The difference between extraction processes
involving a single equilibrium of two bulk phases and chromatography is
that, in the latter technique, species are separated, not by discrete extraction
steps, but by continuous equilibration between two phases, one that is
stationary and the other mobile. The same fundamental laws of phase
equilibrium apply to both, extraction and chromatography. In principle, the
latter is a multiple-extraction process in which the mobile phase moves
continuously over the fixed phase in a chromatographic column.
13.6.1 Liquid–Liquid Partition Chromatography

The liquid–liquid partition chromatography (LLPC) method involves a
stationary liquid phase that is more or less immobilized on a solid support,
and a mobile liquid phase. The analyte is therefore distributed between the
two liquid phases. In conventional LLPC systems, the stationary liquid
phase is usually a polar solvent and the mobile liquid phase is an essentially
water-immiscible organic solvent. On the other hand, in reversed-phase
chromatography (RPC), the stationary liquid is usually a hydrophobic
Fig. 13.7 Structures of cation-selective carriers. (From Ref. 133.)

592 Aguilar et al.
solvent, whereas the mobile liquid is the polar solvent, most frequently
water or an aqueous mixture with polar organic solvents [139–143].
For conventional or ‘‘normal,’’ in contrast with reversed-phase,
LLPC, many materials have been used as the solid support for the sta-
tionary liquid. In addition to silica gel, which was the first and is still the
most popular material, a variety of other adsorbents that adsorb the polar
solvent such as cellulose powder, starch, alumina, and silicic acid have been
used. The more recent practice of HPLC has greatly simplified the technique
in providing column stability for repeated use and for treatment of large
volumes.
Reversed-phase chromatography is the predominant technique in
HPLC, and chemically bonded silica gel supports are made specifically for
the nonpolar stationary phase. In the last decade, as many as 60% of the
published LLPC techniques refer to RPC. The reasons for this involve
the significantly lower cost of the mobile liquid phase and a favorable elu-
tion order that is easily predictable based on the hydrophobicity of the
eluate.
The selection of solvents for LLPC is similar to the selection of sol-
vents in liquid–liquid extraction systems. The solid support has little effect
upon the selection of the solvent pair, except for the obvious fact that a
hydrophilic support for a polar stationary phase requires a hydrophobic
mobile phase with the opposite for a reversed-phase system. Both solvents
should exhibit good solubility for the solute(s); otherwise the column
loading capacity would be too low. Frequently, the separation potential of
LLPC columns can be additionally enhanced by using solvent mixtures
rather than a single solvent as the mobile phase. In RPC, typical mobile
phases are water, aqueous electrolyte solutions, or aqueous mixtures of
one or more water-immiscible organic solvents. Water, the most polar of
mobile phases employed, is the weakest eluent. For specific purposes, such
as the separation of closely related acidic or basic substances, the mobile
phase is adjusted by buffers.
A large variety of analytical separation processes has been reported in
the literature. Table 13.4 [144–146] demonstrates the range of organic com-
pounds for which LLPC has been applied. The method has been limited
essentially to organic compounds, with much less use in the field of separa-
tion of metal ions or complexes. However, chromatographic separation pro-
cedures have been successfully used to separate metals with a combination of
a cation exchanger as a stationary phase and a solution of a chelating reagent
as a selective mobile phase. Thus chromatographic separations of many of
the rare-earth elements [Gd(III), Y(III), La(III), Pr(III), Nd(III), Ho(III),
Er(III), Sc(III)] from acidic solutions have been achieved using different
acidic organophosphorous extractants (e.g., DEHPA, PC88A) as well as
Table 13.4 Some Applications of Liquid–Liquid Partition Chromatography (LLPC)

Eluate Support Stationary phase Mobile phase
Aromatics alumina water n-heptane

Aliphatic alcohols silica acids water chloroform-CCl4
Fatty acids celite aq. sulphuric acid butanol-chloroform
Glycols celite water butanol-chloroform
Phenols silicic acid water isooctane
Amino acids silica gel water dichloromethane
Steroids silica acids methanol/water benzene-chloroform
Alkaloids silica gel water acetonitrile-water
Urine silica gel aq. sulphuric acid dichloromethane
Antibiotics silica gel water dichloromethane
Pesticides silica gel water n-heptane
neutral organophosphorous extractants (e.g., TOPO, TBP) immobilized

onto polymer beads as stationary phases [150,151]. Similar approaches have
been used in environmental control and determination of transuranium
elements (TUE), especially Pu(IV) and Am(III), in various natural samples
[152,153]. Among numerous sorbents for radioanalytical purposes, a new
type of solid sorbent (TVEX) based on introducing an extractant into the
polymer matrix during synthesis has been developed [141,142]. The behavior
of uranium, plutonium, and transplutonium elements (TPE) using TVEXs
with TBP, TOPO, DEHPA, and TOA, etc., has been studied. More recently,
at the Argonne National Laboratory in the United States [156,157], a new
family of impregnated resins based on the impregnation of Amberlite XAD7
with highly specific reagents for actinide analysis in environmental and
biological samples has been intensively studied.
13.6.2 Supercritical Fluid Extraction

Supercritical fluids were soon found to be highly efficient extraction media,
chiefly because of their high solvating power, their low viscosities (inter-
mediate between a gas and a liquid), and their low surface tensions that
enable their penetration deep into the extraction matrix. Supercritical fluid
extraction (SFE) used in isolation is generally not selective enough to sepa-
rate specific solutes from the matrix without further cleanup or resolution
from coextracted species prior to qualitative and quantitative analysis.
Consequently, for analytical applications, SFE is usually used in combination
with chromatographic techniques to improve the overall selectivity in the
isolation of specific solutes. The combined use of SFE with chromatographic
techniques is quite recent.
594 Aguilar et al.
The application of SFE for the preparation of samples in the analytical

laboratory has received serious attention as a sample preparation step for
extracting analytes of interest from a bulk matrix prior to their determi-
nation by other analytical methods including chromatographic, spectro-
metric, radiochemical, and gravimetric techniques.
Much of the current interest in using analytical-scale SFE systems
comes from the need to replace conventional liquid solvent extraction
methods with sample preparation methods that are faster, more efficient,
have better potential for automation, and also reduce the need for large
volumes of potentially hazardous liquid solvents. The need for alternative
extraction methods is emphasized by current efforts to reduce the use of
methylene chloride as an extraction fluid for environmental sample pre-
paration [158]. The potential for applying SFE to a wide variety of envi-
ronmental and biological samples for both qualitative and quantitative
analyses is widely described in reviews [159–161] and the references therein.
Analytical-scale SFE is most often applied to relatively small samples (e.g.,
several grams or less).
A good solvent for extraction should be selective so that it dissolves
the desired analytes to a greater degree than other constituents in the sample
matrix. It should be unreactive and stable, preferably nontoxic and from an
economic point of view noncorrosive to equipment and inexpensive to buy.
Many of these requirements are met by supercritical fluids such as carbon
dioxide.
13.6.2.1 Off-Line and On-Line SFE
Analytical-scale SFE can be divided into off-line and on-line techniques.
Off-line SFE refers to any method where the analytes are extracted using
SFE and collected in a device independent of the chromatograph or other
measurement instrument. On-line SF techniques use direct transfer of the
extracted analytes to the analytical instrument, most frequently a chromato-
graph. While the development of such on-line SFE methods of analysis has
great potential for eventual automation and for enhancing method sensi-
tivities [159–161], the great majority of analytical SFE systems described use
some form of off-line SFE followed by conventional chromatographic or
spectroscopic analysis.
To perform off-line SFE, only the SFE step must be successful, and the
extract can be analyzed at leisure by a variety of methods. The final product
of an off-line SFE experiment typically consists of the extract dissolved in a
few milliliters of liquid solvent, a form that is directly compatible with
conventional chromatographic injectors. In contrast, the successful perfor-
mance of on-line SFE requires that the SFE step, the coupling step (i.e., the
transfer and collection of extracted analytes from the SFE to the chromato-
graphic system), and the final chromatographic conditions be understood
and controlled. With many on-line approaches the sample extract is com-
mitted to a single analysis, and the extract is not available for analysis by
other methods. While off-line SFE has the advantages of simplicity and
availability of commercial instrumentation, and provides extracts that can
be analyzed by several instrumental methods, on-line techniques have
greater potential for enhanced sensitivity and automation.
Off-line SFE is conceptually a simple experiment to perform and
requires only relatively basic instrumentation. The instrumental components
necessary include a source of fluid, most often CO2 or CO2 with an organic
modifier, a means of pressurizing the fluid, an extraction cell, a method of
controlling the extraction cell temperature, a device to depressurize the
supercritical fluid (flow restrictor), and a device for collecting the extracted
analytes.
13.6.2.2 On-Line Coupling SFE with
Analytical Techniques
On-line supercritical fluid extraction/GC methods combine the ability of
liquid solvent extraction to extract efficiently a broad range of analytes with
the ability of gas-phase extraction methods to rapidly and efficiently transfer
the extracted analytes to the gas chromatograph. The characteristics of
supercritical fluids make them ideal for the development of on-line sample
extraction/gas chromatographic (SFE-GC) techniques. SFE has the ability
to extract many analytes from a variety of matrices with recoveries that rival
liquid solvent extraction, but with much shorter extraction times. Addi-
tionally, since most supercritical fluids are converted to the gas phase upon
depressurization to ambient conditions, SFE has the potential to introduce
extracted analytes to the GC in the gas phase. As shown in Fig. 13.8, the
required instrumentation to perform direct coupling SFE-GC includes suit-
able transfer lines and a conventional gas chromatograph [162,163].
Finally, supercritical fluid chromatography, in which a supercritical
fluid is used as the mobile phase, was introduced by Klesper [164–166].
SFE directly coupled to SFC provides an extremely powerful analytical tool.
The efficient, fast and selective extraction capabilities of supercritical fluids
allows quantitative extraction and direct transfer of the selected solutes
of interest to be accomplished to the column, often without the need for
further sample treatment or cleanup. Extraction selectivity is usually
achieved by adjusting the pressure of the supercritical fluid at constant
temperature or, less often, by changing the temperature of the supercritical
fluid at constant pressure. SFE coupled with packed column SFC has found
596 Aguilar et al.
Fig. 13.8 Schematic diagram of a simple SFE-GC system showing all the required
components. (Several manufacturers supply suitable components and specific suppliers are
listed only for the reader’s convenience.) Components are: (A) SFE grade extraction fluid
source; (B) 1.5 mm o.d. stainless steel tubing (0.77 mm or smaller i.d.); (C) shut-off valves
(SSI model 02–120 or equivalent, Supelco, Bellefonte, PA); (D) SFE pump; (E) SFE cell
heater; (F) approx. 0.5 m long coil of 1.5 mm stainless steel tubing for fluid preheater; (G)
1.5 mm1.5 mm tubing union (e.g., Parker or Swagelok brand); (H) finger-tight con-
nectors (e.g. Slip-Free connectors from Keystone Scientific Bellefonte, PA, USA); (I) SFE
cell; (J) restrictor connector ferrule (Supelco M2-A, Bellefonte, PA) which is used to
replace the stainless steel ferrule in the outlet end of the tubing union ‘G’; (K) 15–30 mm
i.d. fused silica tubing restrictor (Polymicro Technologies, Phoenix, AZ); (L) GC injection
port.
many applications in the last decade, particularly in connection with the

determination of food compounds, drug residues, herbicide and pesticide
residues, and polymer additives.
13.6.3 Solvent Extraction in Continuous

Flow Injection Analysis
13.6.3.1 Fundamental of Flow Injection Analysis
Flow injection analysis is based on the injection of a liquid sample into a
continuously flowing liquid carrier stream, where it is usually made to react
to give reaction products that may be detected. FIA offers the possibility
in an on-line manifold of sample handling including separation, pre-
concentration, masking and color reaction, and even microwave dissolution,
all of which can be readily automated. The most common advantages of
FIA include reduced manpower cost of laboratory operations, increased
sample throughput, improved precision of results, reduced sample volumes,
and the elimination of many interferences. Fully automated flow injection
analysers are based on spectrophotometric detection but are readily adapted
as sample preparation units for atomic spectrometric techniques. Flow
injection as a sample introduction technique has been discussed previously,
whereas here its full potential is briefly surveyed. In addition to a few books
on FIA [168,169], several critical reviews of FIA methods for FAAS, GF
AAS, and ICP-AES methods have been published [170,171].
As noted earlier, the essence of flow injection is the controlled physical
dispersion of an injected liquid sample into a continuous flowing unseg-
mented liquid carrier stream that may contain a suitable reagent to produce a
transient reproducible detector response proportional to the analyte con-
centration. Thus, a typical FIA system consists of the following components:
1. A propulsion unit, usually a peristaltic pump, that produces a flow of
one or several solutions (streams), either containing a dissolved reagent
or merely acting as a carrier
2. An injection system whose function is the reproducible introduction of
an accurately measured sample volume into the flow without stopping it
3. A mixing (reaction) unit
4. A flow cell for photometry detection
The simplest flow injection arrangement is the single-line configuration
shown in Fig. 13.9. As can be seen, the carrier stream is propelled by a pump
through a narrow conduit to an injector, then via the reaction manifold to
the detector and finally to waste. Each injection yields a single peak at the
detector, with the height, area, or width proportional to the analyte con-
centration.
598 Aguilar et al.
Fig. 13.9 Schematic diagram of a single-channel flow injection manifold, showing the
transient nature of the signal output. (--) Liquid flow; (.....) data flow.
13.6.3.2 Liquid–liquid Extraction in Continuous Flow

Analysis
A typical extraction manifold is shown in Figure 13.2. The sample is
introduced by aspiration or injection into an aqueous carrier that is seg-
mented with an organic solvent and is then transported into a mixing coil
where extraction takes place. Phase separation occurs in a membrane phase
separator where the organic phase permeates through the Teflon membrane.
A portion of one of the phases is led through a flow cell and an on-line
detector is used to monitor the analyte content. The back-extraction mode
in which the analyte is returned to a suitable aqueous phase is also some-
times used. The fundamentals of liquid–liquid extraction for FIA [169,172]
and applications of the technique [174–179] have been discussed. Pre-
concentration factors achieved in FIA (usually 2-5) are considerably smaller
than in batch extraction, so FI extraction is used more commonly for the
removal of matrix interferences.
Chemical reactions between the analyte and a reagent or combination
of reagents can be implemented in continuous systems to obtain a reaction
product that is capable of facilitating mass transfer and/or continuous
detection that results in increased sensitivity or selectivity. These reactions
could take place in different locations of the schemes defined previously, for
example:
1. In a coil after the sample is merged with a reagent stream.
2. In the solvent segmenter when it is necessary to avoid precipitation of
the extractable compound in the aqueous phase before it reaches the
continuous extractor (SS).
3. In the extraction coil (EC), when the extraction reaction is based on the
formation of metal chelates and the ligand is dissolved in the organic
solvent. Here the reaction and the transfer of the chelate take place
simultaneously.
4. In a coil located after the extractor using a reagent stream that is merged
prior to the detector with the phase containing the analyte.
The derivatizing reactions used in these systems are not substantially
different from those described in classical analytical determinations. The
most common are ion-pair formation reactions, where the analyte forms an
ion pair directly, e.g., determination of anionic surfactants by using ethyl
violet or methylene blue, and the determination of codeine with picric acid
and enalapril with bromothymol blue; or after reacting with a reagent to
form a bulkier, charged species e.g., tetracyanatocobaltate or phosphomo-
lybdate. Metal cations are typically transformed into extractable metal che-
lates by using such common ligands as described in section 13.4.1 (e.g., oxine,
dithizone, dimethylglyoxime, etc.). Some of the most important applications
of the technique are in the determination of surfactants in waters, quinizarin
in hydrocarbons, amines in pharmaceuticals, and drugs in urine [177].
Typical FIA manifolds are shown in Fig. 13.10 with two general
alternatives depending on whether injection takes place before or after the
continuous extractor device. The most common situation is when prior
injection of the sample take place. Figure 13.10(a) depicts a manifold for the
determination of vitamin B1 in pharmaceuticals [178], based on the oxida-
tion of thiamine to thiochrome in a carrier of potassium ferricyanide in a
basic medium (NaOH). The thiochrome is continuously extracted into a
chloroform stream and the fluorescence of the organic phase is measured
continuously.
Continuous multiextraction could be applied for the separation and
determination of carcinogenic polynuclear aromatic compounds (Fig. 13.10b)
in crude-oil-ash residues [179]. The ash sample dissolved in a cyclohexane
stream merges with a DMSO stream in the first solvent segmenter. After
passing through a glass extraction coil, the DMSO phase is mixed with a
water stream of cyclohexane segmented with the aqueous DMSO phase.
After passing through a Teflon extraction coil, the organic phase is carried
through the flow cell of a fluorimeter.
Fig. 13.10 Typical FIA extraction manifolds. Injection in front of the extractor com-
ponents: (a) simple; (b) multiextraction; and (c) injection after the extraction has been
carried out. (Based on systems from Refs. 178, 179.)
600
FIA can be used for the determination of traces of perchlorate in

serum and urine by formation of an ion pair with cuproine-type chelates
(CuL+ 2 ) and subsequent extraction with 6-methylpicolinealdehyde azine
dissolved in MIBK (Fig. 13.10c). The sample is continuously added to the
system and is mixed with another stream containing Cu(II) and ascorbic
acid. This mixture merges with the organic solution in the solvent seg-
menter. The organic phase stream containing the ion pair emerging from a
T-type separator fills the loop of an injector valve. This volume is then
injected into a water carrier in which copper is continuously determined and
hence perchlorate (0.5 to 5 ppb) may be indirectly determined [179].
ACRONYM LIST
AAS atomic absorption spectrophotometry
ASE accelerated solvent extraction
CCFA completely continuous flow analysis
CMC critical micelle concentration
CPE cloud point extraction
CZE capillary zone electrophoresis
FAAS furnace atomic absorption spectrophotometry
FIA flow injection analysis
GC gas chromatography
GC-MS mass chromatography
GF-AAS graphite furnace atomic absorption spectrophotometry
HPLC high-performance liquid chromatography
ICP inductively coupled plasma spectrophotometry
ICP-AES inductively coupled plasma with atomic emission spectrophoto-
metry
ISE ion-selective electrode
ISFET ion-selective field effect transistors
LLPC liquid–liquid partition chromatography
ME micelle-mediated extraction
MIMS membrane introduction mass spectrometry
PAH polynuclear aromatic hydrocarbon
PCDF polychlorinated dibenzofuran
PCB polychlorinated biphenyl
PF preconcentration factor
PONPE polyoxyethylene nonyl phenyl ether
PTFE polytetrafluoroethene
PVC porous polyvinyl chloride
REE rare earth elements
602 Aguilar et al.
RPC reversed-phase chromatography

SC synergistic coefficient
SFA segmented flow analysis
SFE supercritical fluid extraction
SLM supported liquid membranes
SPE solid-phase extraction
SPS solid-phase spectroscopy
THM trihalomethane
TUE transuranium elements
UFA unsegmented flow analysis
ALAMINE 336 mixture of trioctyl- and tridecylamine
ALIQUAT 336 methylated (quaternized) alamine 336
DEHPA di-(2-ethylhexyl)phosphoric acid
MIBK methylisobutylketone
DMSO dimethylsulfoxide
PAN 1-(2-pyridylazo)-2-naphthol
PAR 4-(2-pyridylazo)resorcinol
PC88A 2-ethylhexylphosphonic acid 2-ethylhexyl ester
Oxine 8-hydroxyquinoline
SDVB styrene divinylbenzene
TBP tributylphosphate
TOA tri-n-octylamine
TOPO trioctylphosphine oxide
Thioxine 8-quinolinethiol
REFERENCES
1. Morrison, M. G.; Freiser, H. Solvent Extraction in Analytical Chemistry; John
Wiley and Sons; New York, 1957.
2. Stary, J. The Solvent Extraction of Metal Chelates; Pergamon Press: Oxford,
1964.
3. De, A. K.; Khopkar, S. M.; Chalmers, R. A. Solvent Extraction of Metals;
Van Nostrand Reinhold: London, 1970.
4. Alders, L. Liquid-Liquid Extraction; Elsevier: New York, 1995.
5. Zolotov, Y. A.; Bodnya, V. A.; Zagruzina, A. N. CRC Crit. Rev. Anal.
Chem., 1982, 14, 2.
6. Keller, R.; Mermet, J. M.; Otto, M.; Widmer, H. M. Eds. Analytical Chem-
istry; Wiley-VCH, Verlag GmbH: Weinheim, Germany, 1998; p. 25.
7. Valcarcel, M. Liquid-Liquid Extraction in Continuous Flow Analysis. In
Developments in Solvent Extraction; S. Alegret, Ed., Ellis Horwood Series in
Analytical Chemistry: West Sussex, UK, 1988; p. 135.
8. Preconcentration of Trace Elements. In Comprehensive Analytical Chemistry;

Lobinsky, L., Ed.; Vol. XXV, Elsevier: Amsterdam, 1992.
9. Aguilar, M. Graphical Treatment of Liquid-Liquid Equilibrium Data. In
Developments in Solvent Extraction, Ed. Alegret, S. Ed., Ellis Horwood Series
in Analytical Chemistry, West Sussex, UK, 1988; p. 87.
10. Tagushi, S.; Goto, K. Pre-Concentration Techniques for Trace Elements; Al-
fassi, Z. B.; Wai, C. M. Ed.; CRC Press: Boca Raton, Florida, 1989; p. 31.
11. Torre, M.; Marina, M. L. Crit. Rev. in Anal. Chem., 1994, 24, 327.
12. Cortina, J. L.; Warshawsky, A. Ion Exchange and Solvent Extraction, Mar-
insky, J. A.; Marcus, Y. Ed.; Marcel Dekker: New York, Vol. 13, 1996; 195.
13. Karasek, F. W.; Clement, R. E.; Sweetman, J. A. Anal. Chem., 1981, 53(9),
1051A.
14. Webb, R. C. EPA Report EPA-660/475-003, June 1975.
15. Bowers, W. D.; Parsons, M. L.; Clement, R. E.; Kasarek, F. W. J. Chroma-
togr., 1981, 207, 203.
16. Lobinski, R.; Marczenko, Z. Eds., Spectrochemical Trace Analysis for Metals
and Metalloids. In Comprehensive Analytical Chemistry, Volume XXX; Else-
vier: Amsterdam, Volume XXX, 1996.
17. Cheng, K. L.; Ueno, K.; Imamura,T. Handbook of Organic Analytical Re-
agents; CRC Press: Boca Raton, Florida, 1982.
18. Townshend, A.; Burns, D. T.; Guibault, G. G.; Lobinski, R.; Marczenko, Z.;
Newman, E. J.; Onishi, H. Eds. Dictionary of Organic Analytical Reagents;
Chapman and Hall: London, 1993.
19. Lobinski, R.; Marczenko, Z. CRC Crit. Rev. Anal. Chem., 1992, 23, 55.
20. Kaiyanaraman, S.; Khopkar, S. M. Anal. Chim. Acta., 1975, 78, 231.
21. Murata, M.; Omatsu, M.; Mushimoto, S. X-ray Spectrom., 1984, 13, 83.
22. Maguar, B.; Lobanov, F. L. Talanta, 1973, 20, 55.
23. Cresser, M. Solvent Extraction in Flame Spectroscopic Analysis; Butterworths:
London 1978.
24. Anders, U.; Mailer, D. Z. Anal. Chem., 1976, 278, 203.
25. Cordis, V.; Sigartau, G.; Pop, J. Z. Anal. Chem., 1976, 279, 355.
26. Fujinaga, T.; Nagaosa, Y. Chem. Lett., 1978, 6, 587.
27. Keller, C. Radiochemistry; Ellis Horwood Series in Physical Chemistry: Chi-
chester, UK, 1988.
28. Zorov, N. B.; Gocci,T. F.; Winefordner, J. D. Talanta, 1976, 23, 455.
29. Zolotov, Y A.; Petrukhin, O. M.; Nagy, V. Y.; Volodarskii, L. B. Anal.Chim.
Acta, 1980, 115, 1.
30. Zolotov, Y. A.; Kuzmin, N. M.; Petrukhin, O. M.; Spivakov, B. Y. Anal.
Chim. Acta, 1986, 180, 137.
31. Guter, G. A.; Hammond, G. S. J. Amer. Chem. Soc., 1956, 78, 5166.
32. Fix, R. C.; Irvine, J. W. Jr., Lab. Nucl. Sci. Prog. Rep., May 1956.
33. Shigematsu, T.; Kudo, K. J. Radioanal. Chem., 1978, 45, 373.
34. Hoentsch, E.; Flachowsky, J. Isotopenpraxis, 1977, 13, 432.
35. Iqbal, M.; Ejaz, M.; Chaudhri, S. A.; Zamiruddin, A. J. Radioanal. Chem.,
1978, 42, 335.
604 Aguilar et al.
36. Jauniaux, M.; De Meyer, M.; Lejeune, W.; Levert, J. M. Bull. Soc. Chim.
Belg., 1975, 84, 565.
37. Luke, C. L. Anal. Chem., 1956, 28, 1443.
38. Marczenko, Z.; Kalowska, H. Chem. Anal. (Pol.), 1976, 21, 183.
39. Shtefak, M.; Kirsh, M.; Rais, I. Zh. Anal. Khim., 1976, 32, 1364.
40. Takeda, Y.; Suzuki, S.; Ohyagi, Y.; Chem. Lett., 1978, 12, 1377.
41. Roberts, A. H. C.; Turner, M. A.; Syers, T. K. Analyst, 1976, 101, 574.
42. Karisson, R.; Gorton, L. Talanta, 1976, 23, 672.
43. Carry, J.; Menon, M. P. J. Radioanal. Chem., 1975, 34, 319.
44. Gentry, C. H. R.; Sherrington, L. G. Analyst, 1946, 71, 432.
45. Argollo, R. M.; Schilling, J. G. Anal. Chem. Acta., 1978, 96, 117.
46. Fujinaga, T.; Puri, B. K. Talanta, 1975, 22, 71.
47. Bode, H.; Z. Anal. Chem., 1955, 144, 165.
48. Kaiyanarman, S.; Khopkar, S. M. Anal. Chem., 1977, 49, 1192.
49. Charib, A.; Morris, D. F. C. Talanta, 1978, 25, 569.
50. Tomazic, B.; Branica, M. J. Radioanal. Chem., 1976, 30, 361.
51. Pollock, E. N. Anal. Chim. Acta., 1977, 88, 399.
52. Fantova, J.; Kliz, Z.; Cermakova, Z.; Suk, V. Chem. Listy, 1980, 74, 291.
53. Shimomura, S.; Sakurai, H.; Morita, I.; Mino, Y. Anal. Chem. Acta., 1978,
96, 69.
54. Steinnes, E. Radiochem. Radioanal. Lett., 1978, 33, 205.
55. Shamaev, Y. I.; Bogdanov, N. V. Zh. Anal. Khim., 1976, 31, 72.
56. Hasemann, K.; Bock, R. Z. Anal. Chem., 1985, 274, 185.
57. Huffman, E. H.; Beaufait, L. J. J. Amer. Chem. Soc., 1949, 71, 3179.
58. Wadelin, C.; Mellon, M. G. Anal. Chem., 1953, 25: 1668.
59. Fetchett, A. W.; Daughtrey, D.; Hunter, E. Anal. Chim. Acta., 1975, 79, 93.
60. Kamada, T.; Yamamoto, Y. Talanta, 1977, 24, 330.
61. Nakamura, K.; Ozawa, T. Anal. Chim. Acta., 1976, 86, 147.
62. Svehla, G.; Tolg, G. Talanta, 1976, 23, 755.
63. Lapenko, L. A.; Gibalo, I. M. Zh. Anal. Khim., 1976, 31, 481.
64. Spivakov, B. Y.; Orlova, Y. A.; Malyutina,T. M.; Kirilova, T. L. Zh. Anal.
Khim., 1979, 34, 161.
65. Kamada, T. Shiroishi, T.; Yamamoto, Y. Talanta, 1978, 25, 15.
66. Chao, T. T.; Sanzolone, R. F.; Hubert, A. E. Anal. Chim. Acta., 1978, 96, 251.
67. Kudo, K.; Shigernatsu, T.; Kobayashi, K. J. Radioanal. Chem., 1977, 36, 65.
68. Bhowal, S. K.; Umiand, E. Anal Chem., 1976, 282, 197.
69. Musil, J.; Dolezal, J. Anal. Chim. Acta, 1977, 92, 301.
70. Roberts, R. F. Anal. Chem., 1977, 49, 1861.
71. Tribalat, S.; Beydon, J. Anal. Chim. Acta., 1953, 8, 22.
72. Fujinaga, T.; Lee, H. L. Talanta, 1977, 24, 395.
73. Fujinaga, T.; Puri, B. K. Indian J. Chem., 1976, 14, 72.
74. Zolotov, Y. A.; Petrukhin, O. M.; Shevehenko, U. N.; Dunina, U. Y.; Ru-
khadze, E. G. Anal. Chim. Acta., 1978, 100, 613.
75. Matsuoka, S.; Yoshimura, K.; Waki, H. Mem. Fac. Sci. Kyushu Univ., Ser.
C., 1991, 18, 55–62.
76. Compaño, R.; Ferrer, R.; Guiteras, J.; Prat, M. D. Analyst, 1994, 119, 1225.
77. Kirkbright, G. F.; Narayanaswamy, R.; Welti, N. A. Analyst, 1984, 109, 15.
78. Narayanaswamy, R.; Sevilla, F. Anal. Chim. Acta., 1986, 189, 365.
79. Jönsson, J. Å.; Mathiasson, L. Trac-Trend. Anal. Chem., 1999, 18, 318.
82. Audunsson, G. Anal. Chem., 1986, 58, 2714.
83. Norberg, J.; Zander, Å.; Jönsson, J. Å. Chromatographia, 1997, 46, 483.
84. Nilvé, G.; Stebbins, R. Chromatographia, 1991, 32, 269.
85. Nilvé, G.; Audunsson, G.; Jönsson, J. J. Chromatogr., 1989, 471, 151.
86. Audunsson, G. Anal. Chem., 1988, 60, 1340.
87. Lindegård, B.; Jönsson, J. Å.; Mathiasson. L. J. Chromatogr., 1992, 573, 191.
88. Thordarson, E.; Pálmarsdóttir, S.; Mathiasson, L.; Jönsson, J. Å. Anal.
Chem., 1996, 68, 2559.
89. Norberg, J.; Emnéus, J.; Jönsson, J. Å.; Mathiasson, L.; Burestedt, E.;
Knutsson, M.; Marko-Varga, G. J. Chromatogr. B, 1997, 701, 39.
90. Pálmarsdóttir, S.; Lindegård, B.; Deiningr, P.; Edholm, L. E.; Mathiasson, L.;
Jönsson, J. A. J. Cap. Electrophor., 1995, 2, 185.
91. Pálmarsdóttir, S.; Mathiasson, L.; Jönsson, J. Å.; Edholm, L. E. J. Chro-
matogr. B, 1997, 688, 127.
92. Pálmarsdóttir, S.; Thordarson, E.; Edholm, L. E.; Jönson, J. Å.; Mathiasson,
L. Anal. Chem., 1997, 69, 1732.
93. Bauer, S.; Solyom, D. Anal. Chem., 1994, 66, 4422.
94. Cisper, M. E.; Hembeergeer, P. H. Rapid Commun. Mass Sp., 1997, 11,
1454.
95. Johnson, R. C.; Koch, K.; Cooks, R. G. Ind. Eng. Chem. Res., 1999, 38, 343.
96. Johnson, R. C.; Koch, K.; Cooks, R. G. Anal. Chim. Acta., 1999, 395, 239.
97. Shen, Y.; Obuseng, V.; Grönberg, L.; Jönsson, J. Å. J. Chromatogr. A, 1996,
725, 189.
98. Grönberg, L.; Shen, Y.; Jönsson, J. Å. J. Chromatogr. A, 1993, 655, 207.
99. Djane, N. K.; Bergdahl, I. A.; Ndung’u, K.; Schütz, A.; Johansson, G.; Ma-
thiasson, L. Analyst, 1997, 122, 1073.
100. Djane, K. N.; Ndung’u, F.; Malcus, G.; Johansson, G.; Mathiasson, L. Fre-
senius J. Anal. Chem., 1997, 358, 822.
101. Cox, J. A.; Bhatnagar, A. Talanta, 1990, 37, 1037.
102. Parthasarathy, N.; Pelletier, M.; Buffle, J. Anal. Chim. Acta., 1997, 350, 183.
103. Djane, N. K.; Ndung’u, K.; Johnson, K.; Sartz, H.; Törnström, T.; Ma-
thiasson, L. Talanta, 1999, 48, 1121.
104. Scamerhorn, J. F.; Harwell, J. H. Surfactant-Based Separations Processes;
Marcel Dekker: New York, 1989; p. 139.
105. Hinze, W. L.; Pramauro, E. Criti. Rev. Anal. Chem., 1993, 24, 133.
106. Carabias-Martı́nez, R.; Rodrı́guez-Gonzalo, E.; Moreno-Cordero, B.; Pérez-
Pavón, J. L.; Garcia-Pinto, C.; Fernández Laespada, E. J. Chromatogr. A,
2000, 902, 195.
107. Quina, F. H.; Hinze, W. L. Ind. Eng. Chem. Res., 1999, 38, 4150.
606 Aguilar et al.
108. Huddleston, J. G.; Willauer, H. D.; Griffin, S. T.; Rogers, R. D. Ind. Eng.
Chem. Res., 1999, 38, 2523.
109. Watanabe, H.; Tanaka, H. Talanta, 1978, 25, 585.
110. Garcı́a Pinto, C.; Pérez Pavón, J. L.; Moreno Cordero, B.; Romero Beato, E.;
Garcı́a Sánchez, S. J. Anal. Atomic Spectrometry, 1996, 11, 37.
111. Akita, S.; Rovira, M.; Sastre, A. M.; Takeuchi, H. Sep. Sci. Technol., 1998,
33, 2159.
112. Okada, T. Anal. Chem., 1992, 64, 2138.
113. Horvath, W. J.; Huie, C. W. Talanta, 1993, 9, 1385.
114. Garcı́a Pinto, C.; Pérez Pavón, J. L.; Moreno Cordero, B. Anal. Chem., 1995,
67, 2606.
115. Stangl, G.; Niessner, R. Mikrochim. Acta., 1994, 113, 1.
116. Ferrer, R.; Beltran, J. L.; Guiteras, J. Anal. Chim. Acta., 1996, 330, 199.
117. Eiguren, A.; Sosa, Z.; Santana, J. J. Analyst 1999, 124, 487.
118. Calvo Seronero, L.; Fernández Laespada, M. E.; Pérez Pavón, J. L.; Moreno
Cordero, B. J. Chromatogr. A, 2000, 897, 171.
119. Eiguren, A.; Sosa, Z.; Santana, J. J. Anal. Chim. Acta., 1998, 358, 145.
120. Moreno, B.; Pérez, J. L.; Garcı́a, C.; Fernández, M. E. Talanta, 1993, 40,
1703.
121. Carabias-Martı́nez, R.; Rodrı́guez-Gonzalo, E.; Domı́nguez-Alvarez, J.;
Hernández-Méndez, J. Anal. Chem., 1999, 771, 2468.
122. Eisenman, G. Ion Selective Electrodes; National Bureau of Standards (USA)
Special Publications: Washington, D.C., 1969; p. 314.
123. Janata, J.; Josowitz, M.; de Vaney, D. M. Anal. Chem., 1994, 66, 207R.
124. Camman, K.; Lemke, U.; Rohen, A.; Sander, J.; Wilken, H.; Winter, B.
Angew. Chem. 1991, 109, 519.
125. Szczepaniac, W.; Oleksy, J. Anal. Chim. Acta., 1986, 189, 237.
126. Worf, W. E. The Principles of Ion Selective Electrodes and of Membrane
Transport; Akademiai Kiadó: Budapest, 1981; Elsevier: Amsterdam, 1981.
127. Koryta, J. Ion-Selective Electrodes; Cambridge University Press: London,
1975.
128. Freiser, H. Ion-Selective Electrodes in Analytical Chemistry; Plenum Press:
New York, Vol. 1, 1978.
129. Pretsch, E.; Buchi, R.; Amman, D.; Simon, W. Essays on Analytical Chem-
istry; Pergamon Press: New York, 1977.; p. 321.
130. Svehia, G. Comprehensive Analytical Chemistry; Elsevier: New York, Vol. 11,
Chap. 3, 1981; p. 346.
131. Hobby, P. C.; Moody, G. J.; Thomas, J. D. R. Analyst, 1983, 108, 581.
132. Geissler, M.; Kunze, R. Anal. Chim. Acta., 1986, 189, 245.
133. Simon, W.; Pretsch, E.; Morf, W. E.; Amman, D.; Oesch, U.; Dinten, O.
Analyst, 1984, 109, 207.
134. Simon, W.; Morf, W. E.; Meier, P. C. Struct. Bonding, 1973, 16, 113.
135. Eisenman, G. Anal. Chem., 1968, 40, 311.
136. Xie, R. Y.; Christian, G. D. Analyst, 1987, 112, 61.
137. Pioda, L. A. R.; Stankova, Y.; Simon, W. Anal. Lett., 1969, 2, 665.
138. Scholer, R. P.; Simon, W. Chimia, 1970, 24, 372.

139. Mair, B. J. Treatise on Analytical Chemistry, Part 1, Theory and Practice;
Kolthoff, I. M.; Elving, P. J.; Sandeli, E. B. Eds., Wiley-Interscience: New
York, Vol. 3, 1961: p. 1469.
140. Pecsok, R. L.; Shields, L. D.; Cairns, T.; McWilliam, L. G.Modern Methods
of Chemical Analysis, 2nd ed.; John Wiley and Sons: New York, 1968; pp.
41–70.
141. Huber, J. F. K. Comprehensive Analytical Chemistry, eds. Wilson, C. L.;
Wilson, D. W.; Stroits, C. R. N. Elsevier: Amsterdam, Vol. 11B, 1968; p. 33.
142. Peters, D. G.; Hayes, J. M.; Hieftje, G. M. Modern Chemical Analysis;
W.B. Saunders; Philadelphia, 1976.
143. Hamilton, R. J.; Sewei, P. A. Introduction to High Performance Liquid
Chromatography; Chapman and Hall: London, 1977.
144. Snyder, L. R.; Kirkland, J. J. Modern Liquid Chromatography; John Wiley and
Sons: New York, 1974.
145. Parris, N. H. Instrumental Liquid Chromatography; Elsevier: Amsterdam, 1976.
146. Lawrence, J. F.; Frei, R. W. Chemical Derivitization in Liquid Chromato-
graphy; Elsevier: Amsterdam, 1976.
147. Horvath, C. Ed. High-Performance Liquid Chromatography; Academic Press:
New York, Vols. 1–3, 1980–1983.
148. Mulier, W. Kontakte (Merck), 1986, 3, 3.
149. Braun, R.; Ghersini, G. Extraction Chromatography; Elsevier: Amsterdam,
1975.
150. Wakui, Y.; Matsunaga, H.; Suzuki, T. M. Anal. Sci, 1988, 4, 325.
151. Muscatello, A. C.; Yarbro, S. L.; Marsh, S. F. New Separation Chemistry
Techniques for Radioactive Waste and other Specific Applications; Cecille, L.;
Casarci, M.; Pietrelli, L. Eds.; Elsevier Science Publishing: Amsterdam, 1991;
p. 88.
152. Dietz, M. L.; Horwitz, E. P.; Nelson, D. M.; Walgren, M. Health Physics
1991, 61, 871.
153. Horwitz, E. P.; Dietz, M. L.; Nelson, D. M.; La Rosa, J. J.; Fairman, W.
Anal. Chim. Acta., 1990, 238, 263.
154. Myasoedova, G. V.; Savvin, S. B. CRC Crit. Rev. Anal. Chem. 1988, 17.
155. Kremlyakova, N.; Barsukova, K. V.; Myasoedov, B. F. Radiochem. Radio-
anal. Lett., 1983, 57, 293.
156. Horwitz, E. P.; Dietz, M. L.; Fisher, D. E. Anal. Chem., 1991, 63, 522.
157. Horwitz, E. P.; Chiarizia, R.; Dietz, M. L. Solvent Extr. Ion Exch., 1992, 10,
313.
158. Westwood, S. A. Ed. Supercritical Fluid Extraction and its Use in Chromato-
graphic Sample Preparation; Blackie Academic and Professional, Chapman
and Hall: UK, 1993.
159. Veuthy, J. L.; Caude, M.; Rosset, R. Analysis, 1990, 18, 103.
160. Hawthorne, S. B.; Anal. Chem. 1990, 62, 633.
161. Vannoort, R. W.; Chervet, J. P.; Lingeman, H.; De Jong, G. J.; Brikman, A.
U. Th. J. Chromatogr., 1990, 505, 45.
608 Aguilar et al.
162. Hawthorne, S. B. Coupled (on-line) supercritical fluid extraction-gas chro-

matography. In Supercritical Fluid Extraction and Its Use in Chromatographic
Sample Preparation; Westwood, S. A. Ed., Blackie Academic and Profes-
sional, Chapman and Hall; UK, 1993; p. 65.
163. Hawthorne, S. B.; Miller, D. J. J. Chromatogr. Sci, 1985, 24, 258.
164. Klesper, E.; Corwin, A. H.; Turner, D. A. J. Org. Chem., 1962, 27, 700.
165. Sugiyama, K.; Saito, M.; Hondo, T.; Seda, M. J. Chromatogr. 1985, 332, 107.
166. Anderson, I. G. M. Supercritical fluid extraction coupled to packed column
supercritical fluid chromatography. In Supercritical Fluid Extraction and Its
Use in Chromatographic Sample Preparation; Westwood, S. A. Ed., Blackie
Academic and Professional, Chapman and Hall: UK, 1993; p. 112.
167. Foreman, J. K.; Stockweli, P. B. Automatic Chemical Analysis; Ellis Horwood:
Chichester, UK, 1975.
168. Foreman, J. K.; Stockweli, P. B. Topics in Automatic Chemical Analysis; Ellis
Horwood: Chichester, UK, Vol. 1, 1979.
169. Valcarecel, M.; Luque de Castro, M. D. Flow-Injection Analysis; Ellis Hor-
wood: Chichester, UK, 1987.
170. Furman, W. B. Continuous Flow Analysis: Theory and Practice; Dekker: New
York, 1976.
171. Coakley, W. A. Handbook of Automatic Analysis. Continuous Flow Techni-
ques; Dekker: New York, 1982.
172. Ruzicka, J.; Hansen, E. H. Flow-Injection Analysis; Wiley: New York, 1981.
173. Valcarcel, M.; Gallego, M. Separation techniques. In Flow Injection Atomic
Spectroscopy, Bruguera, J. L. Ed., Marcel Dekker: New York, 1989.
174. Luque de Castro, M. D.; Valcarcel, M. Trends in Analytical Chemistry, 1991,
10, 114.
175. MacLaurin, P.; Andrew, K. N.; Worsfold, P. J. Flow Injection Analysis. In
Process Analytical Chemistry, McLennan, F.; Kowalski, B. R. Eds.; Blackie
Academic and Professional: Glasgow, 1995; p. 159.
176. Valcarcel, M.; Luque de Castro, M. Non chromatographic Continuous Se-
paration Techniques; Royal Society of Chemistry: Cambridge, UK, 1990.
177. Karlberg, B.; Thelander, S. Anal. Chim. Acta., 1978, 98, 1.
178. Shelly, D. C.; Rossi, T. M.; Warner, I. M. Anal. Chem., 1982, 54, 87.
179. Gallego, M.; Valcarcel, M. Anal. Chim. Acta., 1985, 169, 161.
14
The Use of Solvent Extraction in
the Recovery of Waste
MICHAEL COX* University of Hertfordshire, Hatfield, Hertfordshire,
United Kingdom
HANS REINHARDT MEAB Metallextraktion AB, Göteborg, Sweden
14.1 INTRODUCTION
Recently the protection of the environment has become increasingly
important for industry with the requirement that the potential impact on the
environment is considered for all aspects of industrial processes. Such
considerations are supported by environmental legislation that controls all
types of emissions as well as the treatment of wastes. Such legislation is
based on global standards that have largely resulted from developments
within the European Union, Japan, and the United States in collaboration
with international conventions. Of these, the Basel Convention (1989) and
the Earth Summit in Rio de Janeiro (1992) were significant in the control
and prevention of wastes. In the case of liquid wastes that are most
appropriate for treatment by liquid–liquid extraction, limits for discharge
into the aqueous environment have been established by the three countries
already mentioned. These limits depend on the particular country and
sometimes on the industry. (See section 14.6.)
Environmental limits vary according to country, and in the future
pressure will be applied to those countries where limits are considered too
high. Indeed, there is a general trend for discharge limits to be reduced, with
the concept of zero industrial discharge for certain metals being the ultimate
aim for the future.
Metal pollution can arise from a number of different sources, for
example:
*Retired.
609
610 Cox and Reinhardt
Acidic mine waters emerging naturally from underground mines

Leachates from solid residues from the processing of primary minerals
such as flotation and leach residues, and from landfill
Metal treatment processing including coating (zinc), plating (chromium,
nickel, copper), and pickling (hydrofluoric acid, iron)
Chemical wastes from other industries: tanning (chromium), photo-
graphic (silver), electronic (copper, silver, gold, nickel, lead, cadmium)
Finally, there are many metal-containing solid wastes that may
undergo leaching if disposed to land: spent catalysts (cobalt, nickel, vana-
dium); spent batteries (nickel, cadmium, lithium, lead); combustion ashes;
etc.
Although this chapter concentrates on environmental problems asso-
ciated with metals, the discharge of organic compounds is also subjected to
stringent limits. The compounds that cause most concern are pesticides,
herbicides, phenols, biphenyls, polychlorinated hydrocarbons (PCBs), and
polyaromatic hydrocarbons (PAHs).
This chapter concentrates on the applications of solvent extraction. It
should be remembered that other technologies could be applied to these
wastes. These range from the well-established technologies of precipitation,
adsorption, and electrolysis to those more recently developed, e.g., ion
exchange and membrane processes: The flow sheet finally chosen will
depend on a number of factors, including the precise composition of the
waste, local expertise, disposal options, and of course the overall economics.
In addition, this flow sheet may involve more than one technique. Thus one
disadvantage of the use of solvent extraction, the possibility of entrained
organic compounds in the raffinate, can be minimized or eliminated by the
addition of a final adsorbent column before discharge.
14.2 SOLVENT EXTRACTION AND THE

ENVIRONMENT
Before considering particular applications, the potential for the recovery of
values from wastes is discussed. In other chapters the various applications of
solvent extraction have been presented and the technology has been shown
to be highly selective, easy to operate and control, and versatile in terms of
scale. Therefore, it seems to be ideal for application in the treatment of
wastes, both solid (after leaching) and liquid. As noted earlier, however,
there are environmental limitations mainly caused by unintentional release
of the solvent into the environment. These problems can be minimized by
ensuring that the treated solutions are recycled within the plant and that any
solutions discharged into the environment are treated appropriately to
Solvent Extraction in the Recovery of Waste 611
remove entrained organic compounds and residual toxic metals. Thus the
main applications of solvent extraction in treatment of wastes are in the
removal and recovery of values from waste streams to provide a process
water for recycling. This reduces the effect on the environment by mini-
mizing the amount of effluent for treatment and reduction in fresh water
consumption.
The most important metalliferous liquid effluents where solvent
extraction could be applied are from the various metal finishing operations:
plating, pickling, etching, and the wash waters arising from the cleaning of
work pieces. In the case of solids, in addition to scrap metal and alloy wastes
from manufacturing operations, a number of other products use valuable
and toxic metals and offer potential applications, e.g., spent automobile
catalysts, Ni/Cd batteries, etc.
In the treatment of both types of waste, the objectives must be:
To remove the metals effectively so as to produce a liquid waste stream
capable of reuse or finally meeting environmental discharge limits
To recover the metals for recycling within the plant, or at appropriate
purity for sale
To separate other impurities in a form for resale as by-products or safe
environmental discharge
These principles are embodied in the concepts of zero discharge and
sustainable technology.
14.3 RECOVERY OF METALS FROM WASTE:

GENERAL ASPECTS
Successful use of solvent extraction in processes for the recovery of metals
from waste depends extensively on successful pre- and posttreatments. Sol-
vent extraction is a selective separation procedure for metals; however, it is
limited to metals dissolved in aqueous media. It is obvious that solid wastes
containing metals need a primary step of leaching to bring the metals in
solution and it is very seldom that the resulting strip liquor from the solvent
extraction process has a direct use. The technology covering these types of
treatment is termed hydrometallurgy, defined as involving a sequence of
chemical reactions of metals carried out in aqueous solutions.
To discuss the processes for the recovery of metals from waste, more
than one hydrometallurgical procedure has to be considered, to establish
general processing concepts for a broad range of feed materials, such as:
Mine waters and leachates from residue dumps
Plating, pickling, etching solutions, including rinse waters
Neutralization sludge from the plating industries

Flue dust from steel and brass mills
Ash and scrap from industrial production
There are three main kinds of operations used within hydrometallurgy for
metal recovery:
Leaching (mixing and solid/liquid separation)-dissolution with acid,
alkali, or bacteria, using redox, pressure (autoclave), etc.
Purification and separation-solvent extraction, ion exchange, cemen-
tation, adsorption, liquid membranes, reverse osmosis, etc.
Generation of products-evaporation, electrowinning, precipitation,
crystallization
As can be seen from the list, there are many competing procedures for
leaching, selective separation of the metals, and generation of products.
Thus, the detailed design of a specific recovery process often involves a
number of selected procedures, combined in unique ways. This chapter
discusses processes containing a solvent extraction separation procedure.
14.4 WASTE RECOVERY PROCESSES

INVOLVING SOLVENT EXTRACTION
To use solvent extraction for metal separation, it is first necessary to have
the metals dissolved in an aqueous solution. Starting with a solid waste
material, the initial treatment is therefore a leaching procedure to provide an
aqueous solution from which the metals may be recovered [1–4]. The dis-
solved metal ions, together with all other ions in solution, interact with one
another in various ways. The extent to which they do so and the resulting
changes in behavior are of great importance in solution chemistry and are
discussed in Chapter 2. In general, association occurs between ions of
opposite sign, and the commonly used method of describing the reactions is
the complex formation approach.
The interaction between the ions is regarded as a chemical equilibrium.
The resulting product of such an association between two individual kinds
of ions is called a complex species, and the extent to which these various
complex species occurs are described in terms of the equilibrium constants.
It should be noted that in leaching the rate of reaction is of great
importance, since the temperature used is relatively low and many factors
combine to keep the rate slow. For practical reasons, leaching studies are
performed on a homogenized sample, testing the temperature dependency of
the rate of reaction (concentration and time) in various leaching mixtures.
The most important variables in aqueous systems are pH and redox
potential, together with the metal concentration. Oxidation is frequently

used to make the solid to be leached unstable with respect to the solution
and thus promote dissolution.
The importance of an optimal leaching procedure can be illustrated by
considering a solvent extraction process for the production of 10,000 metric
tons of copper. Commonly achieved leach solutions contain 2.5 g dm3 Cu
in the feed solution. This value results in an aqueous feed flow from the
leaching operation to the mixer-settler units of about 500 m3 h1. To this
add 200 m3 h1 of organic solvent and allow organic recycle to achieve
organic continuous conditions in the mixer, and the total liquid flow
through the mixer-settlers will be over 1000 m3 h1. Now, if it is possible to
design the leaching operation to provide a leach solution of 25 g dm3 Cu,
then both the dimensions of the leaching and solvent extraction equipment
can be drastically reduced.
14.4.1 Solvent Extraction from Acid

Feed Solutions
As seen in Chapters 4 and 11, the extraction of many metals with an acid
extractant (H+R) is pH dependent, with, in general, the lower the pH, the
lower the extraction of the metal. In other words, the D-value (or % ex-
traction) decreases with decreasing pH (Fig. 11.8). In addition, the extraction
of each metal ion (Mn+) liberates n hydrogen ions (H+), from the extractant
H+R. This means that the pH decreases with extraction, resulting in
decreased transfer of metal to the organic solvent. Thus, the more metal in
solution and the higher acidity, the less is extracted, no matter how many
stages are used.
For example, the extraction of nickel with di-(2-ethylhexyl)phosphoric
acid (DEHPA) at a pH below 3 is <10%. If the extraction of nickel is
carried out between pH 3–4, the extraction of about 0.3 g dm3 Ni2+ will
decrease the pH to below 3 and the mass transfer of nickel will stop. For
copper, the extraction is performed between pH 2–3 and thus, an amount of
3 g dm3 Cu2+ will decrease the pH from 3 to 2 and again extraction will
stop.
To allow a higher extraction capacity (net transfer) and avoid the
influence of liberated H+ ions, simultaneous neutralization with NaOH
directed into the mixer is used. Such an example is described later, where
nickel is extracted from a cadmium electrolyte. Other means of controlling
the pH includes the use of acid extractants in either alkaline (NH+ 4 R) or
neutral (Na+R) form.
An increase of the net transfer of zinc was reported [9] for the
extraction with DEHPA (50%) in kerosene of zinc from a feed solution
containing 40 g dm3 Zn, emanating from the leaching of steel furnace dust.
The resulting raffinate, with a residual Zn concentration of 15–20 g dm3,
was recycled to leaching. Comparable net transfer conditions were obtained
[5] in the modified Zincex process (MZP) for the processing of spent
domestic batteries.
Similar high transfer of copper in a solvent extraction procedure has
been recently reported [6]. After pressure leaching to liberate copper
(60 g dm3) under conditions that minimize the free acid (pH 1.4) and iron
in solution, copper was extracted to 10 g dm3 in three stages with loading in
the organic phase of 12.5 g dm3.
The situation described is valid in sulfuric acid solutions, when no
strong metal complexes are formed. However with hydrochloric acid the
formation of metal chloride complexes makes the situation quite different.
The influence of the chloride ion concentration on the extraction of some
metal chloride complexes of general interest is shown in Fig. 11.2. For
example, the separation of zinc and iron(III) can be achieved at a low
chloride concentration, and later the separation of cobalt and nickel is
described. The disadvantage of using the chloride system is concerned with
stripping. For example, when a zinc chloride complex ZnCl42 is extracted
with a tertiary amine, the stripped species corresponds to ZnCl2, resulting in
an increased chloride concentration and an interruption of the mass trans-
fer. The consequence of this is that it is difficult to achieve concentrated zinc
chloride strip liquors. Two examples are given next.
14.4.1.1 Extraction of Zinc from Weak Acid Effluents
The first example describes the extraction of zinc from weak acid solutions.
In the manufacture of rayon, rinse waters and other zinc-containing liquid
effluents are produced. The total liquid effluent in a rayon plant may amount
to several m3 per minute with a zinc concentration of 0.1–1 g dm3 and pH
normally 1.5–2. In addition to zinc, the effluent contains surface-active
agents and dirt (organic fibers and inorganic sulfide solids). The use of both
precipitation (OH and S2) and ion exchange has been reported to remove
zinc from such effluents. In addition, solvent extraction has successfully
been used to recycle the zinc back to the operation.
In the process [3,7], zinc is extracted to greater than 95% (pH > 2)
with DEHPA (25%) in kerosene in two or three stages. Stripping is per-
formed with a sulfuric acid solution (Figs. 14.1 and 14.2). By adjusting the
net flow rate of the solution, the concentration of zinc in the strip solution
may be increased to 50 g dm3 or more. Thus, the resulting zinc sulfate
solution can be reused directly in the spinning bath.
The extraction of zinc is pH dependent. The lower the pH the lower
the extraction of zinc. In addition, the extraction of each zinc ion (Zn2+)
Fig. 14.1 Extraction of zinc from weak acid effluents.
liberates two hydrogen ions (H+). Thus, the more zinc in solution, the less is
extracted. This disadvantage means that using solvent extraction without
some neutralization (pH > 2) results in a residual zinc concentration in the
raffinate, which is too high to meet with environmental standards of today.
14.4.1.2 Extraction of Nickel from Cadmium Electrolytes
In the electrowinning of cadmium, nickel is an interfering element that has
to be continuously removed from the electrolyte, a weak acidic cadmium
sulfate solution. To handle the undesirable buildup of nickel contamination
and at the same time obey environmental demands, the industry installed a
solvent extraction ‘‘kidney’’ [8].
The nickel extraction can be performed with DEHPA under defined
conditions. As described earlier for zinc, the extraction of nickel is pH
dependent. The D-value for nickel decreases drastically with decreasing pH
below 3.5. The extraction of each nickel ion (Ni2+) liberates two hydrogen
ions (H+) from the extractant H+DEHPA, which means that only a very
small amount of nickel can be transferred to the solvent before the extrac-
tion stops. However, by adding a neutralization reagent (NaOH) into the
Fig. 14.2 Zinc extraction plant at Enka AKZO, Netherlands. Full-scale plants have been
built and operated at Svenska Rayon AB in Sweden and at Enka AKZO in The Netherlands.
mixer simultaneously with nickel extraction, this disadvantage can be avoid.

Three large mixers in parallel are used to allow for the necessary reaction
time (Fig. 14.3a and b).
The flow sheet is similar to the zinc block diagram with two extraction
stages, containing three parallel mixers followed by a settler, and two
stripping stages with normal mixer-settler arrangement. The pH is mon-
itored and the acidity controlled by the addition of NaOH in the two first
mixers. The extraction efficiency for nickel is better than 98%.
Another reported alternative to eliminate the extraction dependency
on H+ (pH) when extracting nickel with an acid reagent is to use the sodium
form of the reagent (Na+R) [9]. From a weak sulfuric acid rinse water
containing 1–2 g dm3 Ni, the metal could be effectively (>99%) removed in
two extraction stages. By loading the solvent with nickel, the transfer of
sodium to the strip solution was minimized. Nickel was stripped from the
loaded solvent with dilute sulfuric acid and recovery from the strip solution
was best performed by electrowinning. The main problem with this proce-
dure was high loss of extractant in the NaOH extraction reconditioning
wash. A considerable reduction of this loss could be achieved by increasing
the amount of sodium sulfate in the wash solution.
14.4.1.3 Extraction of Zinc and Chromium(III)

from Electroplating Baths
The use of Zn-Cr(III) alloy plating has almost replaced the use of Cr(VI) in
the electroplating industry due to its excellent corrosion resistance and its
lower toxicity. Recently, a solvent extraction procedure for separating and
selectively recovering the two metals, zinc and chromium, from electro-
plating wastewaters has been demonstrated [10].
Zinc is extracted and stripped with DEHPA in the conventional way
as previously described. The major challenge lies in the extraction and
stripping of Cr(III), which requires a higher pH than zinc. Chromium
Fig. 14.3 Nickel extraction plant with simultaneous neutralization.

Fig. 14.3 (Continued ).
hydroxide precipitation places an upper limit on the pH that can be used.

The extraction of Cr(III) with DEHPA is successful, but stripping of the
organic solvent is difficult. Now, it was found that using NH+ 4 DEHPA
(the ammoniated form), the extraction was similar to using the solvent in the
acid form and all stripping problems were overcome.
The suggested flow sheet includes extraction of Zn with H+DEHPA
(0.1 mol dm3) followed by the extraction of Cr(III) with NH+ 4 DEHPA
(0.1 mol dm3). The stripping of both metals was performed with dilute
sulfuric acid. By using the same reagent in both extraction circuits, con-
tamination of the second solvent is avoided. A process for the recovery of
Zn and Cr(III) with good separation from each other has been demon-
strated. The treated wastewater contained the metals at levels meeting
stringent environmental standards.
14.4.1.4 Extraction of Copper and Zinc
from Brass Mill Flue Dust
Another reported example with two metals in solution is the recovery of
copper and zinc from brass mill flue dust [11]. The material contains very little
iron. The solid material is leached with sulfuric acid to produce a weak acid
metal sulfate solution. The operation can be regarded as two separate inter-
connected processes, one for copper recovery and one for zinc (Fig. 14.4).
In the copper circuit, copper-rich material is leached in a pH-controlled
leach at 608C to a final pH of about 2. Oxidizing conditions are maintained,
e.g., by addition of MnO2, to promote copper dissolution and to prevent
cementation of metallic copper when the feed material contains metallic zinc
(brass or iron). The leach solution, containing about 20 g dm3 zinc and
4 g dm3 copper, is filtered and fed to a solvent extraction circuit. Copper is
extracted in four stages with a kerosene solution containing H+R (10%).
Stripping is performed in three stages with a sulfuric acid electrolyte.
Copper metal is recovered from the electrolyte by electrowinning.
The zinc circuit consists of a similar pH-controlled leach at 608C under
oxidizing conditions. Zinc flue dust with low copper content is leached with
the copper barren raffinate and with part of the zinc raffinate. The zinc
leaching operation is maintained at about pH 2 for most of the leaching time
and then slowly raised to a final pH of 4.5, reducing the iron level to below
10 ppm. The leach solution is filtered and cleaned from impurity metals such
as Cu, Ni, and Cd by an ordinary cementation procedure, again filtered and
finally fed to a solvent extraction circuit. Zinc is extracted in three stages with
Fig. 14.4 Extraction of copper and zinc from weak sulfuric acid leach solutions.
a kerosene solution containing DEHPA (50%). The amount of zinc trans-

ferred in the extraction procedure can be maintained as high as 20–25 g dm3,
because the remaining zinc in the raffinate is recycled to leaching. Stripping
of zinc is performed in two stages with sulfuric acid electrolyte, and zinc
metal is recovered from the electrolyte by electrowinning.
The process flow sheet was first tested for direct leaching of steel mill
flue dust and production of zinc metal by electrowinning. The tests were
performed in a continuously operating pilot plant, producing 10–20 kg/day
zinc metal. The same pilot plant was then used for treating copper/zinc-rich
brass mill flue dust in a closed loop operation, recycling all the zinc solvent
extraction raffinate to the copper circuit leach section. In the zinc circuit
leach section, only the amount of zinc rich dust necessary for neutralization
of the copper solvent extraction raffinate was used. The results obtained
from the pilot plant tests indicated contamination problems within the sol-
vent extraction loops. The estimation of economic data showed a weak
return on the assets compared with the alkali route, and sensitivity toward
the raw material price.
A similar flow sheet has been suggested for the treatment of mine
waters. However, the similarity is only apparent. The iron content in these
waters (Fe 3–5 g dm3, Zn *1 g dm3, and Cu 0.5–1 g dm3, pH 2–2.5) is the
reason for the difficulty of adapting the above flow sheet. The extraction of
copper can be performed in the presence of iron, but the extraction of zinc
requires removal of the iron, and the pre-preparation of the solution to
remove iron is costly and complicated, mainly because iron occurs in both its
valences.
14.4.1.5 Extraction of Zinc from Spent Hydrochloric

Acid Pickling Liquors
To avoid coextraction of iron, the iron is initially reduced to its II-valent
state. Then zinc is extracted as a zinc chloride complex into an organic
solution containing tributylphosphate (TBP). Zinc is stripped from the
organic solution with water or dilute hydrochloric acid (Fig. 14.5). The
resulting strip solution is evaporated, either (1) after addition of sulfuric
acid, giving a dilute hydrochloric acid condensate and a zinc sulfate pre-
cipitate, or (2) directly without any addition, giving a dilute hydrochloric
acid condensate and a concentrate zinc chloride product solution.
Three product streams are possible from the process:
1. Iron (II) chloride solution, for possible treatment in a pyrolysis plant or,
more probably, for the production of flocculation chemicals used in
sewage water treatment
2. Dilute hydrochloric acid condensate, mainly used as strip solution
Fig. 14.5 Extraction of zinc from spent hydrochloric acid pickling liquors.
3. Zinc sulfate (solid precipitate) or a concentrated (about 250 g dm3) zinc

chloride solution
A pilot plant, containing two pulsed columns, one for extraction and
one for stripping, and batchwise evaporation was in operation in Sweden
during 1981. Pilot plant operations have also been performed in Holland
(Fig. 14.6) and Germany. The experience from these tests shows that the
process concept is technically practicable and well proven. The economics of
the process, however, are strongly dependent on the cost for disposal of
spent pickling liquors.
14.4.1.6 Extraction of Vanadium and Nickel
from Soot and Fly Ash
In ash and soot residues from Orimulsion-fired power stations, the dom-
inating metal constituents are vanadium, nickel, and iron, together with
Fig. 14.6 Zinc extraction pilot plant in The Netherlands.
high amounts of magnesium. A process [3,12] for the recovery of vanadium

and nickel from these residues starts with a primary water wash to eliminate
the large amount of magnesium (Fig. 14.7). The reason for this prewash is to
recover and recycle magnesium back to combustion and to decrease the
consumption of acid in the following sulfuric acid leaching step. The solu-
bility of vanadium is very low and the dissolution of nickel can be neglected.
The residue after the water wash is leached at an elevated temperature
by sulfuric acid, part of which is the recycled raffinate from the vanadium
extraction. The leaching yield of vanadium (mainly IV-valent) is about 55%
and of nickel about 95%. A final (post) leach with sodium hydroxide dis-
solves the remaining vanadium (mainly V-valent). The resulting leach solu-
tion, containing practically all the vanadium (*25 g dm3) and nickel
(*12 g dm3) is fed to the solvent extraction circuit.
It may be mentioned that starting with ash and soot from crude oil-
fired stations, the resulting metals are the same, but the main leaching
residue is carbon. This residue is initially burned and the ash is leached again
to increase the total yield of vanadium. In the same operation, the con-
centration of iron is reduced by precipitation of jarosite. During leaching,
the redox potential is controlled by SO2 addition to keep vanadium in its IV-
valent state.
The extraction of vanadium is performed with a mixture of DEHPA
and TBP in kerosene (Fig. 14.8). Half of the raffinate, with a somewhat
increased acidity, is returned to leaching. The other half is further treated to
extract nickel and to control the buildup of impurities. Owing to the high
vanadium concentration in the feed, a large extraction capacity can be ob-
tained and, consequently, the process flows will be less for a given vanadium
throughput. The economical importance of this effect is decisive.
Vanadium is stripped from the organic solution with 1.5 mol dm3
sulfuric acid to a concentration of about 50 g dm3 V. After an iron removal
step with strong sulfuric acid, the organic solution is washed with water and
recycled. From the vanadium strip liquor, ammonium polyvanadate (APV)
is precipitated by oxidation and addition of ammonia. The APV slurry is
thickened and pumped to a vacuum belt filter, where the APV cake is
carefully washed with fresh water. The APV filter cake is dried and then
calcined to vanadium pentoxide.
To the second half of the raffinate, ammonia is added to form a nickel
ammonium complex, which is extracted with a hydroxyaryloxime (LIX84).
Due to simultaneous formation of metal hydroxides, the solution is carefully
Fig. 14.7 Extraction of vanadium and nickel from soot and flue ash.
Fig. 14.8 Recovery of vanadium and nickel from soot, Stenungsund, Sweden.
filtered. After extraction of nickel, the resulting raffinate is further treated in

a bleed treatment section. The organic solvent is scrubbed with weak sul-
furic acid and then nickel is stripped by 1.5 mol dm3 sulfuric acid. Flow
rates of solutions and solvents are adjusted to give a very concentrated
nickel sulfate strip liquor.
14.4.1.7 Recovery Cobalt and Nickel from Scrap Alloy
In a pilot plant [2,13], superalloy scrap containing Mo, W, Cr, Fe, Co, and Ni
is pretreated in a furnace with carbon to transfer refractory metals (Mo, W,
etc.) into carbides. The melt is granulated and the resulting material is
charged into titanium baskets. Diaphragm-type electrolytic cells are used for
anodic dissolution of the granulated material. Fe, Co, Ni, and small amounts
of Cr are dissolved into a calcium chloride solution by the current. The metal
carbides are not dissolved and remain as an anodic residue in the baskets.
The anolyte, containing about 75 g dm3 Cl at pH 2, is fed to a
solvent extraction circuit for the separation of Fe-Co and Fe-Ni with a
tertiary alkyl amine, dissolved in kerosene. The separation is based on the
tendency of the metals to form metal–chloride–amine complexes (Fig. 11.2).
At low chloride ion concentration (about 75 g dm3), Fe(III) is extracted to
the organic solvent, while Co(II) and Ni(II) remain in the aqueous raffinate.
If the chloride ion concentration is then increased to about 250 g dm3,
cobalt is extracted, leaving nickel behind in the raffinate.
Using an organic solution containing 25% Alamine 336 and 15%
dodecanol in kerosene, it is possible to separate the metals one after the
other by oxidizing iron to Fe(III) and altering the chloride ion concentration
by evaporation. This was also the original object of the process. However, it
was found that the complete oxidation of iron was complicated and, as the
iron in ferro-cobalt and ferro-nickel did not ruin the market value of the
products, the process finally used is shown in Fig. 14.9.
The anolyte is evaporated to increase the chloride concentration from
75–250 g dm3 and cobalt, and most of the iron, is then extracted by the
organic solvent. The nickel containing raffinate is diluted by condensate
from the evaporation and returned to the Ni-cathode compartments in the
diaphragm electrolytic cells for precipitation of nickel. Stripping of cobalt
from the organic solvent is performed with the weakly acidic condensate.
The strip liquor is fed to the Co-cathode compartment in the electrolytic
cells.
The process described has been running in pilot plant scale operation
for more than a year and the advantages of the process are summarized here:
Using diaphragm-type electrolytic cells, electrical energy is used in an
optimal manner for dissolution and deposition.
Fig. 14.9 Recovery of cobalt and nickel from scrap alloy.
The amine-chloride system is well suited for the separation of the metals.
Using an evaporator for raffinate concentration and production of strip
solutions, a closed wet cycle is obtained and no aqueous waste pollutes
the environment.
The low consumption of chemicals and energy indicate an economic
advantage over similar open cycle processes.
A limited amount of available waste and a price increase of the raw material
resulted in the discontinuation of the project. This situation is not unique.
Many projects suffer from the same problem, i.e., a waste material becomes
a valuable raw material when it is in demand.
14.4.1.8 Extraction of Chromium(VI) from Surface
Finishing Wastewater
Hexavalent chromate [Cr(VI)] is still used within the industry to meet crit-
ical high corrosion control and other metal surface finishing requirements.
Cr(VI) is toxic and its control generates a hazardous, costly waste.
Recently, a solvent extraction process [14] for the recovery of hexa-
valent chromium from surface finishing process water, the A-LLX system,
has been successfully demonstrated. The concentration level of Cr(VI) in the

raffinate is low enough (
0.1 mg dm3) for discharge to surface waters.
Chromium recycling enables continued use of high performance chromium.
A tertiary amine (Alamine 336) is used as a selective extractant of
Cr(VI). The extraction mechanism involves ion pairing between the amine
and the dichromate anion, HCrO 4 . After extraction, the Cr(VI)-loaded
extractant is stripped with alkali to produce a strip liquor, containing 2%
Cr(VI). This solution can be recycled.
14.4.1.9 Recovery of Zinc from Spent Batteries
Spent domestic batteries cause environmental repercussions for a number of
reasons: the presence of soluble heavy metals; the increase in their use; the
short life cycle, and the low efficiency of their recovery and recycling pro-
cedures. To improve the situation, a solvent extraction process plant has
been built and operated in Spain [15]. It is based on the modified Zincex
process (MZP) and has a capacity of processing 2800t/year of batteries in a
continuous operation.
The process starts with pre-preparation of the feed material. The
batteries are classified and pulverized and paper, plastics, and iron com-
ponents are removed by physical procedures, leaving an impure zinc-
manganese dust, which is treated by leaching in sulfuric acid to give a
residue of mainly MnO2 and graphite. Before the pregnant leach liquor,
with a zinc content of 22 g dm3, is purified by solvent extraction, metallic
impurities like Hg, Ni, Co, Cd, etc., are removed by cementation.
Using a solvent extraction procedure, containing one depletion mixer-
settler stage, three extraction, three washing, and two stripping stages, zinc
is transferred to an acid sulfate strip liquor, containing about 140 g dm3
zinc and suitable for electrowinning of high-quality zinc or evaporation to
very pure zinc sulfate. The extractant reagent used is DEHPA, with a
reported good selectivity for zinc over impurities in the feed liquor. In
addition, the washing stages eliminate most of the entrained and coextracted
Mn2+ and Cl. Only coextracted Fe3+ was left in the organic solvent after
washing. Its concentration in the feed solution is <5 ppm and this small
quantity was eliminated by an HCl regeneration stage.
The economic evaluation of the plant indicates that for a feasible
operation with the limited plant capacity, compensations for the treatment
through direct consumers or through official organizations are required.
14.4.2 Extraction of Acids

Undissociated acids form adduct complexes with organic compounds,
containing oxygen donor atoms. If the adducts formed are soluble in
organic diluents like kerosene, the acids can be extracted from an aqueous
solution by solvent extraction. This procedure has been described in more
detail (See Chapter 10).
14.4.2.1 Recovery of Hydrofluoric and
Nitric Acids from Stainless Steel
Pickling Baths
A stainless steel pickling bath initially contains 2.2 mol dm3 HNO3 and
1.6 mol dm3 HF. The bath is used until the iron concentration reaches
40–80 g dm3. At this stage, the bath contains about 50% unused acids in
addition to the dissolved metals: iron, nickel, chromium, and molybdenum.
In a used pickling bath, most of the acids are bound to the metals as
complexes. By adding sulfuric acid to the pickling bath, sulfate will replace
some of the nitrate and fluoride, leading to the formation of extractable,
undissociated nitric and hydrofluoric acids.
The general procedure to remove monovalent acids from the aqueous
solution by adding sulfuric acid and then extracting the released acids with
an organic donor molecule, dissolved in an organic diluent, has been called
acid exchange or the AX process [15,16]. The procedure was first presented
for the treatment of spent stainless steel pickling liquids. The process con-
sists of three main steps (Fig. 14.10):
1. Addition of H2SO4 and extraction of HNO3 and HF with an organic
solution, containing 75% TBP in kerosene, in a pulsed column operation
2. Precipitation of the metals (Cr, Fe, and Ni) left in the aqueous raffinate
3. Stripping HNO3 and HF from the organic solution with water in a
second pulsed column operation and recycling the acids back to the
pickling bath
A plant for treating 600 dm3 h1 spent pickling liquid (corresponding
to an annual production of 25,000 t stainless steel) was in operation for some
years in Sweden (Fig. 14.11). The recovery operation was successful; how-
ever, the steel works are now closed.
Similar extraction procedures [17] using a mixture of trialkylphos-
phine oxides (CYANEX 923) for extraction of the mineral acids from elec-
troplating and stainless steel pickling baths are reported as the ARSEP
procedure.
The possibility of using nitric acid as an oxidizing agent in a leaching
operation is often prohibitive because many reagents used in solvent
extraction are not stable in an oxidizing solution. By the AX method,
however, a nitric acid (nitrate) leach solution can be converted to a weak
sulfuric acid (sulfate) solution, well suited for a solvent extraction treat-
ment.
14.4.2.2 Purification of Green Phosphoric Acid (H3PO4)

By reacting apatite with sulfuric acid, phosphoric acid and gypsum will
result. Most phosphoric acid is produced in this way and is normally used as
raw material (green acid) to produce fertilizers. A small amount, however, is
further refined to phosphoric acid of food grade quality.
The green phosphoric acid from a phosphoric acid plant (PAP) con-
tains 25–30% H3PO4. The acid is heavily entrained with impurity cations,
among which arsenic, cadmium, and uranium are the most toxic. In addition,
anions, like chloride, fluoride, and sulfate must be considered. Selective ex-
traction of phosphoric acid [18] as an adduct complex with TBP according to
the block diagram in Fig. 14.12 has been used to produce acid of food grade.
The purification operation starts with addition of chemicals to pre-
cipitate arsenic and sulfate in a pretreatment step. The acid is then cooled
and left for an appropriate aging time. A flocculent is added and the solid
content is allowed to flock (precipitate) in a clarifier.
The clear phosphoric acid is fed to the solvent extraction operation,
where about 60% of the acid is extracted countercurrently with an organic
solvent containing TBP in kerosene. The resulting raffinate containing most
Fig. 14.10 Recovery of acids from stainless steel pickling baths.

Fig. 14.11 Acid recovery by pulsed plate column operation at Söderfors Steel Works,
Sweden.
of the contaminating substances, together with the slurry from the clarifier,
is returned to the phosphoric acid plant. In this way, the impurities in the
green acid will be returned to fertilizer production without creating addi-
tional environmental problems.
After extraction, the organic solvent, loaded with phosphoric acid, is

scrubbed with a phosphoric acid solution in equilibrium with the organic
solvent. In this way, coextracted contamination is removed. The scrub
liquor is recycled to extraction.
After adjusting the temperature to about 408C, the organic solvent is
treated with clean water to strip the phosphoric acid. This acid, containing
about 30% H3PO4, is heated and water evaporated to produce concentrated
(85%) phosphoric acid of food grade quality.
Fig. 14.12 Purification of phosphoric acid to food grade quality.

In addition, a process for the purification and production of citric acid

has been reported, utilizing a similar flow sheet.
14.4.2.3 Removal of Arsenic Acid (H3AsO4)
from Copper Electrolytes
The use of low-grade ores with large amounts of harmful impurities results
in increased accumulation of these impurities in the electrolyte. One of the
most harmful is arsenic, which seriously affects the quality of the copper
cathodes. Arsenic occurs in the sulfuric acid electrolyte as arsenic acid, and
as such can be extracted as an adduct complex with TBP. The extraction is
favored by increasing acidity and low temperature.
Arsenic control by solvent extraction using TBP has been demon-
strated [19] in copper electrorefining plant practice and recently an arsenic
removal process with undiluted TBP has been suggested. Centrifugal ex-
tractors were used (Fig. 14.13), as the separation conditions (emulsions)
in extraction and stripping were very slow. After filtration of the barren
electrolyte, containing about 8 g dm3 As, the sulfuric acid concentration
was raised to 180 g dm3. The electrolyte was fed to extraction, where about
50% of the arsenic was transferred in two stages to the organic solvent.
After a scrubbing stage with dilute sulfuric acid to minimize copper coex-
traction, arsenic was stripped with water at elevated temperature.
The process was demonstrated on a semi-industrial scale-removal of
1 kg h1 As. The operation of the extractors was reliable. The phase
entrainment in each stage was less than 0.05% at an organic/aqueous phase
ratio of 1.
TBP only extracts As(V) and therefore other extractants active for
As(III), Sb, and Bi have been tested. Some years ago, a two-extractant
procedure [20] was suggested and recently, the use of a mixture of alkyl-
phosphine oxides (R3PO) has been reported [21,22].
14.4.2.4 Cyanide Recovery by Solvent Extraction
A solvent extraction process for the recovery of cyanide from acidified
solutions using organophosphorus solvating extractants has recently been
suggested (Fig. 14.14) [23]. The most favorable extractant for effective HCN
loading is a mixture of trialkylphosphine oxides (R3PO). The DHCN is well
over 10, which indicates that a four-stage extraction would be sufficient to
remove cyanide to less than 10 mg dm3 from a starting concentration of
200 mg dm3 at an aqueous/organic phase ratio of 5. The extraction kinetics
are reported to be fast.
The stripping of the organic loaded solvent is accomplished by neu-
tralization of the organic solvent with an alkali solution (e.g., NaOH) to
reconstitute a strong cyanide solution for recycling.
Fig. 14.13 Centrifugal extractor arrangement with five CENTREK EC-320 units.
14.4.3 Solvent Extraction from Ammoniacal

Feed Solutions
Some years ago a hydrometallurgical route, called the AmMAR concept,
was presented [24]. The concept (Fig. 14.15) represented a general proces-
sing route for a wide variety of feed materials. The detailed design of each
specific process involved a specific number of known chemical operations,
combined in unique ways. The main thread was the ammonia–ammonium
chloride or carbonate leach solution and its extraordinary chemical flex-
ibility.
The general outline of an ammoniacal process for treatment of a
material containing two or more of the metals: iron, chromium, copper,
nickel and zinc, includes the following process steps:
Fig. 14.14 Recovery of cyanide as NaCN.
1. Primary separation occurs in the leaching procedure, where the metals

copper, zinc, and nickel are dissolved as metal ammonium chlorides or
chloro complexes, while iron and chromium remain in the solid residue
as hydroxides.
2. Copper and nickel are subsequently removed from the leach liquor by
solvent extraction. After selective stripping, nickel sulfate is produced
by crystallization and copper metal by electrowinning.
3. Excess ammonia is evaporated from the resulting extraction raffinate
and zinc is precipitated as carbonate by the remaining ammonium
carbonate in the solution or by addition of carbon dioxide. The evap-
orated ammonia is absorbed in the filtrate and recycled to the leaching
step.
Fig. 14.15 General diagram of the AmMAR concept.
4. To maintain the water balance, especially when a water-containing

sludge is treated, an additional effluent treatment with lime, followed by
ammonia evaporation, is necessary.
14.4.3.1 Extraction of Copper and Recycle of

Ammoniacal Etch Liquors
In the etching procedure, copper concentration is continuously increased by
the etching and, to keep the etching efficiency constant and optimal, spent
etchant has to be withdrawn and replaced with fresh etchant (replenisher).
An alternative would be to remove a part of the copper content from the
spent etchant, without changing the other conditions in the solution, in such
a way that the barren etching solution can be recycled. This process is called
the MECER process [2,25].
In the process (Fig. 14.16) spent etchant is withdrawn and fed through
a mixer-settler unit, where about 30% of the copper is transferred to an
organic solution containing the reagent H+R in kerosene. The raffinate
(barren etchant) is recycled back to the etching line. The copper containing
rinse water, from the rinsing of the boards after etching, is also treated in a
second extraction mixer-settler to reduce the copper content (<5 ppm).
Finally, copper is stripped from the organic solution to a sulfuric acid
copper electrolyte, and pure copper metal is produced by electrowinning.
Over 100 MECER installations are in operation at PCB manu-
facturers all over the world, recovering almost all the copper and reducing
their consumption of chemicals to more than 95% (Fig. 14.17a and b).
14.4.3.2 Recovery of Nickel and Cadmium
from Accumulator Waste
Nickel and cadmium are used in the production of NiFe rechargeable
batteries. Using the AmMAR concept, the main leaching procedure to
dissolve these valuable metals from spent accumulator scrap and production
waste is performed in a two-step procedure, first with an ammonium car-
bonate solution and second with diluted sulfuric acid to obtain very high
leaching efficiency (Fig. 14.18).
Due to chemical passivation in ‘‘old’’ waste materials, subsequent
leaching of the leach residue with sulfuric acid, with perhaps some oxida-
tion, is necessary to achieve very high yields. If the pH is kept about 3 in this
second treatment, leaching of iron is negligible. Also, to reduce iron in the
leach solution, spent potassium hydroxide electrolyte from discarded bat-
teries can be used for iron hydroxide precipitation.
The procedure has several advantages: iron will remain in the leach
residue; nickel and cadmium are almost completely leached and form metal-
amine complexes in solution; and, finally, the buffer capacity of the solution
Solvent Extraction in the Recovery of Waste
Fig. 14.16 Recovery of copper and recycle of ammoniacal etching liquors.

637
facilitates solvent extraction of metals with hydrogen ion exchange reagents.

Thus, nickel is extracted from the leach solution using a hydroxyaryloxime
(LIX 84) with significant amounts of coextracted ammonia and cadmium.
Scrubbing is therefore performed, first with an ammoniacal carbonate
solution and then with very dilute sulfuric acid. Nickel is then stripped
with strong sulfuric acid to produce a nickel sulfate solution, containing
90–100 g dm3 nickel. This solution is used directly in the production of new
accumulators.
The presence of carbonate in the ammoniacal system offers a possi-
bility of controlling the conditions for the formation of cadmium carbonate.
After separation of nickel in the solvent extraction loop, cadmium is pre-
cipitated from the raffinate as carbonate. Thermal stripping reduces the
ammonia concentration and some carbon dioxide will also be evaporated.
Precipitation is completed by cooling the solution and by addition of
Fig. 14.17 Copper recovery units (MECER) integrated in the production of printed
circuit boards. This photograph shows the solvent extraction unit.
Fig. 14.17 (Continued ). This photograph shows the electrowinning unit.
a carbon dioxide purge. After filtration of the cadmium carbonate, the fil-
trate is used for absorption of the stripped ammonia and the resulting
solution is recycled to leaching. The cadmium carbonate precipitate is used
directly in the main production.
It is necessary to bleed about 25% of the filtrate from the carbonate
precipitation to maintain material balances. This large bleed is due to the
sulfate added in the sulfuric acid leach and because the waste material
contains considerable amounts of alkali (KOH) and water. The bleed needs
special treatment to recover ammonia and to produce an environmentally
acceptable effluent.
The process has been tested at pilot plant scale with good technical
results. At the time, however, the amount of waste material available was
not sufficient to justify the feasibility of a full-scale plant.
Alternatively, an acid route has also been reported [26], where the
leaching is accomplished using a mixture of HNO3 and HF, and separation
and recovery of the nickel and cadmium is performed with solvent extraction.
First, in the cadmium circuit the leaching solution is adjusted to pH 3 prior to
640
Cox and Reinhardt
Fig. 14.18 Recovery of nickel and cadmium from accumulator scrap.

extraction with an ammonia preequilibrated NH+ 4 DEHPA. Following

scrubbing with cadmium nitrate solution to remove coextracted Co, Ni, and
NH3 impurities, the scrubbed solvent is contacted with nitric acid to strip the
cadmium and produce cadmium nitrate by crystallization. To extract nickel,
the cadmium extraction raffinate is adjusted to about pH 5.5 and again
contacted with NH4+DEHPA. The loaded organic phase is scrubbed with
nickel nitrate and stripping is performed with nitric acid solution at pH 1.0–
1.7 from which nickel nitrate is recovered by crystallization.
14.4.3.3 Extraction of Ni, V, and Mo
from Spent Desulfurizing Catalyst
In this process [27], the metals are extracted successively by the same
extractant, employing suitable pH values for each metal. The spent catalyst
is first roasted in an air stream at 3008C to remove oil and sulfur deposited
on the surface of the catalyst. Then the metals are dissolved in an ammo-
nium carbonate leach liquor at 808C. Finally, the three metals are isolated
and purified from one another using solvent extraction with an alkylmono-
thiophosphoric acid (MTPA) with a structure similar to DEHPA. The
following flow sheet was suggested (Fig. 14.19):
14.4.4 Pyro-Hydrometallurgical Concept

In a centrally located recovery plant for the treatment of oxidic and
hydroxidic metal containing waste (dust and sludge) from steel and metal
works, including the plating industry, the main problem is the dominating
amounts of iron. To avoid a secondary waste problem, special precautions
have to be considered to create an iron product. One possibility is a pyro-
metallurgical treatment to produce pig iron or a master alloy.
Some years ago, the process, shown in Fig. 14.20, was suggested as a
centrally located plant for treatment of metal-containing waste in Sweden.
The operation is based on the use of a pyrometallurgical induction converter
and hydrometallurgical AmMAR process technology.
In the pyrometallurgical treatment, dried and pulverized waste (flue
dust and sludge) together with fluxes and carbon powder for the reduction,
suspended in a small quantity of gas, are blown into the iron melt through
an injection tuyere, situated below the metal surface in the converter. Due to
the fine dispersion of the feed material and the intensive turbulence in the
bath, a rapid reaction occurs between the injected metal oxides and the
reducing carbon in the melt. Thus, the ferruginous powders are converted to
pig iron or master alloys. In addition, the thermal energy in the flue gases
from the furnace is used in the hydrometallurgical operation and for the
drying of the input material.
642
Fig. 14.19 Recovery of nickel, vanadium, and molybdenum from spent desulfurizing catalyst.
Cox and Reinhardt
The first step in the hydrometallurgical treatment is leaching the feed

material with an ammoniacal solution. Metals such as copper, nickel, and
zinc form metal ammonium complexes and will therefore dissolve in the
leach solution. Other metals like iron and chromium form insoluble hydrox-
ides and will be found in the residue. This residue is dried and treated in the
induction converter. Copper and nickel are separated from the filtered leach
Fig. 14.20 Pyro-hydrometallurgical treatment.

solution by solvent extraction, either by sequential extraction of the metals,

or by coextraction followed by selective stripping. Zinc coextraction can be
neglected. Products are copper cathodes and nickel sulfate.
The secondary flue dust from the converter, containing mainly zinc
and lead, is leached with the ammoniacal raffinate solution from the Cu/Ni
extraction. Lead is left in the residue. Finally, zinc carbonate is precipitated
by addition of CO2 followed by thermal stripping of ammonia. The zinc
carbonate is calcined to zinc oxide and ammonia is recycled to leaching.
The incentive to build a pyro-hydrometallurgical plant is often based
on political opinions and somewhat diffuse environmental public demands
and therefore the decision process is complicated. The direct investment of
such a plant is high; the return on assets is relatively low, because of the
unstable value and uncertain situation of feed materials (raw material or
waste) and the fluctuating prices of end products.
14.5 GENERAL COMMENTS

In most countries, solid waste containing metals such as neutralization
sludge from the plating industry and flue dust from the metal and steel
industries is currently collected and dumped in landfill, where it constitutes a
perpetual toxic threat to the environment and a waste of resources. The
alternatives to this landfill disposal are either to reduce the rate of discharge
at source by an individually designed recovery process or to separate and
recover the metals from the collected waste in a centrally located facility. A
presumption for a centrally located facility would be that companies with
metals in their effluents require treatment of their total wastewater streams.
This could be accomplished through the relatively simple process of neu-
tralization, which requires minor investment in sedimentation tanks and
dewatering equipment and involves relatively modest operation costs.
Although the problem of disposal of large amounts of metal waste is
faced by most industrialized countries, relatively few centrally located
operations for waste recovery have, to date, been started. Sweden still
deposits its dewatered metal waste in a simple landfill, although Swedish
industry has been in the forefront of developing both hydrometallurgical
and pyrometallurgical recovery techniques. The same applies to most
European countries; however, interest in environmentally safe recovery has
increased in recent years and recovery plants are now being considered.
The initiative to develop at-source processes very often comes from an
industry and emanates from internal environmental problems. The devel-
opment of the process may be sponsored by the industry, government, or by
private R&D organizations that may be attracted by the general market
potential. The incentive of installing such a process in the industry is, besides
the environmental advantages, the possibility of recycling expensive chem-

icals and of optimizing the main production.
The incentive to build a centrally located facility is often the result of
an environmental problem that is tending to disturb the public at large. The
initial money comes from governmental funds, although industry may
actively participate. As industry usually has limited knowledge of such gen-
eral recovery projects, R&D institutions or organizations predominantly
perform the development work with generally close cooperation with the
industry concerned.
The development conditions that are feasible for at-source processes
are somewhat different from those of the centrally located facility regarding
the freedom in feed materials and end products. This is especially char-
acteristic for at-source processes, where the feed solution is dependent on
optimal working conditions in the main process and therefore the compo-
sition is relatively constant and well defined. In addition, the product
solution must have an accurate specification in order not to disturb the main
production. So in the development work, a good knowledge of the main
operation is necessary.
The strong link between at-source processes and the main operation is
the cause of a large degree of fixed conditions. Absence of easily adjustable
valves and an on/off mode of operation are recommended. The fact that the
industry in question often has neither competent knowledge nor especially
experienced personnel means that services and troubleshooting have to be
arranged by an external organization. Certain independence in utility time of
the auxiliary and main processes is arranged by built-in buffer capacities. On
the contrary, the centrally located facility has to be rather flexible with respect
to the physical conditions and to the composition of the feed material. This is
very important, as this may be the decisive incentive for the investment. In
addition, treatment of bleed streams is necessary and is often extensive.
It was earlier generally believed that the at-source approach with an
individually designed recovery process entailed a considerable risk of
investing in small, nonviable industries. Thus, it would appear that a cen-
trally located facility was likely to be the only feasible way to solve the
disposal needs of a large number of point sources. Such recovery facilities,
based on hydrometallurgical treatment concepts, were already suggested 30
years ago. However, at that time, the feasiblity of these processes was
doubtful. Pyrometallurgical treatment was marketed as an environmentally
safe method for removing a number of different metals, such as copper, lead,
and zinc from flue dust, wastewater sludges, or mining slimes and tailings. It
has, however, been shown that such processes would probably not be eco-
nomically justified in the case of the relatively small volumes of waste that
were likely to be at hand.
In general, for the centrally located facility, direct investment is high

and the decision time for the investment usually drastically increases with
the size of operation. Big investment, a relatively low return, the unstable
value and uncertain situation of feed materials (raw material or waste), and,
finally, fluctuating prices of end products characterize a metal waste
recovery process. These facts indicate the necessity of financial support from
government to secure the realization of such a project.
The direct investment is relatively low for at-source processes, as they
can be built in very small units. In addition, the operation costs are also
comparably low because of the integration with the main operation. For
example, administration costs can be almost neglected. The same is true for
process supervision, especially at night, when it can be shared with the main
operation. Usually, at-source solvent extraction processes have very low
costs for chemicals and energy.
The economic outcome of the operation of a centrally located facility
is comparable with all other hydrometallurgical operations and is highly
dependent on price and quality of the metals produced. All products have to
conform to commercial specifications. Adequate quality and quantity of
cheap feed materials are also essential. It is very important to consider the
fact that the transformation of waste and raw material usually involves a
dramatic increase in the value of the feed material. The economic result
cannot be isolated to an auxiliary process, but is the result of the combined
effects on the total operation.
Considering the general conditions for recovery plants, a conclusion
may be reached-it would be desirable to store all metal waste under con-
trolled conditions and with no mixing or dilution. Future development will
definitely result in new and more economic recovery procedures. As metal-
containing wastes differ widely in nature and complexity, selective separa-
tion techniques such as solvent extraction will be of increasing importance.
14.6 APPENDIX: ENVIRONMENTAL LEGISLATION

Currently, legislation covering the discharge of metals into the environment
is monitored, controlled, and devised by individual countries or regions.
However, there is a move toward harmonizing of such legislation led by the
United States, the European Union, and Japan.
The United States has the most extensive environmental regulatory
programs that aim to minimize the release of metals into the environment.
These regulations seek to control the emission of pollutants from specific
industries and to define the emission limits according to the type of industry,
the age and size of the plant, and the nature of the processes involved. In
addition, different limits are set according to the following criteria: best
practicable technology, best available technology, best conventional tech-

nology, new source performance standards, and pretreatment standards for
existing and new sources. Details of the regulations and criteria, which are
far too complex to be considered here, are given on the Internet at www.epa.
gov/OST/guide.
The European Union is also a major contributor to environmental
policy, with over 450 legislative measures. One of the early key directives
(76/464/EEC) divides dangerous substances into two categories: List I or
Black List and List II or Grey List. This was followed by directive 80/68/
EEC controlling the discharge of these materials into groundwater and
setting up authorization procedures. The drinking water directive (75/440/
EEC) is concerned with the quality of surface water used for drinking and
was followed by the bathing water directive (76/160/EEC). The directive on
integrated pollution prevention and control (IPPC) (96/61/EEC) aims at
preventing or minimizing environmental pollution. This directive not only
covers emissions from a process but also considers energy efficiency, noise
emissions, consumption of raw materials, and restoration of a site following
closure. These factors are integrated into a pollution control strategy (best
practical environmental option, BPEO) that covers the total impact of the
process on the environment and includes the principles of sustainable
development. The strategy does not emphasize the cost implications, and the
emission limits would be based on best available techniques (BAT). Infor-
mation concerning the various directives can be obtained from the Eur-Lex
database, which can be accessed at the website //europa.eu.int/eur-lex/en/
index.html and the various limits for pollutants from the Water Handbook
at //europa.eu.int/comm./environment/enlarg/handbook/handbook.htm.
Environmental legislation in Japan followed a number of serious air
and water pollution incidents, of which the Minimata incident of mercury
pollution was one. It resulted in an environmental control law in 1967. This
obliged the government and industries to control pollution, allocate the
costs for pollution prevention, provide for relief of victims, and set proce-
dures to solve pollution disputes. It also recognized the importance of
economic development that seriously limited its effectiveness. This act was
amended in 1970 and this clause concerning the link between the environ-
ment and economic development was removed. A number of laws followed
controlling specific discharges to air and water, as well as waste management
and soil pollution. In 1971, the Environmental Agency was created, re-
sulting in a comprehensive system of pollution control in Japan. This agency
is responsible for formulating basic legislation and coordinating the actions
of related government departments in implementing environmental laws,
such as the Water Pollution Control Law. It also has jurisdiction
over industrial pollution control. Details of the standards imposed by the
Japanese environmental regulations can be found at www.env.go.jp/en/lar/

blaw/index.html for details of the basic Japanese environmental law and
www.env.go.jp/en/lar/regulations/nes.html for the national effluent stan-
dards.
REFERENCES
1. Andersson, S. O. S.; Reinhardt, H. Handbook of Solvent Extraction; Lo, T.C.
Baird, M. H. I. Hanson, C. Eds.; John Wiley and Sons, 1982; 751 pp.
2. Reinhardt, H. Chem. Ind., 1975, 210.
3. Reinhardt, H. Proc. Int. Waste Treatment and Utilization Conf., Waterloo,
Canada, 1978; 83 pp.
4. Reinhardt, H. Proc. Hydrometallurgy ’81, Soc. Chem. Ind.: London, Section
F1, 1981.
5. Martin, D.; Diaz, G.; Garcia, M. A.; Sánchez, F. Proc. ISEC 2002; Cape
Town, South Africa, 2002, 1045.
6. Sole, K. C. Proc. ISEC 2002; Cape Town, South Africa, 2002, 1033.
7. Reinhardt, H.; Ottertun, H.; Troeng, T. I. Chemical Engineering Symposium
Series 41, I. Chem. E., Section W1, 1975.
8. MEAB Report. Extraction of nickel with simultaneous neutralization, 1989.
www.meab-mx.se.
9. Flett, D. S.; Pearson, D. Chem. Ind., 1975, 639.
10. Sze, Y. K. P.; Xue, L. Sep. Sci. Tech., 2003, 38, 405.
11. Andersson, S. O. S.; Reinhardt, H. Proc. ISEC ’77; Canadian Inst. Min. Metal.:
Toronto, 1979; Vol. 1, 798.
12. Ottertun, H.; Strandell, E. Proc. ISEC ’77; Canadian Inst. Min. Metal.:
Toronto, 1979; Vol. 1, 501.
13. Aue, A.; Skjutare, L.; Björling, G.; Reinhardt, H.; Rydberg, J. Proc. ISEC ’71;
The Hague, Soc. Chem. Ind. London, 1971; Vol. 1, 447.
14. Monzyk, B.; Conkle, H. N.; Rose, J. K.; Chauhan, S. P. Proc. ISEC 2002; Cape
Town, South Africa, 2002; 755.
15. Martin, D.; Garcia, M. A.; Diaz, G.; Falgueras, J. Proc. ISEC ’99; Barcelona,
Spain, Soc. Chem. Ind. London, 2001; Vol. 1, 201.
16. Rydberg, J.; Reinhardt, H.; Lundén, B.; Haglund, P. Proc. Int. Symp. Hydro-
metallurgy; Chicago, 1973; 589.
17. Demster, J. H.; Björklund, P. Proc. 4th Annual Meeting Hydromet. Sec. Met-
allurgical Soc.; CIM; Toronto, 1974; 68.
18. Dias, G.; Martin, D.; Frias, C.; Pérez, O. Proc. ISEC ’99; Barcelona, Spain,
Soc. Chem. Ind. London, 2001; Vol. 2, 1449.
19. Kalujta, V. V.; Kuznetsov, G. I.; Kravchenko, A. N.; Lanin, V. P.; Miraevshy,
G. P.; Pushkov, A. A.; Travkin, V. F.; Shklyar, L. I. Proc. ISEC ’88; Moscow,
1988; Vol. 1, 252.
20. Dreisinger, D. B.; Leong, B. J. Y.; Balint, B. J.; Beyad, M. H. Proc. ISEC ’93;
York, UK: Soc. Chem. Ind. London, 1993; Vol. 3, 1271.
21. Wang, C.; Jiang, K.; Liu, D.; Wang, H. Proc. ISEC 2002; Cape Town, South
Africa, 2002; 1039.
22. Cox, M.; Flett, D. S.; Velea, T.; Vasiliu, C. Proc. ISEC 2002; Cape Town,
South Africa, 2002; 995.
23. Dreisinger, D.; Wassink, B.; Ship, K.; King, J.; Hames, M.; Hackl, R. Proc.
ISEC 2002; Cape Town, South Africa, 2002; 798.
24. Reinhardt, H. Proc. ISEC ’93; York, UK: Soc. Chem. Ind. London, 1993;
Vol. 3, 1625.
25. Reinhardt, H.; Ottertun, H. Eur Patent Nr. 0 005 415, 1982. www.sigmamecer.
com.
26. Ritcey, G. M. Proc. ISEC ’93; York, UK: Soc. Chem. Ind. London, 1993;
Vol. 1, 189.
27. Tsuboi, I.; Kunugita, E.; Komasawa, I. Proc. ISEC ’93; York, UK: Soc. Chem.
Ind. London, 1993; Vol. 3, 1319.
15
Recent Advances in Solvent Extraction
Processes
SUSANA PÉREZ de ORTIZ and DAVID STUCKEY Imperial College,
London, United Kingdom
15.1 INTRODUCTION
Recent developments in separation techniques are a response to the growing
demands imposed by extraction processes that cannot be carried out eco-
nomically using conventional technologies. Typical examples are the treat-
ment of wastewaters containing low concentrations of pollutants, such as
heavy metals, and the downstream separation of biological products. In the
case of wastewaters, environmental regulations impose discharge con-
centrations of the order of a few parts per million, or even per billion, for
certain pollutants. Therefore, treatment using conventional solvent extrac-
tion would require large volumes of solvent and several extraction stages to
achieve the target concentrations, even in systems with large metal dis-
tribution coefficients (see Chapter 4).
The suitability of using solvent extraction for a given separation is
determined by thermodynamic and kinetic considerations. The main ther-
modynamic parameter is the solute distribution ratio, DM, between the
organic and the aqueous phase. This is given by [Eq. (4.3), Chapter 4]:
DM ¼ ½MT;org =½MT;aq (15.1)
where [M]T is the sum of the concentrations of all M-species in a given
phase, and the second subscript indicates the organic and the aqueous
phase. The magnitude of DM determines the feasibility of the separation as
an industrial process; the higher DM the better the solute separation.
Another consideration affecting the design of extraction processes is
the extraction rate as it determines the residence time of the phases in the
651
652 Pérez de Ortiz and Stuckey
contactor, and consequently its size. The extraction rate in a two-phase

system depends on the rate of interfacial transfer of species M, i.e., the
interfacial flux J, and the interfacial area between the two liquid phases, Q.
These are linked by the equation:
d½Mt;aq =dt ¼ JQ=V (15.2)
where V is the total volume of the phases, and the subscript t indicates the
contact time. Introducing the definition of specific interfacial area, as:
as ¼ Q=V (15.3)
Eq. (15.2) becomes:
d½Mt;aq =dt ¼ Jas (15.4)
Taking as an example an extraction process with chemical reaction con-
ducted in a conventional mixer-settler unit, the value of J will depend on the
balance between the overall rate of chemical reaction and the mass transfer
coefficients; i.e., J will in general depend on the concentrations of the
reactants and on the degree of turbulence in the phases. For a given system,
the interfacial area generated in the contactor, as, depends mainly on the
degree of turbulence created by the power input into the mixer: the higher
the power input, the smaller the drop sizes and the higher the value of as.
Therefore, J and as are not independent, which makes the prediction of the
rate of transfer in a contactor difficult.
Equations (15.1) and (15.4) provide the background for the study of
the performance of solvent extraction processes.
15.2 NOVEL SOLVENT EXTRACTION PROCESSES

Conventional solvent extraction is a well-established technology for the
separation of solutes from relatively concentrated feeds such as those found
in the industrial production of chemicals and of metals by hydrometallurgy.
Dilute streams, on the other hand, pose a challenge. Equation (15.1) indicates
that treatment of these feeds using conventional liquid–liquid extraction
requires a very large value of DM, otherwise the organic phase volume would
become unacceptably high from environmental and safety considerations.
The novel solvent extraction technologies developed in the last few decades
try to address these limitations. Their potential to improve the performance
of conventional solvent extraction can be analyzed in a systematic way using
Eqs. (15.1) and (15.4). Four novel technologies that have been developed in
the last few decades will be discussed in the following sections:
Liquid membranes
Nondispersive solvent extraction
Recent Advances in Solvent Extraction Processes 653
Microemulsions
Colloidal liquid aphrons
15.3 LIQUID MEMBRANES

One way of achieving both a large distribution coefficient and a reduction
in the solvent duty was proposed and patented by Norman Li in 1968 [1].
The main feature of this new separation technique, called the liquid mem-
brane process, was that it was a three-phase system consisting of two
phases of a similar nature but different composition (aqueous–aqueous,
organic–organic, gas–gas) separated by a third phase of a different nature
and as insoluble as possible into the other two. The middle phase is the liquid
membrane. Figure 15.1a shows the first configuration presented by Li, the
single-drop liquid membrane, in which the membrane is formed by coating
liquid drops or bubbles with a liquid film layer and by subsequently dispers-
ing the resulting particles into a continuous liquid phase containing a solute.
This configuration is now only of historical importance. There are two other
configurations that have been more widely investigated due to their potential
industrial application: (1) the emulsion liquid membrane, also called the
surfactant liquid membrane; and (2) the supported liquid membrane.
In the emulsion liquid membrane configuration, the liquid membrane
is formed by dispersing into the feed (phase 1) an emulsion of the stripping
Fig. 15.1 Liquid membrane configurations: (a) Single-drop liquid membrane;

(b) emulsion globule; (c) supported liquid membrane.
phase (phase 3) in an organic phase (phase 2) containing an emulsifying

agent [2]. This configuration is shown in Fig. 15.1b. Here the liquid mem-
brane is the continuous phase of the emulsion and the viability of the
process depends primarily on the stability of the emulsion.
In the supported liquid membrane process, the liquid membrane phase
impregnates a microporous solid support placed between the two bulk
phases (Figure 15.1c). The liquid membrane is stabilized by capillary forces
making unnecessary the addition of stabilizers to the membrane phase. Two
types of support configurations are used: hollow fiber or flat sheet mem-
brane modules. These two types of liquid membrane configuration will be
discussed in the following sections.
The formulation of the three phases must be such that the liquid
membrane extracts the solute from one of the phases and the third phase
strips it from the membrane. Thus extraction and stripping take place in the
same contactor, and the stripping phase is where the solute is accumulated,
instead of the organic phase as in the case of conventional solvent extrac-
tion. This allows for a middle phase of small volume that, being thin,
behaves like a membrane.
The main advantage of this process becomes clear when applying
Eq. (15.1) to the three-phase system. As the liquid membrane phase does not
accumulate the solute, the distribution ratio that is relevant to the efficiency
of this process is that between phases 1 and 3. At equilibrium, this equation
applies to both pairs of phases: 2–1 and 3–2:
DM 21 ¼ ½M2 =½M1 (15.5)
DM 32 ¼ ½M3 =½M2 (15.6)
The distribution ratio between phases 3 and 1 is then given by:
DM 31 ¼ ½M3 =½M1 ¼ DM 21 D M32 (15.7)
As the distribution ratio between phases 1 and 3 is the product of those in
the two pairs of fluids, the potential effectiveness of the liquid membrane
process is considerably greater than that of conventional solvent extraction.
Thus the liquid membrane process is particularly suitable for the treatment
of dilute feeds. In addition, if the liquid membrane is an organic phase, its
small volume reduces the solvent duty considerably.
15.3.1 Mechanisms of Solute Transfer in

Liquid Membranes
As in most membranes, the liquid membrane must have selective per-
meability to specific solutes. The overall mechanism of solute transfer con-
sists of three steps: (1) extraction of the solute into the liquid membrane;
(2) diffusion of the extracted species through the membrane; and (3) strip-
ping into the third phase. As in conventional solvent extraction, solute
transfer into the membrane can be achieved by selective solubility or by
chemical reaction with a component in the liquid membrane. Stripping at
the interface of the liquid membrane with the third phase can be equally
achieved by either physical or chemical mechanisms. The mechanism of
solute separation requires analysis as it affects the conditions that control
the overall rate of transfer between the first and the third phase.
Application of Eq. (15.1) to the liquid membrane process highlights
one of the main advantages of the process, i.e., the high solute distribution
coefficient that can be obtained between phases 3 and 1. However, another
factor that must be considered when evaluating a separation process per-
formance is the kinetics of transfer, which is given in a general form by
Eq. (15.4). This equation indicates that the transfer rate in the contactor
increases with both the interfacial flux and the specific interfacial area.
The two main mechanisms of solute separation by liquid membranes
involve chemical reactions. In the case illustrated in Fig. 15.2a, the solute
first dissolves in the liquid membrane, then diffuses toward phase 3 due to the
buildup of a concentration gradient, and finally transfers to phase 3 at the
Fig. 15.2 Basic mechanisms of liquid membrane extraction: (a) type I facilitated
transport (A + B?AB); (b) type II facilitated transport (A + B , B + C).
second interface. Phase 3 contains a reagent that reacts irreversibly with

the solute to form products that are insoluble in the membrane, and therefore
incapable of diffusing back through the membrane, thus maintaining the
solute gradient in the membrane. Ammonia, sulfidric acid, and phenol are
among the various compounds that have been successfully removed using
H2SO4 and NaOH as reactants in phase 3 [3,4]. A major disadvantage
associated with this mechanism is the difficulty of achieving selective
separations of solutes of similar size and chemical properties, as membranes
are usually permeable to all of them.
The other type of chemical mechanism is more selective and is used
when the solute is not soluble in the membrane phase, therefore requiring the
addition of a selective reactant into the membrane to form a complex or an
ion pair with the solute. The reaction product then diffuses across the
membrane and at the second interface it reacts with a species added to phase
3 so that stripping also takes place by chemical reaction (Fig. 15.2b). This
mechanism is called carrier-mediated membrane transfer. The reagent re-
covered from the reversed reaction then transfers back to the extraction
interface. This is usually called the reagent shuttle mechanism.
A typical application of carrier-mediated transfer is the recovery of
metal cations from aqueous phases. The overall reactions involved in the
extraction and stripping stages can be represented by the following revers-
ible reaction:
þ
ðaqÞ þ n.RHðorgÞ $ Rn MðorgÞ þ n.HðaqÞ
Mnþ (15.8)
where Mn+ is a metal cation of valence n, RH is an oil-soluble liquid ion-

exchange reagent, and RnM is the metal complex. In this example, the for-
ward reaction takes place at the interface between phase 1 and the membrane,
and the reverse reaction at the other membrane interface. Equation (15.8)
provides the guidance for the formulation of both the liquid membrane and
the stripping phase, as for a given concentration of metal ion in the feed a
high concentration of reactant favors the forward reaction, whereas a high
complex concentration and a low pH facilitate the reverse reaction. The latter
indicates that one of the conditions required in order to improve the flux
through the membrane is that the pH of phase 3 must be substantially lower
than that of phase 1.
15.3.2 Emulsion Liquid Membrane Process

The emulsion liquid membrane (Fig. 15.1b) is a modification of the single
drop membrane configuration presented by Li [2] in order to improve the
stability of the membrane and to increase the interfacial area. The membrane
phase contains surfactants or other additives that stabilize the emulsion.
Depending on the mechanism of extraction, the liquid membrane may also

contain a carrier that reacts with the solute; the internal phase of the emul-
sion is the stripping phase and must be formulated accordingly. In this
configuration, the liquid membrane is the continuous phase of the emulsion,
and the extent of the interface between the feed and the liquid membrane
depends on the size of the emulsion globules dispersed into the feed, which in
turn depends on the physical properties of the phases and the mode and
intensity of the mixing. The interfacial area on both sides of the liquid
membrane is relevant to the rate of extraction, as indicated by Eq. (15.4).
This equation shows that knowledge of the relevant interfacial area and of
the interfacial flux J would allow the rate of extraction in the contactor to be
calculated. However, both the size of the emulsion globules dispersed in a
stirred contactor, which provides the extraction interfacial area, and the size
of the emulsion droplets that leads to the calculation of the extent of the
stripping interface are difficult to measure. Reported emulsion globule sizes
measured in stirred tanks are of the order of 0.2–0.4 mm [5,6], whereas
emulsion droplet sizes are in the range of 1–10 mm [3,7].
Figure 15.3 illustrates schematically the different stages of a con-
tinuous separation process using the emulsion liquid membrane. There are
four main stages in the flow sheet: (1) emulsification of the stripping phase
Fig. 15.3 Emulsion liquid membrane process.

with the liquid membrane phase; (2) dispersion of the emulsion into the feed;
(3) separation of the emulsion from the raffinate phase; and (4) demulsifi-
cation. This final stage separates the stripping solution that contains the
species extracted from the feed, from the liquid membrane phase, which is
recycled to the emulsification stage.
In common with the supported liquid membrane, the emulsion liquid
membrane yields a solute partition coefficient of a higher order of magni-
tude than that obtained with the conventional solvent extraction process,
thus allowing a high separation percentage from dilute feeds, and con-
centrated stripping solutions in just one contact stage. However, the process
has its disadvantages; one is that the need to produce a stable emulsion
requires the use of additives that slow down the rate of extraction and, even
if their solubility is negligible, they may contaminate the raffinate. Another
disadvantage is the problem posed by emulsion rupture in the contactor.
Emulsion rupture is usually due to emulsion swelling caused by the
transport of the external phase into the emulsion. Three different mecha-
nisms have been identified as causes of emulsion swelling: (1) occlusion due
to entrainment of the external phase [8]; (2) secondary emulsification of the
external phase caused by an excess of surfactant in the liquid membrane
phase; and (3) external phase permeation through the liquid membrane
[9,10]. The latter includes osmosis and, in the case of external aqueous feed,
the transport of hydration water attached to complexes and water transport
due to the presence of reverse micelles in the organic membrane. Swelling
and emulsion rupture can be greatly decreased by including additives and
especially designed components to the membrane phase.
15.3.3 Supported Liquid Membrane

As shown in Figure 15.1c, the supported liquid membrane consists of a
microporous solid support impregnated with the membrane phase and
placed between the two bulk phases [11–13]. In this case, the interfacial area
and the thickness of the liquid membrane can be selected by choosing the
solid membrane porosity, size, and thickness. The main advantages of sup-
ported liquid membranes over emulsion liquid membranes are their well-
defined and easily measurable interfacial mass transfer area and membrane
thickness, and the absence of surfactant additives that in general reduce the
membrane flux and may contaminate the bulk phases. The main dis-
advantages are the difficulty in controlling the pressure on both sides of the
membrane in order to avoid blowing the membrane out of the support, and
the washing of the membrane from the support caused by shear forces. The
former leads to contamination of the separated phases, and the latter to the
need to reimpregnate the membrane.
15.4 NONDISPERSIVE SOLVENT EXTRACTION

Nondispersive solvent extraction is a novel configuration of the conventional
solvent extraction process. The term nondispersive solvent extraction arises
from the fact that instead of producing a drop dispersion of one phase in the
other, the phases are contacted using porous membrane modules. The
module membrane separates two of the immiscible phases, one of which
impregnates the membrane, thus bringing the liquid–liquid interface to one
side of the membrane. This process differs from the supported liquid mem-
brane in that the liquid impregnating the membrane is also the bulk phase at
one side of the porous membrane, thus reducing the number of liquid–liquid
interfaces between the bulk phases to just one.
There are two different arrangements for the process. One uses two
modules: one for extraction and the other for stripping, making it formally
closer to conventional solvent extraction. The other configuration is closer to
the liquid membrane process, as the three phases flow through the same
module: the liquid membrane phase in the shell, and the feed and the strip-
ping phase through the lumen of different fibers in the module. Therefore,
this is a three-liquid phase system and although the liquid membrane may not
be as thin as in the emulsion or supported liquid membrane configurations,
extraction and stripping take place simultaneously in the same contactor,
thus keeping the thermodynamic advantages of the three-phase system. A
review of membrane-based nondispersive solvent extraction has been pub-
lished by Pabby and Sastre [14].
The main benefits of nondispersive solvent extraction over the con-
ventional process are: (1) it avoids the need of a settling stage for phase
disengagement and the consequent risk of dispersed phase carryover; (2) the
value of the interfacial area per unit volume can be much higher than in a
liquid–liquid dispersion as there is no risk of phase inversion; and (3) the
interfacial area is easily calculated and scale-up of the process is straight-
forward.
15.5 MICROEMULSIONS AND REVERSE MICELLES

The term microemulsion is applied in a wide sense to different types of
liquid–liquid systems. In this chapter, it refers to a liquid–liquid dispersion
of droplets in the size range of about 10–200 nm that is both thermo-
dynamically stable and optically isotropic. Thus, despite being two phase
systems, microemulsions look like single phases to the naked eye. There are
two types of microemulsions: oil in water (O/W) and water in oil (W/O). The
simplest system consists of oil, water, and an amphiphilic component that
aggregates in either phase, or in both, entrapping the other phase to form
the dispersion. The aggregates formed in the aqueous phase, called micelles,
have their molecules orientated with their hydrophobic tails pointing to the
interior of the aggregate and their hydrophilic head toward the continuous
aqueous phase, whereas the aggregates in the organic phase, called reverse
micelles, have the opposite orientation, as shown in Fig. 15.4. The micro-
emulsion droplets are therefore the cores of either micelles or reverse
micelles, stabilized by a surfactant layer. If more components besides the
surfactant are present in the system, they may also be incorporated into
the micelles or reverse micelles; these are then called mixed micelles or
reverse micelles.
The ternary equilibrium diagram in Fig. 15.4 illustrates the effect of
component concentration on the structure and the number of phases in
Fig. 15.4 Schematic ternary-phase diagram of an oil-water-surfactant microemulsion

system consisting of various associated microstructures. A, normal micelles or O/W mi-
croemulsions; B, reverse micelles or W/O microemulsions; C, concentrated microemulsion
domain; D, liquid-crystal or gel phase. Shaded areas represent multiphase regions.
Fig. 15.5 Types of Winsor microemulsions.
a system consisting of an organic solvent, water, and a surfactant soluble in

both. In the regions where the concentration of either water or organic
solvent tends to zero, the system consists of one phase containing reverse
micelles or micelles, respectively, with a negligible content of the third com-
ponent in their cores. As the concentration of this component increases,
more of its molecules transfer to the core of the micelles, forming either an
O/W or a W/O microemulsion. In the multiphase region the system has two
or more phases, depending on the components of the system and the
characteristics of the amphiphilic molecule.
Winsor [15] classified the phase equilibria of microemulsions into four
types, now called Winsor I–IV microemulsions, illustrated in Fig. 15.5.
Types I and II are two-phase systems where a surfactant rich phase, the
microemulsion, is in equilibrium with an excess organic or aqueous phase,
respectively. Type III is a three-phase system in which a W/O or an O/W
microemulsion is in equilibrium with an excess of both the aqueous and the
organic phase. Finally, type IV is a single isotropic phase. In many cases, the
properties of the system components require the presence of a surfactant
and a cosurfactant in the organic phase in order to achieve the formation of
reverse micelles; one example is the mixture of sodium dodecylsulfate and
pentanol.
15.5.1 Critical Micelle Concentration

Critical Micelle Concentration (cmc) is the surfactant concentration below
which the formation of reverse micelles does not occur, while the number of
surfactant molecules per micelle is referred to as the aggregation number, n.
The cmc is obtained through physical measurements, and varies from 0.1–
1.0 mmol dm3 in water or the nonpolar solvents.
15.5.2 Water Solubilization Capacity

The amount of water solubilized in a reverse micelle solution is commonly
referred to as Wo, the molar ratio of water to surfactant, and this is also a
good qualitative indicator of micelle size. This is an extremely important
parameter since it will determine the number of surfactant molecules per
micelle and is the main factor affecting micelle size. For an (AOT)/iso-
octane/H2O system, the maximum Wo is around 60 [16], and above this
value the transparent reverse micelle solution becomes a turbid emulsion,
and phase separation may occur. The effect of salt type and concentration
on water solubilization is important. Cations with a smaller hydration size,
but the same ionic charge, result in less solubilization than cations with a
large hydration size [17,18]. Micelle size depends on the salt type and con-
centration, solvent, surfactant type and concentration, and also tempera-
ture.
15.5.3 Mechanisms of Solute Extraction

with Microemulsions
The Winsor II microemulsion is the configuration that has attracted most
attention in solvent extraction from aqueous feeds, as it does not affect the
structure of the aqueous phase; the organic extracting phase, on the other
hand, is now a W/O microemulsion instead of a single phase. The main
reason for the interest in W/O microemulsions is that the presence of the
aqueous microphase in the extracting phase may enhance the extraction of
hydrophilic solutes by solubilizing them in the reverse micellar cores. How-
ever, this is not always the case and it seems to vary with the characteristics of
the system and the type of solute. Furthermore, in many instances the
mechanism of extraction enhancement is not simply solubilization into the
reverse micellar cores. Four solubilization sites are possible in a reverse
micelle, as illustrated in Fig. 15.6 [19]. An important point is that the term
solubilization does not apply only to solute transfer into the reverse micelle
cores, but also to insertion into the micellar boundary region called the
palisade. The problem faced by researchers is that the exact location of
the solute in the microemulsion phase is difficult to determine with most of
the available analytical tools, and thus it has to be inferred.
Some insights can be obtained from the mechanisms of extraction in
two-phase systems. As in conventional solvent extraction, the mechanism of
transfer of the solute is either physical or chemical. In conventional solvent
extraction, physical transfer is used for species that prefer the organic phase,
i.e., their distribution coefficient D allows the use of conventional solvent
extraction. In some cases of low solubility in the organic phase, micro-
emulsions have proved to enhance extraction. An important example in this
Fig. 15.6 Possible solubilizate locations in a micelle. (From Ref. 19.)
category is the extraction of biological molecules, which is discussed in detail

later.
15.5.3.1 Microemulsion Extraction with
Chemical Reaction: Metal Ion Extraction
Conventional extraction with chemical reaction is used for solutes that are
insoluble in the organic phase unless they react with a reagent present in that
phase. An example of this is the extraction of metal ions described by
Eq. (15.8). In this case, if the organic phase is replaced by a W/O micro-
emulsion containing the reactant, there is usually extraction enhancement
due to the solubilization of the metal complex in the microphase. There are
two possible ways of forming a W/O microemulsion in the solvent phase:
1. The reactant forms reverse micelles in the organic phase leading to the
formation of a microemulsion when this phase is contacted with the
aqueous one (in which case, although perhaps unknown, the organic
phase cannot be anything but a microemulsion).
2. The reactant does not form reverse micelles under the conditions of the
process, in which case a surfactant, and sometimes also a cosurfactant,
must be added to the organic phase in order to produce a reverse
micellar phase. In this case the reverse micelles are usually mixed, i.e.,
they include in the micellar shell the reactant and the additives.
Only case (2) can provide a comparison between conventional and micellar
extraction. The few comparisons reported in the literature on the metal
extraction performance of microemulsions containing an extractant with
that of the extractant on its own are, at first sight, contradictory. In
some cases microemulsions produce both synergism and extraction rate
enhancement with respect to the single surfactant, whereas in others they
substantially reduce the metal distribution coefficient and the extraction rate,
or leave them unchanged. An interesting example is found in the extraction
with di(2-ethylhexyl)phosphoric acid (DEHPA). DEHPA does not form
microemulsions in aliphatic solvents at pH 4 or below; however, it forms
microemulsions on addition of a surfactant and a cosurfactant, e.g., sodium
dodecylbenzene sulfonate and n-butanol. Bauer et al. [20] reported sub-
stantial improvements in the extraction of trivalent and quadrivalent metals
with respect to the conventional system with a DEHPA microemulsion.
However, Brejza and Pérez de Ortiz [21] obtained improvements in the
extraction of Al(III) using the same microemulsion, in contrast with the
extraction of Zn(II), which was reduced significantly with respect to the
single DEHPA system. A similar different behavior of a microemulsion in
the extraction of Bi(III) and Zn(II) was observed by Pepe and Otu [22]; they
observed an increase in the extraction of the trivalent metal, but not for zinc.
Brejza and Pérez de Ortiz [21] attempted to explain the contrasting effect of
the microemulsion in the extraction of aluminum and zinc with DEHPA
based on the different interfacial behavior of their complexes: the aluminum
complex is more hydrophilic, therefore it may have a greater desorption
energy than the zinc one, making the interface its preferred location, whereas
the zinc complex is more soluble in the organic phase. Therefore, in the
microemulsion system it would be preferentially solubilized in the palisade
rather than in the micellar cores, leading to an enhancement of extraction due
to the large increase in interfacial area produced by the microemulsion. On
the other hand, the zinc complex is more soluble in the organic phase than in
the aqueous phase or the palisade, and consequently the presence of the
aqueous microphase does not improve its solubility in the extracting phase.
This hypothesis remains to be confirmed.
One important point regarding microemulsion extraction is that the
complexity of the system with its three phases, two interfaces, and usually
unknown phase morphology makes the prediction of its performance quite
difficult, particularly when dealing with solutes of diverse properties. The
literature indicates that it does not always improve metal extraction, and in
cases it may even hinder it due to the effect of the emulsifying additives.
15.5.3.2 Extraction of Biological Molecules
The mechanism of separation of biological molecules such as proteins and
amino acids, and the parameters that affect the extraction distribution coef-
ficient and the kinetics of extraction have been studied more extensively than
the extraction of inorganic solutes. This is mainly due to the variety of size
and structure of these molecules and, furthermore, to the fact that their
characteristics may be adversely affected by their contact with solvents and
surfactants.
15.5.3.2a Effect of System Parameters

on Forward Transfer
The distribution of biomolecules between the aqueous and reverse micelle
phases depends on system parameters such as pH, ionic strength, and salt
type, all of which affect the physicochemical state of the protein and its
interaction with surfactant head groups specifically, and the water pool in
general. In addition to these factors, solvent structure and type, tempera-
ture, surfactant concentration, and cosurfactant play a significant role in
determining the aggregation properties of a surfactant in the solvent such as
size, and will influence protein partitioning behavior as much as they will
affect the cooperative formation of the protein-micelle complex. Further-
more, protein size and hydrophobicity are also important in determining
protein partitioning behavior in the reverse micelle phase. The following
sections deal with the parameters known to influence protein extraction.
Effect of pH: The pH of a solution affects the solubilization char-
acteristics of a protein primarily in the way in which it modifies the charge
distribution over the protein surface. At pH values below its isoelectric point
(pI), or point of zero net charge, a protein acquires a net positive charge,
while above its pI the protein will be negatively charged. Thus, if electro-
static interactions are the dominant factor, solubilization should be possible
only with anionic surfactants at pH values less than the pI of the protein;
because at values above pI, electrostatic repulsion would inhibit solubili-
zation. The opposite effect would be anticipated in the case of cationic
surfactants.
Effect of salt type and concentration: The ionic strength of the aqueous
solution in contact with a reverse micelle phase affects protein partitioning
in a number of ways [18,23]. The first is through modification of electrostatic
interactions between the protein surface and the surfactant head groups by
modification of the electrical double layers adjacent to both the charged
inner micelle wall and the protein surface. The second effect is to ‘‘salt out’’
the protein from the micelle phase because of the increased propensity of the
ionic species to migrate to the micelle water pool, reduce the size of the
reverse micelles, and thus displace the protein.
Effect of surfactant type and concentration: An increase in surfactant
concentration results in an increase in the number of micelles rather than any
substantial change in size, and this enhances the capacity of the reverse
micelle phase to solubilize proteins. Woll and Hatton [24] observed in-
creasing protein solubilization in the reverse micelle phase with increas-
ing surfactant concentration. In contrast, Jarudilokkul et al. [25] found
that at low ‘‘minimal’’ concentrations (6–20 mmol dm3 AOT), reverse
micelles could be highly selective in separating very similar proteins from
fermentation broths, and the recovery in activity of up to 95% could be

achieved from broths.
Effect of temperature: Luisi et al. [26] reported that the temperature
markedly affected the transfer of a-chymotrypsin in a chloroform-trioctyl-
methylammonium chloride (TOMAC) system. By increasing the tempera-
ture from 25–408C, about 50% higher transfer yield was realized. No
appreciable transfer of glucagone took place at room temperature, whereas
transfer at 378C was possible. These results contradict work by Dekker et al.
[27], who studied the back stripping (desolubilization) of a-amylase from a
TOMAC/isooctane/octanol/Rewopal HV5 system by increasing the tem-
perature. This caused a decrease in Wo with increasing temperature and, as a
result, the a-amylase was expelled from the reverse micelle phase.
Effect of affinity ligand: Woll et al. [28] reported that the solubilization
of concanavalin A in AOT/isooctane increased by introducing an affinity
surfactant, octyl-b-D-glucopyranoside as a cosurfactant. Further studies by
Kelley et al. [29] using an affinity cosurfactant such as octyl glucoside for
concanavalin A, lecithin for mycelin basic protein, and alkyl boronic acids
for chymotrypsin were carried out, and the results show an increase in the
amount of protein extracted at the same operating pH and salt concentra-
tion. Because of the large number of system parameters that can influence
separation efficiency, a method is needed to optimize the parameters to
maximize removal and the recovery of activity (not necessarily the same
thing). Jarudilokkul et al. [25] used response surface methodology (RSM) to
optimize the separation of three proteins (cytochrome c, ribonuclease A, and
lysozyme) from a fermentation broth; the parameters found to be the most
influential were AOT concentration, pH, and temperature.
15.5.3.2b Effect of System Parameters on
Backward Transfer
There are two classes of parameters that influence the efficiency of back
extraction: first, the parameters that govern the forward extraction such as
pH, salt type and concentration, surfactant type and concentration, and
protein type and concentration; and second, the pH, salt type and con-
centration of stripping solutions, and extraction temperature.
Effect of pH: It is obvious that in order to recover the protein from
reverse micelles, the pH of the stripping solution needs to change toward
the pI, which will result in a reduction of the protein interaction with the
oppositely charged head groups. The extent of protein recovery from reverse
micelles increases with increasing pH for anionic surfactants; however, for
cationic surfactants the opposite is true.
Effect of surfactant concentration: Increases in the surfactant con-
centration will lead to an enhancement of protein extraction due to an
increase in the number and size of the reverse micelles. The work of Hentsch
et al. [30] showed that the back transfer of a-chymotrypsin decreased with
decreasing AOT concentration. They suggested that this decrease was due
either to a-chymotrypsin being trapped in the emulsion, or to denaturation.
Salt type and concentration: For back-extraction, increases in pH are
not enough to strip the protein out from reverse micelles; this is also due to
the size exclusion effect resulting from a decrease in the reverse micelle size
[31,32]. This means that high salt concentration and salts that form small
reverse micelles favor back transfer. Most of the work reported in the lit-
erature used KCl solution, normally 1.0 mol dm3 KCl coupled with a pH
around 7.5. Marcozzi et al. [23] also showed that the back transfer efficiency
of a-chymotrypsin depends on the salt type and concentration used in the
forward transfer.
Counterion extraction: Due to the relative slowness of back extraction
based on the methods above, the back-extraction of proteins encapsulated in
AOT reverse micelles was evaluated by adding a counterionic surfactant,
either TOMAC or DTAB, to the reverse micelles [33]. This novel backward
transfer method gave higher backward extraction yields compared to the
conventional method. The back-extraction process with TOMAC was found
to be 100 times faster than back-extraction with the conventional method,
and as much as three times faster than forward extraction. The 1:1 complexes
of AOT and TOMAC in the solvent phase could be efficiently removed using
adsorption onto montmorillonite so that the organic solvent could be reused.
15.5.4 Extraction Kinetics with Micellar Systems

For the scale-up of reverse micelle extractions, it is important to know which
factors determine the mass transfer rate to or from the reverse micelle phase.
So far most work has concentrated on the kinetics of solubilization of water
molecules [34,35], protons [36], metal ions [20,35,37,38–40], amino acids
[41], and proteins [8,35,42,43]. There are two separate processes: forward
transfer, which is transfer of solute from the aqueous to the reverse micelle
phase, and back transfer, which is the antithesis of the first one.
The most commercially important mechanism of all is the kinetics of
solute transfer from an aqueous to a reverse micelle phase. The kinetics of
extraction of metal ions have not received the same research attention as the
extraction capacity of W/O microemulsions. As the mechanism of extraction
of metal ions is chemical, the effect of creating a microemulsion in an
organic phase that contains the reactant can be measured experimentally.
Results indicate that, as in the case of extraction equilibrium, the rate of
extraction may increase substantially by the presence of the microemul-
sion as compared with the conventional system [20,38,44] or decrease it to
negligible levels [44]. Brejza found that the effect of the microemulsion on
the kinetics of extraction depends on the characteristics of the metal ion and
its complex, e.g., valence and degree of complex hydration, and on the
solubility of the complex in the organic phase [44]. Thus there are no general
rules as to the advantage of forming a microemulsion in a system that
contains a reagent.
In the case of protein transfer, early work by Bausch et al. [42] found
that transport was controlled by convective processes in the aqueous phase,
and the mass transfer coefficient increased with increasing surfactant but
depended strongly on stirring speed. Poppenborg et al. [45] evaluated the
kinetic separation of lysozyme and cytochrome in a Graesser contactor and
Lewis cell, and with a low rotor speed (2–3 rpm), low temperature (48C),
and a pH close to the pI of both proteins (pH 9–10), about 83% of the
lysozyme and only 11% of the cytochrome c was extracted into the reverse
micellar phase after 30 minutes. This optimum separation was based on the
effect pH changes have on the extraction kinetics, and the rate of cyto-
chrome c extraction was reduced much more than for lysozyme when the
pH approached the pI of cytochrome c, and differed markedly from the
effect the pH had on phase distribution. The extraction rate measured
in the Graesser contactor differed from that measured in the Lewis
cell, i.e., lysozyme was extracted faster than cytochrome c and this obser-
vation indicates that different steps of the reverse micellar transfer mecha-
nism are controlling the transfer, depending on the way the phases are
contacted.
The kinetics of back extraction are equally important to obtain a
better understanding of the mechanism of solute transfer, and to determine
the rate-limiting step for the process. Such information is crucial for the
rational design of an extraction apparatus and, as discussed above, Jar-
udilokkul et al. [33] showed that counterion extraction resulted in remark-
able increases in the back-extraction of proteins.
15.5.5 Micellar Extraction Potential Applications

15.5.5.1 Extraction from Synthetic Mixtures
The ease that certain protein mixtures can be separated using reverse micelle
extraction was clearly demonstrated by Goklen and Hatton [46], Goklen
[31], and Jarudilokkul et al. [25], who investigated a series of binary and
ternary protein mixtures. In two cases, they were able to quantitatively
extract cytochrome c and lysozyme from a ternary mixture of these proteins
with ribonuclease A. Woll and Hatton [24] investigated the separation of a
mixture of ribonuclease A and concanavalin A, and showed that the system
behaved ideally and that there was no interaction between the proteins.
15.5.5.2 Extraction of Extracellular Enzymes

Rahaman et al. [47] demonstrated the use of reverse micelles (AOT/
isooctane) for the recovery of an extracellular alkaline protease (MW, 33 k
Dalton; pI, 10) from a whole fermentation broth. Purification factors as high
as 6 and yields of 56% were achieved in a three-stage cascade. The combi-
nation of a cascade with a higher aqueous/organic ratio, and the use of true
cross-flow designs, shows promise for purification without dilution. Krei and
Hustedt [48] also demonstrated the application of the reverse micelle tech-
nique by extracting an a-amylase broth of Bacillus licheniformis using a
CTAB/isooctane/5% octanol reverse micelle system. In a two-step extrac-
tion, they were able to reduce the protein concentration by a factor of 10 with
a purification factor of 8.9, and a maximum yield of 89% a-amylase activity.
Jarudilokkul et al. [49] extracted lysozyme from egg white, and while
extractions as high as 98% were achievable, a variety of demulsifiers added to
the mixture could actually enhance yields substantially.
15.5.5.3 Extraction of Intracellular Enzymes

Reverse micelles of CTAB in octane with hexanol as cosurfactant were
reported to be able to lyse whole cells quickly and accommodate the liber-
ated enzyme rapidly into the water pool of surfactant aggregates [50,51]. In
another case a periplasmic enzyme, cytochrome c553, was extracted from
the periplasmic fraction using reverse micelles [52]. The purity achieved in
one separation step was very close to that achieved with extensive column
chromatography. These results show that reverse micelles can be used for
the extraction of intracellular proteins.
15.5.6 Process Consideration and Scale-Up

The liquid–liquid extractors developed for conventional liquid–liquid
extraction are, in principle, also suitable for this application. At present,
process development has only centered on the use of mixer-settlers, cen-
trifuges, spray columns, and membrane extractors, and Lye [53] demon-
strated that it was possible to carry out protein extraction with reverse
micelles in a spray tower. He used a small tower to carry out batch ex-
tractions and showed that existing correlations predicting mass transfer
were an order of magnitude too high due to the rigid interface of the
reverse micelle droplet. However, Jarudilokkul et al. [54,55] have evalu-
ated the use of a Graesser contactor (‘‘raining bucket’’) to extract lyso-
zyme from egg white. This type of contactor allows for the countercur-
rent flow of the reverse micelle phase and the pregnant aqueous mother
liquor under very low shear conditions, thereby minimizing emulsion forma-
tion. These workers characterized the mass transfer performance of the
contactor and developed an integrated system of separation and back-

extraction [56].
Nevertheless, despite many advances in understanding the basic pro-
cesses controlling the separation of organic solutes using reverse micelles,
and in the design and operation of contactors, the use of reverse micelles has
still not been scaled up to an industrial-sized unit.
15.6 COLLOIDAL LIQUID APHRONS

Colloidal liquid aphrons (CLAs), obtained by diluting a polyaphron phase,
are postulated to consist of a solvent droplet encapsulated in a thin aqueous
film (‘‘soapy-shell’’), a structure that is stabilized by the presence of a
mixture of nonionic and ionic surfactants [57]. Since Sebba’s original reports
on biliquid foams [58] and subsequently ‘‘minute oil droplets encapsulated
in a water film’’ [59], these structures have been investigated for use in
predispersed solvent extraction (PDSE) processes. Because of a favorable
partition coefficient for nonpolar solutes between the oil core of the CLA
and a dilute aqueous solution, aphrons have been successfully applied to the
extraction of antibiotics [60] and organic pollutants such as dichlorobenzene
[61] and 3,4-dichloroaniline [62].
15.6.1 Preparation, Structure, and Stability of CLAs

Polyaphron phases are prepared by gradually introducing the organic
solvent/nonionic surfactant solution from a burette into a stirred 2 cm3
reservoir of aqueous/surfactant solution. Due to the influence of prepara-
tion conditions on the stability of the polyaphron phase, a consistent pre-
paration procedure needs to be used; the typical organic addition rate is
*0.5 cm3 min1, with a stirring speed of 400 rpm using an 18 mm magnetic
stirrer bar. The typical container in which polyaphron preparation takes
place is a 30 cm3 glass jar with an internal diameter of 35 mm, and a tem-
perature of *22±28C. The resulting polyaphron phases are highly viscous
and cream colored, with a phase volume ratio (PVR=Vorg/Vaq) of around 4
(f=0.8), and are very stable, with no phase separation evident over a period
of months.
Although potential applications for CLAs have been investigated,
little work has been carried out to either confirm or refute Sebba’s proposed
structures for polyaphron or CLA phases (Fig. 15.7). Note the terminology
that is used in this chapter.
Upon manufacture, the initial creamy-white phase consists of an
aggregate of individual aphrons having a structure resembling that of a
biliquid foam termed a polyaphron [57,58], and data on the structure and
Fig. 15.7 Proposed structure of CLA. (From Ref. 57.)
stability of this phase is obtained when the polyaphron is not dispersed in an

aqueous phase. Upon dispersion of a polyaphron in a continuous aqueous
phase, however, the individual aphrons become separated to form spherical
droplets [58]; these are termed CLAs. Obviously the properties of the poly-
aphron will depend on the aphron formulation and method of manufacture
[58,63], while the properties of the CLAs will, in addition, be influenced by
the nature of the continuous phase in which they are dispersed. Light scat-
tering has previously been used to determine the size of dispersed CLAs and
colloidal gas aphrons (CGAs) [63,64] though, due to their opaque nature, it is
not applicable to the study of polyaphron phases.
Lye and Stuckey [65] used the techniques of cryoultramicrotome TEM
and DSC to analyze polyaphrons to test Sebba’s proposed structure for
these phases (Table 15.1). Results from these methods were also compared
with those obtained by light scattering of dispersed CLAs to see if any
Table 15.1 Comparison of the Sizes and Shapes of Polyaphrons as Determined by

Electron Microscopy (Cryo-TEM) with Those of Dispersed CLAs as Determined by Light
Scattering
Cryo-TEM (polyaphron) Light (CLA) scattering
Initial
Phase Shell
volume ratio doila thickness dovb dovb
(PVR) (mm) (mm) (mm) Shape (mm) Shape
10 3.5–4.6 0.03 3.6–4.7 polyhedral 4.0 spherical
4 1.5–5.5 0.15 1.8–5.8 ovoid 7.0 spherical
a
Mean diameter of oil core measured for major and minor axes.
b
Overall diameter of oil core and soapy shell.
correlation existed. In addition, they extended their initial investigation on

dispersed CLA stability to study the influence of continuous phase prop-
erties on CLA half-lives, and examined to what extent this data could be
used to further elucidate the structure of CLAs.
The effect of NaCl concentration in the continuous phase on the half-
lives of dispersed CLAs at 258C was examined by Scarpello and Stuckey
[66]. Compared to CLAs dispersed in deionized water, which have a half-life
of approximately 60 min, they found that the values of t1/2 decreased with
increasing ionic strength reaching a constant value of around 15 min at ionic
strengths greater than 0.3 mol dm3. The data presented by Matsushita and
co-workers [64] for CGAs dispersed in 3.4 mmol dm3 NaCl also indicates a
decrease in t1/2 for CGAs formulated from cationic or anionic surfactants,
but no significant change for those formulated from a nonionic surfactant.
Their values of t1/2 are estimated to be around 4 min, which indicates that
dispersed CGAs are considerably less stable than the CLAs investigated
here. Addition of salts to CGA or CLA polyaphrons, especially those of
polyvalent ions, has previously been shown to reduce the stability of non-
dispersed polyaphron phases, and even break them [67].
Since the stability of CLAs displays a strong dependence on ionic
strength, due to electrostatic interactions associated with the surfactant head
groups, pH should also have an influence on CLA half-lives. The effect of
continuous phase pH on the stability of CLAs dispersed in deionized water
at 258C was investigated by Lye and Stuckey [65]. Above a pH of 6–7, t1/2
values were essentially constant, but began to decline as the continuous
phase became more acidic. At low pH values, the excess hydrogen ion
concentration led to protonation of the sulfonate head groups of the SDS
molecules located at the outer soapy-shell interface. This would have two
effects on aphron stability. First, it would reduce the surface charge on the
aphrons and hence the energy barrier to droplet coalescence. It had pre-
viously been found that the zeta potential of a CLA suspension falls from
45 mV at pH 8.4 to 36 mV at pH 4 [68]. Secondly, protonation of the
head groups reduces the polarity of the surfactant monomers, making it
energetically less favorable for the hydrophobic tails of the surfactant
molecules to remain in an aqueous environment. Experimentally, it is easy
to show that the solubility of SDS in 20 mmol dm3 buffer solutions rapidly
decreases below pH 7, which may also be responsible for decreasing CLA
stability at low pH. If the collision-coalescence mechanism proposed is
correct, then the stability of CLAs should depend upon the temperature at
which this process occurs. This was indeed found experimentally with t1/2
values falling from 96 to 4 min for a corresponding increase in temperature
of the continuous phase from 108C to 608C [65]. An Arrhenius plot resulted
in a linear relationship between Ln k and 1/T. Performing linear regression
yields a single value of the activation energy, Ea, for the collision process of
50 kJ mol1. A value for Ea of this magnitude would suggest that, over the
temperature range investigated, the collision-fusion process of the CLAs is
controlled by both diffusion and chemical reaction.
Concerning the structure of dispersed CLAs, the model originally
proposed by Sebba [57] of a spherical oil-core droplet surrounded by a thin
aqueous film stabilized by the presence of three surfactant layers is, in our
opinion, essentially correct. However, there is still little direct evidence for
the microstructure of the surfactant interfaces. From an engineering point of
view, however, there is now quantitative data on the stability of CLAs
which, together with solute mass transfer kinetics, should enable the suc-
cessful design and operation of a CLA extraction process.
15.6.2 Formation of CLAs

On a more practical level, to use CLAs and CGAs in PDSE it is important
to understand the influence of key parameters such as solvent type and
polarity, and surfactant type (hydrophilic/lipophilic balance, HLB) and
concentration, on the formulation and stability of CLAs and CGAs. These
are discussed next.
A variety of nonpolar to moderately polar solvents has been evaluated
for their ability to form stable CLAs (Table 15.2) [68]. As can be seen, as the
solvent becomes more polar the aphron size increases and it becomes more
unstable. Hence, the influence of the surfactant HLB number was evaluated
using a moderately polar solvent, n-pentanol [68]. This demonstrated that if
the HLB number of the surfactant is high enough, then it is possible to
formulate stable CLAs.
Table 15.2 Preparation of CLAs from Different Solvents

Solubility in CLA size PVR
Solvent water (wt %) CLA stabilitya (diameter, mm) obtained
n-Decane 52 ppb Very stable 14.0 20
n-Octane 6.6.107 Very stable 10.8 20
Isooctane - Very stable 9.6 20
n-Hexane 0.00123 Very stable 9.6 20
Decalin <0.02 Very stable 13.1 20
Kerosene - Very stable 11.1 20
p-Xylene 0.0156 Stable 29.8 20
Toluene 0.0515 Stable 30.8 20
Benzene 0.1791 Stable 7.2 20
Decan-1-ol Insoluble Stable 25.1 10
Octan-1-ol 0.0538 Stable 27.9 5
n-Pentanol 2.19 Unstable - -
a
CLAs made with 0.4% (w/v) sodium dodecyl benzene sulphonate (SDBS) in the aqueous phase, and
1% (w/v) Softanol-30 in the organic phase.
15.6.3 Kinetics of Solute Extraction

and CLA Separation
Measuring the kinetics of solute extraction using CLAs is difficult for two
reasons: first, due to their complex interfacial structure, i.e., the soapy shell,
they are impossible to mimic in an apparatus with a defined interfacial area
such as a Lewis cell; second, due to their large interfacial area, when they
are dispersed with a solute with a high partition coefficient such as ery-
thromycin, the partition equilibrium is virtually instantaneous. For example,
at a pH of 10, erythromycin (pKa=8.6) has a partition coefficient of 170, and
with 0.5 g dm3 of erythromycin at this pH and a phase ratio of 100 parti-
tioning is extremely rapid, and accurate rates of extraction are impossible to
measure. Within the errors of measurement, as expected, mixing does not
seem to have any effect on mass transfer or partitioning since the system is
likely to be interfacially controlled. Using ‘‘typical’’ interfacial mass transfer
rates for this type of system, together with a surface area obtained through
particle size analysis, partition equilibrium should be achieved within 102
seconds. Hence, due to the large surface area, contact times for complete
extraction of nonpolar solutes could be reduced to seconds in an in-line pipe
contactor using CLAs. Even at a phase volume ratio of 50:1, 82% extraction
was found, and a two-stage contactor would result in 96% extraction with
concentration factors of around 100.
The kinetics of back-extraction of erythromycin at pH 6 are an order

of magnitude slower and measurable, with incomplete recovery of the solute
[69]. The reason for this is not entirely clear, but it is likely that the ery-
thromycin forms a complex with the surfactant, which results in a dis-
sociation reaction, slowing the back-extraction and resulting in less than
100% recovery. In addition, since the viscosity of the organic phase is 13
times greater than water, the diffusivity of the antibiotic in the organic
phase is low, resulting in a longer extraction time from the stagnant solvent
droplet.
After extraction, the solute-laden CLAs need to be separated from the
mother liquor so that they can be back stripped. Hence attempts were made
to filter the solute-rich CLAs from the aqueous phase using cross-flow
microfiltration [70]. The filtration characteristics of the CLAs as indicated by
the flux, CLA size, and concentration showed that they are completely
retained by the membrane and do not foul the membrane surface. Using this
system, the CLAs could easily be concentrated up to 30% w/v at low pres-
sures, and the permeate stream remained totally clear. The CLAs appear to
maintain their structural integrity because only 3–4 mg dm3 of SDS was
measured in the permeate.
15.6.4 Potential Applications of CLAs and CGAs

As previously discussed, CLAs can be used to extract any nonpolar solute
from antibiotics to polluting chlorinated organics; however, polar solutes
that do not partition well cannot be extracted. Because of this, ‘‘liquid ion
exchangers’’ such as the alkylamines (Aliquat and Alamine 336) have been
incorporated into the solvent phase to extract these polar solutes. Initial
studies have shown that it is possible to incorporate as much as 50% of these
reagents into a CLA without drastically influencing its stability [66]. In
addition, the intriguing phenomenon of increased water inclusion in the
CLA during formulation has been observed. Some evidence seems to sug-
gest that this may be due to the formation of reverse micelles. If this was the
case, then there is a possibility that quite polar solutes could be extracted
using these ‘‘mini-liquid emulsion membranes,’’ but more work on this
phenomenon is required.
Another potential use for CLAs is in two-phase reactors where the
substrate is very nonpolar, and the product is also poorly water soluble and
perhaps unstable. Dispersing CLAs containing high concentrations of sub-
strate into a fermenter will enable the fermentation to proceed rapidly
without substrate limitation, while the product is removed rapidly back into
the CLA. Problems with cells accumulating at the interface of the CLAs
should not occur due to the strong negative charge on the CLA, and the
generally negative charge on most bacterial cells. This situation is commonly
faced in new biotransformation reactions, and even when the product is
polar the use of CLAs will enhance the transfer of substrate.
The final intriguing use of CLAs is in the immobilization of enzymes in
the soapy shell in order to carry out an enzymatic reaction. Thus the hydrol-
ysis of p-nitrophenyl acetate to p-nitrophenol has been demonstrated by
immobilizing a lipase into the shell of a CLA. The CLAs were then pumped
through a cross-flow membrane, where they were separated and recycled,
with the product appearing in the permeate [70].
15.7 CONCLUDING REMARKS

All the novel separation techniques discussed in this chapter offer some
advantages over conventional solvent extraction for particular types of feed,
such as dilute solutions and the separation of biomolecules. Some of them,
such as the emulsion liquid membrane and nondispersive solvent extraction,
have been investigated at pilot plant scale and have shown good potential for
industrial application. However, despite their advantages, many industries
are slow to take up novel approaches to solvent extraction unless substantial
economic advantages can be gained. Nevertheless, in the future it is probable
that some of these techniques will be taken up at full scale in industry.
REFERENCES
1. Li, N. N. U.S. Patent 3,310,794, November 12, 1968.
2. Li, N. N. Ind. Eng. Chem. Process Des. Dev., 1971, 10(2), 215–221.
3. Cahn, R. P.; Franfeld, J. W.; Li, N. N.; Naden, D.; Subramanian, K. N. Recent
Developments in Separation Science; Li, N. N. Ed., CRC Press: Boca Raton,
Florida, Vol. VI, 1981; p. 51.
4. Cahn, R. P.; Li, N. N. Sep. Sci., 1974, 9(6), 505–519.
5. Sharma, A.; Goswami, A. N.; Rawat, B. S. J. Membr. Sci., 1991, 60, 261.
6. Gallego Lizon, T.; Perez de Ortiz, E. S. Ind. Chem. Eng. Res. Dev., 2000, 39,
5020–5028.
7. Marr, R.; Kopp, A. Int. Chem. Eng., 1982, 22, 44–60.
8. Kinugasa, T.; Tanahashi, S. I.; Takeuchi, H. Ind. Eng. Chem. Res.,1991, 30,
2470.
9. Colinart, P.; Delepine, S.; Trouve, G.; Renon, H. J. Membr. Sci.,1984, 20, 167–
187.
10. Ding, X. C.; Xie, F. Q. J. Membr. Sci.,1991, 59, 183–188.
11. Bloch, R. Hydrometallurgical Separations by Solvent Membranes; Flynn, J. E.
Ed., Membrane Science and Technology, Plenum Press; New York, 1970.
12. Danesi, P. R. Sep. Sci. Technol., 1984–1985, 19(11–12), 857–894.
13. Danesi, P. R. Proc. ISEC ’86, Munich: DECHEMA, 1986; 527–535 pp.
14. Pabby, A. K.; Sastre, A-M. Ion Exchange and Solvent Extraction; Marcus, Y.
and Sengupta, A. K. Eds., Marcel Dekker: New York, 2002; vol. 15, 331–469.
15. Winsor, P. A. Solvent Properties of Amphiphilic Compounds; Butterworth Sci-
entific: London, 1954.
16. Zulauf, M.; Eicke, H. F. J. Phys. Chem., 1979, 83(4), 480.
17. Leodidis, E. B.; Hatton, T. A. Langmuir, 1989, 5, 741.
18. Leodidis, E. B.; Hatton, T. A. The Structure and Reactivity in Reverse Micelles;
Pileni, M. P. Ed., Elsevier; Amsterdam, 270, 1989, 270p.
19. Osseo-Asare, K.; Kenney, M. E. Proc ISEC ’80, Liege, 1980, 1, paper no. 80,
121.
20. Bauer, D.; Cote, G.; Komornicki, J.; Mallet-Faux, S. Can. Soc. Chem. Eng.,
1989, 2, 425.
21. Brejza, E. V.; Perez de Ortiz, E. S. J. Colloid Interface Sci., 2000, 227, 244–246.
22. Pepe, E. M.; Otu, E. O. Solv. Extr. Ion Exch., 1996, 14(2), 247.
23. Marcozzi, G.; Correa, N.; Luisi, P. L.; Caselli, M. Biotechnol. Bioeng., 1991,
38, 1239.
24. Woll, J. M.; Hatton, T. A. Bioprocess Eng., 1989, 4, 193.
25. Jarudilokkul, S.; Poppenborg, L. H.; Stuckey, D. C. Sep. Sci. Tech., 2000, 35,
503–517.
26. Luisi, P. L.; Bonner, F. J.; Pellergrini, A.; Wiget, P.; Wolf, R. Helv. Chim. Acta,
1979, 62, 740.
27. Dekker, M.; van’t Riet, K.; Van Der Pol, J. J.; Baltussen, J. W. A.; Hilhorst, R.;
Bijsterbosch, B. H. J. Chem. Eng., 1991, 46, B69.
28. Woll, J. M.; Hatton, T. A.; Yarmush, M. L.; Biotechnol. Prog., 1989, 5(2), 57.
29. Kelley, B. D.; Wang, D. I. C.; Hatton, T. A. Biotechnol. Bioeng., 1993, 42(10),
1199.
30. Hentsch, M.; Menoud, P.; Steiner, L.; Flaschel, E.; and Renken, A. Bio/
Technol., 1992, 6(4), 359.
31. Goklen, K. E. Ph.D. Thesis, Massachussetts Institute of Technology, 1986.
32. Dekker, M.; van’t Riet, K.; Baltussen, J. W. A.; Bijsterbosch, B. H.; Hilhorst,
R. Laane, C. in Proc. 4th European Conf. on Biotechnology, Amsterdam, 1987,
2, 507.
33. Jarudilokkul, S.; Poppenborg, L. H.; Stuckey, D. C. Biotech. Bioeng., 1999,
62(5), 593–601.
34. Battistel, E.; Luisi, P. L. J. Colloid Interface Sci., 1989, 128(1), 7.
35. Nitsch, W. Plucinski, P. J. Colloid Interface Sci., 1990, 136(2), 338.
36. Albery, W. J.; Choudhery, R. A.; Atay, N. Z.; Robinson, B. H. J. Chem. Soc.
Faraday Trans. 1, 1987, 83(8), 2407.
37. Savastano, C. A.; Osseo-Asare, K.; Pérez de Ortiz, E. S. Separation Processes in
Hydrometallurgy; Davis, G. A. Ed., SCI; London, 1987; p. 89.
38. Kim, H. S.; Tondre, C. Sep. Sci. Technol., 1989, 24, 485.
39. Plucinski, P.; Nitsch, W. Ber. Bunsenges. Phys. Chem., 1989, 93, 994.
40. Plucinski, P.; Nitsch, W. J. Phys. Chem., 1992, 97, 8983.
41. Plucinski, P.; Nitsch, W. Langmuir, 1994, 10, 371.
42. Bausch, T. E.; Plucinski, P. K.; Nitsch, W. J. Colloid Interface Sci., 1992,
150(1), 226.
43. Dekker, M.; van’t Riet, K.; Bijsterbosch, B. H.; Fijneman, P.; Hilhorst, R.
Chem. Eng. Sci., 1990, 45(9), 2949.
44. Brejza, E. V. Doctoral Thesis, University of London, 1994.
45. Poppenborg, L. H.; Brillis, A.; Stuckey, D. C. Sep. Sci. Tech., 2000, 35, 843–858.
46. Goklen, K. E.; Hatton, T. A. Proc. ISEC ’86, Munchen, 1986, 3, 587.
47. Rahaman, R. S.; Lee, J. Y.; Cabral, J. M. S.; Hatton, T. A. Biotechnol. Prog.,
1988, 4(4), 218.
48. Krei, G. A.; Hustedt, H. Chem. Eng. Sci., 1992, 47(1), 99.
49. Jarudilokkul, S.; Paulsen, E.; Stuckey, D. C. Biosep., 2000, 9, 81–91.
50. Laane, C.; Dekker, M. Proc. 6th Int. Symp. on Surfactants in Solution; Mittal,
E. L., Ed., Plenum Press: New York, 9, 1989; 9p.
51. Giovenco, S.; Verheggen, F.; Laane, C. Enzyme Microb. Technol., 1987, 9, 470.
52. Jarudilokkul, S.; Poppenborg, L. H.; Valetti, F.; Gilardi, G.; Stuckey, D. C.
Biotech. Techniques, 1999, 13, 159–163.
53. Lye, G. Y. Ph. D Thesis, University of Reading, 1993.
54. Jarudilokkul, S.; Paulsen, E.; Stuckey, D. C. Biotech. Bioeng. 2000, 69, 618–
626.
55. Jarudilokkul, S.; Paulsen, E.; Stuckey, D. C. Biotech Progress, 2000, 16, 1071–
1078.
56. Jarudilokkul, S.; Stuckey, D. C. Sep. Sci. Technol., 2001, 36, 657–670.
57. Sebba, F. Foams and Biliquid Foams-Aphrons; John Wiley: New York, 1987.
58. Sebba, F. J. Colloid Interface Sci., 1972, 40, 468.
59. Sebba, F. Colloid Polymer Sci., 1979, 257, 392.
60. Lye, G. J.; Stuckey, D. C. Separations for Biotechnology III; Pyle, D. L. Ed.,
SCI: London, 1994; 280–286.
61. Wallis, D. A.; Michelsen, D. L.; Sebba, F.; Carpenter, J. K.; Houle, D. Biotech.
Bioeng. Symp. 1985, 15, 399.
62. Lye, G. J.; Poutiainen, L. V.; Stuckey, D. C. Biotechnology ’94, 2nd Int. Sym-
posium on Environmental Biotechnology; I. Chem. E., II, 1994; 25–27.
63. Matsushita, K.; Mollah, A. H.; Stuckey, D. C.; del Cerro, C.; Bailey, A. I.
Colloids Surfaces, 1992, 69, 65–72.
64. Chaphalkar, P. G.; Valsaraj, K. T.; Roy, D. Sep. Sci. Technol., 1993, 28, 1287.
65. Lye, G. J.; Stuckey, D. C. Colloids Surfaces, A: Physiochem. Eng. Aspects,
1998, 131(1–3), 113–130.
66. Scarpello, J. T.; Stuckey, D. C. J. Chem. Tech. Biotech., 1999, 74, 409–416.
67. Save, S. V.; Pangarkar, V. G. Chem. Eng. Comm., 1994, 127, 35.
68. Stuckey, D. C.; Matsushita, K.; Mollah, A. H.; Bailey, A. I. Third Int. Conf. on
Effective Membrane Processes: New Perspectives; University of Bath, U.K.,
1993.
69. Lye, G. J.; Stuckey, D. C. Chem. Eng. Sci., 2001, 56, 97–108.
70. Lye, G. J.; Pavlou, O. P.; Rosjidi, M.; Stuckey, D. C. Biotech Bioeng., 1996, 51,
69–78.
16
Computational Chemistry in Modeling
Solvent Extraction of Metal Ions
JERZY NARBUTT Institute of Nuclear Chemistry and Technology,
Warsaw, Poland
MARIAN CZERWIŃSKI Zawiercie University of Administration and
Management, Zawiercie, Poland
16.1 INTRODUCTION
Great progress in computational techniques and methods in the last decade
of the 20th century has made it possible to perform quantum mechanical
(QM) calculations on large molecular systems composed of several dozens
or even hundreds of atoms. General ab initio schemes based on Hartree-
Fock-Roothaan (HFR) methods [1], formerly used for small molecules
composed of light elements, now are becoming applicable for metal com-
plexes, including even heavy metals. Calculations based on the density-
functional theory [2] have appeared particularly fruitful in this respect.
These methods have also been applied in solvent extraction chemistry.
Reviews have been presented, e.g., [3,4], on the computational methods-
from purely empirical molecular mechanics (MM) approaches to advanced
ab initio calculations rendered accessible in the form of commercial software
packages: Cerius2 [5], Gaussian 98 [6], ADF [7], Spartan [8], HyperChem [9],
etc. These methods and programs have been widely used for structural
optimization of metal complexes and calculation of their energy of forma-
tion, problems of primary importance in solvent extraction.
Numerical methods have been used in solvent extraction chemistry for
treating experimental data for many years. As shown in Chapter 4 (see
section 4.14.3), modeling of extraction processes in terms of a set of assumed
equilibria with adjustable (best-fit) coefficients was widely used to identify
and characterize the major species formed in these processes. The im-
provement of this approach, directed toward studying systems of greater
diversity and complexity, including corrections for thermodynamic activity
679
680 Narbutt and Czerwiński
effects, is still an object of advanced research [10,11]; this problem is,

however, outside the scope of this chapter. Another direction has been
focused on data mining methods coupled with theoretical combinatorial
chemistry, i.e., the analysis of vast collections of experimental data in order
to establish a relationship between the structure of an ionophore and its
extraction properties, as well as to design the most selective separation
systems for the ions. Attempts to create an information system on solvent
extraction based on large experimental databases and on structure-property
relationships are continuing making possible empirical evaluation of the
distribution ratios of metal ions in given extraction systems [12]. However,
the main goal of computational methods in this field is nowadays the anal-
ysis of the extraction process both on the molecular level and from the
thermodynamic point of view. The knowledge of the species involved, charge
distribution, bond character, energies, extraction mechanism, etc. allows the
interpretation and understanding of the phenomena studied, prediction of
properties of the compounds, and the optimization of solvent extraction
systems, e.g., by designing new extractants (ligands) with improved selec-
tivity. Selective separations of metal ions are due to differences in solvation
energy of metal-containing species in both liquid phases. Therefore, solva-
tion phenomena are crucial when analyzing solvent extraction equilibria.
Equally important is the formation of the metal-containing species, which
can undergo different solvation (hydration) in both phases. These two broad
fields-thermodynamics of solutions and coordination chemistry-are the
main subjects of this chapter.
16.2 COMPUTATIONAL STUDIES ON

STRUCTURES OF METAL COMPLEXES
To characterize the composition and structure of metal complexes formed in
extraction processes (either in the aqueous phase or at the interface), various
experimental methods are used. Theoretical methods become helpful in
complementing the results if the spectroscopic data are not sufficient to fully
describe the structure, if crystals suitable for diffraction studies are not
available, etc. Moreover, the calculations can result in reliable structures of
the complexes or ligands in solution, which are often different from those
observed in the solid state.
16.2.1 Molecular Mechanics Modeling

Significant successes in studying structures of metal complexes have been
achieved by using MM modeling-simple enough to carry out calculations
using computers of low computing power, but not very accurate. The MM
Modeling Solvent Extraction of Metal Ions 681
methods, based upon classical physics, use simple expressions and param-
eters describing interatomic forces in the molecule and also between mole-
cules. To find the optimized structure of a molecule, the total strain energy
accompanying the formation of the molecule must be minimized. This total
strain energy includes the sum of the energies of deformation of all bond
lengths and bond angles; the torsional energies related to barriers to rotation
of the molecule fragments about bonds; and the energies related to non-
bonded interactions between the molecule fragments, i.e., terms describing
van der Waals (Lennard-Jones potential) and long-range electrostatic
(Coulomb potential) interactions [Eq. (16.1)]:
X X X
E¼ Kr (r req )2 þ K ( eq )2 þ Vn [1 þ cos (n)]
bonds angles dihedrals
" #
X Aij Bij qi qj
þ 6þ (16.1)
nonbonded
R12
ij Rij "Rij
The initial set of parameters for a selected force field, including the ideal
bond lengths and bond angles, the related force constants, the effective
barriers to rotation, point charges and van der Waals parameters of the
atoms, and effective dielectric constant of the medium, are obtained from
experimental data and/or QM calculations, and their performance is tested
by comparing the calculated and experimental structures and energies. If
necessary, the procedure is repeated to minimize the errors in the calculated
output [13]. In some MM models a separate term is added, describing
hydrogen-bonding potentials; however, the hydrogen bond energies calcu-
lated in such a way are usually approximate.
Optimizing the geometry of metal complexes in the gas phase neglects
environmental effects (solvent, concentration), important in the analysis of
both structure and stability of the complexes because of their interactions
with neighboring molecules. To study complex formation in solution, some
MM models use the parameterization of the entire force field, based on
experimental data. In this case, the environment is implicitly included in the
model and the environmental effects are isotropic [14]. Another way to
account for the environment is to calculate through-space attractive and
repulsive intermolecular interactions, by the use of the last right-hand-side
term of Eq. (16.1), parameterized for the whole system, and, if related, also
the term describing hydrogen bonding.
MM methods, originally developed for organic compounds, have been
modified to describe metal complexes [13–17] where polarization interac-
tions must be considered and where a variety of geometrical configuration is
observed. This structural diversity reflecting the existence of multiple local
energy minima is due to ligand-dependent effects observed in complexes of
certain transition metal (TM) ions: Jahn-Teller effect, spin crossover [18,19],
and stereochemical activity of valence s-electron pairs in some p-block heavy
metal ions in lower oxidation states [20]. Various types of potential energy
equations, and methods developing the parameters, are used to determine
force fields for coordination compounds. It is generally assumed and con-
firmed for a variety of coordinated ligands that the parameters for metal
complexes are transferable from the organic force fields [13]. Modification
and extension of MM by the effect of ligand field stabilization energy on the
strain energy in complexes of certain TM ions (the LFMM model) makes it
possible automatically to take into account Jahn-Teller distortions, and to
reproduce with a reasonable accuracy both the structures and the associated
d-d transition energies [18,19]. The not very high but often sufficient accu-
racy of MM calculations is recompensed by lesser requirements for com-
putational resources, shorter calculation time, and a possibility to deal with
larger systems.
16.2.2 Quantum Mechanical Calculations

Accurate QM calculations (based on ab initio methods) on molecular sys-
tems as large as metal complexes in condensed phases and/or including
conformational analysis are still limited by the resources of even the most
powerful supercomputers. Solution of HFR equations [1], essential in the ab
initio method, requires the calculation of n4 integrals, where n denotes the
number of basis orbitals. This very large number (in the case of large sys-
tems) causes serious problems and requires simplification of the method.
Thus a significant reduction of the number of electrons (and the basis
orbitals) in a molecule is possible by assigning all its electrons to two sub-
sets: (1) electrons in the inner shells (core electrons), and (2) valence elec-
trons in the atoms. The calculations are then carried out only for the much
lower number of valence electrons; the effect of core electrons being in-
cluded in the effective core potential (ECP), which represents the interaction
of the core electrons with the valence electrons.
Semiempirical methods are widely used, based on zero differential
overlap (ZDO) approximations which assume that the products of two
different basis functions for the same electron, related to different atoms, are
equal to zero [21]. The use of semiempirical methods, like MNDO, ZINDO,
etc., reduces the calculations to about n2 integrals. This approach, however,
causes certain errors that should be compensated by assigning empirical
parameters to the integrals. The limited sets of parameters available, in
particular for transition metals, make the semiempirical methods of limited
use. Moreover, for TM systems the self-consistent field (SCF) procedures
are hardly convergent because atoms with partly filled d shells have many
energetically close electronic states, in contrast to systems consisting of s-

and p-block atoms only.
Because of these difficulties, great interest arose in the last decade in
methods free of such limitations, based on the density functional theory
(DFT). The DFT equations contain terms that evaluate-already at the
SCF level-a significant amount (ca. 70%) of the correlation energy. On the
other hand, very accurate DFT methods require calculation of much fewer
integrals (n3) than ab initio, which is why they have been widely used in
theoretical studies of large systems. The DFT [2] is based upon Hohenberg-
Kohn (HK) theorems, which legitimize the use of electron density as a basis
variable [22].
In 1976 Warshel and Levitt introduced the idea of a hybrid QM/MM
method [23] that treated a small part of a system (e.g., the solute) using a
quantum mechanical representation, while the rest of the system, which did
not need such a detailed description (e.g., the solvent) was represented by an
empirical force field. These hybrid methods, in particular the empirical
valence bond approach, were then used to study a wide variety of reactions
in solution. The combined QM/MM methods use the MM method with the
potential calculated ab initio [24].
All these methods have found applications in theoretical considera-
tions of numerous problems more or less directly related to solvent
extraction. The MM calculated structures and strain energies of cobalt(III)
amino acid complexes have been related to the experimental distribution of
isomers, their thermodynamic stability, and some kinetic data connected
with transition state energies [15]. The influence of steric strain upon chelate
stability, the preference of metal ions for ligands forming five- and six-
membered chelate rings, the conformational isomerism of macrocyclic
ligands, and the size-match selectivity were analyzed [16] as well as the
relation between ligand structures, coordination stereochemistry, and the
thermodynamic properties of TM complexes [17].
16.2.3 Structures of Ligands and Complexes,

Ligand Design
Extensive MM calculations on metal complexes, supplementing extraction
experiments, helped to explain how the structure of various crown ether
ligands affects their binding and extraction ability toward alkali and alkaline
earth metal ions, contributing to the design of new ligands [25,26] and
synergistic systems [27] of improved selectivity. Complementarity of bi- and
polydentate ligands to the metal ion, crucial for ligand design, is controlled
by numerous factors including appropriate donor atoms (HSAB principle
and the nature of metal-ligand bond), the number of donor atoms that
satisfy the coordination and geometric requirements of the ion, the geom-
etry of ligands (e.g., bite size), and orientation of the donor atoms [28].
More advanced semiempirical molecular orbital methods have also
been used in this respect in modeling, e.g., the structure of a diphosphonium
extractant in the gas phase, and then the percentage extraction of zinc ion-
pair complexes was correlated with the calculated energy of association of
the ion pairs [29]. Semiempirical SCF calculations, used to study structure,
conformational changes and hydration of hydroxyoximes as extractants of
copper, appeared helpful in interpreting their interfacial activity and the rate
of extraction [30]. Similar (PM3, ZINDO) methods were also used to model
the structure of some commercial extractants (pyridine dicarboxylates,
pyridyloctanoates, b-diketones, hydroxyoximes), as well as the effects of
their hydration and association with modifiers (alcohols, b-diketones) on
their thermodynamic and interfacial activity [31–33]. In addition, the
structure of copper complexes with these extractants was calculated [32].
Semiempirical and ab initio calculations were used to interpret experi-
mental data and to find structure-complexation relationships in the extrac-
tion of transition metals and lanthanides [34–38]. The calculations contribute
to understanding the factors that affect the strength of metal–ligand bond-
ing, and to designing new ligands. For example, the extraction ability and
selectivity of some b-diketonate ligands toward lanthanides is shown to be a
function of ligand structure (bite size) and the energy of intramolecular
hydrogen bonds [37,38]. Semiempirical PM3-tm and DFT B3LYP calcula-
tions in the gas phase made it possible to determine the experimentally
unavailable molecular structure of some heterometallic TM complexes
extracted by trioctylphosphine oxide into hexane [39].
Semiempirical QM calculations of the geometry, charge distribution,
bond orders, dipole moments and HOMO energy for a series of monoamide
extractants in the gas phase, carried out at the Hartree-Fock level, were
supplemented by MM calculations on uranyl complexes with these ligands
[40,41]. The observed qualitative relationships between the calculated values
and the experimental distribution ratios of the metal have emphasized
the importance of electron density on the coordinating atoms or groups,
and steric effects in the ligands. Ligand basicity is a factor that affects the
complex stability. Semiempirical calculations appeared insufficiently accu-
rate to calculate basicities of several malonamides, the extractants of tri-
valent lanthanides and actinides; therefore advanced ab initio methods
were used, which brought about a good agreement between the calculated
(relative) values and the experimental data [42]. Conformational ab initio
analysis of the protonated terpyridine extractant used for lanthanide/
actinide separation (ion-pair extraction from strongly acidic solution) has
confirmed stabilization of the trans,trans form of the terpyridine cation in
aqueous solution by the formation of intermolecular hydrogen bonds to

solvent and/or to accompanying anions [43].
Gas phase QM calculations at the HF level, tested by DFT B3LYP
studies and completed by molecular dynamic simulations (see section 16.3),
made it possible to estimate the stability of lanthanide complexes with
various neutral mono- and bidentate ligands bearing amide and phosphoryl
groups, which were considered potential extractants of lanthanide and
actinide cations from nuclear waste solutions [44]. It has been demonstrated
that the stability of the complexes depends on the basicity of the ligand
binding sites, on the ligand structure (i.e., on the chelate effect and the size of
the chelate ring), on the hardness of the lanthanide cations, and on the
properties of the counterions. Ab initio calculations on various carbonyl
and phosphoryl compounds have allowed the determination of factors
influencing the selectivity of Am3+–Eu3+ separations in extraction by dial-
kyldithiophosphinic acid (CYANEX 301) with the abovementioned species
as coextractants [45].
16.2.4 Complexes of Heavy Metals

QM studies on the compounds of heavy metals, actinides in particular,
require the incorporation of relativistic effects that disturb the periodicity of
physical and chemical properties observed for lighter elements [46]. Accurate
relativistic calculations on compounds containing heavy atoms can be per-
formed only for very small molecules; the calculations on larger species
require a certain approximation. This can be achieved not only by the
approximate solution of the Dirac-Schrödinger equation adjusted to a many-
electron system, but also in a simpler way through the use of quasi-relativistic
or fully relativistic effective core potentials (RECP). According to this ap-
proximation, the inner shells of the atom (core) are omitted and an additional
potential energy term replaces the core electrons in the Hamiltonian (see
section 16.2.2). The application of relativistic DFT methods, including new
RECPs to molecular systems containing actinides, has been reviewed [47].
New methods for calculating RECPs are being tested [48]. This problem is
not only theoretical but also has practical significance, because partitioning
of trivalent actinides and lanthanides from liquid nuclear waste is a ‘‘hot’’
research task due to the necessity to isolate long-lived highly radiotoxic
minor actinides (mainly Am and Cm) for further treatment [45].
The novel RECP integrals have been implemented inside the quantum
chemistry DFT-based program MAGIC developed for studying large
molecules containing heavy atoms, in order to design ligands with better
selectivity for extractive separation of actinides from aqueous nuclear waste
solutions [49,50]. The program also calculates the electrostatic solvent effects
exerted on complex molecules placed in a cavity embedded in a dielectric

continuum, which represents the bulk solvent (shown later). It has been
shown that the calculated formation energies of fully optimized thenoyl-
trifluoroacetonate complexes of plutonium and uranium in different oxi-
dation states correspond to the experimental extraction data [51]. The DFT
calculations on hydration of uranyl and plutonyl ions in the gas and aque-
ous phases have shown that the short-range interactions between the ions
and their closest water molecules are very strong and involve an appreciable
amount of charge transfer that could not be included in simple solvent
cavity models. The MO2þ 2 5H2 O hydrates appeared to be the most stable,
their binding energy being, however, significantly greater in vacuo than in
the aqueous solution [52]. This shows how important solvent effects are in
the calculations of the formation energy of metal complex with lipophilic
ligands, which are then extracted to the organic phase.
A fully relativistic, self-consistent field method based on the numerical
solution of the Hartree-Fock equations with the approximation of local
exchange instead of the HF nonlocal exchange potential, and with the self-
consistent charge approximation, was used to calculate the electronic
structure of anionic chloride and oxychloride complexes of Group 5 metals
including the heaviest, dubnium (element 105). The calculation results
explain the experimental extraction behavior of the metal ions in strongly
acidic solutions. Based on the radii of the oxychloride complexes and using
the Born model [Eq. (16.12)] to describe the extraction equilibria of the ion
pairs, it was possible to predict the position of dubnium in the series of
metals extracted to an organic phase [53]. Similar calculations for Group 6
metal ions lead to predicting the relative extraction behavior of seaborgium
(element 106) within the group. It has been concluded that the difference
between geometrical configurations and electron density distribution of
Group 5 MO2þ 2þ
2 ions and UO2 , their actinide pseudoanalog, would result in
different extraction behavior of Sg and U complexes [54]. Correct prediction
of extraction properties is particularly important in studying chemistry of
the heaviest elements, obtained on the scale of ‘‘one atom at a time,’’
because solvent extraction is one of few experimental methods that can be
used to study these ‘‘exotic’’ elements [55].
16.2.5 Solvent Effects

Solvent effects can be incorporated into two kinds of solvation models,
either those that consider each solvent molecule as an individual molecular
species (explicit models), or those that deal with the averaged effect of the
solvent molecules through use of a coarse-grained description of solvent
(e.g., dielectric models, implicit solvent models, etc.).
The explicit solvation models that consider solute molecules in a dis-

crete environment of a number of individual solvent molecules can hardly be
used by the available QM methods and hardware because such systems are
too large. QM calculations involving solvent effects can be performed using
implicit solvation models based on the continuum solvation theory, as
reviewed, for example, by Cramer and Truhlar [56]. Instead of a discrete
solvent medium consisting of a large number of solvent molecules, a con-
tinuous medium is considered. Its macroscopic properties, imposed by the
dielectric constant of the solvent modeled, correspond to the properties of
the bulk solvent. The medium is, however, homogenous, isotropic, and
nonpolarizable, so it can well be replaced by an electric field. Into this field a
solute molecule is placed, and its interactions with the field are studied. The
substitution of this field for a large number of solvent molecules drastically
decreases the number of orbitals in the system, i.e., the number of integrals
to be calculated. The polarizing action of this field on the solute molecule
modifies its charge distribution, dipole moment, energy of electrons, and to
some extent its geometry. The modified solute affects in turn the field
parameters, etc.; this is the principle of the self-consistent reaction field
(SCRF) approach. Most current works on the implicit solvation models
treat the solute either as quantum mechanically flexible, i.e., including all the
previously mentioned effects, or as quantum mechanically rigid, including
only the electronic relaxation.
The implicit solvation models do not take into account, however,
interactions of the solute with solvent molecules that lead to formation of
the first solvation shell around the solute. This shell may be defined either in
terms of an integer number of solvent molecules, based on the solute–sol-
vent distance or interaction energy, or in terms of distribution functions that
result in an averaged noninteger number of solvent molecules. The inter-
action of bulk solvent with such a ‘‘supermolecule’’ (the solute with its first
solvation shell) can be expressed by the terms due to cavity formation in the
solvent, to electrostatic interactions, and, in certain models, also to disper-
sion and repulsion contributions. The presence of explicit solvent molecules
in the first solvation shell requires the knowledge of their number and
orientations after the supermolecule has been placed in the continuous
medium. This can be done by the QM/MM methods based on Monte Carlo
or molecular dynamics simulations, where the solute molecule, originally
modeled by QM, is afterward treated as rigid and nonpolarizable. The
introduction of the solute polarizability is possible but with a high com-
putational cost [57]. Molecular modeling by molecular dynamics (MD)
simulations are discussed in the following section. Solvent effects can be
taken into account in MM calculations by parameterization of the force
field, based on solution data [14,56].
QM calculations provide an accurate way to treat strong, long-range

electrostatic forces that dominate many solvation phenomena. The errors
due to the use of the approximate continuum solvation models can be small
enough to allow the quantitative treatment of the solute behavior; therefore,
this approach is widely used also for evaluating solvent extraction equilibria
of organic molecules [56].
16.3 MOLECULAR SIMULATIONS

16.3.1 Monte Carlo and Molecular Dynamics Methods
Molecular dynamics simulations, which afford possibilities for microscopic
insight into various chemical processes at the molecular level, play a partic-
ular role in solvent extraction, where they are used to model both complex
formation in solution and its distribution between two liquid phases [58].
The advantage of MD consists in not being restricted to harmonic motion
of a molecule (a set of atoms) about local minima. This allows the molecule
under study to cross energy maxima and to adopt other stable configura-
tions. Successive configurations of the system are generated by implement-
ing Newton’s laws of motion. The solution of these equations represents the
evolution of positions of the molecule in time, i.e., its dynamic ‘‘trajectory,’’
which can be used then to study time-dependent properties of the molecule,
such as diffusion or reaction progress [59]. On the contrary, Monte Carlo
(MC) techniques randomly move atoms (molecules) in such a way that
alternative configurations are generated, which do not represent a time-
dependent trajectory but correspond to a given energy [60].
Software packages such as AMBER [61] are widely used in MD
simulations with the classical molecular force field where the potential
energy is represented by Eq. (16.1). If the simulations relate to a solution,
the intermolecular interactions are modeled using the last term, which
describes nonbonded interactions (Coulomb and Lennard-Jones). In order
to correctly treat long-range interactions that may become cut off to some
extent at the boundaries of a finite size system, corrections are being
introduced, e.g., by using the Ewald summation method for electrostatic
interactions, or techniques allowing the user to avoid a discontinuity in
potential energy and forces near the cutoff distance (shifted potentials,
switching functions) [59,60]. The Ewald approach applying periodic
boundary conditions makes it possible to obtain correct free energies of
hydration from molecular calculations by extrapolation of relatively small
simulation systems to infinite size [62].
An important advantage of the simulation methods consists in their
explicit treatment of solvent effects as the effects of a set of individual
molecules. To calculate the free energy of hydration of small hydrophobic

solutes, the simulations have been successfully used, based on the energy of
cavity formation in bulk water [62–65] (see section 16.4.2). The results
correspond well with experiment in spite of simplifications of the model
used. The use of a more sophisticated model that incorporates information
on the locally tetrahedral structure of water improves the results [65,66].
Most applications, however, require calculations for large amphiphilic
molecules, with both polar and nonpolar regions in close vicinity. In this
respect, the water-structuring effects of carbohydrates were studied [67].
Such an approach may appear useful in solvent extraction studies for the
analysis of interactions of metal chelates in the aqueous phase, where
competition between hydrophobic hydration and outer-sphere hydration
determines solvent extraction equilibria of coordinatively saturated metal
complexes (see section 16.4.4).
16.3.2 Solvent Extraction Systems and Mechanisms

of Extraction Processes
MD simulations are used to study solvent extraction processes of metal ions
in two-phase systems modeled by two adjacent assemblies (boxes) consisting
of hundreds of molecules of water (one) and of an immiscible organic solvent
(the other). The contact surface of these two boxes represents the interface,
which locally is molecularly sharp, but some mutual penetration of both
kinds of solvents is observed [68,69]. MD simulations performed at the water/
chloroform interface for a period of a few hundred picoseconds directly
model, at the molecular level, the phenomena accompanying extraction of
metal ions. It has been shown that large amphiphilic ionophore ligands-
functionalized calix[4]arenes, phosphoryl-containing podands, cryptands-
both free and complexed, adsorb at the interface (Fig. 16.1). This is due to the
decrease in the free energy of the system when the hydrophilic parts of the
solute interact with water, while the lipophilic parts interact with the organic
solvent [70,71].
MD simulations coupled with free energy perturbation (FEP) calcu-
lations allow the calculation of the relative binding affinities of two different
metal ions by the same ligand or of the same metal ion by two different
ligands. The method consists in stepwise mutation of one solute to the other.
Mutation in two solvents, e.g., water and chloroform, with the use of a
thermodynamic cycle makes it possible to compute the changes in the free
energy of solvation of the two solutes [58]. The relative free energies of
transfer for two metal complexes can be calculated in this way (see section
16.4.2), thus enabling the determination of the extraction selectivity of cer-
tain extractants toward different metal ions. For example, such theoretical
Fig. 16.1 Complex of Cs+ Pic with calix-[4]-bis-crown-6 at the chloroform (left) and
water (right) interface. A snapshot at 500 ps of MD. (A. Varnek and G. Wipff, unpub-
lished).
calculations not only confirm the experimental fact that calix[4]-crown-6

ligands selectively extract Cs+ over Na+ ions from aqueous solutions, but
also reliably explain this selectivity as the effect of the higher dehydration
energy of the latter, required for complex formation [70]. This demonstrates
again that solvent effects really play an important role in complexation
selectivity. The role of the counterion (picrate, chloride) in the interfacial
behavior of the crown [70], cryptand, and phosphoryl podand [71] complexes
has also been pointed out. The structures of the complexes with alkaline
earth and uranyl ions have been calculated [72].
MD simulations at the water/organic liquid interfaces give a powerful

tool for studying mechanisms of the solvent extraction of metal ions, in
particular described by the ion-pair model with complex formation at the
interface. Snapshots of solutes taken at picosecond intervals allow us to
observe the progress of the reaction, steady states and intermediate species
(if any), and finally dynamic equilibrium. In many cases, this theoretical
approach is irreplaceable as interfacial phenomena are very difficult to study
experimentally [73] (see Chapter 5). MD simulations are also carried out to
model extraction processes successfully without taking solvent effects into
account. For example, gas phase MD studies on the mechanism of insertion
of monovalent metal ions into calix[4]arene tubes have shown how the ions
enter and migrate along the tube, and show the reason for the selectivity of
the host molecule for the given ions [74].
It must be pointed out that in studies on the complexation of metal
ions by extractants, ion solvation phenomena are of primary importance
because upon complex formation the solvent molecules from the primary
coordination sphere are replaced by binding sites of the ligands. MD
simulations that poorly describe charge transfer and polarization effects are
sometimes coupled with QM methods. For example, lanthanide solvation in
acetonitrile solution was studied by both QM and MD simulations. The
former, taking into account charge transfer and polarization, which are so
important for this system, gives the structure, coordination numbers, and
charge distribution on the Ln3+ solvates. The latter gives a remarkable
dynamic feature of the primary solvation sphere of the ions, where the
lifetime of acetonitrile molecules in the complex increases from Yb3+ to
La3+, inversely to the energy of cation–solvent interaction and unlike the
aqueous phase behavior. The ion-binding selectivity by the pyridinedi-
carboxamide ligand, studied by MD-FEP simulations, has appeared to be
determined to a great extent by the change in the free energy of cation
solvation [75].
The limitations of MD simulations stem from the approximate char-
acter of MM methods; however, their relatively low computing require-
ments allow the study of large systems consisting of even 15,000–20,000
atoms. The size of such systems is, nevertheless, too small to take into
account formation of aggregates (micelles, vesicles, etc.) that may play an
important role in the mechanism of extraction in some special systems. This
relatively small size of simulating boxes requires corrections to treat long-
range interactions correctly (see section 16.3.1). Also, the force fields used in
MD methods typically do not treat nonadditive effects correctly. The
necessity of very short time simulation steps (in order not to lose important
energy minima) makes the calculations long lasting.
16.4 EQUILIBRIA IN SOLVENT EXTRACTION OF

METAL IONS
The results of QM calculations (except some semiempirical parameterized
methods, e.g., AM1 and PM3) on the geometry of metal complexes (see
section 16.2) are much more accurate and much better correlated with
experiments than the calculated energy of intermolecular interactions, e.g.,
the energy of complex formation. That is because the latter is calculated as
the difference between large total energies of products and substrates, and
hence constitutes only a small fraction of the total energies. On the other
hand, relative calculations for analogous complexes (formed by, e.g., dif-
ferent metals with the same ligand or different ligands with the same metal)
are much more accurate [3]. To calculate Gibbs free energy, the entropy
term must be calculated and deducted from the energy [Eq. (3.13)], which
increases the error. Calculations on systems consisting of many molecules
are even more difficult; therefore new methodological approaches are sought
to overcome the limitations. Moreover, in relation to the solvent extraction
of metal ions, the current QM calculations of the energy are usually frag-
mentary and do not correctly reflect the complicated process. This problem
is discussed next using the example of a commonly accepted model of
extraction.
16.4.1 Modeling of Solvent Extraction of Metal Ions

The process of transferring a metal ion Mn+ from an aqueous solution to a
water-immiscible organic solvent by an acidic chelating extractant HL that
forms a lipophilic neutral complex MLn with the ion can be represented as:
Mnþ þ nHLorg $ MLn;org þ nHþ (16.2)
Both in this and in the subsequent equations the subscript org relates to a
species in the organic phase (or to its concentration), while the lack of such a
subscript denotes a species in the aqueous phase.
It has already been shown in Chapter 4 (section 4.2.1) that from the
thermodynamic point of view the process described by Eq. (16.2) can be
modeled by the sum of its partial processes (extraction steps), irrespective of
whether they really proceed or not. That is because Gibbs free energy is the
function of state and its total change does not depend on the reaction path.
According to the complex formation–partition model [76], one can distin-
guish two main steps in extraction of metal ions:
1. Formation of a neutral complex by a given metal ion in the aqueous
phase

MðH2 OÞnþ
k þ nL $ MLn ðH2 OÞkm þ mH2 O (16.3)
with the stability constant, bn, of the complex, defined by Eq. (3.5). The
concentration of ‘‘free’’ ligand in the anionic form, [L], depends on the
distribution of the extractant HL between the phases and its dissocia-
tion in the aqueous phase:
HL $ HLorg (16.4)
HL $ Hþ þ L (16.5)
with the partition constant (KD,HL) and the dissociation constant (Ka) of
the extractant, defined by Eqs. (4.17) and (4.18), respectively.
2. Partition, i.e., transfer of the complex from the aqueous to the organic
phase, accompanied [77] by at least partial dehydration
MLn ðH2 OÞkm $ MLn;org þ ðk mÞH2 O (16.6)
with the equilibrium constant equal to the partition constant, KDM, of
the complex, also called the distribution constant, and defined by Eq.
(4.1).
Taking into account all these relationships, one can express the
extraction constant [the equilibrium constant in Eq. (16.2)] as:
Kex ¼ n KDM ðKa Þn ðKD;HL Þn (16.7)
When extraction of coordinatively unsaturated chelates proceeds in the
presence of a coextractant (synergist), i.e., a lipophilic neutral electron
donor, B, which additionally coordinates the metal ion, a third step
must be added:
3. Adduct formation in the organic phase:
MLn;org þ pBorg $ MLn Bp;org (16.8)
with the equilibrium constant (the stability constant, bn,p, of the adduct)
given by:
½MLn Bp org
n;p ¼ (16.9)
½MLn org ½Bporg
The role of synergist is often played by undissociated HL molecules (self-

adduct formation) when present in the system in a sufficient concentra-
tion.
16.4.2 Partition of Hydrophobic and

Amphiphilic Solutes
The equilibrium distribution of a solute B between water and an organic
solvent is described by the partition constant or by its function, the free
energy of partition equal to the difference between the free energies of

solvation and hydration:
G0p ¼ RT ln KD;B ¼ G0solv G0hydr (16.10)
The free energy of solvation (hydration) is defined as the free energy of
transfer of the molecule from an ideal gas phase into a given solvent.
Insertion of a solute molecule into a solvent requires the formation of a
cavity of appropriate size and shape to accommodate the solute molecule.
When using explicit solvation models we can relate the free energy of cavity
formation in a solvent, e.g., water, to the excess free energy of equilibrium
fluctuations of bulk water density, in particular, of excluding all the nw bulk
water molecules from a volume of a given size and shape which correspond
to the size and shape of the solute molecule
Gex ¼ RT ln½ pðnw ¼ 0Þ (16.11)
where p(nw =0) denotes the probability of observing a solute-sized cavity in
the bulk liquid [78]. The probability can be calculated either by Monte Carlo
simulations (section 16.3.1) or from the information theory approach (sec-
tion 16.5). Apart from cavity formation in water, there is a second com-
ponent of the free energy of hydration, which originates in van der Waals
interactions [typically approximated by the Lennard-Jones term in Eq.
(16.1)] of the hydrophobic solute with surrounding water molecules. The
free energy of hydrophobic hydration of numerous solutes-small neutral
molecules-was calculated in this way [64,65]. For ionic solutes, the
cumulate expansion method has been used to calculate free energies of
charging [62,63]. A similar approach, based on the calculation of free energy
of hydrophobic hydration in salt solutions, was used to study the effect of
concentration of different salts (ion pairs) in aqueous solution on the chem-
ical potential of a nonpolar solute. Molecular dynamic simulations of salt
solutions resulted in reasonable values of Setchenow’s salting-out and
salting-in coefficients [79].
The hydrophobic hydration has been considered by molecular theories
[80,81] in terms of water structuring. The challenge to the theories of
hydrophobic effects is that the real solutes, in particular those dealt with in
solvent extraction, are never entirely nonpolar. It has been recognized that
hydrogen bonding to amphiphilic solutes in aqueous solutions disturbs the
hydrophobic water structure around the solute. This is probably responsible
for the observed nonadditivity of the effects of hydrophobic and hydrophilic
fragments of numerous solutes, e.g., acetals and diethers, on the thermo-
dynamic functions of their distribution between water and organic solvents,
which strongly depend on the position of the hydrophilic fragments in the
molecule [82].
The free energy of solvation of a solute can also be calculated using

implicit solvation models (section 16.2.5), after parameterization. Carrying
out such calculations for both water and an organic solvent allows the
prediction of the partitioning behavior of a solute between these two liquid
phases. Various continuous solvation models were used to evaluate partition
constants of a large number of simple organic solutes in numerous two-
phase systems [56]. The calculations were performed using various methods
of calculation at different levels of theory, from MM [83] to HF-SCRF [84].
The calculations of the relative values of partition constants of two similar
organic species in water/organic solvent systems were also made by MD
simulations using statistical perturbation theory (section 16.3.2). An im-
portant advantage of this method consists in providing molecular-level
details of the process and an insight into the structure of the species studied,
and in making it possible to evaluate the mechanism of the process. How-
ever, given correct parameterization, the results of continuous solvation
calculations seem to be more accurate than those obtained from MD-FEP
simulations [85].
16.4.3 Partition of Ion Pairs

Extraction processes that proceed according to the model of ion-pair
extraction are described by a formalism different from that presented in
section 16.4.2, and are based on partition of single ions and their association
in the organic phase [76] (see also section 2.6). The Born equation has been
widely used to describe the transfer of an ion of the charge q and radius r
from vacuum to the liquid (water) of the dielectric constant e:

q2 1
G ¼ 1 (16.12)
2r "
Numerous QM and MM calculations refer to this model. The free energy of

hydration of the tetramethylammonium cation is equal to the sum of
free energy of cavity formation in water [Eq. (16.11)], calculated by Monte
Carlo simulations, and free energy of ion–water interactions [Coulomb and
Lennard-Jones, Eq. (16.1)]. Small, in particular highly charged, ions
strongly interact with water; therefore their free energy of hydration is large
and negative [63]. Because the respective free energy of solvation in a given
organic solvent (smaller e) is much less negative, the free energy of transfer
of a small single ion from water to the organic solvent is positive and resists
the transfer. The driving force in this process is the negative free energy
that accompanies breaking the hydrophobic hydration of large hydrocar-
bon fragments of, for example, tetraalkylammonium ions (tetrapropyl-
and larger). Because of the principle of electroneutrality, the hydrophobic
ions drag hydrophilic counterions into the organic phase in equivalent

amounts; hence the experimental partition constant of an ion depends on
the properties of the counterion. Concentration-dependent association of
ions in an organic solvent of low dielectric constant shifts the equilibrium
of the ion transfer toward the organic phase. Another important factor
is interactions of ions with the components of any liquid phase. For
example, the free energy of transfer of a complex polyanion from water
to chloroform has been calculated (QM) based on the charge transfer
theory [86].
The fact that ions of the same charge are never extracted alone but
together with the equivalent amounts of oppositely charged ions, makes direct
experimental determinations of single-ion partition constants impossible. In
order to overcome this difficulty, the extrathermodynamic assumption was
widely used that the partition constants of complex ions of the same structure
but of opposite charge (e.g., the tetraphenylarsonium cation and the tetra-
phenylborate anion) are equal to each other [87]. Monte Carlo simulations of
hydration of tetramethylammonium ion as a function of the ion charge have
shown that the quadratic charge dependence is fulfilled as expected from the
Born model, but that hydration is asymmetric in respect to the ion charge.
Due to the structure of liquid water, well modeled in the calculations,
negative ions are more favorably hydrated than positive ones [63]. This
observation affords the possibility of modifying the extrathermodynamic
assumption, although it is not clear how this computational result pertains to
the very large tetraphenyl ions.
The combination of computational methods for calculations of the
free energy of single-ion partition and the free energy of association [29] of
ions in organic solvents can be helpful in further developing the partition–
association model of extraction. A hypothetical thermodynamic cycle show-
ing the role of ligand strain in ion-pair extraction has been presented for the
example of lithium salts extraction by crown ethers [26]. The relative free
energies of extraction of ion pairs in water/organic solvent systems can be
calculated by using the MD-FEP simulations, as exemplified by several
mono- and divalent cations with polydentate ionophores; it has been found
that the calculated extraction selectivity corresponds to the experimental
data [70].
16.4.4 Partition of Metal Chelates

In our opinion, the differences in partition constants of neutral complexes of
metal ions of the same charge with the same number of the same ligands,
extracted to inert organic solvents, are almost entirely due to the differences
in hydration of these complexes in the aqueous phase or, more precisely, to
the differences in the free energy of the dehydration accompanying the

transfer of these complexes from the aqueous to the organic phase [77]. It
must be pointed out that changes in hydration of metal complexes will be
considered afterward not in terms of an integer number of water molecules
attached or released, but rather in terms of free energy released upon
transferring the complex from one liquid phase to the other. This can also be
explained in terms of an averaged picture of dynamic equilibria between
various differently hydrated forms of the same complex.
The hydrophobicity of metal complexes with organic ligands promotes
their easy transfer from the aqueous to the organic phase. According to
molecular theories, water structuring around nonpolar hydrocarbon regions
of the solute (hydrophobic hydration) is accompanied by negative entropy
and positive free energy changes, which depend on the volume or rather on
the hydrophobic surface of the solute molecule [80,81]. In another approach,
it is assumed that the free energy of cavity formation is more positive in
water than in common organic solvents (see section 16.4.2). Both models
predict a negative free energy change of removal of hydrophobic solutes
from water, which means their spontaneous transfer to organic solvents. In
contrast to hydrophobic hydration, specific hydration of metal complexes in
the aqueous phase, either by water coordination to the central metal ion or
by hydrogen bonding to the coordinated ligands, decreases the partition
constant of the complex.
16.4.4.1 Inner-Sphere Hydration
The main contribution to hydrophilicity of coordinatively unsaturated
metal complexes results from direct coordination of water molecules to the
central metal ion, i.e., inner-sphere hydration of the complexes. The extent
of the inner-sphere hydration depends on the properties of both the ion and
ligand. For example, the smaller the ion within the Ln(III) series, the
stronger its complex (e.g., tris-acetylacetonate) and the weaker its hydration
in the inner sphere. This means that less energy is required to dehydrate the
ion (and the complex) in the process of transferring from water to an
organic phase. This conclusion may appear surprising in the light of increas-
ing hydration energy across the series of Ln3+ ions [88] (see also Table 2.4).
It may, however, be easily understood in terms of competition for the metal
ion between ligands as, for example, between the strongly bonded acetyl-
acetonate and rather weakly bonded water. The stronger the bonds with
the chelating ligand [strengthened across the Ln(III) series], the weaker the
bonds with the remaining water molecules. The effect of the strength of the
metal–ligand bond on the inner-sphere hydration of the complex and,
therefore, on its partition constant can also be illustrated by an example of
zinc(II) extraction with chelating extractants belonging to the homologous
series of b-diketones. Minute differences in the basicity of the ligands result

in enormous changes in partition constants of the complexes. This has been
explained in terms of decreasing coordinative unsaturation of the complexes
with increasing ligand basicity, and the resulting decrease in their inner-
sphere hydration [89].
In the calculations of the energy of hydration of metal complexes in
the inner coordination sphere, one must consider hydrogen bond formation
between the first-shell water molecules and those in bulk water, which leads
to chains of hydrogen-bonded water molecules. Such hydrogen-bonded
chains of ethanol molecules attached to the central metal ion have been
found as a result of DFT B3LYP calculations on ethanol adducts to nickel
acetylacetonate, where the calculated energy of hydrogen bonds correlated
well with experimental data [90].
16.4.4.2 Outer-Sphere Hydration
Not only differences in the inner-sphere hydration of the complexes con-
tribute to differences in partition constants. Even strongly lipophilic metal
chelates with oxygen-donor organic ligands show some hydrophilic prop-
erties, because not all electron pairs on the donor atoms are engaged in
complex formation. Because of this, the solvent (water) molecules easily
form hydrogen bonds with the donor atoms of coordinated ligands [91].
This outer-sphere hydration in the aqueous phase not only makes the
complexes much more hydrophilic than hydrocarbons of similar molar
volume, but also differentiates the complexes in respect to partition. This is
illustrated by examples of the partition of neutral coordinatively saturated
acetylacetonates of the 3d metal(III) ions (from Sc to Co) [92] and 4d–6f
metal(IV) ions (from Th to Zr) [93]. In both series, the partition constants
decrease with decreasing radius of the metal ion by over two orders of
magnitude. A similar relationship has been observed in the series of 2,4-
hexanedionates of 3d metal(III) ions [92]. According to our recent DFT
calculations [94] based on the self-consistent isodensity polarized continuum
model (SCI-PCM) [95], these differences are due to solvent effects exerted
on the supermolecule (hydrate). The hydrate of a more covalently bonded
chelate has a larger dipole moment; thus it more strongly interacts with bulk
water.
It is worth mentioning that organic solvents with an acidic character
also form hydrogen bonds with metal chelates. For example, chloroform
and chlorophenols solvate acetylacetonates of trivalent metal ions with
the same preference as water [96,97]. In the case of extraction of metal
chelates into such solvents, the negative effect of their outer-sphere hydra-
tion is cancelled to a large extent by the hydrogen bonding in the organic
phase.
16.4.4.3 Adduct Formation

Coordinatively unsaturated metal chelates can be additionally coordinated
in the inner sphere of the metal ion not only by water but also by other
electron donor molecules. The presence in the extraction system of such
lipophilic neutral species which form molecular adducts with the chelate in
the organic phase [Eq. (16.8)] shifts the equilibrium of extraction [Eq. (16.2)]
to the right. These coextractants, which alone poorly extract the metal ions,
synergistically enhance the extraction by chelating ligands; therefore they
are often called synergists. Synergistic extraction of metal ions by adduct
(and self-adduct) formation to a significant degree compensates the effect of
their inner-sphere hydration.
The authors’ DFT calculations [98] have shown that the energy of
adduct formation between metal(III) tropolonates and trioctylphosphine
oxide (TOPO) depends not only on the size of the central metal but also on
the electronic structure of the ion, in particular on the availability of empty d
orbitals to form bonding molecular orbitals in the chelate. The bonding
MOs in the adducts are easily formed with the participation of the unoc-
cupied (n1)d orbitals of metal(III) ions of Group 3. The coordination bond
between the TOPO molecule and the yttrium ion in the chelate appears to be
much more complex (Fig. 16.2) than an expected simple s bond. On the
other hand, the energy of virtual nd orbitals of the Group 13 ions is too high
to allow them to participate in bonding. The adducts (CN 7) of the p-block
metal ions are therefore hypervalent compounds, of the bond order less
than one [98]. This explains well the experimental observation that the
TOPO adducts with tropolonates of Group 3 metals (Sc and Y) are much
stronger than those of Group 13 metals (In and Tl), irrespective of the ionic
radii. The contribution from the deformation energy into the energy of
fully optimized structures of the adducts confirms the important role in
the synergism of the ligand bite size, concluded in an earlier experimental
work [99].
16.4.5 Selectivity of Metal Ion Separations

The selectivity of extractive separations of metal ions Mn+ is commonly
described by the separation factor (SF), defined as the ratio of the distribu-
tion ratios, DM [Eq. (4.3)], of the ions in the same system. In the case of
two metal ions An+ and Zn+ extracted from the same aqueous solution by
an acidic bidentate extractant HL, with no synergist in the system, one
obtains:
P
n;A KDM;A nþ1
i¼0 i;Z ½L
i
SFA=Z ¼ Pnþ1 i
(16.13)
n;Z KDM;Z i¼0 i;A ½L
where bi denotes the stability constant of MLi, M = A or Z; KDM =

[MLn]org/[Mln]aq denotes the partition constant of the neutral MLn complex;
and [L] the equilibrium concentration of the anion L in the aqueous phase.
For simplicity the charges of the species have been omitted.
The often used formulation of SF as the ratio of extraction constants
of the metal ions [26,36–38] is justified only under special conditions, i.e.,
when the content of [MLi]ni complexes in the aqueous phase as compared
to Mn+ is insignificant (at low pH) and there is no adduct formation. The
only (practically) form of the metal in the aqueous phase is then the un-
complexed ionic form, [M(H2O)m]n+, while in the organic phase only the
neutral complex [MLn] is present. Then Eq. (16.13) simplifies to the ratio of
the respective bn and KDM values:
Kex;A n;A KDM;A
SFA=Z ¼ ¼ (16.14)
Kex;Z n;Z KDM;Z
However, even this simplified formula does not justify the use of the ratio
of stability constants of the extracted complexes as the only measure of se-
lectivity of extractive separations. Such a widely used approach is obviously
based on an implicit assumption that the partition constants of neutral
complexes MLn of similar metal ions are similar, so that their ratio should
be close to unity. This is, however, an oversimplification because we have
shown that the KDM values significantly differ even in a series of coordi-
natively saturated complexes of similar metals [92,93]. Still stronger differ-
ences in the KDM values have been observed in the series of lanthanide
acetylacetonates, due to different inner-sphere hydration of the complexes
(shown earlier), but in this case, self-adduct formation acts in the opposite
direction [100,101] and partly compensates the effect of the differences in
KDM on SF (see also Fig. 4.15). Such compensation should also be observed
in extraction systems containing coordinatively unsaturated complexes and
a neutral lipophilic coextractant (synergist).
One can easily conclude that under certain conditions the differences
in partition constants may exert a decisive effect on SF. The effect of dif-
ferences in the stability constants of the complexes is significantly reduced if
the aqueous phase concentration of the uncomplexed metal ions is not much
higher than that of the [MLi]ni complexes. In the extreme case, when (at a
moderate acidity) practically the only form of the metal in the aqueous
3
Fig. 16.2 3D contours of bonding molecular orbitals in the tris(2,3-butanedionato)
yttrium(III)-TMPO adduct-a model of tris(tropolonato)yttrium(III)-TOPO: (a) MO of s
symmetry, Es = 0.4006 a.u.; (b) MO of p symmetry, Ep = 0.2874 a.u. (From Ref. 98;
with permission from Wiley-VCH Verlag Weinheim, Germany.)
phase is the neutral chelate, MLn, the bn values influence neither the dis-
tribution ratios, nor the SF value, which then becomes dependent only on
the KDM ratio [77].
Of course, one has to recognize that for practical separation a high SF
value is the prerequisite but not the sufficient condition. To separate selec-
tively two metal ions, we must ensure their distribution ratios in the system to
be respectively equally greater and less than unity. All these considerations
show how many factors one must control to assure high selectivity of solvent
extraction separations of metal ions. In the case of coordinatively unsatu-
rated metal chelates, the increasing concentration of a synergist in the
extraction system decreases, as a rule, the selectivity. For example, scandium
and yttrium tris-tropolonates undergo partition with KDM = 0.381 and
0.0020, respectively, in the system toluene/102 mol dm3 tropolone/water
( pH = 5 ± 1) at 298 K [98]; therefore SFSc/Y = 190. Taking into account the
stability constants [Eq. (16.9)] of their TOPO adducts [logb3,1 = 4.27 for
Sc(trop)3; logb3,1 = 4.99 and logb3,2 = 7.39 for Y(trop)3] we arrive at the
following calculated values for DSc, DY and SFSc/Y: 1.09, 0.022 and 50 at 104
mol dm3 TOPO; 7.48, 0.246 and 30 at 103 mol dm3 TOPO; 71.3, 6.87 and
10.4 at 102 mol dm3 TOPO, respectively. In spite of the decrease in SFSc/Y
with increasing TOPO concentration, the optimum separation can be
achieved at ca. 103 mol dm3 TOPO. The effects become greater for ions
that show greater differences in b3,1, e.g., Sc3+ and In3+ (logb3,1 = 4.27 and
0.97 [98]). However, the unexpectedly increased selectivity in the synergistic
Am3+–Eu3+ system with CYANEX 301 and some O-bearing weak-base
coextractants [45] shows that still other factors, for example different partic-
ipation of the lanthanide and actinide f-orbitals in bonding, must be taken
into account when considering the selectivity of separation.
There is, therefore, a good deal of work for theoretical chemists,
dedicated to merging the calculations of the separate steps of extraction of
metal ions into a whole. Only such a united approach will allow us to
analyze correctly all the factors that affect the thermodynamics of extrac-
tion. The greatest challenge stems from the need to evaluate solvent effects
by the use of more accurate, explicit solvation models. That will require
quantum mechanical calculations on extremely large systems consisting of
many hundred molecules, thousands of atoms.
16.4.6 Actual Calculations

16.4.6.1 MM and QM Methods
The reported QM and MM calculations on extraction of metal ions do
not comprise as a rule the whole process described by the complex forma-
tion–partition model. The calculations rarely aim at the determination of the
equilibrium of the process, because of the complexity of such a task. Most

often, they are limited to the determination of the relative extractability of a
given ion in a series of ions, or in a series of ligands. Therefore, in the
majority of studies, the partition step becomes neglected and only the energy
of complex formation is calculated as a qualitative or semiquantitative
measure of metal extractability [25,32,36,51]. Sometimes the calculations
become even more simplified and only the formation energy of a given
conformation of the ligand is calculated as the measure of extractability of a
given metal [26,37,38,45]. Another simplification consists in neglecting sol-
vent effects on the stability of complexes and ligands, whereas both structure
and stability of these species depend on solvent effects [26]. The calculations
usually relate to complex formation in vacuo. Even those works that take
into account dehydration of a given metal ion in the aqueous phase upon
formation of the neutral complex [Eq. (16.3)] [36] still neglect bulk solvent
effects, i.e., interactions of the complex with water molecules. These inter-
actions that depend on the central metal ion affect to a different degree, not
only the stability of the complexes but also their inner-sphere hydration (see
section 16.4.4.2). Solvent effects exerted on simple molecules or ions were
studied using both implicit and explicit solvation models (see section 16.2.5).
The partition constants obtained (see section 16.4.2) remain in reasonable
agreement with experiment [56,58]. The relative KD values of simple organic
solutes, calculated with the use of continuum solvation models, proved to be
more accurate than those obtained from MD-FEP simulations, the average
errors in logKD for seven solute pairs being 0.50 and 1.01, respectively [85].
Unfortunately, to the authors’ knowledge, similar calculations of partition
constants have not yet been carried out on metal complexes.
We know of only one attempt to calculate all extraction equilibria for
metal complexes, i.e., to analyze the whole metal ion extraction process
including both complex formation and its two-phase distribution. In this
attempt, Varnek and Wipff computed the relative free energy changes of
extraction of several mono- and divalent cations with polydentate iono-
phores and counterions, using MD-FEP simulations [70]. For example, the
calculated selectivity of extraction for the Ba2+/Sr2+ pair from water to
chloroform by a phosphoryl-containing podand (DDG0extr = 1.5 kcal mol1)
proved to be close to the experimental value (DDG0extr = 1.2 kcal mol1) [70].
Despite the limitations of the method, this is the only methodologically
correct approach used that makes it possible to deal with solvent extraction
of metal ions. Other correlations based on the energy (or free energy)
changes of complex formation as a measure of relative extractability of
metal ions studied are oversimplified because they neglect the transfer of the
complex from the aqueous to an organic phase. They are, however, often
used because the QM methods applied for large systems that contain bulk
solvent molecules still fail, whereas it is tempting to compare the calculation

results with the experiment and to seek a correlation. We hope that the use
of QM/MM methods and continuum solvation models may be helpful in
overcoming such limitations. Also, statistical treatment of MD data for
simple complexes that reversibly distribute between two liquid phases can
contribute to the problem, provided the simulation systems (boxes) are large
enough to model the properties of the bulk solvents.
To illustrate the importance of considering the whole extraction pro-
cess, especially when coordinatively unsaturated complexes are extracted,
let us consider extraction of trivalent lanthanides with b-diketones. The
lanthanide contraction and the accompanying decrease in the coordination
number (CN) make the complexes less and less coordinatively unsaturated
across the lanthanide series, therefore less and less hydrated in the inner
sphere of the metal ion [102]. This makes the partition constants of these
complexes larger and larger, so that the KDM values for the two extreme ions
in the series (La–Lu) differ from each other by four orders of magnitude,
approximately as much as the respective b3 values [100,101,103]. (see also
Fig. 4.15). It has been shown in a series of papers [37,38] that the ability of
some b-diketones to extract lanthanides selectivity depend on the structure
of the bidentate ligand, in particular on bite size, and on the energy of the
intramolecular hydrogen bond in the enol form of the ligand. However, the
experimental extraction data were correlated only with the calculated
(semiempirical MNDO/H and ab initio) heats of formation of the anionic
form of ligands, which affect only the thermodynamic stability of the lan-
thanide complexes. Because the respective differences in their KDM values
were disregarded, the observed agreement was of only qualitative character.
In another interesting work, the experimental extraction constants
[Eq. (16.7)] of all the lanthanides, relative to that of lanthanum, have been
correlated with the differences between the calculated (MM) total strain
energies required for the formation of the respective metal(III) complexes
with bis-alkylhydrogenphosphate ligands in the aqueous phase [36]. The
calculations included the term responsible for dehydration of the metal ions,
necessary to bind the ligands [Eq. (16.3)], but they are concerned with the
process in vacuo, neglecting solvent effects. The partition step was com-
pletely disregarded. Therefore, the slope of the calculated energies vs.
logarithm of the experimental extraction constants (1.26) has appeared
much less than expected, which is 2.3RT = 5.71 (the value reported in [36] is
incorrect). This discrepancy shows that an important component of the
energy has been omitted in the calculations. In fact, the inclusion of the term
reflecting the differences in the partition constants of the complexes, most
probably similar to those in the acetylacetonates (shown earlier), would
significantly increase the slope and improve the correlation between the
calculations and the experiment. It has been argued before that in the case of
coordinatively unsaturated lanthanide complexes, the contribution from the
partition step increases the differences in the Kex values as well as the
selectivity of lanthanide separation.
The necessity of considering the whole extraction process has been
pointed out, however, calculations dealing with one particular extraction
steps can also be important. For example, the authors’ DFT calculations on
adduct formation in extraction systems [98] well correlate with experimental
data on the synergistic enhancement of partition constants of metal(III)
tropolonates by trioctylphosphine oxide. In spite of simplifications (in con-
trast to the experimental conditions, the calculations were performed with
models for tropolonate and TOPO, in the gas phase, and for 0 K), the com-
puted order of the energies of adduct formation (Y > Sc > Tl > In; no Ga
adduct) agreed with the experimental order of the adduct stability constants.
The calculated energies of adduct formation for tris-tropolonates of scan-
dium and yttrium are significantly less negative than the experimental free
energy changes of adduct formation (which also include the entropy term),
but the differences between the respective values, calculated DEadd(Y)
DEadd(Sc) = 5.5 kJ mol1 and experimental DGadd(Y) DGadd(Sc) = 4.2 kJ
mol1 [98] are comparable. Another work [94] points to a reasonable
agreement between the relative values of standard enthalpy and free energy
changes of transfer (heptane ? water) of tris-acetylacetonates of scandium
and cobalt(III); the experimental [92] and the calculated (DFT, continuous
solvation model; see section 16.2.5) for model compounds.
16.4.6.2 Artificial Neural Networks
Another, very different calculation method, which belongs to the field of
artificial intelligence, is till now only occasionally used in solvent extraction.
It consists in the operation of artificial neural networks (NN), which are
subject to appropriate procedures of construction and learning. NN com-
puting techniques arose from attempts to model the functioning of the
human brain and have evolved into a powerful tool for solving even the
most complex, nonlinear problems, which are difficult to handle by standard
modeling. This nonparametric method offers universal approximators able
to represent any, even the most complex, implicit functional relationship.
The calculations are based on the analysis of large sets of (empirical) data,
part of which is used for training the system, the rest for testing. The
widespread back propagation networks consist of large numbers of com-
putational units, ‘‘neurons,’’ connected by means of artificial synapses
characterized by matrices of adjustable weights. These connections and the
transfer functions of the neurons form distributed representations between
the input and output data. The required representation is selected during the
training procedure, when the weights are adjusted in a self-consistent pro-

cess based on the minimization of errors generated by the discrepancy
between the actual output of the network and the expected value used for
training. An important advantage of NN is the universality of the applied
methods, which are independent of detailed definition and of the data,
within a given group of problems [104,105]. Numerous software packages
related to NN operation, e.g., SNNS [106] and FlexTool [107], have been
developed and used. The artificial neural networks are able to generalize the
knowledge acquired and to self-improve the effectiveness of their perfor-
mance by an evolutionary process.
In an excellent early review, Zupan and Gasteiger gave an essential
background of the NN method and focused on its applications to various
chemical problems, including process control and structure/activity rela-
tionships [108]. A few further works have shown the usefulness of the NN
approach in various problems of solvent extraction, which are particularly
difficult for modeling with the use of parametric empirical methods because
of either complexity of the processes or lack of understanding of their
nature. In fact, mass transfer processes involve a wide range of interactions
between the chemical species in both liquid phases, and kinetics problems
additionally entangle their course.
The ability of NN to recognize the structure/property relationships of
crown ethers, widely used as extractants of alkali metal ions, was used for
prediction of complexing properties of these ligands, which depend on their
structure (ring size, number of benzo units) and on the metal ion. The
reported average error of the training sets (10–11 ligands) was ±0.27 logK
units, and that of the testing sets (4–5 ligands) was ±0.34 logK units [109].
The equilibrium distribution ratios of lanthanides in the HEH(EHP)/
kerosene/HCl extraction system have been successfully modeled by NN as a
complex function of the metal ion, its concentration, concentration of the
extractant, phase ratio, and acidity of the aqueous phase [110,111]. A data
set of 200 experimental points was used for training and testing. The system
predicted the test data with an absolute average error of 11%, much better
than most regression models based on the system parameters.
Such applications of NN as a predictive method make the artificial
neural networks another technique of data treatment, comparable to para-
metric empirical modeling by, for example, numerical regression methods
[e.g., 10,11] briefly mentioned in section 16.1. The main advantage of NN is
that the network needs not be programmed because it learns from sets of
experimental data, which results in the possibility of representing even the
most complex implicit functions, and also in better modeling without pre-
scribing a functional form of the actual relationship. Another field of
NN applications, perhaps more important, is process control. Processes that

are poorly understood or ill defined can hardly be simulated by empirical
methods. The problem of particular importance for this review is the use of
NN in chemical engineering to model nonlinear steady-state solvent
extraction processes in extraction columns [112] or in batteries of counter-
current mixer-settlers [113]. It has been shown on the example of zirconium/
hafnium separation that the knowledge acquired by the network in the
learning process may be used for accurate prediction of the response of
dependent process variables to a change of the independent variables in the
extraction plant. If implemented in the real process, the NN would alert the
operator to deviations from the nominal values and would predict the
expected value if no corrective action was taken. As a processing time of a
trained NN is short, less than a second, the NN can be used as a real-time
sensor [113].
16.5 NEW INSIGHTS AND PERSPECTIVES

Various instruments of theoretical chemistry have been widely to describe
separate steps of solvent extraction of metal ions. Because of the complexity
of solvent extraction systems, there is still no unified theory and no suc-
cessful approach aimed at merging the extraction steps. It has already been
pointed out that the challenging problem for theoreticians dealing with
solvent extraction of metals, in particular with thermodynamic calculations,
is to evaluate correctly solvent effects by the use of the most accurate explicit
solvation models and QM calculations. However, such calculations on
extremely large sets consisting of hundreds or even thousands of molecules,
necessary to model all aspects of the extraction systems, are still impossible
due to both hardware and software limitations.
The methods applied to the theoretical description of complex
formation, hydration, solvation, etc. discussed so far are characterized
by different degrees of approximation made when solving the equations of
Dirac-Schrödinger and Newton mechanics. But even the introduction of
QM/MM-MD-MC methods does not allow one to make effective calcula-
tions. This is because it is common in this approach to neglect nonbonding
interactions for pairs of atoms separated by distances greater than a selected
cutoff value. The range of interactions in the solution is thus limited by the
size of the system studied. This arbitrarily assumed division of the liquid
phase into molecules that interact with one another and molecules which are
neglected in the calculations and whose contribution is estimated by cor-
rections makes the calculations possible. Nevertheless, in spite of this sim-
plification such calculations are still very difficult, time-consuming, and
expensive; moreover, the results obtained are often insufficiently accurate,

inconsistent with the experiment, and therefore inconclusive.
The development of other methods with less computing requirements
and free of such limitations seems to be necessary in order to understand
better the processes that take place in the liquid phase, understand their
molecular mechanism, predict the influence of various factors on their
equilibria and kinetics, and evaluate correctly the thermodynamic functions
for the whole process and its particular steps.
The recent theoretical approach based on the information theory (IT)
in studying aqueous solutions and hydration phenomena [62–66] shows such
a direction. IT is a part of the system based on a probabilistic way of
thinking about communication, introduced in 1948 by Shannon and sub-
sequently developed [114]. It consists in the quantitative description of the
information by defining entropy as a function of probability
X1
1
¼ pi ln (16.15)
i¼1
pi
The probability of cavity formation in bulk water, able to accommodate a

solute molecule, by exclusion of a given number of solvent molecules, was
inferred from easily available information about the solvent, such as the
density of bulk water and the oxygen–oxygen radial distribution function
[65,79].
MacKay’s textbook [114] offers not only a comprehensive coverage of
Shannon’s theory of information but also probabilistic data modeling and
the mathematical theory of neural networks. Artificial NN can be applied
when problems appear with processing and analyzing the data, with their
prediction and classification (data mining). The wide range of applications
of NN also comprises optimization issues. The information-theoretic capa-
bilities of some neural network algorithms are examined and neural net-
works are motivated as statistical models [114].
The applications of NN to solvent extraction, reported in section
16.4.6.2., suffer from an essential limitation in that they do not apply to
processes of quantum nature; therefore they are not able to describe metal
complexes in extraction systems on the microscopic level. In fact, the net-
works can describe only the pure state of simplest quantum systems, without
superposition of states. Neural networks that indirectly take into account
quantum effects have already been applied to chemical problems. For ex-
ample, the combination of quantum mechanical molecular electrostatic
potential surfaces with neural networks makes it possible to predict the
bonding energy for bioactive molecules with enzyme targets. Computational
NN were employed to identify the quantum mechanical features of the
inhibitory molecules that contribute to bonding. This approach generates

the relationships between the quantum mechanical structure of the inhibi-
tory molecule and the strength of bonding. Feed-forward NN with back-
propagation of error were trained to recognize the quantum mechanical
molecular electrostatic potential at the entire van der Waals surface of a
group of training molecules and to predict the strength of interaction
between the enzyme and novel inhibitors [115]. On the other hand, the
interactions on the microscopic level can be treated by a new kind of neural
networks, quantum neural networks (QNN) [116]. Early publications con-
cerned problems different from those discussed in this chapter. For example,
QNN was applied to the recognition of handwritten numerals, and a
reliability over 99% has been achieved [117]. One may expect that the QNN
approach will also be helpful in describing intermolecular interactions
accompanying the processes of extraction of metal ions.
The methods developed from either NN or IT undoubtedly reduce the
requirements concerning the hardware that must be applied. In the NN
approach it is sufficient to determine electrostatic potential, for which far
less computer resources are required than for solving HFR equations sup-
plemented with the energy of correlations, necessary for reliable calculation
of interaction energy. Similarly, the calculation of information entropy (the
IT case) based on the electron density is possible, and can be done with
much shorter calculation time than in the case of HFR equations with the
correlation energy.
Unfortunately, the application of these methods to the description of
processes that take place in solutions, such as solvation, still requires the
rejection of certain interactions between molecules that build the systems.
This results from the tremendous number of interactions of the order of 2N,
where N denotes the number of molecules in the system. Optimization of
such large systems consisting of mutually interacting elements requires
special numerical procedures. The problem can be divided into classes
according to the time required for their solution on digital computers. If
there is an algorithm that solves a problem in a time which increases
polynomially (or slower) with the increase in the problem, we call the pro-
blem a polynomial problem and classify it as belonging to the class P
(polynominal). The class P is a subclass of the class NP (nondeterministic
polynominal), which describes the processes taking place in solutions.
The computer time needed to solve an NP problem increases expo-
nentially with the increase in N and it is impossible to solve it using classical
computers. However, a solution has been theoretically possible since 1994,
when Shor elaborated an algorithm that changed the increase of the com-
putational time with the increase in the system size from exponential to
polynomial [118]. The realization of the algorithm is possible basing on the
construction of a ‘‘quantum computer’’ [119]. However, as long as the idea

of construction of the quantum computer is far from being realized, it seems
necessary to apply approximate algorithms that, in spite of some dis-
advantages, bring us closer to the understanding of the processes that take
place in solutions. As far as the authors are aware, no QM approach to the
explicit solvation models, in particular to solvation of metal complexes, has
been used as yet.
ACKNOWLEDGMENTS
The authors would like to thank Alexandre Varnek for fruitful discussion,
and also Mariusz Bogacki, Yizhak Marcus, Bruce Moyer, Claude Musikas,
Jan Rydberg, and Slawomir Siekierski for their valuable comments on the
draft of this chapter at various stages of writing. The work was supported in
parts by the Institute of Nuclear Chemistry and Technology and by the
Polish Committee for Scientific Research (KBN) under grant number 4
T09A 110 23.
REFERENCES
1. Roothaan, C. C. J. Rev. Mod. Phys. 1951, 23, 69.
2. Parr, R. G.; Yang, W. Density-Functional Theory of Atoms and Molecules;
Oxford University Press: Oxford, 1989.
3. Deeth, R. J. Struct. Bonding 1995, 82, 1.
4. Rambusch, T.; Hollmann-Gloe, K.; Gloe, K. J. Prakt. Chem. 1999, 341, 202.
5. Cerius2 Version 3.5; Molecular Simulations Inc., San Diego, CA.
6. GAUSSIAN 98; Revision A.5, Gaussian Inc., Pittsburgh, PA, 1998.
7. ADF Version 2.3; Scientific Computing and Modeling, Vrije Universiteit,
Amsterdam.
8. Spartan Version 5.0.3; Wavefunction Inc., Irvine, CA.
9. HyperChem Version 5.02; Hypercube Inc., Gainesville, Florida.
10. Baes, C. F. Jr., Solv. Extr. Ion Exch. 2001, 19, 193.
11. Moyer, B. A.; Baes, C. F. Jr.; Case, F. I.; Driver, J. L. Solv. Extr. Ion Exch.
2001, 19, 757.
12. Varnek, A.; Wipff, G.; Solov’ev, V. P. Solv. Extr. Ion Exch. 2001, 19, 791.
13. Hay, B. P. Coord. Chem. Rev. 1993, 126, 177.
14. Boyens, J. C. A.; Comba, P. Coord. Chem. Rev. 2001, 212, 3.
15. Brubaker, G. R.; Johnson, D. W. Coord. Chem. Rev. 1984, 53, 1.
16. Hancock, R. D. Progr. Inorg. Chem. 1989, 37, 187.
17. Comba, P. Coord. Chem. Rev. 1993, 123, 1.
18. Deeth, R. J. Coord. Chem. Rev. 2001, 212, 11.
19. Deeth, R. J.; Foulis, D. L. Phys. Chem. Chem. Phys. 2002, 4, 4292.
20. Hancock, R. D.; Martell, A. E. Chem. Rev. 1989, 89, 1875.
21. Pople, J. A.; Santry, D. P.; Segal, G. A. J. Chem. Phys. 1965, 43, 129.
22. Hohenberg, P.; Kohn, W. Phys. Rev. B 1964, 136, 864.
23. Warshel, A.; Levitt, M. J. Mol. Biol. 1976, 103, 227.
24. Weiner, S. J.; Kollman, P. A.; Case, D. A.; Singh, U. C.; Ghio, C.; Alangona,
G.; Profeta, S. Jr.; Weiner, P. J. Am. Chem. Soc. 1984, 106, 765.
25. Hay, B. P. Metal Ion Separation and Preconcentration: Progress and Oppor-
tunities; Bond, A. H. Dietz, M. L. Rogers, R. D. Eds.; ACS Symposium Series
716; American Chemical Society: Washington, D.C., 1999; p. 102.
26. Sachleben, R. A.; Moyer, B. A. Metal Ion Separation and Preconcentration:
Progress and Opportunities; Bond, A. H. Dietz, M. L. Rogers, R. D. Eds.;
ACS Symposium Series 716; American Chemical Society: Washington, D.C.,
1999, p. 114.
27a. Sella, C.; Bauer, D. Solv. Extr. Ion Exch. 1992, 10, 579.
27b. Sella, C.; Bauer, D. Solv. Extr. Ion Exch. 1993, 11, 395.
28. Bond, A. H.; Chiarizia, R.; Huber, V. J.; Dietz, M. L.; Herlinger, A. W.; Hay,
B. P. Anal. Chem. 1999, 71, 2757.
29. Hay, B. P.; Hancock, R. D. Coord. Chem. Rev. 2001, 212, 61.
30. Kopczyński, T.; Łozyński, M.; Prochaska, K.; Burdzy, A.; Cierpiszewski, R.;
Szymanowski, J. Solv. Extr. Ion Exch. 1994, 12, 701.
31. Bogacki, M. B.; Jakubiak, A.; Prochaska, K.; Szymanowski, J. Colloids
Surfaces A 1996, 110, 263.
32. Bogacki, M. B.; Jakubiak, A.; Cote, G.; Szymanowski, J. Ind. Eng. Chem.
Res. 1997, 36, 838.
33. Kyuchoukov, G.; Bogacki, M. B.; Szymanowski, J. Ind. Eng. Chem. Res.
1998, 37, 4084.
34. Krueger, T.; Gloe, K.; Stephan, H.; Habermann, B.; Hollmann, K.; Weber, E.
J. Mol. Model. 1996, 2, 386.
35. Stephan, H.; Gloe, K.; Krüger, T.; Chartroux, C.; Neumann, R.; Weber, E.;
Möckel, A.; Woller, N.; Subklew, G.; Schwuger, M. J. Solv. Extr. Res. De-
velop. Japan. 1996, 3, 43.
36. Comba, P.; Gloe, K.; Inoue, K.; Krüger, T.; Stephan, H.; Yoshizuka, K.
Inorg. Chem. 1998, 37, 3310.
37. Le, Q. T. H.; Umetani, S.; Suzuki, M.; Matsui, M. J. Chem. Soc. Dalton
Trans. 1997, 643.
38. Umetani, S.; Kawase, Y.; Le, Q. T. H.; Matsui, M. Inorg. Chim. Acta. 1998,
267, 201.
39. Torgov, V.; Erenburg, S.; Bausk, N.; Stoyanov, E.; Kalchenko, V.; Varnek,
A.; Wipff, G. J. Mol. Struct., 2002, 611, 131.
40. Rabbe, C.; Madic, C.; Godard, A. Solv. Extr. Ion Exch. 1998, 16, 1091.
41. Rabbe, C.; Sella, C.; Madic, C.; Godard, A. Solv. Extr. Ion Exch. 1999, 17, 87.
42. Spjuth, L.; Liljenzin, J. O.; Hudson, M. J.; Drew, M. G. B.; Iveson, P. B.;
Madic, C. Solv. Extr. Ion Exch. 2000, 18, 1.
43. Drew, M. G. B.; Hudson, M. J.; Iveson, P. B.; Russel, M. L.; Liljenzin, J. O.;
Skalberg, M.; Spjuth, L.; Madic, C. J. Chem. Soc. Dalton Trans. 2002, 2973.
44. Coupez, B.; Boehme, C.; Wipff, G. Phys. Chem. Chem. Phys. 2002, 4, 5716.
45. Ionova, G.; Ionov, S.; Rabbe, C.; Hill, C.; Madic, C.; Guillaumont, R.;
Krupa, J. C. Solv. Extr. Ion Exch. 2001, 19, 391.
46. Pyykkö, P. Chem. Rev. 1988, 88, 563.
47. Schreckenbach, G.; Hay, P. J.; Martin, R. L. J. Comput. Chem. 1999, 20, 70.
48. Schautz, F.; Flad, H.-J.; Dolg, M. Theor. Chim. Acta 1999, 99, 231.
49. Skylaris, C.-K.; Gagliardi, L.; Handy, N. C.; Ioannou, A. G.; Spencer, S.;
Willets, A.; Simper, A. M. Chem. Phys. Lett. 1998, 296, 445.
50. Willets, A.; Gagliardi, L.; Ioannou, A. G.; Skylaris, C.-K.; Simper, A. M.;
Spencer, S.; Handy, N. C. Int. Rev. Phys. Chem. 2000, 19, 327.
51. Gagliardi, L.; Handy, N. C.; Skylaris, C.-K.; Willets, A. Chem. Phys. 2000,
252, 47.
52. Spencer, S.; Gagliardi, L.; Handy, N. C.; Ioannou, A. G.; Skylaris, C.-K.;
Willets, A.; Simper, A. M. J. Phys. Chem. A 1999, 103, 1831.
53. Pershina, V.; Fricke, B.; Kratz, J. V.; Ionova, G. V. Radiochim. Acta 1994, 64,
37.
54. Pershina, V.; Fricke, B. J. Phys. Chem. 1996, 100, 8746.
55. Pershina, V. Chem. Rev. 1996, 96, 1977.
56. Cramer, C. J.; Truhlar, D. G. Chem. Rev. 1999, 99, 2161.
57. Gao, J. L. Acc. Chem. Res. 1996, 29, 289.
58. Kollman, P. Chem. Rev., 1993, 93, 2395.
59. Leach, A. R. Molecular Modeling: Principles and Applications, 2nd Ed.;
Pearson Education Limited, 2001.
60. Allen, M. P.; Tildesley, D. J. Computer Simulations of Liquids; Clarendon:
Oxford, 1987.
61. Case, D. A.; Pearlman, D. A.; Caldwell, J. C.; Cheatham, T. E. III; Ross, W. S.;
Simmerling, C. L.; Daren, T. A.; Mertz, K. M.; Stanton, R. V.; Cheng, A. L.;
Vincent, J. J.; Crowley, M.; Ferguson, D. M.; Radmer, R. J.; Seibel, G. L.;
Singh, U. C.; Weiner, P. K.; Kollman, P. A. AMBER5; University of Cali-
fornia: San Francisco, 1997.
62. Hummer, G.; Pratt, L. R.; Garcia, A. E. J. Phys. Chem. A 1998, 102, 7885.
63. Garde, S.; Hummer, G.; Paulaitis, M. E. J. Chem. Phys. 1998, 108, 1552.
64. Hummer, G.; Garde, S.; Garcia, A. E.; Paulaitis, M. E.; Pratt, L. R. J. Phys.
Chem. B 1998, 102, 10469.
65. Hummer, G.; Garde, S.; Garcia, A. E.; Pratt, L. R. Chem. Phys. 2000, 258,
349.
66. Gomez, M. A.; Pratt, L. R.; Hummer, G.; Garde, S. J. Phys. Chem. B 1999,
103, 3520.
67. Behler, J.; Price, D. W.; Drew, M. G. B. Phys. Chem. 2001, 3, 588.
68a. Benjamin, I. Chem. Rev. 1996, 96, 1449.
68b. Benjamin, I. Solvent Extraction for the 21st Century, Proc. ISEC’99; Cox, M.
Hildalgo, M. Valiente, M. Eds.; Soc. Chem. Ind.: London, 2001, 1, 3.
69. Varnek, A.; Wipff, G. J. Mol. Struct. (Theochem) 1996, 363, 67.
70a. Varnek, A.; Wipff, G. J. Comput. Chem. 1996, 17, 1520.
70b. Varnek, A.; Wipff, G. Solv. Extr. Ion Exch. 1999, 17, 1493.
71. Varnek, A.; Troxler, L.; Wipff, G. Chem. Eur. J. 1997, 3, 552.
72. Nazarenko, A. Y.; Baulin, V. E.; Lamb, J. D.; Volkova, T. A.; Varnek, A. A.;
Wipff, G. Solv. Extr. Ion Exch. 1999, 17, 495.
73. Szymanowski, J. Solv. Extr. Ion Exch. 2000, 18, 729.
74. Felix, V.; Matthews, S. E.; Beer, P. D.; Drew, M. G. B. Phys. Chem. Chem.
Phys. 2002, 4, 3849.
75. Baaden, M.; Berny, F.; Madic, C.; Wipff, G. J. Phys. Chem. A 2000, 104, 7659.
76. Siekierski, S. J. Radioanal. Chem. 1976, 31, 335.
77. Narbutt, J. J. Radioanal. Nucl. Chem. 1992, 163, 59.
78. Reiss, H.; Frisch, H. L.; Lebowitz, J. L. J. Chem. Phys. 1959, 31, 369.
79. Kalra, A.; Tugcu, N.; Cramer, S. M.; Garde, S. J. Phys. Chem. B 2001, 105,
6380.
80. Nemethy, G.; Scheraga, H. A. J. Chem. Phys. 1962, 36, 3382.
81. Madan, B.; Sharp, K. J. Phys. Chem. B 1997, 101, 11237.
82. Gniazdowska, E.; Narbutt, J. J. Mol. Liquids 2000, 84, 273.
83. Sithoff, D.; Ben-Tal, N.; Honig, B. J. Phys. Chem. 1996, 100, 2744.
84. Li, J.; Hawkins, G. D.; Cramer, C. J.; Truhlar, D. G. Chem. Phys. Lett. 1998,
288, 293.
85. Best, S. A.; Merz, K. M. Jr.; Reynolds, C. H. J. Phys. Chem. B 1999, 103, 714.
86. Osakai, T.; Ebina, K. J. Electroanal. Chem. 1996, 412, 1.
87. Parker, A. J.; Alexander, R. J. Am. Chem. Soc. 1968, 90, 3313.
88. Marcus, Y. Ion Properties; Dekker: New York, 1997.
89. Narbutt, J. Polish J. Chem. 1993, 67, 293.
90. Polyakov, V. R.; Czerwiński, M. Inorg. Chem. 2001, 40, 4798.
91a. Narbutt, J. J. Inorg. Nucl. Chem. 1981, 43, 3343.
91b. Narbutt, J. J. Phys. Chem 1991, 95, 3432.
92. Narbutt, J.; Bartoś, B.; Siekierski, S. Solv. Extr. Ion Exch. 1994, 12, 1001.
93. Narbutt, J.; Fuks, L. Radiochim. Acta, 1997, 78, 27.
94. Czerwiński, M.; Narbutt, J. Phys. Chem. Chem. Phys. submitted.
95. Foresman, J. B.; Keith, T. A.; Wiberg, K. B.; Snoonian, J.; Frisch, M. J. J.
Phys. Chem. 1996, 100, 16098.
96. Katsuta, S.; Imura, H.; Suzuki, N. Bull. Chem. Soc. Jpn 1991, 64, 2470.
97. Katsuta, S.; Yanagihara, H. Solv. Extr. Ion Exch. 1997, 15, 577.
98. Narbutt, J.; Czerwiński, M.; Krejzler, J. Eur. J. Inorg. Chem. 2001, 3187.
99. Narbutt, J.; Krejzler, J. Inorg. Chim. Acta 1999, 286, 175.
100. Albinsson, Y.; Mahmood, A.; Majdan, M. Rydberg, J. Radiochim. Acta 1989,
48, 49.
101. Albinsson, Y. Acta Chem. Scand. 1989, 43, 919.
102. Habenschuss, A.; Spedding, F. H. J. Chem. Phys. 1980, 73, 442.
103. Suzuki, N.; Nakamura, S. Inorg. Chim. Acta 1985, 110, 243.
104. Kohonen, T. Self-Organizing Maps; Springer Verlag: Berlin, 1995.
105. Hayki, S. Neural Networks: A Comprehensive Foundation; Prentice Hall:
Englewood Cliffs, NJ, 1994.
106. Zell, A.; Mamier, G.; Vogt, M.; Mache, N.; Hübner, R.; Döring, S.; Herr-
mann, K.-U.; Soyez, T.; Schmalzl, M.; Sommer, T.; Hatzigeorgiou, A.; Pos-
selt, D.; Schreiner, T.; Kett, B.; Clemente, G. SNNS-Stuttgart Neural
Network Simulator User Manual, Version 4.0; University of Stuttgart, Institute

for Parallel and Distributed High Performance Computing (IPVR), Depart-
ment of Computer Science; Report June 1995. wwwra.informatik.uni-tue-
bingen.de/SNNS/.
107. FlexTool; Flexible Intelligence Group, LLC: Tuscalloosa, AL; 2003; www.
flextool.com.
108. Zupan, J.; Gasteiger, J. Anal. Chim. Acta 1991, 248, 1.
109. Gakh, A. A.; Sumpter, B. G.; Noid, D. W.; Sachleben, R. A.; Moyer, B. A. J.
Incl. Phenom. Mol. Recogn. Chem. 1997, 27, 201.
110. van Deventer, J. S. J.; Aldrich, C. Value Adding Through Solvent Extraction,
Proc. ISEC’96; Shallcross, D.C. Paimin, R. Prvcic L.M. Eds. University of
Melbourne, 1996, 1, 831.
111. Giles, A. E.; Aldrich, C.; van Deventer, J. S. J. Hydrometallurgy 1996, 43, 241.
112. Slater, M. J.; Aldrich, C. Application of Neural Network and Other Learning
Technologies in Process Engineering; Mujtaba, I. M. Hussain, M. A. Eds.;
Imperial College Press: London, 2001; 3.
113. Boger, Z.; Ben-Haim, M. Solvent Extraction in the Process Industry, Proc.
ISEC’93; Logsdail, D. H. Slater, M. J. Eds.; Elsevier Applied Science, 1993, 2,
1198.
114. MacKay, D. J. C. Information theory, inference and learning algorithms; Draft
3.1415, January 2003; www.inference.phy.cam.ac.uk/mackay/itprnn/book.
html.
115. Braunheim, B. B.; Bagdassarian, C. K.; Schramm, V. L.; Schwartz, S.; D. Int.
J. Quantum Chem. 2000, 78, 195.
116. Purushothaman, G.; Karayiannis, N. B. IEEE Trans. Neural Networks 1997,
8, 679.
117. Zhou, J.; Gan, Q.; Krzy_zak, A.; Suen, C. Y. Int. J. Docum. Anal. Recogn.
1999, 2, 30.
118. Shor, P. W. Polynomial time algorithms for prime factorization and discrete
logarithms on a quantum computer. Proc. 35th Annual Symposium on the
Foundations of Computer Science; Goldwasser, S. Ed.; IEEE Computer So-
ciety Press: Los Alamos, CA, 1994, p. 124.
119. Bouwmeester, D.; Ekert, A.; Zeilinger, A. Eds. The Physics of Quantum In-
formation; Springer Verlag: Berlin, 2001.
Appendix
Appendix
A. FUNDAMENTAL CONSTANTS, BASIC AND

DERIVED SI UNITS
A.1 Fundamental Constants
e electron (proton) charge, 1.6022 1019 coulomb (C)
F Faraday constant, 96,485 C mol1
NA Avogadro number, 6.022 1023 mol1
kB Boltzmann constant, 1.3807 1023 joule/kelvin (J K1)
R molar gas constant, 8.314 J K1mol1 (0.08206 L atm mol1 K1)
vM molar volume of ideal gas, 0.02241 m3 mol1
A.2 Fundamental SI Units
Physical quantity Symbol Unit name Symbol
Length l meter m
Mass m kilogram kg
Time t second s
Electric current I ampere A
Thermodynamic T Kelvin K
temperature
715
716 Appendix
A.3 Derived SI Units as Used in Chemistry
Physical quantity Symbol Unit name, symbol, explanation
Physical mass (weight) m(w) gram, g; mA is weight of pure substance A

Molar mass M molecular weight in atomic mass units (amu)
Amount of substance n moles, mol; nA = mAMA
1 mol = NA molecules
Volume V liter, L, or dm3 (SI-unit is m3)
Concentration c or [ ] molarity, M, mol L1, or mol dm3
Density r g/cm3 = 103 kg/m3 (SI unit is kg/m3)
Pressure P megapascal, MPa; 1 atm = 0.1013 MPa; 1 bar =
0.1 MPa
Temperature T 8C degrees Celsius; 0 T8C = 273.15 K
Energy E joule (watt seconds, Ws) J; 1 kilocalorie
(Kcal) = 4.187 J; 1 electron volt (eV) =
1.602 2 1019 J corresponds to 96.485 kJ
mol1
B. SYMBOLS USED IN CHEMICAL REACTION

FORMULAS
Attempts have been made to adhere to the following set of abbreviations for
the reacting species to simplify their identification:
M metal ion (central atom)
A (preferably lipophilic) extractant anion (anion of HA)
L (preferably hydrophilic) ligand in aqueous phase (monobasic)
X any (likely noncomplexing) anion in aqueous phase
B uncharged donor molecule (adduct former)
S organic solvent molecules (solvating or nonsolvating)
Ionic charges are omitted, when not necessary for understanding.
Other abbreviations commonly used:
A, B, C for various solutes in general
Az anion in general, charge z (may be omitted)
Cz+ cation in general, charge z+ (may be omitted)
Cn+An electrolyte in general, with n+ atoms of C and n atoms of A in
a neutral salt (e.g., NaCl or NaClO4).
e, and ne for electrolyte and nonelectrolyte, respectively
Appendix 717
Physical state:
Gaseous (g)
Solid (s)
Liquid (l), not recommended in this text
Aqueous [i.e., dissolved in water (aq)]
Organic [i.e., dissolved in organic solvent (org); not recommended (s)]
Water (w), to specifically refer to water molecules
B.1 Abbreviations for Organic Compounds

AA: acetylacetone; 2,4-pentanedione
BA: benzoylacetone
DCTA: 1,2-diaminocyclohexane tetraacetic acid
DBP: dibutyl phosphoric acid
DTPA: diethylenetriamine pentaacetic acid
DEHPA: di-2-ethylhexyl phosphoric acid
PC88A: (and Ionquest 801) 2-ethylhexyl phosphonic acid 2-ethylhexyl ester
EDTA: ethylene diaminetetraacetic acid
FAA: hexafluoroacetylacetone
HIBA: a-hydroxyisobutyric acid
LIX64: o-hydroxybenzophenone oxime
NTA: nitrilotriaminoacetic acid
TTA: thenoyltrifluoroacetone
TOA: trioctylamine; Aliquat-336 is mainly TOA
TDA: tridecylamine; Adogen 464 is C8–C10 amine
TLA: trilaurylamine; C12
TBP: tributyl phosphate
TOPO: trioctylphosphine oxide
RNH+ 3: organic primary ammonium ion
R2NH+ 2: organic secondary ammonium ion
R3NH+: organic tertiary ammonium ion
R4N+: organic quaternary ammonium ion
R: organic radical in general
Protolytic compounds are written as monobasic acids (e.g., HA), but
the proton may be excluded as an abbreviation (e.g., TTA or EDTA). For
example, in the text HTTA and TTA may mean the same if not specified, as
by the reaction HTTA ! H+ + TTA.
718 Appendix
C. TERMINOLOGY FOR SOLVENT

EXTRACTION EQUILIBRIA
Most chapters in this book use the following symbols, which follow IUPAC
recommendations (see Refs. 1– 4) where such recommendations are avail-
able.
C.1 Reference to Aqueous and Organic Phases

In the reaction below a metal ion Mn+ reacts with n ligand anions L to
form an uncharged complex MLn. If the ions dissolve only in the aqueous
phase, and the metal complex and undissociated acid HL dissolve only in
the organic phase, IUPAC allows the reaction to be written in four different
ways:
Mnþ (aq) þ nHL(org) Ð MLn (org) þ nHþ (aq) Kex (C.1a)
Mnþaq þ nHLorg Ð MLn;org þ nHþ
aq Kex (C.1b)
þ
M nþ
þ nHL(org) Ð MLn (org) þ nH Kex (C.1c)
þ
M nþ
þ nHL Ð MLn þ nH Kex (C.1d)
Kex is the overall extraction constant [2]. Different authors favor different
notations, all being used here (as in other literature). For example, the con-
centration of HL in the organic phase may be written ½HL, [HL]org, or [HL]o,
and in the aqueous phase [HL] (with no index), [HL]aq or [HL]a. Freiser and
Nancollas [2] recommend against the combination [HL]o/[HL]a, which
invites confusion.
Total concentration of species A in a phase is indicated either by [A]t,
[A]tot, or cA. ‘‘Original’’ value (e.g., total weight of substance, volume, con-
centration, or such, in the beginning of an experiment) may be indicated by
an upper index 0 (e.g., [A]0 or cA0. Total concentration and original con-
centration may mean the same.
C.2 IUPAC Terminology

IUPAC prefers ‘‘liquid–liquid distribution’’ or ‘‘liquid–liquid extraction’’
rather than ‘‘solvent extraction.’’ We justify our preference for ‘‘solvent
extraction’’ by its more common usage. IUPAC defines the following terms:
Distribution ratio is the ‘‘total analytical concentration of a substance
in the organic phase to its total analytical concentration in the aqueous
phase, usually measured at equilibrium.’’ Symbol D. D shall be defined and,
preferably, specified by an index; if the distribution of mercury is measured,
the distribution ratio is written D(Hg) or DHg. The term ‘‘partition ratio’’ is
not used for the distribution ratio.
Appendix 719
Distribution constant is symbolized by KD. Thus for

A(aq) Ð A(org); KD ; KD (A); KD;A ; KDA ; or KDR ¼ ½Aorg =½Aaq
MAz (aq) Ð MAz (org); KD ; KD (MAz ); KDMAz ; or KDC ¼ ½MAz =½MAz
In this book we prefer a suffix to the index to identify if the distribution
constant refers to the neutral reagent (or extractant) or to the uncharged
metal complex (i.e., KDR and KDC, respectively).
Extraction equilibria (including adduct formation) must be defined by
text, for example,
Mnþ þ nHA(org) þ aS(org) Ð MAn Sa (org) þ nHþ Kex
Recovery factor is defined as the percentage (R%), or fraction R, of the
total quantity of a substance extracted under specified conditions. (In our
text we also use percentage extraction, E%.)
Solvent is ‘‘a liquid phase (usually an organic liquid) or the solution of
an extractant in an organic diluent which is used to extract a substance from
another liquid (usually aqueous).’’
Diluent is ‘‘an inert (organic) solvent used to improve the physical
properties (density, viscosity, etc.) or the extractive properties (e.g., selec-
tivity) of the extractant. The diluent has negligible extractant properties for
the substance to be extracted.’’
Extractant is ‘‘the reagent which forms a complex or other adduct in
the solvent with the substance which partitions across the phase boundary of
the extraction system. The extractant (extracting agent) may also partition.’’
Extract is ‘‘the separated phase (usually organic) containing the sub-
stance extracted from the other phase.’’
Stripping is ‘‘the process of back-extraction of the distribuend from the
extract (usually into the aqueous phase).’’
Distribuend is ‘‘the substance that is distributed between two immis-
cible liquids or liquid phases.’’
C.3 Complex Formation in Aqueous Solutions

The nomenclature used in this book is presented in section 3.1. It should be
noted [2] that ‘‘For polynuclear complexes and complexes with several kinds
of ligands, it may sometimes be practicable to use b with . . . multiple sub-
scripts. Their general meaning must then be defined very clearly with a full
reaction formula. . . . Strict standardization in this field has not yet been
achieved.’’
720 Appendix
C.4 Other Equilibrium Constants

Commonly Used in This Text
QM distribution ratio of M between nonaqueous and aqueous phase
DM distribution ratio of M between organic solvent and aqueous phase
(e.g., DLa for distribution of element La)
Ki stepwise formation constant for ith complex (e.g., MLi)
Kex overall extraction constant
Ka acid dissociation (concentration) constant; pKa = log Ka
Kal first acid dissociation constant for multibasic acid
Kadl first adduct formation constant (organic phase)
KDx distribution constant for x (specified); for example, x = R, or = HA,
for undissociated extractant (reagent), = C for neutral
(e.g., metal-containing) complex
Ksp solubility product
Kw ionic product of water
bn overall complex formation, or stability, constant for formation of nth
complex (e.g., MLn).
P thermodynamic distribution constant
sAC solubility of AC in moles or grams per liter (mL1 or gL1)
Kass association constant for ion pair formation
KBS Bunsen coefficient for solubility of gas in a liquid
D. FORMULA STRUCTURES
Structural formulas of a number of commonly used extractants, and some
typical chelate compounds.
1. Dinonylnaphtylsulfonic acid 2. Benzoic acid
3. Salicylic acid 4. Isopropyltropolone

Appendix 721
5. b-Diketones:
(a) ketoform; (b) enol form; (c) enolate ion;
(d) acetylacetone, 2,4-pentanedione, HAA: R = R0 = CH3;
(e) trifluoroacetylacetone, HTFA: R = CH3, R0 = CF3;
(f) benzoylacetone, HBA: R = CH3, R0 = C6H5;
(g) thenoyltrifluoroacetone, HTTA: R = formula (g), R0 = CF3;
(h) bis-acetylacetonatocopper(II) complex; and
(i) dibenzoylmethane: R = R0 = C6H5.
6. 8-Hydroxoquinoline, oxine 7. 1-Nitroso-2-naphthol

722 Appendix
8. Oxime: R and R0 are different long-chained

and branched aliphats or aromates
10. N-Nitrosophenylhydroxylamine
9. N-Phenylbenzo-hydroxamic acid (ammonium salt is cupferron)
11. (a) Amide R = R0 = R@ = alkyl; and

(b) carbamoylphosphinate: R1 = R2 = alkyl, R = phosphinate
12. Carbamoylmethylene phosphonate 13. Dialkyl phosphonate

Appendix 723
14. Di(2-ethylhexyl) phosphoric acid, HDEHP, D-2EHPA; R is C8H17:

(a) monomeric form;
(b) dimeric form;
(c) M(HA2)3 complex formed with M(III); and
(d) UO2(HA2)2 complex.
15. Tributylphosphate, TBP 16. Trioctyl phosphine oxide, TOPO
17. Methyl isobutyl ketone, MIBK, hexone 18. o-Phenanthroline

724 Appendix
19. Dialkyl dithiophosphonic acid 20. Diethyl dithiocarbamate
21. Crown ethers:

(a) dicyclohexyl-18-crown-6, DC18-C6; and
(b) 15-crown-5, 15-C5
REFERENCES
1. Högfeldt, E. Stability Constants of Metal-Ion Complexes. Part A: Inorganic
ligands; IUPAC Chemical Data Series No 22; Pergamon Press: New York, 1982.
2. Freiser, H.; Nancollas, G. H. Compendium of Analytical Nomenclature. Definitive
Rules 1987; IUPAC; Blackwell Scientific Publications (www.IUPAC.com/
publications): Oxford, 1987.
3. McNaught, A. D.; Wilkinson, A. IUPAC Compendium of Chemical Terminology,
2nd Ed.; Blackwell Science, 1997.
4. IUPAC. Quantities, Units and Symbols in Physical Chemistry, 2nd Ed.; Blackwell
Science, 1993.
Index of Compounds
METALS AND INORGANIC [Actinides]

COMPOUNDS thermodyn, 105
water exch rate, 218
Actinides Actinium
compl w phosphates, chelate, amides, hardness, 99
178, 182, alkylphosphoric acids, oxid states, 511
235, b-diketones, 100 stereochem, 93
coord number, 105, 118, 170 Aluminum
extrn 531ff, by D2EHPA, 218, analysis, 572
TLA/NO3, 160ff, compl w triphenylmethane dyes, 568,
TTA/TBP, 118, bifunctional 8-hydroxyquinoline, 572
organophosphorus reagents, 532, in crud, 313, 317, 321
monofunctional organophos- extn by D2EHPA, 664
phorus reagents, 536, flowsheet, (microemulsion)
539, carbamoyl-organophosphorus hardness, 99
reagents, 532, diamides, ionic radius, 45
534, tris-pyridine, 550, stereochem, 93
triazine, 550 water exch rate, 218
inner/outer sphere compl, 105 Americium
oxidation states, 511 adduct extrn w bromodecanoic acid, 183
sepn proc actinide-lanthanide, 547ff, compl w diamides, 178, terphenyl, 182–5
DIAMEX, 534, 538, TRUEX, extrn, 531ff, by CMPO, 532ff, TTA/
532ff, 543, SANEX, 548ff MIBK, 540, CYANEX-301
stereochem, 93 (modeling), 685, 702
The key for the abbreviations contained within the index can be found on page 739.
725
726 Index of Compounds
[Americium] Bismuth
sepn from Ln, 547ff, flowsheet, 543 analysis w diphenyldithiocarbazone,
hardness, 99 568, dithiocarbamate, 573,
ionic radius, 45 thioxine, 573
oxid states, 511 extn by D2EHPA, 664
partitioning, 539 (microemulsion)
solub param w decanoic acid, 173, hardness, 99
terpyridine, 173, 182–5 ionic radius, 45
struct w triazine and pyridine adducts, stereochem, 93
182–5 Boron
Antimony boron trifluoride/HF extractant, 431
acceptor acid as SbCl5, 101 Bromine
analysis w dithiocarbamate, 573, basicity, 137
thioxine, 573 elemental, 101
extn by dibutylcarbitol/Cl, 486, extrn, 124
MIBK/Cl, 487 hardness, 99
hardness, 99 ionic radius, 45
stereochem, 93 trioctylamine salts, 159
SbCl3 distribution, 124
Arsenic Cadmium
analysis tetrachloromethane, 573, analysis, 572, w 8-(benzene-
thioxine, 573 sulfonamido)quinoline, 575,
extn by dibutylcarbitol, 486, MIBK, diethyldithiocarbamate, 572,
487, TBP, 632, phosphine oxide, D2EHPA, 581 (SLM),
632 diphenyldithiocarbazone, 568
extn from copper electrolyte, 632 cloud point extn, 586
stereochem, 93 compl w amm, 85, 96, 106, Cl, 89,
AsCl3 distrib, 124 en, 106
extrn by alkyl phosphoric acids,
Barium 179, D2EHPA from
analysis w crown ethers, 572 accumulator waste, 636ff
extrn by crown ether in SREX hardness, 99
process, 538 ionic radius, 45
hardness, 99 stereochem, 93
ionic radius, 45 Calcium
stab const TTA/TBP, 148 analysis, 572, w diethyldithiocarbamate,
stereochem, 93 572, phosphates/
water exch rate, 218 phosphonates, 590, (ISE),
Beryllium w thioxine, 572
analysis w b-diketones, 572, compl w F, 90, TTA/TBP, 148, TBP,
triphenylmethane dyes, 568 572
compl w oxalate, 165 in crud, 313, 468
extrn by TTA/MIBK, 165 extrn by D2EHPA, 474
stereochem, 93 hardness, 99
water exch rate, 218 ionic radius, 45
[Calcium] [Chromium]
stereochem, 93 hardness, 99
water exch rate, 218 ionic radius, 45
Californium stereochem, 93
compl w F, 104 water exch rate, 218
extrn by TTA/MIBK, 544 Cobalt
oxid states, 511 analysis w diethyldithiocarbamate,
Calomel 573, diphenyldithiocarbazone,
complexation, 68 568, 1-nitroso-2-naphthol, 573
Carbon compl w AA, 172, Alamine-336, 6,
activated, 312, 319 amm, 95, BFA, 172, BZA, 172,
Carbon dioxide Cl, 91, 461, en, 95, F, 90, FTA,
supercritical fluid extn, 444ff 172, PTA, 172, SCN, 9, 462, TFA,
solubilities in, 448 172, TOA/Cl, 156, TTA, 172
Cerium extrn by Acorga P50, 463, amine, 5,
extrn by TLA/NO3, 161, TBP/NO3, 138, 461ff, 625ff, carboxylic acid,
525, in Thorex process, 531 463, 465, CYANEX-272, 466,
hardness, 99 468, b-diketones, 464, D2EHPA,
ionic radius, 45, 494 280ff, 286, 289ff, 305, 322, 465ff,
Cesium hydroxyoximes, 464ff, LIX84I,
analysis, 572, w nitromethane, 572 467, LIX63, 465, Kelex100, 308,
extn w crown ethers (modeling), 463, PC88A, 466, TBP, 290,
690, fr radioactive waste, 541ff 465, 468, TOPO, 465,
in nitrobenzene, 77 Versatic acid, 304, 465, 468
hardness, 98, 99 extn fr scrap alloy, 625
ionic radius, 45 hardness, 99
nuclear waste management, 528 hydration, 88, 91
partitioning, 545ff ionic radius, 45
precip by phosphotungstate, 545 oxidn of diluents by, 311
water exch rate, 218 redox, 458, 464, 468
Chlorine stereochem, 93, 458
basicity, 137 synergistic extn, 465
basicity of ClO4, 137 water exch rate, 218
hardness, 99 Cobalt/Nickel
ionic radius, 45 Caron process, 467
TOA salts, 159 extrn, 458ff, by Kelex/LIX, 308,
Chromium extrn proc fr laterite ores, 466ff,
analysis w Aliquat336/D2EHPA flowsheets: Bulong, 469, Cawse,
(SLM), 582, 8-hydroxyquinoline, 470, Murrin Murrin, 471
573, MIBK, 573, nitromethane, sepn, 471ff
573 Copper
extrn by Alamine-336, 6, 627, analysis, 572, w diethyldithiocarbamate,
D2EHPA NHþ 4 , 618 572, D2EHPA, 581, (SLM),
extn from electroplating baths, 617, diphenyldithiocarbazone, 568,
surface finishing wastewater, 626 dithizone, 572
[Copper] Erbium
cloud point extrn, 586 cloud point extrn, 586
compl w amine, 96, amm, 95, en, ionic radius, 494
95, dithizone, 235, Europium
hydroxyoxime, 235 adducts w bromodecanoic acid,
dump leach, 329 pyridine, terphenyl, terpyridine,
extrn w AA, 177, hydroxyaryloxime, triazine, 182ff, AA, hexone,
11, IPT, 144–5, TTA, 144–5, 188, ITP, 156, quinoline, TBP,
b-diketones, 478ff, LIX54, 479, TOPO, TTA, 149, TBP/BFA,
amine/Cl, 463, carboxylic acid, /FTA, /BZA, /TFA,
473, D2EHPA, 474, hydroxy- 149
oxime, 474ff, Kelex-100, 476ff, compl w acetate, 86, b-diketones,
LIX, 474ff, pyridine-carboxylic 149, F, 90, alkylphosphoric acids,
acid ester (Acorga CLX50), 479, 179
ACORGA, 475, 8-quinolinol extrn by D2EHPA, 500, CYANEX-
reagents, 476ff 301 (modeling), 685, 702
extrn from ammoniacal etch soln, ionic radius, 494
636, brass mill flue dust, 618
extrn costs (LIX64N), 327 Fluorine
extrn eff of org solv, 77, 141–3, 146, basicity, 137
174, of temp, 177 hardness, 99
extrn plants, 477ff ionic radius, 45
hardness, 98, 99
ionic radius, 45 Gadolinium
kinetics/hydroxyoxime, 474 cloud point extrn, 586
ore, 11 ionic radius, 494
solub param, 174 Gallium
stereochem, 93 analysis, 572, w diphenyldithiocarba-
struct, 169, 721 zone, 568, di-isopropyl ether,
water exch rate, 218 572, 8-hydroxyquinoline, 572,
Curium thioxine, 57
compl w Cl, 104 cloud point extrn, 586
extrn, 531ff, by CMPO, 532ff, TTA/ compl w F, 90, OH, 166,
MIBK, 544 Cl in diethyl ether,
extn flowsheet, 543, PUREX, 520, 110
DIAMEX, 537 extrn w AA, 166
oxid states, 511 hardness, 99
partitioning, 539 ionic radius, 45
Cyanide stereochem, 93
recovery using phosphine oxides, water exch rate, 218
632 Germanium
analysis, w dibutyl ether, 572
Dysprosium distribution of GeCl4,
compl w F, 90 124
ionic radius, 494 stereochem, 93
Gold [Indium]
analysis, 572, w diphenyldipyridyl- stereochem, 93
methane, 572, diphenyldithio- water exch rate, 218
carbazone, 568, dithizone, 572, Iodine
polyoxyethylene nonyl phenyl in nitrobenzene, 77
ether (PONPE), 584 basicity, 137
cloud point extrn, 584, 586 elemental, 101, extrn, 124
compl, 219, w Cl, 485 hardness, 99
extrn, 485ff, by amide, 486, ionic radius, 45
amine/Cl (TOA), 485, /CN, 487, Iridium
/S2O3, 488, CYANEX-471X/Cl, analysis w diphenyldithiocarbazone, 568
485, butylcarbitol/Cl, 485, comp w Cl, 481
2-ethylhexanol/Cl, 485, guanidine extrn, 493
(LIX79)/CN, 487, MIBK/Cl, 485, extrn flowsheet INCO, 482, Lonhro,
8-quinolinol (Kelex 100, LIX26), 483, MRR, 484
486, TBP/S2O3, 488 redox, 489
extn flowsheet INCO, 482, Lonhro, stereochem, 93
483, MRR, 484 Iron
hardness, 99 analysis w AA, 572, diethyldithio-
solub of Cl compl, 9 carbamate, 573,
stereochem, 93 triphenylmethane dyes, 568
cloud point extrn, 586
Hafnium compl w amm, en, 95, TOA/Cl,
analysis w TTA, 573 157, F, 90, hydroxyoxime, 235,
extrn by MIBK, 322, TBP/NO3, 317, hydroxamic acid, 235
329 extrn by trialkylamm/Cl, 236, amine/
hardness, 99 Cl, 463, calixarenes, 519,
ionic radius, 45 dibutylcarbitol/Cl, 486,
stereochem, 93 D2EHPA, 474, MIBK/Cl, 487,
Hydrogen ion TBP/NO3,Cl, 512
hardness, 99 extrn w Cu, 473ff
ionic radius, 45 in crud, 313, 317
HCl in ethylether, 136 hardness, 99
HClO4/TBP, 136 ionic radius, 45
HF extn by TBP, 628, CYANEX-923, oxidn of CYANEX-302, 311
628, fr pickling acid baths, 628 stereochem, 93
sulfamate as reductant, 520
Indium water exch rate, 218
analysis, 572 w diphenyldithiocarba-
zone, 568, 8-hydroxyquinoline, Lanthanides
572, b-mercaptoquinoline, 572, analysis, 572, in chromatogr sepn, 59,
thioxine, 572 D2EHPA, 572
compl w F, 90, F, Cl, Br, I, 100 charge density, 168, 178
hardness, 99 compl, 179, w b-diketones, 100, OH
ionic radius, 45 150, TTA/TBP, 118
[Lanthanides] Magnesium
compl distr const, adducts w AA, 153, analysis, 572, w 8-hydroxyquinoline,
155, alkylphosphoric acids, 235 572
contraction, 169, 494 compl w F, 90
coord number, 100, 118, 170, 179 in crud, 313, 468
in crud, 317, 319 extrn by D2EHPA, 474
electron shells, 178 hardness, 99
extn, 493ff, w AA, 210, 215, 225, 701, ionic radius, 45
alkylphosphonic acid ester, 501, stereochem, 93
(modeling), 706, amine, 497, water exch rate, 218
b-diketones (modeling), 698, Manganese
carboxylic acid, 497, 500, analysis w tetraethylene-
CMPO, 532, CYANEX-272, dithiocarbamate, 573,
501, D2EHPA, 329, 498, 500ff, 8-hydroxyquinoline, 573
TBP/NO3, 496, tropolone, 702 compl w amm, en, 95
extrn proc DAPEX, 512, 514 extrn by Alamine-336 6, amine/Cl,
hardness, 99 463, CYANEX272/TBP, 468
ionic radius, 45, 168, 494 hardness, 99
partitioning sep from An, 539 ionic radius, 45
modeling, 691, 697, 698, 701ff, 704, 706 stereochem, 93
scale-up, 333 water exch rate, 218
stereochem, 93 Mercury
water exch rate, 218 analysis, 572, w diphenyldithiocarba-
Lanthanum zone, 568, dithizone, 572
compl w F, 90 compl w Cl, 9, 10, F, Cl, Br, I, 100,
extrn w alkyl phosphoric acids, 179, 125
D2EHPA, 218 extrn by alkyl phosphoric acids,
self-adducts w TTA, ITP, 156 179, monothiophosphoric acid,
Lead 641
analysis w diethyldithiocarbamate, hardness, 99
572, D2EHPA (SLM), 581, ionic radius, 45
diphenyldithiocarbazone, 568, stereochem, 93
dithizone, 573, thioxine, 573 HgCl2 distribution, 124
compl w F, 90 water exch rate, 218
hardness, 99 Molybdenum
ionic radius, 45 analysis w N-benzoyl-N-phenyl-
stereochem, 93 hydroxylamine, 573,
water exch rate, 218 8-hydroxyquinoline, 573
Lithium in crud, 321
analysis, 572, w dipivaloylmethane, extrn by Alamine-336, 6, w U
572, TOPO (ISE), 590 (DAPEX process), 512,
hardness, 99 514, from spent catalyst,
in org solv, 77 641
water exch rate, 218 stereochem, 93
Neodymium Niobium
compl w F, 90 analysis w tetraphenylarsonium
extrn by TBP/NO3, 496 chloride, 573, 8-hydroxyquino-
ionic radius, 494 line, 573
Neptunium stereochem, 93
compl w F, 90, SO4, 104, AA, 146, 169 Nitrate, NO3/Nitric acid
extn, 531ff, by CMPO, 532ff, analysis w tributyl octadecylphos-
diamides, 534ff, TBP/NO3, 525ff, phonium nitrate (ISE), 590
in partitioning, 539 compl w trioctylamine salts, 159
oxidn states, 511 basicity, 137
Nickel extrn by TBP/HNO3, 138, 628,
analysis w diethyldithiocarbamate, CYANEX-923/HNO3, 628
573, dimethylglyoxime, 573, extrn from acid pickling baths,
diphenyldithiocarbazone, 568, 628
TTA, 573
cloud point extn, 586 Osmium
compl, 219, w amm, en, 92, 95, Cl, 92, extn, 481, 485
94, 463, dithizone, 235, TOA/Cl, flowsheets INCO, 482, Lonhro, 483,
156, CN, 92 MRR, 484
extrn by Alamine-336, 6, 138, 463 hardness, 99
extrn by carboxylic acids, 463, 468: stereochem, 93
Versatic acid, 322, 468, OsO4 compl w AA, 124
naphthenic acid, 322
extrn by hydroxyoximes: ACORGA- Palladium
P50, 463, LIX84I, 467, 468, analysis w diethyldithiocarbamate,
LIX84, 623ff, 638ff 574, diphenyldithiocarbazone,
extrn by organophosphorus acids: 568, diphenyldithiourea, 574,
D2EHPA, 279ff, 289, 465, 613, 8-hydroxyquinoline, 574
615, 641, CYANEX272, 466, 471, cloud point extrn w Triton X-100,
PC88A, 466, monothiophos- 585
phoric acid, 641 compl, 219, w Cl, 480ff,
extn proc, 469–71, fr accumulator extrn, 488ff, by amine/Cl, 491,
waste, 638ff, cadmium aminoacid, 490, CYANEX-
electrolytes, 615, fly ash and 471X/Cl, 490, di-n-octylsulphide/
soot, 621, laterites, 466ff, scrap Cl, 488, hydroxyoxime,
alloy, 625, spent catalyst, 488ff, quinolinol (LIX26),
641 491
extrn rate, 463 extrn flowsheet INCO, 482, Lonhro,
hardness, 99 483, MRR, 484, fr nucl waste
ionic radius, 45 (OMEGA proj), 545
redox, 458 hardness, 99
scrubbing, 286ff stereochem, 93
sepn from Co, 458ff Phosphate PO4/Phosphoric acid
stereochem, 93, 94, 458 analysis w butanol, 573
water exch rate, 218 in crud, 317
[Phosphate PO4/Phosphoric acid] Radon

extrn by TBP/H3PO4, 629 elemental, 101
purification from ‘green acid’, 629 extrn elemental, 124
Platinum Rare Earth (see Lanthanide)
analysis w diethyldithiocarbamate, Rhenium
574, diphenyldithiocarbazone, analysis w aliquat-336, 573
568, 572, diphenyldithiourea, stereochem, 93
574, 8-hydroxyquinoline, Rhodium
574 analysis w diphenyldithiocarbazone,
compl, 219, w Cl, 480ff 568, diphenyldithiourea, 574,
extrn, 488ff, by alkylamides/Cl, 492, 8-hydroxyquinoline, 574, TTA,
amine/Cl, 488, 491, 574
aminoacid, 490, MIBK/Cl, 487, compl w Cl, 481
quinolinol (LIX26), 491, TBP, extrn, 480ff, 493, by 8-quinolinol, 493
488ff extrn flowsheet INCO, 482, Lonhro,
extrn proc flowsheet INCO, 482, 483, MRR, 484
Lonhro, 483, MRR, 484 hardness, 99
hardness, 99 recovery from nuclear waste
stereochem, 93, 488, 489 (OMEGA proj), 545
Plutonium stereochem, 93
compl w DOBA, DOiBA, 182, TBP, water exch rate, 218
TiBP, 182, TBP/NO3, 100, 104, Rubidium
trialkylamm/NO3, 236 analysis, 572, w nitromethane, 572
extrn by amide/NO3, 182, phosphate/ hardness, 99
NO3, 182, diamides, 534–5, ionic radius, 45
CMPO, 532–3, D2EHPA, 537, Ruthenium
TBP/NO3, 519ff, PUREX proc, analysis w diphenyldithiocarbazone,
519ff, in partitioning, 544 568, 8-hydroxyquinoline, 574
hardness, 99 cloud point extrn, 586
ionic radius, 45 compl w Cl, 481
oxidn states, 511 extn, 480ff, flowsheet INCO, 482,
redox (PUREX), 520 Lonhro, 483, MRR, 484
Potassium hardness, 99
analysis, 572, w crown ether, 572, stereochem, 93
valinomycin (ISE), 590 water exch rate, 218
extrn by tetraphenylborate, 162–163 RuO4 distribution, 124
hardness, 99
ionic radius, 45 Samarium
water exch rate, 218 compl w acetic, benzoic, chloroacetic,
Precious metals fluorobenzoic, iodoacetic, nitro-
compl w Cl, 480 benzoic and propionic acid, 91
extrn, 480ff ionic radius, 494
extrn flowsheets, 482–4 Scandium
Promethium analysis w mesityl oxide, 572
compl w AA, 152–3, 192 hardness, 99
[Scandium] Technetium
ionic radius, 45 analysis w tetraphenylarsonium
stereochem, 93 chloride, 573
water exch rate, 218 extrn by CMPO, 532, diamides, 535,
Selenium in PUREX proc, 527, SREX
analysis w diethyldithiocarbamate, proc, 538
573, triphenylmethane dyes, partitioning, 527, 539
568 stereochem, 93
Silica Tellurium
in crud, 313ff, 317, 319, 468 analysis w MIBK, 573
Silver extrn by dibutylcarbitol/Cl, 486,
analysis w diethyldithiocarbamate, 572, MIBK/Cl, 487
diphenyldithiocarbazone, 568, stereochem, 93
dithizone, 572 Thallium
cloud point extrn, 586 analysis w diethyldithiocarbamate,
compl w Cl, 480ff 572, 8-hydroxyquinoline, 572
extrn by CYANEX-301, 488, hardness, 98, 99
CYANEX-471X, 488 ionic radius, 45
hardness, 99 stereochem, 93
ionic radius, 45 Thorium
stereochem, 93 analysis w 8-hydroxyquinoline, 572,
Sodium TOPO, 572
in nitrobenzene, 77 extn by AA, 186ff, amines, 518,
extrn by crown ethers (modeling), TBP/NO3,Cl, 518, 528, TTA,
690 118
hardness, 99 extrn from ores, 518, fuels (THOREX
ionic radius, 45 process), 528ff, reproc, 552
water exch rate, 218 hardness, 99
Strontium ionic radius, 45
analysis w polyethylene glycol, 572 Thulium
compl w F, 90, by TTA/TBP, 148, compl w F, 90
TBP/NO3, 525 Tin
extrn w crown ether (SREX process), analysis w diisopropyl ether/SCN,
538, from nuclear waste, 537ff, 572
541–2 extrn by dibutylcarbitol/Cl, 486,
hardness, 99 TOA/Cl, 485
stereochem, 93 stereochem, 93
water exch rate, 218 Titanium
Sulfate analysis w monolaurylphosphoric
trioctylamine salts, 159 acid, 573, triphenylmethane dyes,
568
Tantalum hardness, 99
analysis w MIBK, 573 stereochem, 93
stereochem, 93 water exch rate, 218
Tungsten [Uranium]
analysis w MIBK, 573 oxidn states, 511
stereochem, 93 production, 11, 511
redox reactn, 260
Uranium sepn fr Pu, 520
activity coeff in ClO4, 253–275, Cl, struct w dibromosalicylate, 92,
259 water exch rate, 218
adduct compl, chelation, 114, 155
analysis w diethyldithiocarbamate, Vanadium
572, 8-hydroxyquinoline, 572, analysis w cupferron, 573,
quaternary amine, 572, TTA, 572 tetraphenylarsonium
cloud point extn, 586 chloride, 573, TTA, 573
compl w carbonate, 272, F, 90, 104, extrn by D2EHPA/TBP, 622ff,
Cl, SO4, 104, NO3, 10, AA, 189ff, monothiophosphoric acid, 641
selfadd, 156, HFA, 101, TTA, extrn fr fly ash and soot, 621, spent
116, selfadd, 156, D2EHPA, 723, catalyst, 641
NO3/TBP, inner/outer sphere, hardness, 99
180, struct, 181 stereochem, 93
composn of irrad U, 518ff water exch rate, 218
coord, 180–1
in crud, 311, 314, 316ff Xenon
dissolution, 519, 522 elemental, 101
extrn by alkylpyrophosphoric acids, extrn, 124
516, amide/NO3, 189, 322, 325,
calixarenes, 517–8, diamides, Ytterbium
178, 528, D2EHPA/TOPO, compl w F, 90
515ff, DOBA, DOiBA, 182, ionic radius, 494
organophosphate/NO3, 182, Yttrium
salicylic acid, 92, TBP, TiBP, 182, analysis w diantipyrylmethane,
TBP/NO3, 100, 113ff, 322, 513, 572
515, TBPO/NO3, 116, TLA/SO4, extrn by NO3/TBP, 496, 497
157, TTA/TBP, 116, 185, TTA/ modeling, 699
TOPO, 116, 185, NO3+TBP, hardness, 99
236, fr phosphoric acids, 319, ionic radius, 45, 494
515, fr sea-water, 517 stereochem, 93
extrn plants, 524 water exch rate, 218
extrn proc AMEX, 512, PUREX,
518ff, DAPEX, 512, Thorex, Zinc
528ff analysis w 8-(benzene-sulfonamido)-
hardness, 99 quinoline, 575,
hydration, 114 diethyldithiocarbamate, 572,
ionic radius, 45 diphenyldithiocarbazone, 568,
modeling, 690 dithizone, 572,
nucl fuels, 507 thioxine, 572
ore leaching, 511ff cloud point extrn, 586
[Zinc] ORGANIC COMPOUNDS

compl w amm, b-diketones/
phosphoryls, 187, 698, en, Acetic acid, acetate, 135, 137, 436,
95, 172, F, 90, 172, BFA, 438
BZA, FTA, HFAA, PTA, 172, Acetone, 101
TBP, TFAA, TOPO, TTA, Acetonitrile, 101, 570
147, 187, TOPO/CN, 150, 151 Acetylacetone (AA or HAA), 30, 39, 75,
dithizone, 235 125ff, 130–132, 140, 146, 164,
extrn by AA, 146, 176, AA/TOPO, 175, 573, 698, 721
Alamine-336, 6, alkylphosphoric Acid-base indicator dyes, 576
acids, 179, amine/Cl, 463, 485, Adogen-306, 368, 436
CYANEX-272/TBP, 468, Adogen-364, 436, 512ff
dibutylcarbitol, 486, D2EHPA, Alamine-336, 6, 306, 436, 512ff, 627
468, 613ff, 617ff, 619ff, 627, 664, Aldoxime, 460, 475
(microemulsion), TBP, 620, Aliquat-336, 436, 497ff, 573, 582
TOPO/HFA, TOPO/TFA, 147, Alkaloids, 593 (LLPC)
177 Alkylammonium salts, 220
extrn proc fr acid pickling liquor, 620, Alkylarylsulfonic acid, 220
brass mill flue dust, 618, electro- Alkylhydroxamic acid, 220
plating baths, 617, spent bat- Alkylphosphoric acids, 220
teries, 627, weak acid effluents, Alkylnaphthalene isomer sepn, 432
614 Alkyl sulfides/sulfonates, 462, 583
hardness, 99 Amberlite LA-2, 430
modeling, 697 Amides, 462, 492
selfadd w IPT, 156 Amines, 96, 137, 311, 322, 323
solub param AA, 174 Amine extrn, 279, 462ff, 490, 493, 518,
solv eff on distrn AA, 146, 172, 509, 614, 625
174 Amino acids, 437, 490, 593 (LLPC)
stereochem, 93 AOT (sodium sulfosuccinic acid
struct w AA, 169 di-octyl ester), 665ff
water exch rate, 218 Aromatic/aliphatic sepn proc, 427–428
Zirconium Arsenyls/arsines, 137
analysis w TOPO, 573, Azo dyes, 568
triphenylmethane dyes, 568
cloud point extn., 586 Benzoic acid, 135, 720
extrn by NO3/TBP, 318, 320, 325, Benzonitrile, 101
329, 525, diamides (DIAMEX), Benzoylacetone (BA, BZA, HBA),
534 125–7, 130–1, 721
extrn w U (AMEX), 515, in N-Benzoyl-N-phenyl hydroxylamine,
THOREX proc, 528ff 573
extrn in crud, 318ff Benzoyltrifluoroacetone (BTFA),
hardness, 99 125–7, 130–1
ionic radius, 45 Butanol syst, 372, 410, 438, 573
partitioning, 543 Butyl acetate syst, 410, 429, 434, 569
stereochem, 93 Butylamine, 101
n-Butylnitrile, 101 Diethylether, 101, 136

Butyric acid, 135 Diethyldithiocarbamate, 724, 572ff
Diethyldithiophosphoric acid, 586
Calixarenes, 511, 517ff, 541ff, 689 Di-2-ethylhexylphosphoric acid,
Carbamoylmethylene phosphonate, (HDEHP, D2EHPA) 130, 140,
532, 722 217, 304–5, 321–3
Carbamoylphosphinate, 532, 722 D2EHPA extrn w, 217, 459, 723, Al,
Carbon tetrachloride extrn, 573 664; amino acids, 437; An, 509,
Carbonyls, 137 537, 540, 543; Bi(III), 664; Cd,
Carboxylic acids, 140, 304, 463, 459 641; Co/Ni, 287, 290, 466; Cr,
Chelating dyes, 568 582, 618; Ln, 498ff (sepn), 499,
a-Chymotrypsin, 666 592; M, 474, 572, 581; Ni, 613,
Cinnamic acid, 140 615ff, 641; U, 280, 512ff; V, 622;
Citric acid, 435 Zn, 613, 614, 617, 619ff, 627, 664
CMPO (see Octylphenyldiethylcarba- D2EHPA/TOPO, 515ff
moylphosphine oxide) Diethylene glycol dibutyl ether (see
Cresol isomer sepn, 433 Dibutylcarbitol)
Crown ethers, 220, 511, 538, 568, 572, Dihexyldiethylcarbamoylmethylenepho-
582, 706, 724 sphonate (DHDECMP), 509, 722
Cryptands, 689 Di-n-hexylsulfide, 489, 490
Cupferron, 130, 140, 573ff, 722 Di-iso-propyl ether, 78, 572
CYANEX-272, 324, 459, 466, 468, 471, b-Diketones, 140, 460, 464, 476, 684,
501 572, 704
CYANEX-301, 311, 459, 488, 551, 685 Dimethylalkylphosphine oxide, 583
CYANEX-302, 311, 459 Dimethylethylacetamide, 101
CYANEX-471X, 459, 485, 488, 490 Dimethylformamide, 101
CYANEX-923, 459, 628, 632 Dimethylglyoxime, 573, 599
Cyclodextrin, 432 Dimethylsulfoxide, 427
Dinactin, 590 (ISE)
iso-Decanol, 323 Dinonylnaphthalene sulfonic acid
Dextran, 440 (DNNS), 164, 437, 459, 720
N,N 0 -Dialkylamides, 528ff Dinonylnaphthyl sulfonic acid (see
Dialkyl phosphonate, 722 Dinonylnaphthalene sulfonic
Diantipyrylmethane, 572 acid)
Dibenzo-p-dioxines, 566 Dioctylpyrophosphate, 140
Dibenzoylmethane (DBM), 125–7, 721 Dioxin, 101
2,6-tert-butyl-4-methylphenol (BHT) Diphenyldipyridylmethane, 572
[antioxidant], 311 Diphenylthiocarbazone, 140
Dibutylbutylphosphonate (DBBP), 536ff Diphenyldithiourea, 573ff
Dibutylcarbitol, 485ff, 508 Diphosphinates, 532
Dibutylether, 572 Diphosphine oxides, 532
N,N 0 -di-n-butyloctanamide, 485 Diphosphonates, 532
Dicarbollide, 541ff Dipivaloylmethane, 571
Dicarboxylic acids, 140 Dithiocarbamate, 140, 573, 724
Di-iso-decylphosphoric acid, 545 Dithiophosphonic acid, dialkyl-, 724
Dithizone, 140, 568, 572ff, 599 LIX63, 460, 465, 474, 489
Dodecane, 427 LIX64N, 460, 474
Dyes, 568 LIX79, 488, 487
LIX84, 460, 476, 623, 638
Enzyme extrn, 442ff, 669ff LIX84I, 460, 467, 468, 475
Ethanol extrn, 438
Ethane, supercrit, 446 b-Mercaptoquinoline, 572
Ethers, 462 Mesityl oxide, 572
Ethyl acetate/acetic acid/water system, Methanol, 570
426 Methyl isobutyl ketone, (MIBK/
Ethylenediamine (en), 96, 101 hexone), 485, 487, 508, 569,
Ethylenediamine tetraacetate, (EDTA), 573ff, 723
96, 217 Methyl carbonate, 427
Ethylene carbonate, 101 4-Methyl-2-pentanone (see Methyl
2-Ethylhexanol, 281, 310, 485 isobutyl ketone)
2-Ethylhexylphosphonic acid 2- N-Methylpyrrolidone, 427
ethylhexylester (PC88A), 459, MIBK (see Methyl isobutyl ketone)
466, 501, 592 Modifiers, 310
Monolaurylphosphoric acid, 573
Fatty acids, 125–6, 593 (LLPC) Monothiophosphoric acid (see also
N-Formylmorpholine, 427 CYANEX-302), 641
Glycerides sepn (SCF), 449 Naphthalenes, 432, 446

Glycols, 593 (LLPC) Naphthenic acid, 315, 322, 459, 465
Nitrobenzene, 162
Hexafluoroacetylacetone (HFAA), 125–8 Nitromethane, 101, 572
Hexanol, 438 Nitrosohydroxylamine, 140
Hexone (see Methyl isobutyl ketone) 1-Nitroso-2-naphthol, 130, 140, 573, 721
Hydrocarbon diluent, 466 N-Nitrosophenylhydroxylamine, 722
Hydroxamic acid, 140 Nitrosyls, 137
a-Hydroxyoxime ext, 460 Nonactin, 590 (ISE)
b-Hydroxyaryloxime, 11, 460, 489, 684 Nonyl phenol, 466
8-Hydroxyquinoline, (oxine HOQ or
8-quinolinol), 125–127, 130–131, Octylphenyldiethylcarbamoylphosphine
140, 460, 466, 476, 491, 493, 569, oxide (CMPO), 509, 532ff
572, 599, 721 Octylphenylphosphoric acid (DOPA),
516
Kelex-100, 308, 311, 322ff, 464, 476, 486 Octylpyrophosphoric acid (OPPA), 516
Ketoximes, 460, 475 Org amines, 157ff
Oxime, 465, 722
LIX (hydroxyoxime extractants), 308, Oxine (see 8-Hydroxyquinoline)
310, 315, 322, 323, 460, 466
LIX26, 460, 486, 491ff PAH (see Polyaromatic hydrocarbon)
LIX34, 460 PAMP (see 2-(2-Pyridylazo)-
LIX54, 460, 479, 586, 636 5-methylphenol)
PAN (see 1-(2-Pyridylazo)-2-naphthol) Pyridine carboxylic ester ext (ACORGA

PAP (see 2-(2-Pyridylazo)-phenol) CLX-50), 479, 684
PC88A (see 2-Ethylhexylphosphonic Pyrophosphates, 140
acid 2-ethylhexylester)
PCB (see Polychlorinated biphenyl) Quaternary amines, 158, 461, 572
PCDF (see Polychlorinated 8-Quinolinol (see 8-Hydroxyquinoline)
dibenzofuran)
PEG (see Polyethylene glycol) Salicylic acid, 130, 140, 720
Penicillin extrn, 429 Secondary amines, 158, 461, Pd, 491
Perfluorobutyric acid, 140 Sodium sulfosuccinic acid di-octyl ester
Pesticides anal, 593 (LLPC) (see AOT)
Phenanthroline, 723 Sulfolane, 427
Phenols, 434, 579, 585, 593 (LLPC) Sulfonic acids, 459, 463
Phenolsolvan proc, 434 Sulfuryls, 137
1,3-Propanediamides, 534, 534ff, 537
Podands, 689 (modeling) TATB (see Tetraphenylarsonium
N-Phenylbenzohydroxamic acid, 722 tetraphenylborate)
Phosphates, 462, 589 (ISE) TAN (see 1-(2-Thiazoylazo)-2-naphthol)
Phosphines, 137 Terpyridyl, 684
Phosphine oxides, 462 Tertiary amines, 158, 461, 493
Phosphine sulfides, 462 Tetraalkyl ammonium, 77
Phosphinic acids, 140, 459, 499 Tetraethylene glycol, 427
Phosphobetaine, 583 Tetrahydrofuran, 101
Phosphonic acids, 459, 499 Tetramethylene dithiocarbamate, 573
Phosphoric acids, 140, 459, 465, 499 Tetramethylene sulfone, 101
Phosphonates, 722, 724, Ln, 497 Tetraphenylarsonium chloride, 573
Phosphoryls, 137 Tetraphenylarsonium tetraphenylborate
Polyaromatic hydrocarbon (PAH), 566, (TATB), 46
585 Tetraphenyl compounds, 46, 162
Polychlorinated biphenyl (PCB), 566, Thenoyltrifluoroacetone (TTA), 130,
585 140, 164–5, 572ff, 721
Polyethylene glycol, 427, 440ff, 572 1-(2-Thiazoylazo)-2-naphthol (TAN),
Polyoxyethylene ethers, 582 586
Polyoxyethylene nonyl phenyl ether Thioacids, 140
(PONPE), 585ff Thiocyanate (SCN), 9, 137
Primary amines, 158, 461 Thiophosphoric acids, 459
Propionic acid, 135 Thioxine, 572ff
Protein partitioning, 441, 666ff THM (see Trihalomethane)
iso-Propylether, 434 Toluene, 571
iso-Propyltropolone (IPT), 720 Toluene/water, 371ff, -acetone, 410,
1-(2-Pyridylazo)-2-naphthol (PAN), 586 -phenol, 368ff, 372ff
2-(2-Pyridylazo)-5-methylphenol Tributyl-octadecyl-phosphonium
(PAMP), 586 nitrate, 590 (ISE)
2-(2-Pyridylazo)-phenol (PAP), 586 Tributylphosphate (TBP), 11, 30, 39,
Pyridine, 101 304, 309, 315, 521ff
Tributylphosphate (TBP) extrn w HNO3, agg aggregation

78, 138, 220, 628, Purex, 100ff, amm ammonium
w acids, 136, 436, 723; Au/S2O3, anal analysis, analytical
488; Co, 465; H3AsO4, 632; HF, appl applications
628; H3PO4, 629ff; Ln, 495ff, 593; aq aqueous
M, 572; Pt, 489; Pu, 508, 519ff; calc calculation(s)
U, 515, 519ff, Zn, 620 centr centrifugal, centrifuge,
iso-Tridecanal oxime, 465 centrifugation
Triethylamine, 101 chem chemical(stry)
Trifluoroacetylacetone (TFAA), 721 chromat chromatography
Trihalomethane (THM), 566 class classification
Trilaurylamine (TLA), 158, 509 coeff coefficients
Trilaurylmethylamine, (TLMA), 161 comp compound(s)
Trimethylsulfoxide, 101 compl complexe(s)
Trioctylamine (TOA), 11, 158–9, 437, compln complexation
485, 491 composn composition
Trioctylmethylammonium chloride conc concentration
(TOMAC), 666 const constant(s)
Trioctylphosphine oxide (TOPO), 128, contt contactor
129, 723, 590 (ISE), 684, 702 coord coordination
extrn w carboxylic acid, 580; Co, corrl correlations
465; H3AsO4, 632; HCN, 632; def definition
Ln, 590; M, 572ff degrad degradation
Triton-X, 585ff detn determination
tris-Pyridyl-triazine, 550, 547ff, 551 diag diagrams
diff diffusion(-al)
Union Carbide proc, 427, 430 displ displacement
diss dissociation
Valeric acid, 135 dissol dissolution
Valinomycin, 590 (ISE) distrn distribution
Versatic acid, 304, 322, 323, 459, 468, DN donor number(s)
497, 500 econ economics
eff effect
Xanthates, 140 effic efficiency
Xylene, 431, 571 electr electron, electrode,
electrolyte
en ethylenediamin
ABBREVIATIONS eql equilibrium
eqn equation(s)
Abbreviation Meaning equip equipment
AAS atomic absorption exch exchange
spectrophotometry exp experiments
act activity extrn extraction(s)
add adduct fact factor
ads adsorption/absorption form formation
fr from purif purification

funct function reactn reaction
gravit gravitational rel relations
hydrn hydration reproc reprocessing
interact interaction SCF supercritical fluid
irrad irradiated sel selective, -ity
ISE ion selective electrodes selfadd selfadduct
kin kinetic sepn separation(s)
LLPC liquid partition SLM supported liquid
chromatography membranes
mech mechanics, mechanisms soln solution
met metal, metallic, solub solubility
metallurgical solv solvent, solvation
meth method spectr spectroscopy, -metry
misc miscibility stab stabilit(y)ies
mixn mixing struct structure(s),
mixt mixture(s) stereochemistry
nucl nuclear syst system/systematics
org organic(s) techn technique,
oxidn oxidation technological
param parameters temp temperature
pot potential thermod thermodynamic
prep preparation tox toxicity
proc process(es) transf transfer
proj project vap vaporization
prop properties w with
Subject Index
Accelerated solvent extraction, 567 Analytical chem (see Extraction in

Acceptor atoms, 64, 97 analytical chemistry)
Acid adducts, 135ff, 150 Anion exch extrn, 121, 461
diss, 120, 133 Aphrons (see Colloidal liquid aphrons)
hydrn, 136, 152 Aprotic, 28
Acid-base reactions, 101 Aqueous two-phase extrn (see
Activation analysis, appl, 570 Extraction: aqueous
Activation energy of extrn, 224 two-phase systems)
Activity coeff/fact, 50ff, 84, 110, ARD contactor (see Equipment/
253–275 differential columns)
model for sepn eql, 420ff Archimedes number, 388
Adduct formation, 64, 100, 113ff, Artificial intelligence/neural network,
119ff, 148–9, (modeling), appl, 705ff
699 Association const, 61
Affinity partitioning (see Extraction: Associative mech, 214
aqueous two-phase systems)
Agitated extractors (see Equipment: Back-extraction (see Stripping)
differential contactors) Baes and Mesmer act coeff, 254,
Aggregation number, 64, 661 267, 270
AKUFVE techn, 166, 195, 350 Basicity, 64, 119, 137, 149, 684
Alchemy, 6ff Batch exp, 193, 301, 329, 344, 356,
Alloy scrap (see Extraction of ) 365, 450, 485, 494, 538,
Amex process (U), 512ff 562, 598, 621, 669
Amine base, 157 Batteries (see Extraction of )
The Key for the abbreviations contained within the index can be found on page 739.
741
742 Subject Index
Bidentate, 96 [Colloidal liquid aphrons]

Binary syst, 65 appl, 674
Biological molecules (see Extraction of ) extrn kin, 674
Biotechn appl, 19, 21, 429, 435ff, 449, form, 673
664ff, 675ff prep/struct/stab, 670ff
Bjerrum, J., 87, 185, theory, 61 Column contactors (see Equipment/
Bond orbitals, 91ff differential columns)
Born eqn, 89 Compensation eff, 103
Brønsted-Guggenheim-Scatchard, 254 Complex const, def, 83
Bunsen coeff, 41 form, 227ff
param, 168
Calculation of eql const, 185 Component def, 34, 65
Calorimetry, 102 Compressibility, 29
Capital costs (see Process economics) Concentration scales, 53
Carboxylic acid Ka param, 168 Conditional eql const, 59
Carrier-mediated membrane transfer, Conductometry, 248
656 Const ionic medium, 86, 110
Cascade (see Counter-current; Contactor, def, 12, 284ff, 290, 292ff,
Cross-current) (see also Equipment)
Cavity form, 38, 40, 76, 125, 170, 689, 694 class, 295
Cementation, 455 Continuous flow centr, 195
Central atom, def, 82 Coordination number, 87, 93, 118ff
Centr sepn, 194 (see also Equipment/ Coordination sphere, 212, 215,
centrifugal extractor) 217
Charge density, 98, 119, 149, 168ff, 179 Correlation extrn param, 167ff
Cheland, def, 117 Correlation Kex and struct, 177ff
Chelate, 96, 120ff, 139ff Correlation KD and mole volume,
bite, 177, 684 169
ring form, 106, 177ff Counter-current, def, 14
Chelation thermod, 106 calc, 347ff
Chelator def, 117 Covalency, 91, 98
Chem pot, 48ff, 262 Critical temp, 27, 35
Chi-square test, 192 Critical micelle concentration (cmc),
Clathrates, 162 582, 661
Chromatographic appl, 590ff Cross-current, (cross-flow) def, 14
Claw, 178 calc, 344ff
Cloud point extrn, 582ff Crown ethers, 162, 683
appl in analysis, 584ff Crud def, 313
Cluster integral method, 268 causes, 313ff
cmc (see Critical micelle concentration) description, 317
Coalescence (see Drop coalescence) treatment, 318
Co-current, def, 14 prevention, 318ff
Cohesive energy, 31ff, 170 Cubic compl, 93
Colloidal gas aphrons (CGA), 671 CURE concept (clean use of reactor
Colloidal liquid aphrons (CLA), 670 energy), 545
Subject Index 743
Dapex proc (U), 512 Donor-acceptor, 64, 77

Davies eqn, 254, 271 Drop form, 226, 245, 370ff
Debye-Hückel, 57, 254ff coalescence, 288, 292ff, 406ff
Debye unit, 29 size, 292, 371ff
Decaffeination, 27
Degradation (see Extractant Effective core potential (ECP),
degradation) 685ff
Degrees of freedom, 34, 71 Effluent treatment, 614ff (see also
Dehydration, 103 Extraction from)
Density, solv, 29–31 Electrolytes, 55ff
Density functional theory (DFT), 683 Electron distr, 77
Diamex proc (trans-uranic), 543ff Electron shell, 178
Dielectric const, 30ff, 89 Electronegativity, 98, 149
Differential extrn (see Extraction) Electroneutrality, 55
Diffusion, 204ff Electron–pair donation, 40, 42, 65,
coeff, 207 137
diff cells, 243 Electrolytes (see Extraction from)
film, 205, 208, 218ff Electrostatic eff, 89
layers, 205 bond, 98
regime, 223ff, 234ff Electrowinning, def, 468
Diluents, 12, 28, 175, 288 Element analysis, meth, 572ff
aq solub, 310 Emulsion (surfactant) liquid membrane,
stab, 311 653, 656ff
Dimerization, 9, 134 Energy of hydrn, 45
Dimers, cyclic, 33, 63 Energy of mixn, 36, 47ff
Dipole bonds, 32, 63 Enthalphy of compln, 102
moment, 29ff distrn, 176,
Dipole-dipole interaction, 63 react, 101, 180
Dispersion, of phases, 292ff, 370ff Entrainment, solv, 312
Dispersion forces, 32ff, 170 Entropy of compln, 102
Dissociation, 68 mixn, 47
Dissociative interchange mech, 213 reactn, 180
Distillation equip, 7 solv, 41, 70
Distribution (see also Partition) Entropy-entropy compensation, 103
Distribution law/ratio/const/rel, def, Entropy driven react, 103
1, 2, 10, 50, 73, 78, 110ff, 172, Environmental appl, 20, 302ff, 433ff,
341, 651 610ff
isotherm, 282 Environmental legislation–metal
ratio, exp detn, 193ff discharge, 646ff
temp eff, 177 Enzyme extrn, (see Extraction: aqueous
Distribuend, def, 9, 110 two-phase systems; Micellar
Donor atoms, 97 extraction)
ligand eff, 179, Equation of state, 421
numbers (DN), 64, 101, 119, 149 Equilibria (modeling), 692ff
prop, 28, 64, 77 Equilibrium proc, def, 4
744 Subject Index
Equipment/continuous contact [Extraction of: alloy scrap]

differential extrn, 360ff flue dust, 618, 621ff
stagewise, mixer-settler, 12, 294, 356ff; pickling liquors, 620, 628
494, 616, 624, 638 Extraction const, def, 115
Equipment, extractor design, 368ff Extraction from: effluents, 614, 626
agitated extractors, 379ff electr, 615, 617, 632
differential columns, types of, 295 printed circuit board etchant,
Graesser/RTL/raining bucket, 298, 636ff
316, 669 Extraction diag/curves, 111, 123
Karr column, 375 differential, 360ff
Kuehni/Kuhni column, 380 effic, 363
Oldshue-Rushton, 360 fact, coeff/def, 3, 10
pulse columns, 375ff, 621, 630 of inert substance, 124
rotating disc (RDC/ARD), 316, 380 org acids, 129
sieve tray/plate extractors, 316, mech (modeling), 689ff
372ff param, 147, 167
spray column, 369, 669 polynomials, 198ff
Equipment/centr extractor, 296ff, 315ff, principles, 112
324, 382ff, 397, 632 proc overviews, 119
Equipment, micellar extrn, 669 proc modeling, 689ff, 702ff
Ewald summation meth, 688 reactn steps, 120
Excess chem pot, 51 reagent, def, 121
Excess energy of mixn, 52 stage-wise, 356ff
Exchange mech, 214ff vessel, 194
External errors, 192 w adduct form, 45
Extractant, def, 10, 121 Extraction: aqueous two-phase
aq solub, 303–309 systems, 439ff
commercial extrn, 459–61 econ, 443
degrad, 303, 310 for enzyme extrn, 442
losses, 320 for protein extrn, 441, 668
screening, 278 proc, 443
tox, 322ff Extraction kinetics: aphron extrn, 674ff
Extraction in anal chem, 560ff micellar extrn, 667
preconcentration, 561ff Extraction regime, 223
sample prep, 560ff Extractive bioconversion, 442
sepn, 563ff
of met, 565 Feed, def, 13
of org comp, 566 Fick’s first law, 207
trace elements, 564ff Fick’s second law, 207
Extraction of: alloy scrap, 625ff, 633ff Film layers, 204
batteries, 627, 636 Film thickness, 207
biological molecules, 435ff, 664ff First order reactn, 210f
aminoacids, 437 Flooding (of extractors), 389ff
carboxylic acids, 435 Flotation, 20
ethanol, 438 Flow injection analysis, appl in, 597ff
Subject Index 745
Flow-sheet def, 13 Hybrid bonds, 91ff

Flue dust, extrn, 618 Hydration, 38, 103ff, 119, 134
Formation const def, 82 energy, 45
curves, 84ff, inner sphere, 697ff
estimation of, 186 layers, 46
Four-phase systems, 5 number, 87
Fragment model, 73ff outer sphere, 698ff
Free energy of change, 114 Hydrogen bonding def, 28ff, 33, 63
compln, 102ff Hydrometallurgy, def, 19
hydrn/solv, 694 appl, 611ff, 641
Free ligand conc, 83, 113, 133 copper acid leach SX-EW plants,
Fugacity, def, 50 477
Function of state, def, 41 flow sheet, 457
laterite ores, 467ff
Gas solub, 66 Hydrophilic compl, 162, 164ff
extrn, 124 Hydrophobic compl, 73ff, 109
Geometric eff, 91 Hydrophobic-hydrophilic molecules,
Gibbs energy of form, 69 218
dissol, 68 Hydroxy compl, 123
mixn, 36
solv, 40, 68 Ideal dilute soln, 47
transf, 60, 69 Ideal gas law, 40, 66
Gibbs excess energy, 423 Ideal mixing, soln, 47
Gibbs eqn, 219 Immiscibility def, 35
Gibbs-Duhem rel, 57 Inert compl, 212
Graesser contactor, 298 In-line mixer, 298
appl in micelle extrn, 669 Inner/outer sphere compl, 105, 180
Graphic slope anal, 188 Inorg acids, 136
Guggenheim eqn, 254ff In-situ leaching, 20
Gutman donor number (DN), 64 Interface, 204, 217ff, 229ff
Interfacial ads, 231
HAB compl extrn (type IIB), 135ff area, 292
Hansch and Leo, 74 compl, 228
Hansen solub param, 170 conc, 208, 219ff, 229, 236, 238
Hard/soft acid/bases, 87, 97ff extrn model, 136, 690ff
(see also HSAB theory) flux, 652
Harned’s rule, 58 pressure, 220
Heat of soln, 70 reactn, 128, 204, 235ff
vap, 33, 170 struct, 222
Heat transf rate, 224 tension, 218, 404
Henry’s law, 49ff water, 221
Hildebrand param, 33 Internal consistency, 192
Hohenberg-Kohn (HK) theorems, 683 Ion interaction coeff, 259ff, 263ff, 271
Homogenous reactn, 232 Ion pair extrn (type IIIE), 121, 157,
HSAB theory, 179, 510, 683 162ff, 461
746 Subject Index
Ion pair form, 61ff, 214 Liquid membranes, 653ff

Ion sel electr (ISE), 21, 585ff appl in anal, 576ff
Ionic diss, 68 clean-up, 578
Ionic medium method, 58 conc, 579
Ionic radii, 45, 89, 93 mech of solute transf, 654
Ionic strength, 56ff, 85ff, 104, 132, Liquid–liquid partn chromat, 20, 591ff
255 Liquid prop, 28
Ionization pot, 97 Liquid scintillation counting, appl,
ISEC conf, 23 570
Isotope dilution radiometry, 570 Liquid–liquid distr def, 1, 27, 112
Isotropic def, 34 Liquid–liquid interface, 204
LISOL, 196
Jahn-Teller eff, 95, 682 Literature, general, 23
Job’s method, 185 LLPC (see Chromatography:
liquid–liquid partn chromat)
KDR and KDC corrl, 169 Loaded, def, 13
Karr column (see Equipment/
differential columns) Mallinckrodt Chem. Co., 11, 508
Kinetic regime, 223, 226ff MAN compl extrn (type IIIB), 139ff
Kinetics, 17, 203ff, Manhattan proj, 11, 508
of extrn, 284 Marangoni convection/effect, 404ff
Kolmogoroff theory, 381 Marine chem, 272
Kuhni/Kuehni column (see Equipment/ Mass transf, def, 34, 368ff, 398
differential columns) coeff, def, 401
in centr field, 402ff
Labile compl, 212 in gravit field, 398ff
Langmuir ads law, 230 MAzBb compl extrn (type IIIC),
Lattice energy, 44 145ff
enthalpy, 70 McCabe–Thiele diag, 283ff, 350ff
Least squares method, 191 McGowan–Abraham volume, 76
Levich eqn, 247 Mean activity coeff, 255
Lewis acids/bases, def, 65, 136 Mechanism of extrn, 689ff (modeling)
Lewis cell, 225, 245 liquid membrane extrn, 654
Ligand displ, 217 microemulsion extrn, 662
Ligand, def, 82 Medium eff, 60–61
basicity, 684 Membrane extrn, 18 (see also Liquid
diss const, 167 membranes)
field stabilization, 94 Membrane filter, 246
number method, 186 Metal chelate form, 141
Limiting diss mech, 213 Metal compln def, 81, 97
Limiting value method, 188 Micellar extrn, 659ff
Linear plots, 188 appl, 668ff
Lipophilic, 73, 109 biological molecules, 664
Liquid anion exch (type IIID), enzymes, 669
156ff eff of system param, 665ff
Subject Index 747
[Micellar extrn] Nernst distr law, 9ff, 50,

in anal, 582ff eqn, 262
kin, 667 Nernst films, 205
metals, 663ff Nomenclature, 22
Microemulsions, 659ff Nondispersive solv extrn, 659
biological molecule, 664 Nonelectrolyte extrn, 76, 120,
solute extrn, 662 solub, 66ff, 73
with chem reactn, 663 Nonideal mixt, 50
Migration exp, 180 Nonpolar, def, 29
Miscibility, 33ff Nonreactive syst, def, 11
gap, 71 NRTL eqn. (L-L eql), 423
Mixed compl, 123 Nuclear fuels, 100
Mixed regimes, 241 Nucleophilic subst, 213
Mixer-settler (MS) (see Equipment/ Numerical methods, 191
continuous contact: stagewise)
Mixing (see Phase mixing) Octahedral compl, 93, 217
Modifiers, def, 158, 279, 282, 288ff, OECD-NEA-TDB proj, 253
292, 314, 465, 475 OMEGA concept, 545
aq solub, 309 Oldshue–Rushton column (see
in crud, 320 Equipment/differential
optimization, 327 column)
oxidation, 317 Oligomers, 63
tox, 323 Operating costs (see Process econ)
Molal conc, 53ff Operating line, 349
Molar scale, 53, 54 Optical fiber chem sensors, 575
Molar volume, 29ff Orbitals 3d, 95
Mole fraction, def, 36 Org amines, 157ff
Mole percent, def, 34 Organic solv prop, 60
Molecular dynamics modeling (MD), Organic phase reactn model, 137
688ff Osmotic coeff, 57, 258
Molecular mech modeling (MM), Outer sphere compl, 105, 214
680ff Overall stab const, def, 82
Molecular simulations, 688ff
Molecular vol, 169 Partition (see also Distribution)
Monte Carlo method, 688 Partition coeff, def, 10
Moving drops, 243, 249 ion-pair, 695
Multistage extraction (see Counter- metal-chelates, 696
current; Cross-current; radioactive wastes, 544
Extraction) solute, modeling, 694ff
Murphree stage effic, 285 Peng-Robinson eqn (see Equation
Mutual misc, 36 of state)
MXzBb compl 150–2 Perals method (anal), 544
Percentage extrn, 3, 112
Nano exp, 18 Petrochemical processing, appl, 426ff
NEA thermochem data base, 253, 273 Pharmaceutical appl, 428ff
748 Subject Index
Phase boundary, 4 Quantum mechanical (QM) calc, 682ff

in countercurrent flow Quaternary mixt, 71
centr field, 402ff
gravit field, 398ff Radioactive, half-lives, 196,
diagr, 70–73, 660 treatment proc, 531ff
eql, 419 waste, 19
flow, 387ff Raffinate, def, 13
mixn, 393ff Raining bucket contactor
ratio, 341 (see Equipment/differential
rule, 34, columns)
sepn, 406ff Raoult’s law, 49
Phase dispersion (see Dispersion) Rate laws, 209ff
Pickling liquors (see Extraction Rational conc scale, 53
from) RDC contactor (see Equipment/
Pi-electrons, 125, 131 differential columns)
Pilot plant, 299ff, 409 Reactive extrn, 11
Pitzer eqn, 254ff, 267 Reactn models, 113
Pitzer–Brewer eqn, 264 Reference electr/solv, 46
Planar compl, 93, 217 Reflux configuration in mixer-settler,
Polarity, 29ff 494
Polarizability, 98 Refractive index, 32
Polarography, appl, 570 Regeneration, 18
Polyaphron, 670 Regression, anal, 191
Polydentate molecules, 140 Regular mixt, 36
Polynuclear, def, 82 soln def, 170
Preconcentration, 564ff Relative permittivity, 31, 61ff, 68, 125
Pressure-volume work, 33 Reprocessing, 11, 507
Printed circuit board etchant Residence time, 394ff
(see Extraction from) in centr extractors, 397
Process econ, (see also Extraction: in gravit extractors, 395
aqueous two-phase systems) Reverse micelles, 659ff
capital costs, 324 Reynolds number, 388
operating costs, 325 Rotating diff cell, 243, 246
Process optimization, 327ff Rotating disc contactor (see
Process testing–small scale, 297 Equipment/differential columns)
Protein extrn, (see Extraction: aqueous RTL contactor (see Equipment/
two-phase systems; Micellar differential columns)
extrn)
Protic, protogenic, 28, 40, 42 Salt solub, 68,
Protonated form const, 82 adducts, 150
Protonation, 133 Salting in/-out, 59, 130, 176
Purex proc, 509, 519ff SANEX proc (radioactive waste
Purification of phosphoric acid, 629 treatment), 548ff
Pyro-hydrometallurgical concept, Saturated interface, 230ff
641ff Saturation, 65, 73
Subject Index 749
Sauter diameter (of drop), 371ff [Solvent extrn]

Scale-up, 291, 332 flowsheet, 340
of equip, 408 losses, 302, 364
Scrubbing def, 13, pretreatment/conditioning, 279
calc of, 353ff recovery, 302
Sea water, 258, 517 regeneration, 522
Self-adduct, 121–2, 145, 152–6, 176 Solvent impregnated resins, 574ff
Self-consistent field: calc (SCF), 682ff Solvent-in-pulp, 18, 302, 332
reactn (SCRF), 687ff Specific ion interact mod, 254ff
Semiempirical calc, 423ff (phase equil), Spectroscopic determination, appl in,
682ff (modeling) 567ff
Separation factor, 699ff AAS, 569
Sequential interfacial reactn, 232 flame emission, 569
Setchenov eqn, 59 uv/vis, 567ff
Settler design, 359ff X-ray fluorescence, 569
Short-range interact, 256 SPIN programme (radioactive waste
Shuttle mech, liquid membrane, 656 treatment), 547
Sieve tray/plate extractor (see Spray column (see Equipment/
Equipment/differential columns) differential columns)
SIMPLEX, 193 Srex proc (Sr), 537
SISAK, 196, 540 Stab const def, 82, 91
Soave-Redlich-Kwong eqn correl, 90, 167
(see Equation of state) Stagewise sepn, 289ff
Soft species, 98 Stagnant layers, 205
Solid-phase spectr, 575 Standard chem pot, 48
Solubility, mutual of aq and org, 39 molar energy, 40ff
Solubility param, 30ff, 75ff, 170ff, pot eqn, 192
product, 68ff Standard deviation, 262
Solute agg, 63 Standard state, 84
Solute solub def, 65 of infinite dilution, 60
classes, 120ff Stationary reactn state, 212
standard state, 253 Statistical eff, 87–89
Solute-solvent interact, 38, 40ff, 49 Steady reactn state, 212
Solution prop, 28–33 Stepwise form const, def, 82
Solution theories, 15 Steric hindrance, 177, 180
Solvation, 38, 44, 690ff (modeling) Steric struct, 92 and App.C
effects/models, 686ff Stirred cell, 193
number, 150, Stirred vessel, 244ff
shell, 44, 62 Stirring rate, 206, 224ff
Solvatochromic param, 42–43, 64 Stripping (back-extraction)
Solvent def/class, 5, 7, 12, 28ff, countercurrent calc, 354
misc, 37ff isotherm, 284
Solvent extrn, def, 1 Strong acid, 136
degrad, 521 Structure factor F, 74
equip, 18 Sublimation, 9
750 Subject Index
Substitution reactn, 227 Tielines, 72

Supercritical extrn, 19, 444ff Trace element detn, 564
fluid, 27ff Transfer unit, 361, height of (HTU),
in anal, 593ff 362
in proc, 445ff Transition metal compl, 93
for sepn of glycerides, 449 Transmutation, 19
Supported liquid membrane, 653, 658ff Truex process (trans-uranic), 532ff
Surfactant liquid membrane Two-film theory, 204
(see Emulsion liquid membrane) Two-parameter meth, 189
Surface tension, 29 Two-phase liquid distrn, 50ff
Synergistic extrn, 115ff, 465, 501, 563
Systematics of KDC, 169 UNIQUAC eqn/model (L-L eql), 423ff
Taft-Kamlet param, 64
Teflon phase sepn, 244 van der Waal forces, 221
Temperature eff, 176 volume, 67
Terdentate, 96 Vapor pressure relations, 49ff
Ternary systems, 70ff Virial coeff, 254, 267
Tetrahedral compl, 93, 458 Viscosity, 29, 329
Thermodynamic funct/const, 41, 46,
84–85 Water activity, 258, Ch.6
of chelation, 105, Water exchange, 214, 216,
of complexn, 102ff rate, 216
of extrn syst, 113ff Water solubilization capacity,
of interaction, 83 micelles, 662
Third phase, 288 Weber number, 381
Thorex proc (Th), 528ff Weighting, 192
Throughput (of phases), 387ff Winsor microemulsions, 661
колхоз8/1/06

Solvent Extraction Principles and Practice PDF

Uploaded by

Copyright:

Available Formats

Solvent Extraction Principles and Practice PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Solvent Extraction Principles and Practice PDF

Uploaded by

Copyright:

Available Formats

What are some of the applications of solvent extraction?

What are some of the applications of solvent extraction?

What are some techniques used in solvent extraction?

What are some techniques used in solvent extraction?

Solvent Extraction

The partition of a solute between two immiscible solvents is a major tech-

expert authors, each of whom has decades of experience in research,

1. Introduction to Solvent Extraction 1

2. Principles of Solubility and Solutions 27

3. Complexation of Metal Ions 81

4. Solvent Extraction Equilibria 109

5. Solvent Extraction Kinetics 203

6. Ionic Strength Corrections 253

7. Development of Industrial Solvent Extraction Processes 277

8. Principles of Industrial Solvent Extraction 339

9. Engineering Design and Calculation of Extractors for

10. Extraction of Organic Compounds 415

11. Solvent Extraction in Hydrometallurgy 455

12. Solvent Extraction in Nuclear Science and Technology 507

13. Analytical Applications of Solvent Extraction 559

14. The Use of Solvent Extraction in the Recovery of Waste 609

15. Recent Advances in Solvent Extraction Processes 651

16. Computational Chemistry in Modeling Solvent

Manuel Aguilar Department of Chemical Engineering, Universitat Poli-

Eckhart F. Blass* Professor Emeritus, Lehrstuhl A für Verfahrenstechnik,

Gregory R. Choppin* Department of Chemistry, Florida State University,

José Luis Cortina Department of Chemical Engineering, Universitat

Michael Cox* Department of Chemical Sciences, University of Hertford-

Marian Czerwiński Zawiercie University of Administration and Manage-

Pier Roberto Danesi Seibersdorf Laboratories, International Atomic

Ingmar Grenthe* Professor Emeritus, Department of Inorganic Chemistry,

Jan-Olov Liljenzin* Professor Emeritus, Department of Nuclear Chemistry,

Philip J. D. Lloyd Energy Research Institute, University of Cape Town,

Yizhak Marcus* Professor Emeritus, Department of Inorganic and Ana-

Claude Musikas* Département des Procedes de Retraitement, Commis-

Jerzy Narbutt Institute of Nuclear Chemistry and Technology, Warsaw,

Susana Pérez de Ortiz Department of Chemical Engineering, Imperial

Hans Reinhardt MEAB Metallextraktion AB, Göteborg, Sweden

Gordon M. Ritcey Gordon M. Ritcey & Associates, Inc., Nepean, Ontario,

Jan Rydberg* Department of Nuclear Chemistry, Chalmers University of

Ana Marı́a Sastre Department of Chemical Engineering, Universitat

Wallace W. Schulz Consultant, Albuquerque, New Mexico, U.S.A.

Tatsuya Sekine{ Department of Chemistry, Science University of Tokyo,

David Stuckey Department of Chemical Engineering, Imperial College,

Hans Wanner Swiss Federal Nuclear Safety Inspectorate, Villingen-HSK,

Ronald Wennersten Department of Chemical Engineering, Royal Institute

1.1 WHAT IS SOLVENT EXTRACTION?

Fig. 1.1 A schematic representation of solvent extraction (liquid-liquid distribution).

is defined as the ratio of “the total analytical concentration of the substance in

1.2 SOLVENT EXTRACTION IS A FUNDAMENTAL

Solutes have differing solubilities in different liquids due to variations in

made more selective than the conventional gas–liquid–solid separation tech-

1.3 EARLY STEPS TOWARD THE USE

supposed to mean “art of transmutations as practiced by the Egyptians.” Fruit

Fig. 1.4 (a) Equipment used by alchemists, according to an Alexandrian manuscript

1.4 SOLVENT EXTRACTION IN THE DEVELOPMENT

1.4.1 The Distribution Ratio

K D = [HBz]org /[HBz]aq (1.3)

for which one defines an extraction constant, Kex