Progress in Theoretical Chemistry and Physics 31

Series Editors: J. Maruani · S. Wilson

Yan A. Wang
Mark Thachuk
Roman Krems
Jean Maruani Editors

Concepts, Methods
and Applications of
Quantum Systems in
Chemistry and Physics
Selected Proceedings of QSCP-XXI
(Vancouver, BC, Canada, July 2016)
Concepts, Methods and Applications of
Quantum Systems in Chemistry and Physics
 Progress in Theoretical Chemistry and Physics
VOLUME 31
Honorary Editors
Rudolph A. Marcus (California Institute of Technology, Pasadena, CA, USA)
Roy McWeeny (Università di Pisa, Pisa, Italy)

Editors-in-Chief
J. Maruani (formerly Laboratoire de Chimie Physique, Paris, France)
S. Wilson (formerly Rutherford Appleton Laboratory, Oxfordshire, UK)

Editorial Board
E. Brändas (University of Uppsala, Uppsala, Sweden)
L. Cederbaum (Physikalisch-Chemisches Institut, Heidelberg, Germany)
G. Delgado-Barrio (Instituto de Matemáticas y Física Fundamental, Madrid, Spain)
E. K. U. Gross (Freie Universität, Berlin, Germany)
K. Hirao (University of Tokyo, Tokyo, Japan)
Chao-Ping Hsu (Institute of Chemistry, Academia Sinica, Taipei, Taiwan)
R. Lefebvre (Université Pierre-et-Marie-Curie, Paris, France)
R. Levine (Hebrew University of Jerusalem, Jerusalem, Israel)
K. Lindenberg (University of California at San Diego, San Diego, CA, USA)
A. Lund (University of Linköping, Linköping, Sweden)
M. A. C. Nascimento (Instituto de Química, Rio de Janeiro, Brazil)
P. Piecuch (Michigan State University, East Lansing, MI, USA)
M. Quack (ETH Zürich, Zürich, Switzerland)
S. D. Schwartz (Yeshiva University, Bronx, NY, USA)
O. Vasyutinskii (Russian Academy of Sciences, St Petersburg, Russia)
Y. A. Wang (University of British Columbia, Vancouver, BC, Canada)

Former Editors and Editorial Board Members

I. Prigogine (†) W. F. van Gunsteren (*)
J. Rychlewski (†) H. Hubač (*)
Y. G. Smeyers (†) E. Kryachko (*)
R. Daudel (†) M. P. Levy (*)
M. Mateev (†) G. L. Malli (*)
W. N. Lipscomb (†) P. G. Mezey (*)
Y. Chauvin (†) N. Rahman (*)
H. W. Kroto (†) S. Suhai (*)
H. Ågren (*) O. Tapia (*)
V. Aquilanti (*) P. R. Taylor (*)
D. Avnir (*) R. G. Woolley (*)
J. Cioslowski (*)

†: deceased; *: end of term

More information about this series at http://www.springer.com/series/6464

Yan A. Wang ⋅ Mark Thachuk
Roman Krems ⋅ Jean Maruani
Editors

Concepts, Methods
and Applications of
Quantum Systems in
Chemistry and Physics
Selected Proceedings of QSCP-XXI
(Vancouver, BC, Canada, July 2016)

123
Editors
Yan A. Wang Roman Krems
University of British Columbia University of British Columbia
Vancouver Vancouver
Canada Canada

Mark Thachuk Jean Maruani

University of British Columbia Laboratoire de Chimie Physique
Vancouver CNRS & UPMC
Canada Paris
France

ISSN 1567-7354 ISSN 2215-0129 (electronic)

Progress in Theoretical Chemistry and Physics
ISBN 978-3-319-74581-7 ISBN 978-3-319-74582-4 (eBook)
https://doi.org/10.1007/978-3-319-74582-4
Library of Congress Control Number: 2017964436

© Springer International Publishing AG, part of Springer Nature 2018

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part
of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission
or information storage and retrieval, electronic adaptation, computer software, or by similar or
dissimilar methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this
publication does not imply, even in the absence of a specific statement, that such names are exempt
from the relevant protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this
book are believed to be true and accurate at the date of publication. Neither the publisher nor the
authors or the editors give a warranty, express or implied, with respect to the material contained
herein or for any errors or omissions that may have been made. The publisher remains neutral with regard
to jurisdictional claims in published maps and institutional affiliations.

Printed on acid-free paper

This Springer imprint is published by the registered company Springer International Publishing AG
part of Springer Nature
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
PTCP Aim and Scope

Progress in Theoretical Chemistry and Physics

A series reporting advances in theoretical molecular and material sciences, including

theoretical, mathematical and computational chemistry, physical chemistry and chemical
physics and biophysics.

Aim and Scope

Science progresses by a symbiotic interaction between theory and experiment: Theory is

used to interpret experimental results and may suggest new experiments; experiment
helps to test theoretical predictions and may lead to improved theories. Theoretical
chemistry (including physical chemistry and chemical physics) provides the conceptual
and technical background and apparatus for the rationalization of phenomena in the
chemical sciences. It is, therefore, a wide ranging subject, reflecting the diversity of
molecular and related species and processes arising in chemical systems. The book series
Progress in Theoretical Chemistry and Physics aims to report advances in methods and
applications in this extended domain. It will comprise monographs as well as collections
of papers on particular themes, which may arise from proceedings of symposia or invited
papers on specific topics as well as from initiatives from authors or translations.
The basic theories of physics—classical mechanics and electromagnetism, rela-
tivity theory, quantum mechanics, statistical mechanics, quantum electrodynamics—
support the theoretical apparatus which is used in molecular sciences. Quantum
mechanics plays a particular role in theoretical chemistry, providing the basis for the
valence theories, which allow to interpret the structure of molecules, and for the
spectroscopic models, employed in the determination of structural information from
spectral patterns. Indeed, quantum chemistry often appears synonymous with theor-
etical chemistry; it will, therefore, constitute a major part of this book series.
However, the scope of the series will also include other areas of theoretical chemistry,

v
vi PTCP Aim and Scope

such as mathematical chemistry (which involves the use of algebra and topology in
the analysis of molecular structures and reactions); molecular mechanics, molecular
dynamics and chemical thermodynamics, which play an important role in rational-
izing the geometric and electronic structures of molecular assemblies and polymers,
clusters and crystals; surface, interface, solvent and solid state effects; excited-state
dynamics, reactive collisions and chemical reactions.
Recent decades have seen the emergence of a novel approach to scientific
research, based on the exploitation of fast electronic digital computers. Computation
provides a method of investigation which transcends the traditional division
between theory and experiment. Computer-assisted simulation and design may
afford a solution to complex problems which would otherwise be intractable to
theoretical analysis, and may also provide a viable alternative to difficult or costly
laboratory experiments. Though stemming from theoretical chemistry, computa-
tional chemistry is a field of research in its own right, which can help to test
theoretical predictions and may also suggest improved theories.
The field of theoretical molecular sciences ranges from fundamental physical
questions relevant to the molecular concept, through the statics and dynamics of
isolated molecules, aggregates and materials, molecular properties and interactions,
to the role of molecules in the biological sciences. Therefore, it involves the
physical basis for geometric and electronic structure, states of aggregation, physical
and chemical transformations, thermodynamic and kinetic properties, as well as
unusual properties such as extreme flexibility or strong relativistic or quantum field
effects, extreme conditions such as intense radiation fields or interaction with the
continuum, and the specificity of biochemical reactions.
Theoretical chemistry has an applied branch (a part of molecular engineering),
which involves the investigation of structure–property relationships aiming at the
design, synthesis and application of molecules and materials endowed with specific
functions, now in demand in such areas as molecular electronics, drug design or
genetic engineering. Relevant properties include conductivity (normal, semi- and
super-), magnetism (ferro- and ferri-), optoelectronic effects (involving nonlinear
response), photochromism and photoreactivity, radiation and thermal resistance,
molecular recognition and information processing, biological and pharmaceutical
activities, as well as properties favoring self-assembling mechanisms and combi-
nation properties needed in multifunctional systems.
Progress in Theoretical Chemistry and Physics is made at different rates in these
various research fields. The aim of this book series is to provide timely and in-depth
coverage of selected topics and broad-ranging yet detailed analysis of contemporary
theories and their applications. The series will be of primary interest to those whose
research is directly concerned with the development and application of theoretical
approaches in the chemical sciences. It will provide up-to-date reports on theoretical
methods for the chemist, thermodynamician or spectroscopist, the atomic, molecular
or cluster physicist, and the biochemist or molecular biologist who wish to employ
techniques developed in theoretical, mathematical and computational chemistry in
their research programs. It is also intended to provide the graduate student with a
readily accessible documentation on various branches of theoretical chemistry,
physical chemistry, and chemical physics.
Preface

This volume collects 20 selected papers from the scientific contributions presented
at the Twenty-first International Workshop on Quantum Systems in Chemistry,
Physics, and Biology (QSCP-XXI), organized by Yan Alexander Wang at the
University of British Columbia in Vancouver, BC, Canada, on July 02–09, 2015.
Over 160 scientists from 30 countries attended this meeting. The participants dis-
cussed the state of the art, new trends, and future evolution of methods in molecular
quantum mechanics and their applications to a broad variety of problems in
chemistry, physics, and biology.
The high-level attendance attained in this conference was particularly gratifying.
It is the renowned interdisciplinary nature and friendly feeling of QSCP meetings
that make them so successful discussion forums.
Highly ranked among the world best universities, the University of British
Columbia (UBC) holds an international reputation for excellence in advanced
research and teaching. Only 30 min from the vibrant heart of downtown Vancouver,
the spectacular UBC campus is a ‘must-see’ for any visitor in the world: Snow-
capped mountains can be seen meeting the ocean, and breathtaking vistas greet you
around every corner. The UBC campus also boasts some of the city’s best attrac-
tions and recreation facilities, including the Chan Centre for the Performing Arts,
the Museum of Anthropology, the UBC Botanical Garden, and endless opportu-
nities to explore forested trails in the adjoining 763-ha Pacific Spirit Regional Park.
Details of the Vancouver meeting, including scientific and social programs, can be
found on the Web site: https://groups.chem.ubc.ca/qscp/. Altogether, there were 30
morning and afternoon sessions, where 48 plenary lectures and 36 invited talks were
given, and one evening poster session, with 18 posters being displayed. We are grateful
to all 166 participants for making the QSCP-XXI workshop a stimulating experience
and a great success. QSCP-XXI followed the traditions established at previous
workshops:
QSCP-I, organized by Roy McWeeny in 1996 at San Miniato (Pisa, Italy);
QSCP-II, by Stephen Wilson in 1997 at Oxford (England);

vii
viii Preface

QSCP-III, by Alfonso Hernandez-Laguna in 1998 at Granada (Spain);

QSCP-IV, by Jean Maruani in 1999 at Marly-le-Roi (Paris, France);
QSCP-V, by Erkki Brändas in 2000 at Uppsala (Sweden);
QSCP-VI, by Alia Tadjer and Yavor Delchev in 2001 at Sofia (Bulgaria);
QSCP-VII, by Ivan Hubac in 2002 near Bratislava (Slovakia);
QSCP-VIII, by Aristides Mavridis in 2003 at Spetses (Athens, Greece);
QSCP-IX, by J.-P. Julien in 2004 at Les Houches (Grenoble, France);
QSCP-X, by Souad Lahmar in 2005 at Carthage (Tunisia);
QSCP-XI, by Oleg Vasyutinskii in 2006 at Pushkin (St Petersburg, Russia);
QSCP-XII, by Stephen Wilson in 2007 near Windsor (London, England);
QSCP-XIII, by Piotr Piecuch in 2008 at East Lansing (Michigan, USA);
QSCP-XIV, by G. Delgado-Barrio in 2009 at El Escorial (Madrid, Spain);
QSCP-XV, by Philip Hoggan in 2010 at Cambridge (England);
QSCP-XVI, by Kiyoshi Nishikawa in 2011 at Kanazawa (Japan);
QSCP-XVII, by Matti Hotokka in 2012 at Turku (Finland);
QSCP-XVIII, by M.A.C. Nascimento in 2013 at Paraty (Brazil);
QSCP-XIX, by Cherri Hsu in 2014 at Taipei (Taiwan);
QSCP-XX, by Alia Tadjer and Rossen Pavlov in 2015 at Varna (Bulgaria).
The lectures presented at QSCP-XXI were grouped into seven areas in the field
of Quantum Systems in Chemistry, Physics, and Biology, ranging from Concepts and
Methods in Quantum Chemistry through Relativistic Effects in Quantum Chemistry,
Atoms and Molecules in Strong Electric and Magnetic Fields, Reactive Collisions and
Chemical Reactions, Molecular Structure, Dynamics and Spectroscopy, and Molecular
and Nano-materials, to Computational Chemistry, Physics, and Biology.
The width and depth of the topics discussed at QSCP-XXI are reflected in the
contents of this volume of proceedings in Progress in Theoretical Chemistry and
Physics, which includes four sections:
I. Quantum Chemistry Methodology (five papers);
II. Molecular Structure and Dynamics (ten papers);
III. Biochemistry and Biophysics (two papers);
IV. Fundamental Theory (three papers).
In addition to the scientific program, the workshop had its usual share of cultural
events. There was a music concert, Amor & Pasión, by the Vancouver International
Song Institute, in the afternoon of the opening day, and in the evening of July 5
Prof. Jean Maruani, accompanied by his pianist wife Marja Rantanen, delivered an
entertaining lecture on Science and Music.
The award ceremony of the CMOA Prize and Medal took place during the
banquet in the premises of the UBC Golf Club. The CMOA Prize for junior
scientists was shared between the two selected nominees: Roman V. Krems (UBC,
Canada) and Erin R. Johnson (Dalhousie, Canada). The prestigious CMOA Medal
for senior scientists was awarded to Prof. Weitao Yang (Duke, USA).
During the banquet, we also celebrated the 80th birthdays of Prof. Ernest R.
Davidson (born October 1936) and Prof. Josef Paldus (born November 1935).
Preface ix

Birthday cakes were presented to these two great scientists, accompanied by a

Happy Birthday sung by guests attending the banquet.
Following a QSCP tradition, the venue of the next workshop was presented at
the end: Changsha, Hunan, China, in October 2017.
We are most grateful to the members of the Local Organizing Committee: Mark
Thachuk, Roman Krems, Delano Chong, Peter Chung and Jane Cua, as well as to
the members of the Wang team, especially Jianxiong Yang, Yuzhe Chen, Dongmei
Luo, Junqing Yang, Miguel Garcia-Chavez, Yiming Wang and Lin Bryan Zhang,
for their work and dedication which made the stay and work of participants both
pleasant and fruitful. Last but not least, we thank the members of the International
Scientific and Honorary Committees for their invaluable expertise and advice.
We hope the readers will find as much interest in consulting these proceedings as
the participants in attending the meeting.

The Editors
Contents

PTCP Aim and Scope v

Preface vii

Part I Quantum Chemistry Methodology

A Portal for Quantum Chemistry Data Based on the Semantic Web . . . 3
Bing Wang, Paul A. Dobosh, Stuart Chalk, Keigo Ito, Mirek Sopek
and Neil S. Ostlund
Matrix Elements for Explicitly-Correlated Atomic Wave Functions . . . . 29
Frank E. Harris
Effective Bond-Strength Indicators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
Gui-Xiang Wang, Yuzhe Stan Chen, Ya-Kun Chen
and Yan Alexander Wang
Advanced Relativistic Energy Approach in Electron-Collisional
Spectroscopy of Multicharged Ions in Plasmas . . . . . . . . . . . . . . . . . . . . 55
Alexander V. Glushkov, Vasily V. Buyadzhi, Andrey A. Svinarenko
and Eugeny V. Ternovsky
Relativistic Quantum Chemistry and Spectroscopy of Exotic Atomic
Systems with Accounting for Strong Interaction Effects . . . . . . . . . . . . . 71
O. Yu. Khetselius, A. V. Glushkov, Yu. V. Dubrovskaya,
Yu. G. Chernyakova, A. V. Ignatenko, I. N. Serga and L. A. Vitavetskaya

Part II Molecular Structure and Dynamics

Difference of Chirality of the Electron Between Enantiomers of H2X2 . . . . . 95
Masato Senami, Ken Inada, Kota Soga, Masahiro Fukuda
and Akitomo Tachibana

xi
xii Contents

A Crystallographic Review of Alkali Borate Salts and Ab Initio

Study of Borate Ions/Molecules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
Cory C. Pye
An Ab Initio Study of Boric Acid, Borate, and their
Interconversion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
Cory C. Pye
Construction of a Potential Energy Surface Based on a Diabatic
Model for Proton Transfer in Molecular Pairs . . . . . . . . . . . . . . . . . . . . 179
Yuta Hori, Tomonori Ida and Motohiro Mizuno
Ab Initio Investigations of Stable Geometries of the
Atmospheric Negative Ion NO3−(HNO3)2 and Its Monohydrate . . . . . . . 193
Atsuko Ueda, Yukiumi Kita, Kanako Sekimoto and Masanori Tachikawa
A Theoretical Study of Covalent Bonding Formation Between
Helium and Hydrogen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203
Taku Onishi
Small Rhodium Clusters: A HF and DFT Study–III . . . . . . . . . . . . . . . 213
M. A. Mora and M. A. Mora-Ramírez
Spectroscopy of Radiative Decay Processes in Heavy Rydberg
Alkali Atomic Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229
Valentin B. Ternovsky, Alexander V. Glushkov, Olga Yu. Khetselius,
Marina Yu. Gurskaya and Anna A. Kuznetsova
Enhancement Factors for Positron Annihilation on Valence and
Core Orbitals of Noble-Gas Atoms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243
D. G. Green and G. F. Gribakin
Geometric Phase and Interference Effects in Ultracold Chemical
Reactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 265
N. Balakrishnan and B. K. Kendrick

Part III Biochemistry and Biophysics

Adducts of Arzanol with Explicit Water Molecules: An Ab Initio
and DFT Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 281
Liliana Mammino
Computational Study of Jozimine A2, a Naphthylisoquinoline
Alkaloid with Antimalarial Activity . . . . . . . . . . . . . . . . . . . . . . . . . . . . 305
Mireille K. Bilonda and Liliana Mammino
Contents xiii

Part IV Fundamental Theory

Functional Derivatives and Differentiability in Density-Functional
Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 331
Ping Xiang and Yan Alexander Wang
The Dirac Electron and Elementary Interactions: The Gyromagnetic
Factor, Fine-Structure Constant, and Gravitational Invariant:
Deviations from Whole Numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 361
Jean Maruani
A Simple Communication Hypothesis: The Process of Evolution
Reconsidered . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 381
Erkki J. Brändas
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 405
 Part I
Quantum Chemistry Methodology
A Portal for Quantum Chemistry Data
Based on the Semantic Web

Bing Wang, Paul A. Dobosh, Stuart Chalk, Keigo Ito, Mirek Sopek
and Neil S. Ostlund

Abstract Chemical Semantics, Inc. (CSI) is a new start-up devoted to bringing the
Semantic Web to chemistry and biochemistry. The semantic web is referred to as
Web 3.0 or alternatively the Web of Data or the Web of Meaning. It does not
replace the existing World Wide Web but augments it, placing data on the web in a
structured form such that the data has “meaning” and computers can understand it.
CSI has created a demonstration portal for exploring this new technology, specif-
ically at this point for data created by quantum chemistry calculations. This paper
describes the basics of a semantic web portal and the fundamental technology we
have used in developing it.

1 Overview of the Semantic Web

Data on the existing World Wide Web is encoded in documents containing text,
table, images, etc. This data is not really recognized by computers but has to be
interpreted by humans. Because computers cannot recognize the data in existing web
documents, the data cannot properly be shared or even found. The semantic web [1]
puts data onto the web in a form that allows computers to properly recognize it along
with its meaning. Computers can then perform intelligent operations on the data,
ultimately creating new data by using inference from existing data. The future of
scientific data involves structuring the data via the semantic web or equivalent
technologies so that computers can find data and understand data on our behalf.
Chemistry generates enormous amounts of data [2]. Because existing publication
channels do not give credence to the data in the way they give credence to the text
of a “scientific publication” much of this data is lost or discarded and not made
available to other scientists. There is a trend towards journals requiring authors to

B. Wang ⋅ P. A. Dobosh ⋅ K. Ito ⋅ M. Sopek ⋅ N. S. Ostlund (✉)

Chemical Semantics Inc, 2135 NW 15th Ave, Gainesville, FL 32605, USA
e-mail: [email protected]
S. Chalk
Department of Chemistry, University of North Florida, Jacksonville, FL 32224, USA

© Springer International Publishing AG, part of Springer Nature 2018 3

Y. A. Wang et al. (eds.), Concepts, Methods and Applications of Quantum Systems
in Chemistry and Physics, Progress in Theoretical Chemistry and Physics 31,
https://doi.org/10.1007/978-3-319-74582-4_1
4 B. Wang et al.

submit data files along with the text of a publication. However, since there is as yet
no standard and little infrastructure adopted by mainstream publishers for dealing
with this data, most of it remains in inaccessible files (e.g. PDF documents). An
appropriate answer to this dilemma is the semantic web.
In conjunction with the data, the semantic web includes a vocabulary for
describing the data. This is where semantics comes to the fore. This vocabulary for
a field such as Quantum Chemistry needs to be encoded in a formal language. The
Web Ontology Language (OWL) [3] is such a language and the “vocabulary” is
really an ontology describing the formal language of a specific domain such as
computational or quantum chemistry. Chemical Semantics has created such an
ontology which is referred to as the Gainesville Core (http://purl.org/chem/gc). This
first ontology for quantum chemistry will need to be modified and extended by
scientists in the field as basic ontology ideas become more prevalent in chemistry.
Our ontology is meant to be placed in the public domain so that the quantum
chemistry community can both contribute to it and use it. Developing the Gai-
nesville Core is an ongoing project with release numbers.
The semantic web allows scientists to publish their data in a structured way such
that it can be found and used by anyone with access to the World Wide Web.
Chemical Semantics, Inc. is creating client and server software that allows scientists
to automate their publishing of data into a modern graph database [4], i.e. a Giant
Global Graph (GGG), so that they and others can share their data and use it in a way
that is not now possible with the existing World Wide Web (WWW). Adding
semantics to scientific data and making that data available on the semantic web
makes it possible to do science in a new collaborative way that has the potential to
change science forever [1].

1.1 Publishing Quantum Chemistry Data

To make the technology of the semantic web available to scientists, one has to
“publish data” in a way that is related to the standard model of journal publishing.
That is, one puts the data (not journal text) into an appropriate form, sends it to a
publisher (of data not journal text), waits for its publication, and then informs
colleagues that they can access the data (not journal article) in a standard fashion
(more and more via the web rather than via hard copy).
The appropriate form for data described above has been clearly defined by the
World Wide Web Consortium (W3C) [5] as the Resource Description Framework
(RDF) standard [6]. This standard is a Graph Database where RDF statements all
take the “triple” form (subject, predicate, and object). For example, (water,
has_boiling_point, 100) is of this triple form. That is, there is a graph arc called
“has_boiling_point” which points from a “water” subject node to a “100” object
node. This form surpasses the normal relational database in its applicability to the
web. Any and all scientific data can be put into this form and data on the semantic
web is stored in servers referred to as triple stores. These triple stores may contain
billions of triples.
A Portal for Quantum Chemistry Data … 5

Chemical Semantics, Inc. operates servers that store scientific data in these triple
stores. Our client software allows scientists to automate the publishing of their data
to these triple stores. Initially, we are focusing on data produced by quantum
chemistry packages although the basic ideas apply to any chemical data including
experimental data. The data produced from these quantum chemistry packages
depends somewhat upon the specific package but for illustrative purposes let’s
assume a “CompChem” package that produces results from ab initio wave function
calculations, such as Self-Consistent-Field (Hartree-Fock SCF) calculations, post
Hartree-Fock correlated calculations, etc. Examples of such packages are Gamess
[7], Gaussian [8], NWChem [9], etc.
The publishing of the results of these calculations ought not to be more difficult
than having a “Publish” button in the GUI or text input of the package. For initial
demonstration, we have implemented such a button in HyperChem [10] as shown in
Fig. 1. The specific use of HyperChem is not relevant and any CompChem [11]

Fig. 1 Dialog box for submitting calculated quantum data to a portal

6 B. Wang et al.

package ought perhaps to have such a button! The founders of Chemical Semantics,
Inc. have a historical tie to Hypercube, Inc. and as such have used HyperChem to
illustrate the basic ideas. What this button in a “CompChem” package does is create
an XML file structure that includes the information about the current molecular
system and any current calculation results resident in the package and sends the data
to the Chemical Semantics, Inc. portal where it is published, given the Authors,
Title, Abstract, Login information and other details that are part of the global setup
prior to hitting the “Publish” button. Our current XML file structure is called CSX
and is related in historical terms to the Chemical Markup Language (CML) file
structure [12]. We use CSX because CML currently does not have certain properties
that we consider not only desirable, but mandatory, such as the ability to deal with
residues as independent fundamental units of biological molecules. Also, CML
does not define many of the quantum chemistry concepts that we consider critical,
We believe CSX encompasses CML as a subset and we could generate a CML file
from CSX easily.
Our Chemical Semantics portal accepts data using a REST or SOAP [13] pro-
tocol and then publishes the data on its servers. The data is available to anyone
around the world with an account at the portal. The details of the portal are
described in another section below.

1.2 Searching the Data

Our portal allows users to access data at the portal based upon various rules, search
criteria, etc. Semantic web data is usually searched for using a SPARQL Protocol
and RDF Query Language (SPARQL) [14]. Note the recursive acronym.
A SPARQL end point is maintained at the portal which allows queries of various
kinds. SPARQL has features in common with the SQL query language and is
relatively easy to use but requires some experience in forming queries. A natural
language front end would be desirable and many groups are involved in developing
such front ends. A query, for example, could ask how many Density Functional
Theory (DFT) calculations have been done on a specific molecule and with which
functionals and what final total energies. As opposed to querying relational data-
base silos, the query could potentially survey the whole world’s set of such cal-
culations and return with a table of these.
Because the data stored by Chemical Semantics includes a relatively unlimited
number of triples, searching can be very exhaustive while still being very focused.
The data is defined by the ontology so that a search does not return irrelevant
results. Because the data is held by a graph database where a graph node (resource)
uses an arrow to point to another resource, the arrow can simply point to a resource
in a second graph from the first graph and unlike a relational database the data from
different graphs can be merged trivially. This “federation” allows searches to really
use a Giant Global Graph (GGG). As part of the publish activity, data on our portal
can be tagged as private, protected, or public. Any search obviously can return
A Portal for Quantum Chemistry Data … 7

public data. For protected data the author can pass a key to researchers to enable
them to access his/her protected data. An alternative would be to label the publi-
cation private so that only the original authors can access the data.
Data also comes with tags set by the authors that help define the search. For
example an author might add a tag, “used_32bit_gpu” if he/she thought it worth-
while to distinguish calculations on a 32-bit only graphical processing unit from
those on a normal CPU. Once data has been put into a semantic web form elaborate
searches can be performed and the retrieved summary data could also be added as
new data if so desired. Examples of SPAROL queries are shown below after the
SPARQL query language is described.

2 Portal Technologies

The semantic web and our portal use a number of “new” technologies that are
briefly described here. Most of these are new to chemists at this point and the
following therefore provides somewhat of a primer on the semantic web so that the
next section describing the actual operation of publishing and querying using the
facilities of Chemical Semantics, Inc. can be better understood. There are many
good reference books on the semantic web but they generally are written for
computer scientists not chemists.
This tutorial will focus on describing quantum chemistry data as applied to the
semantic web.

2.1 CSX—A Chemical Semantics Markup Language

The semantic web uses RDF and triple stores to hold data in a graphical database.
The data of Quantum Chemistry needs to be put into this form using an ontology
for Quantum Chemistry. As a practical matter it makes sense to have a way of
structuring computed quantum data prior to converting it into RDF semantic web
data. A fundamental issue here is that the data should always be put into a struc-
tured form from its inception so that computers can be taught this structure. We
have introduced a markup language that we call a Common Standard for eXchange
of quantum data (CSX).

2.1.1 Overview

Chemical Semantics, Inc. uses an Extensible Markup Language (XML) format file
to capture information about quantum chemistry calculations. Specifically, a
Common Standard for eXchange (CSX) file is used to transfer structured data and
8 B. Wang et al.

metadata about calculations to our web portal where it is converted to the Resource
Description Framework (RDF) format appropriate to the semantic web.
While CSX has been developed to allow publication pf quantum chemistry
calculations onto the semantic web, it is a useful standard in and of itself because it
organizes all the important information about a calculation in a format that is
readable by both humans and computers. We suggest that CSX could become a
standard output format for the computational chemistry community and invite
interested readers to contribute to its development.
This section thus describes the current CSX standard for describing data from
quantum chemistry calculations (and possibly related computational or experi-
mental data). The standard specifically includes information describing a publica-
tion, i.e. title, author, etc. because our portal essentially accepts “data publications”
and places them onto the semantic web. CSX is still currently under development
and so please be aware that the description below is dated. A CSX file has a version
number and the current version described here is Version 1.0. Ongoing develop-
ment is aimed at creating Version 2.0 in the near future. For example, aligning our
CSX with JSON for Linked Data (JSON-LD) [15] is appropriate as this new
standard for linked data is closely related to the semantic web.

2.1.2 CSX Components

A CSX file is shown in Fig. 2. The fundamental components are described below.

NameSpaces

The CSX file above includes the basic XML components including a comment that
describes where the CSX file was created. The root element of the XML file is
cs:chemicalSemantics. The attributes at the root include this particular version of
CSX, the local file name for the CSX file, and the relevant namespaces being used.

Fig. 2 A basic CSX file

A Portal for Quantum Chemistry Data … 9

The principal namespace is cs: which signifies the XML elements defined by
Chemical Semantics, Inc. Other namespaces like xsd: and xsi: are associated with
the XML schema (blueprint) that describes CSX. The namespaces dc: and dcterms:
are part of Dublin Core, a metadata standard that is used for some of the publication
parameters of CSX like title and abstract. The namespace bse: is used to describe
basis sets and is subject to modification as we collaborate with Pacific NW Lab-
oratories on semantic definitions of standard basis sets.

2.2 CSX—Sections of a File

The description given here corresponds to CSX Version 1.0. The definition of CSX
is ongoing. While other sections could be added in future versions of CSX, the
current standard includes three sections:
• molecularPublication
• molecularSystem
• molecularCalculation.
The Chemical Semantics portal essentially accepts “Data Publications”, i.e. data
from computational chemistry computations that are being published at the portal.
Thus, the first section of a CSX file describes the publication itself, including copies
of the input and output files used in the calculation (if available). The second
section describes the molecular system (a set of molecules) that calculations were
performed on. The final section describes the calculation (or calculations) that were
performed on the molecular system.

2.2.1 Molecular Publication

The molecularPublication section of a CSX file is shown in Fig. 3.

A publication has a title and an abstract as indicated as defined in the Dublin
Core specification. We use the Dublin Core namespace, dcterms:, to describe these.
The “publisher” here is empty but would contain the name of the company/
institution of the lead author.

Authors

A publication can have a number of authors described by their type. The sole author
here is described as cs:corresponding, i.e. the Corresponding Author who is usually
the Principal Investigator (PI) or someone with authority over the publication. The
attribute “type” can also have a value of cs:submitting indicating the author sub-
mitting the publication, or could be empty indicating another author or co-author.
Each author has a name described by cs:creator and an organization described by
cs:organization and an email address described by cs:email.
10 B. Wang et al.

Fig. 3 The molecularPublication section of a CSX file

Source

The data being published should have an indication of its source, i.e. the software
package that created the data such as Gamess, NWChem, PSI4, etc. The version of
these software packages should also be indicated. In the example above, the data
came from Release 9 of HyperChem.
For archival purposes, it is also possible to add the text that constitutes the input
file for the calculation as well as the text of the output file. The publication data may
have been extracted by parsing the output file or directly from the software package.
In any event, the input and output file (if present) constitute additional archival data
about the calculation that can be recovered later, if so desired.

Tags and Flags

Finally, the publication can indicate a set of arbitrary tags that can apply to the
publication and provide an additional way to search for a set of publications. These
tags are independent of searches based on SPARQL and are up to the authors to create.
They may or may not be commonly used depending upon the preference of users.
A set of allowed values for flags is used to characterize publications. These are:

Status

• Preliminary
• Draft
• Final.
A Portal for Quantum Chemistry Data … 11

A publication can be considered preliminary (very raw data perhaps) or draft

(not fully reviewed) or final. This status can be changed (edited) at any later time.

Visibility

• Private
• Protected
• Public.
Depending upon the desires of the authors, a publication can be set to private
which means only the submitting author with a password can see the publication.
Alternatively, a public publication can be seen by anyone with access to the portal.
Finally, an intermediate visibility is available that is termed “protected”. A pro-
tected publication requires a key (similar to a password) that an author can pass to a
collaborator so that they can see an author’s publication.

Category

Finally, it is possible to categorize a publication from one of the areas of chemistry

below.
• Analytical Chemistry
• Biochemistry
• Computational Chemistry
• Inorganic Chemistry
• Materials Chemistry
• Material Science
• Molecular Chemistry
• Nanotechnology
• Neurochemistry
• Nuclear Chemistry
• Organic Chemistry
• Other
• Petro Chemistry
• Physical Chemistry
• Polymer Chemistry
• Synthetic Chemistry
• Theoretical Chemistry.

2.2.2 Molecular System

The molecularSystem section of a CSX file is shown in Fig. 4.

A molecularSystem is a collection of molecules. Each molecule is a collection of
possibly residues (monomers) or perhaps just atoms. The intermediate level of
12 B. Wang et al.

Fig. 4 The molecularSystem section of a CSX file

“group” is also possible. That is, a molecule or residue might be considered as

having groups that are made up of atoms.
Normally a molecule is just a set of atoms, unless we are describing a protein or
dna-like structure where the residues would be amino acids or nucleic acids. In the
above example, there is only one molecule (water) made up of three atoms.

System

The molecular system currently has three properties. The traditional charge and
multiplicity of a quantum calculation are two of these. In addition, we define the
system temperature, as well, for statistical mechanical calculations such as
molecular dynamics (MD), Monte Carlo, etc.

Molecule

Each molecule has an id which is the name or other identifier of the molecule.
A default id of “m1”, “m2”, etc. for molecule 1, molecule 2, … is suggested.
A Portal for Quantum Chemistry Data … 13

A molecule has an atomCount describing the number of atoms in the molecule

and an InChI key [16] that is meant to be a unique identifier for each molecule in
the system. The InChI string and InChI key are defined by IUPAC and may or may
not exist for each molecule. The value “nil” or “” indicates that no InChI key was
returned from IUPAC software. Each molecule is made up of atoms.

Atom

All atoms in a molecule must have an id attribute of the form “a1”, “a2”, “a3”, etc.
These identifiers are used to describe the resulting bonds.
An atom has a number of children including the elementSymbol, elementName,
etc.
• elementSymbol—this is just normal symbol such as C, Cl, etc.
• elementName—this is the full name for the element such as Carbon, Chlorine,
etc.
• atomName—for macromolecules an atom may have a pertinent name such as
CA, CB for the alpha, beta carbons in a chain. For normal molecules, the default
names are:
– H
– MainGroup
– Metal
– Row1TM
– Row2TM
– Row3TM
– Lanthanide
– Actinide
– NobleGas
• atomMass—the mass in amu
• formalAtomCharge—integer charge such as +1 for the N in NH4
• calculatedAtomCharge—as used in molecular mechanics coulomb interactions
• x/y/zCoord3D—the three Cartesian coordinates of an atom
• basisSet—the basis set is a property of each atom and may be different for
different atoms
• coordination—describes the other atoms to which this atom is connected.

Bond

Bonds in CSX are a property of atoms. The XML coordination element describes
these, as children of coordination. The bondCount attribute of coordination is the
number of bonds that this atom participates in. Each bond is a child of the atom’s
coordination with attributes id1 and id2 that describe the two connected atoms. This
means that each bond is described twice—as a grandchild of the atom with attribute
14 B. Wang et al.

id1 and as a grandchild of the atom with attribute id2. The content of the XML bond
element describes the bond as single, double, triple, aromatic, or dative. We believe
that having a bond be defined in the context of defining an atom provides better
functionality that defining bonds as an isolated property as per CML.

2.2.3 Molecular Calculation

The molecularCalculation Section of a CSX file is shown in Fig. 5.

There are different types of potential calculations that need to be placed in a CSX
file although many commonalities exist. In particular scf and dft calculations have
much in common. The following section of a CSX file describes a simple scf
calculation.
The first XML element describes the calculation as a quantum mechanical one
(as opposed to a molecular mechanics). Secondly, it uses a singleReferenceState (as
opposed to a multipleReferenceState such as MCSCF). Then the calculation
describes a singleDeterminant as opposed to a multipleDeterminant such as CISD).

Fig. 5 The molecularCalculation section of a CSX file

A Portal for Quantum Chemistry Data … 15

Finally, the calculation is characterized as abInitioSCF. Alternatives might be

semiempiricalSCF or dft.

Attributes

The attributes of such a calculation are:

• Methodology—allows for deviations from the normal type of SCF calculation.
• spinType—RHF, UHF, or ROHF
• basisSet—as defined by the PNNL Basis Set Exchange.

Child Elements of Calculation

The child elements of an SCF calculation are:

• energies—core nuclear-nuclear interaction, electronic energy, total energy, etc.
• properties—system properties and atom properties
• waveFunction—orbital energies, orbital symmetry, coefficients, etc.
The energies are all labeled by their “type” attribute. For example, cs:to-
talPotential is a potential energy for nuclear motion in the Born-Oppenheimer
approximation and is commonly just called the total energy in quantum calcula-
tions. The advantage of using a Uniform Resource Identifier (URI) here is that it
gives uniqueness to the variable being described.
A large variety of properties (possibly expectation values) could be associated
with a quantum calculation. We divide these properties into system properties like
dipole moment and atom properties like mullikenCharges, A systemProperty has
attributes—“name” and “unit” where name, for example, is cs:dipoleMomentX
denoting the X component of the total dipole moment and unit is cs:debye.
An atomProperty also has attributes “name” and “unit” in addition to the attri-
bute propertyCount which indicates how many atoms follow with properties. The
attributes of an atom property are the moleculeId and the atomId which uniquely
identify the atom (atom indices or id’s are unique only for the specific molecule that
the atom is a member of). The value of the atomProperty is the content of the
associated XML property element.
With quantum calculations there may be no connection table identifying a
“molecule” since a molecule of the molecular system is defined by CSX as the
collection of atoms forming a connected graph. For certain quantum calculations
there may only be a molecularSystem with child atoms and no “molecule” or
“coordination”, “bond”, etc. This is acceptable as valid CSX. It is preferred,
however, to define a connection table, if possible.

waveFunction

The wave function example in Fig. 6. shows the result for the above 3–21 G
calculation with its orbitals, etc. The far right side of the display is cut off and not all
orbital energies, symmetries, etc. are shown.
16 B. Wang et al.

Fig. 6 The wave function section of a CSX file

The wave function has attributes orbitalCount and basisCount. These are gen-
erally the same but some calculations such as those from Gamess may differ here
because of the treatment of 5 spherical or 6 Cartesian d-orbitals.
The child XML elements of the waveFunction include the orbitalEnergies, cs:
orbitalSymmetry, cs:orbitalOccupancy and the orbitals themselves. Each orbital is
identified by its id.

Other Calculations

If the calculation is a density functional calculation, then cs: abinitioSCF is replaced

by cs:dft and two new attributes appear—cs:exchangeFunctional and cs:correla-
tionFunctional. If the calculation is MP2, then cs:abinitioSCF is replaced by cs:
secondOrderMoellerPlesset and an energy with type cs:correlation is added. Other
calculations such as CCD, CCSD(T), etc. may have energies and properties but not
a cs:waveFunction.
If the calculation is a molecular mechanics calculation then cs:molecu-
larMechanics replaces cs:quantumMechanics and new attributes describing the
cs:forceField and cs:parameterSet appear.

2.3 Usefulness of the CSX Description

The CSX proposed file standard is a convenient way to encapsulate computational

data coming from various quantum chemistry packages, for example. It will make
possible publication of the data onto the semantic web as well as portability of
A Portal for Quantum Chemistry Data … 17

molecular structures and results among computational and other chemists. It is a

beginning and other options may become available but we believe it is a valuable
contribution.

2.4 A Quantum Chemistry Ontology

The RDF standard for a graph database and the associated serialized files, like RDF/
XML or Turtle, contain the scientific data (consisting of URI’s and literals). The
interpretation or meaning of that data, however, requires a vocabulary for defining
the data. That vocabulary is an ontology, in this case an ontology for computational
chemistry. Simple ontologies use a schema for RDF (RDFS) but in general the Web
Ontology Language (OWL) and OWL files are better used to describe the ontology.
An OWL file is usually formatted as RDF/XML for convenience although an OWL
ontology is not to be confused with the fundamental data described by RDF. It is
just simply convenient to use RDF to describe an ontology. This shows the power
of RDF in that the ontology can be described by the same structure as the data.

2.4.1 Simple OWL File

An OWL file begins by describing the classes (and subclasses) that are basic
entities of the ontology. For example the classes might be MolecularSystem,
Molecule and Atom, where the MolecularSystem is what a calculation is performed
on and which is assumed to be a collection of atoms and molecules. These classes
have ObjectProperties that relate one class to another. The ObjectProperty, has-
Molecule, relates the class MolecularSystem (the Domain) to the class Molecule
(the Range). The ObjectProperty, hasAtom, similarly relates the class Molecule to
the class Atom. The following is a portion of such an ontology expressed in Turtle.
The Gainesville Core ontology is reflected in the prefix gc:
@prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#> .
@prefix owl: <http://www.w3.org/2002/07/owl#> .
@prefix gc: http://purl.org/gc>.

gc:hasMolecule rdf:type owl:ObjectProperty ;

rdfs:domain gc:MolecularSystem ;
rdfs:range gc:Molecule .

gc:hasAtom rdf:type owl:ObjectProperty ;

rdfs:range gc :Atom ;
rdfs:domain gc :Molecule .

Using such an ontology allows the trivial inference that a MolecularSystem has
atoms although that is not explicitly stated! An ontology also has DataType
Properties that relate a class to a Literal rather than another class. For example, the
18 B. Wang et al.

dataType property, hasMultiplicity, might relate a class of SCF calculations to the

Literal 1 or 3 to indicate singlet or triplet.
Chemical Semantics, Inc. is in the middle of defining an ontology for compu-
tational chemistry. The current Release is usable but certainly not final. The Gai-
nesville Core web site describes the current status. A very professional version of
this will not be available quickly. A view of a very preliminary current version is
shown below.
One of the most important efforts of Chemical Semantics, Inc. is the develop-
ment of a proper ontology for computational chemistry. The effort will require
contributions from many members of this particular scientific community. Funding
for such an effort is being sought.

2.5 SPARQL—Searching

The acronym SPARQL Protocol and RDF Query Language (SPARQL) is officially
recursive but sometimes is referred to as Simple Protocol and RDF Query Lan-
guage. It bears some similarity to SQL but is a W3C standard for querying data on
the semantic web (graph databases) rather than data in relational databases.
There are a number of commercial SPARQL software packages including Vir-
tuoso which Chemical Semantics’ portal uses. The following is an example of a
SPARQL query that searches for all molecules that have 5 atoms and include a
Chlorine atom.
prefix gc: <http://purl.org/gc/>
prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#>
select
?molecule ?moleculeLabel ?inchikey
where {
graph ?graph {
?molecule rdf:type gc:Molecule ;
rdfs:label ?moleculeLabel ;
gc:hasNumberOfAtoms “5”;
gc:hasAtom ?atom;
gc:hasInChIKey ?inchikey .

?atom gc:isElement “Cl” .

}
}
order by ?molecule

The query begins with a definition of the namespaces gc: (Gainesville Core defined
by Chemical Semantics, Inc.) and rdfs: (RDF Schema) that it will use. The quantities ?
molecule, ?moleculeLabel, etc. are simply variables with arbitrary names.
The query searches for Molecules that have certain properties such as a label,
number of atoms, etc. The “;” is essentially an “and” and the search ends with a “.”
In addition, to the required Molecule properties, a ?atom found in the molecule
must also be a Chlorine according to the last requirement (Fig. 7),

?atom gc:isElement “Cl”.

A Portal for Quantum Chemistry Data … 19

Fig. 7 Preliminary version of ontology

20 B. Wang et al.

Fig. 8 Result of a SPARQL query

An example of using the query at the portal of Chemical Semantics, Inc. with a
small demonstration RDF graph results in a table as shown in Fig. 8.
The columns are labelled by the variables used in the query select. It returns here
only the inchiKey of the two molecules but an extension of the query could return
any computed properties of the two molecules that are part of our graph database on
the semantic web.

3 The Web Portal

There are three ways to publish on the portal. The first is to publish directly from
various software packages that produce computational chemistry results. The
number of these will expand as time progresses. For example, versions of PSI 4,
NWChem and HyperChem can publish this way. An example used to develop the
basic ideas is HyperChem, Release 9 with its Publish Button. The other packages
without a GUI require a simple script for publishing.
The second way to publish is to independently create a CSX file that defines the
publication, the molecular system and the calculations and just upload that file to
the portal where it will be translated to a Turtle (*.TTL) file and the data placed
onto the semantic web. One way to do this is to parse an output file with software
such as CSI’s own ChemicalPublisher to create the CSX file as shown in Fig. 9.
Thirdly, one might directly upload to the portal the output file from a compu-
tational package and have that output parsed at the portal where it is put into CSX
form initially and then translated to a TTL file.

3.1 Using a Publish Button

The first publication procedure, which uses HyperChem 9 or later, will not only
publish HyperChem calculations but any third-party calculations that HyperChem
has imported such as those from Gamess, Gaussian, Mopac, etc.
For example, HyperChem can parse a Gamess output file. The screen shot in
Fig. 9 shows three Gamess output files containing results for a single point ab initio
SCF calculation, a geometry optimization of structure and a vibrational analysis
A Portal for Quantum Chemistry Data … 21

Fig. 9 Using parser software to parse an output file to create a CSX file

calculation. Once imported into HyperChem 9, these results can be published just as
if they we computed by HyperChem.
While parsing output files is certainly possible, Chemical Semantics, Inc.
expects to work with developers of these computational chemistry packages to help
them install their own “Publish Button”. The Publish Button in HyperChem is
shown in Fig. 1.
In addition to the title of the publication, the authors, their organizations and
e-mail, and the publication abstract, pushing the publication button (which initially
creates a CSX file) adds a number of other things to the publication. The Login
Data… Button allows entering data so that the Publisher Package can use a login ID
and password to directly publish results. The Content Button allows choices to be
made of what is published among the available results as shown in Fig. 10. The
Flags button allows the author to choose to define the current state of the publi-
cation as shown in Fig. 11 or the Visibility (Private, Protected, and Public).
A private publication can be seen only by the authors, a protected publication can
be show to anyone that the authors send a URI to with a key, and a public
publication can be seen by anyone logged into the portal.
It is also possible to add tags (essentially keywords) to any publication as shown
in Fig. 12. These may help in searching. A common set of tags is available as well
as custom tags set by the authors.
22 B. Wang et al.

Fig. 10 Choosing content to publish

Fig. 11 Publishing flags

A Portal for Quantum Chemistry Data … 23

Fig. 12 Publishing tags

3.2 Logging in and Manual Upload of CSX

In addition to publishing by hitting a “Publish Button,” many scientists will publish

by creating a CSX file and uploading it to the portal site. As described earlier above,
Chemical Semantics, Inc. has created a new CSX standard, similar to CML, for
holding all the required details of a computational chemistry publication. This
includes details about the authors, the title of the publication, etc. as more or less
just shown in Fig. 1. The information transferred to the portal by hitting the Publish
Button is that stored in a transferred CSX file. Multiple pathways can be expected to
produce these CSX files as time progresses. The conversion from CSX to RDF
occurs at the portal server.
One first has to log into the Portal as shown in Fig. 13. One can register if one
does not yet have an ID and Password. After entering the portal one is met with a
24 B. Wang et al.

Fig. 13 Logging into the portal

list of one’s own publications but one can inspect all publications as well depending
upon their visibility (Private, Protected, Public).
One can peruse all publications based upon Author, Title, Category, Tag, etc.
and then view any publication.

3.3 Viewing Publications

Clicking on any publication allows you to view details of a publication. Each

publication requires a Unique Name to be used in generating the URI for this
publication. Normally these will be assigned at the server level by the portal. An
example URI is,
https://staging.chemsem.com/pub/ostlund-20161221102029/

This uniquely identifies this publication. It is dereferenceable and can be passed

to friends to access the publication if so desired. The amount of data displayed with
a publication depends on the type of calculation being published and is evolving
with the portal.
A Portal for Quantum Chemistry Data … 25

3.4 Data Federation

One of the problems with existing databases is that the data exists in silos of
isolation. The individual databases are difficult to merge and there is general dif-
ficulty in sharing data because of a lack of universal agreement on the database
schema, column names, etc. A fundamental aspect of the semantic web is its ability
to federate data, i.e. make data available globally. This comes about because of the
global data standards that have been set, because one can merge individual
ontologies easily and because two separate graph databases can be merged just by
adding a single link (predicate) from one graph to another (Fig. 14).
An elementary example of this federation is available at the Chemical Semantics
portal by clicking on the Data Federation Tab. If the molecule for the current
publication is Methyl Chloride, then clicking on the tab brings up something like
that shown in Fig. 15.
This displays the information about Methyl Chloride that exists at the
ChemSpider site of the Royal Society of Chemistry (RSC) and the Chemical
Entities of Biological Interest (ChEBI) site of the European Molecular Biology Lab
(EMBI-EBI).

Fig. 14 List of portal publications

Fig. 15 Federation of a portal publication

26 B. Wang et al.

4 Conclusion

The semantic web offers a new way to publish the data created by Quantum
Chemistry calculations that matches the capabilities of the World Wide Web. As
opposed to isolated silos of data that are difficult to find and/or share, the semantic
web makes sharing of data a fundamental attribute. Our portal is a demonstration of
this new technology and hopefully is a precursor of technology allowing scientist to
finally have a vehicle for the proper sharing of scientific data leading to new and
enhanced capabilities.

References

1. Berners-Lee T, Hendler J (2001) Publishing on the semantic web—the coming internet

revolution will profoundly affect scientific information. Nature 410:1023–1024
2. Alvarez-Moreno M, de Graaf C, Lopez N, Maseras F, Poblet J, Bo C (2015) Managing the
computational chemistry big data problem: the ioChem-BD platform. J Chem Inf Model
55:95–103
3. Web Ontology Language. https://www.w3.org/OWL/. Accessed 1 Dec 2016
4. Graph Databases. http://GraphDatabases.com. Accessed 1 Dec 2016
5. The World Wide Web Consortium. https://www.w3.org/. Accessed 1 Dec 2016
6. Resource Description Framework. https://www.w3.org/RDF/. Accessed 1 Dec 2016
7. Gordon MS, Schmidt MW (2005) Advances in electronic structure theory: gamess a decade
later. In: Dykstra CE, Frenking G, Kim KS, Scuseria GE (eds) Theory and applications of
computational chemistry: the first forty years. Elsevier, Amsterdam, pp 1167–1189
8. Gaussian 09, Revision A.1, Frisch MJ, Trucks GW, Schlegel HB, Scuseria GE, Robb MA,
Cheeseman JR, Scalmani G, Barone V, Mennucci B, Petersson GA, Nakatsuji H, Caricato M,
Li X, Hratchian HP, Izmaylov AF, Bloino J, Zheng G, Sonnenberg JL, Hada M, Ehara M,
Toyota K, Fukuda R, Hasegawa J, Ishida M, Nakajima T, Honda Y, Kitao O, Nakai H,
Vreven T, Montgomery JA Jr, Peralta JE, Ogliaro F, Bearpark M, Heyd JJ, Brothers E,
Kudin KN, Staroverov VN, Kobayashi R, Normand J, Raghavachari K, Rendell A, Burant JC,
Iyengar SS, Tomasi J, Cossi M, Rega N, Millam JM, Klene M, Knox JE, Cross JB, Bakken V,
Adamo C, Jaramillo J, Gomperts R, Stratmann RE, Yazyev O, Austin AJ, Cammi R,
Pomelli C, Ochterski JW, Martin RL, Morokuma K, Zakrzewski VG, Voth GA, Salvador P,
Dannenberg JJ, Dapprich S, Daniels AD, Farkas Ö, Foresman JB, Ortiz JV, Cioslowski J,
Fox DJ (2009) Gaussian, Inc., Wallingford, CT
9. Valiev M, Bylaska EJ, Govind N, Kowalski K, Straatsma TP, van Dam HJJ, Wang D,
Nieplocha J, Apra E, Windus TL, de Jong WA (2010) NWChem: a comprehensive and
scalable open-source solution for large scale molecular simulations. Comput Phys Commun
181:1477
10. HyperChem, Release 9, Hypercube, Inc. Gainesville, FL. www.hyper.com
11. The Nobel Prize in Chemistry (1998). http://www.nobelprize.org/nobel_prizes/chemistry/
laureates/1998/. Accessed 1 Dec 2016
12. Murray-Rust P, Rzepa H (1999) Chemical markup, XML, and the Worldwide Web. 1. Basic
principles. J Chem Inf Comput Sci 39:928–942
13. Service Station—More on REST. https://msdn.microsoft.com/en-us/magazine/dd942839.aspx
. Accessed 1 Dec 2016
A Portal for Quantum Chemistry Data … 27

14. SPARQL Protocol and RDF Query Language. https://www.w3.org/TR/rdf-sparql-query/.

Accessed 1 Dec 2016
15. JavaScript Object Notation for Linked Data. http://json-ld.org/. Accessed 1 Dec 2016
16. The IUPAC International Chemical Identifier. http://www.inchi-trust.org/. Accessed 1 Dec
2016
Matrix Elements for Explicitly-Correlated
Atomic Wave Functions

Frank E. Harris

Abstract We refer to atomic wave functions that contain the interelectron dis-
tances as “explicitly correlated”; we consider here situations in which an explicit
correlation factor rij can occur as a power multiplying an orbital functional form
(a Hylleraas function) and/or in an exponent (producing exponential correlation).
Hylleraas functions in which each wave-function term contains at most one linear
rij factor define a method known as Hylleraas-CI. This paper reviews the analytical
methods available for evaluating matrix elements involving exponentially-correlated
and Hylleraas wave functions; attention is then focused on computation of inte-
grals needed for the kinetic energy. In contrast to orbital-product and exponentially-
correlated wave functions, no general formulas have been developed by others to
relate the kinetic-energy integrals in Hylleraas-CI (or its recent extension by the
Nakatsuji group) to contiguous potential-energy matrix elements. The present paper
provides these missing formulas, obtaining them by using relevant properties of vec-
tor spherical harmonics. Validity of the formulas is confirmed by comparisons with
kinetic-energy integrals obtained in other ways.

1 Introduction

Ever since the first days of quantum mechanics investigators have sought meth-
ods for describing the electronic structures of atoms and molecules that are more
rapidly convergent than superposition-of-configurations (also called configuration-
interaction) expansions of the electronic wave function in orbital products. Probably

It is a pleasure to present this work as part of a tribute to Professor Josef Paldus in celebration
of his eightieth birthday.

F. E. Harris (✉)
Department of Physics, University of Utah, Salt Lake City, UT, USA
e-mail: [email protected]fl.edu
F. E. Harris
Quantum Theory Project, University of Florida, Gainesville, FL, USA

© Springer International Publishing AG, part of Springer Nature 2018 29

Y. A. Wang et al. (eds.), Concepts, Methods and Applications of Quantum Systems
in Chemistry and Physics, Progress in Theoretical Chemistry and Physics 31,
https://doi.org/10.1007/978-3-319-74582-4_2
30 F. E. Harris

the earliest endeavor of this type was that of Hylleraas, whose study of the He atom
[1] used a wave function that included as a multiplicative factor the explicit appear-
ance of the interelectron distance r12 . Although it was many years before wave func-
tions of this type came into widespread use (probably awaiting the availability of
digital computers), a few landmark studies using such wave functions (which we
identify as traditional Hylleraas functions) were soon carried out, including in par-
ticular a study of the hydrogen molecule by James and Coolidge [2], published in
1936. There followed in 1960 a further study of the hydrogen molecule ground state
by Kolos and Roothaan [3], in 1968 a detailed study of the lithium atom by Larsson
[4], and in 1994 an essentially quantitative computation of the ground state of the
hydrogen molecule by Kolos [5].
Attempts to apply Hylleraas methods to larger systems revealed that the occur-
rence of a wide variety of combinations of rij factors (and higher powers thereof)
led to exceedingly complicated computations, and as early as 1971 it was proposed
by Sims and Hagstrom [6], and independently by Woźnicki [7], to consider con-
figurations (wave function terms) that contained at most a single, linear rij factor.
Methods based on wave functions of this type, now usually referred to as Hylleraas-
CI (Hy-CI), were over time more fully developed and applied to a variety of atomic
problems. Representative work in this area is in [8–11].
An alternative to the Hy-CI development is the use of exponentially-correlated
wave functions. This type of wave function was proposed in the mid-1960s by Bon-
ham [12, 13], but at that time calculations based on it seemed impractical. However,
it was practical to use exponentially-correlated Gaussian orbitals, and work in that
area has been pursued by Rychlewski et al. [14]. Calculations based on exponentially-
correlated Slater-type orbitals finally became practical for small atomic systems after
publication of an extraordinary paper by Fromm and Hill [15].
The possibility of a practical extension to the Hy-CI method (identified by its
proposers as E-Hy-CI) has been examined by the Wang et al. [16], who developed
formulas for the “unlinked” integrals (defined in Sect. 4) that are encountered when
the single rij of a Hy-CI wave function is generalized to a form of the generic type
p
rijij exp(−𝛽ij rij ).
The present contribution reviews some aspects of electronic-structure compu-
tations by these Hylleraas-inspired methods, including a discussion of recently-
discovered methods for simplifying the computation of the kinetic-energy matrix
elements in both Hy-CI and E-Hy-CI.

2 Wave Functions

Before incorporation into an antisymmetrized space-spin function, the spatial parts

of the exponentially-correlated three- and four-body spatial wave functions, written
in terms of their internal coordinates (in which one particle, often a nucleus, is at the
origin of the coordinate system), can take the form
Matrix Elements for Explicitly-Correlated Atomic Wave Functions 31

m m
𝛹 (1, 2) = Yl 1 (𝛺1 )Yl 2 (𝛺2 )𝛷(1, 2) , (1)
1 2
m m m
𝛹 (1, 2, 3) = Yl 1 (𝛺1 )Yl 2 (𝛺2 )Yl 3 (𝛺3 )𝛷(1, 2, 3) , (2)
1 2 3

with

𝛷(1, 2) = r11 r22 r12 e−𝛼1 r1 −𝛼2 r2 −𝛽r12 ,

n n p
(3)
r11 r22 r33 r123 r132 r231 e−𝛼1 r1 −𝛼2 r2 −𝛼3 r3 −𝛽3 r12 −𝛽2 r13 −𝛽1 r23
n n n p p p
𝛷(1, 2, 3) = . (4)

Here Ylm are spherical harmonics, at Condon-Shortley phase [17], and 𝛺i stands for
the angular coordinates of Particle i. Particle i, at position 𝐫i , has radial coordinate ri ,
𝐫ij = 𝐫i − 𝐫j , and rij = |𝐫i − 𝐫j | is the distance between Particles i and j. Traditional
Hylleraas methods use wave functions with all 𝛽 equal to zero; Hylleraas-CI methods
additionally require each wave-function term to have at most one pi nonzero; that pi ,
if present, has the value unity.

3 Exponential Correlation

For three-body systems exponentially-correlated wave functions do not present major

problems because the three interparticle distances r1 , r2 , and r12 can be chosen as
internal coordinates, with the position and orientation of the three-particle triangle
described by specifying the center of mass (or the position of one particle) and the
Euler angles of the triangle orientation. In those coordinates, the internal “volume
element” is proportional to r1 r2 r12 , and the main computational issue is to deal with
the ranges of these distances.
For four-body systems it is still possible to use all the ri and rij as explicit coordi-
nates, but the internal volume element (contrary to a claim in Ref. [18]) is extremely
complicated, leading to integrals long thought to be intractable. Evaluation of the
four-body integrals needed for exponentially-correlated electronic-structure compu-
tations was the problem solved by Fromm and Hill [15]. Fromm and Hill’s solu-
tion, however, suffered from the difficulty that its implementation required laborious
tracking of the branches of the dilogarithm functions it contained. Avoidance of this
inconvenience and various other technical improvements were described in a paper
by the present author in 1997 [19]. That 1997 paper also resolved another issue:
Remiddi [20] had published completely analytical formulas for certain integrals aris-
ing in Hylleraas calculations, and they agreed numerically with the corresponding
special cases of the Fromm/Hill formula. But an analytical demonstration to that
effect was missing. A formula presented in [19] was shown to lead to the desired
analytical result.
As indicated above, the integrals needed for Hylleraas wave functions were spe-
cial cases of those needed for exponentially-correlated wave functions, so in prin-
ciple all such integrals for four-body systems could now be computed analyti-
cally. However, the formulas involved were rather unweildy, and the computational
32 F. E. Harris

process was greatly simplified by the development of recurrence formulas enabling

all four-body integrals to be obtained from a single starting value. This objective
was achieved, first for traditional Hylleraas integrals only, by Pachucki et al. [21],
and later, for general exponential correlation, by the present author [22].

4 Hylleraas Integrals

Integrals containing various combinations of rij as factors (we call these potential-
energy integrals) will occur in atomic Hylleraas calculations; each integral can be
identified with a diagram that is formed by
(1) Introducing a vertex corresponding to each particle not at the origin of the
coordinate system (usually the nucleus);
(2) For each factor rij or 1∕rij in the integral, drawing a line in the diagram that
connects vertices i and j.
Any vertex not connected to other vertices by a line corresponds to a one-particle
integral that is easily evaluated. Diagrams containing closed loops (with three or
more vertices) are termed linked, and any diagram or part of a diagram that is not
contained in a closed loop is called unlinked. Unlinked integrals and parts can be
evaluated in closed form after introducing the Laplace expansion of each 1∕rij [23]
and/or its generalization to the related quantity rij [24]. An alternative to the Laplace-
type expansion for unlinked integrations is to use a coordinate system in which the
unlinked rij are coordinates. That approach has been followed by Ruiz in [25–28] and
in other papers. Integrations over the particles in closed loops can also be treated
using Laplace-type expansions, but such linked integrations lead to infinite series
that are usually evaluated numerically.
For atomic Hy-CI, the limitations in the occurrence of rij factors cause the
potential-energy integrals to consist only of completely unlinked integrals involv-
ing four or fewer vertices, except for one three-vertex integral containing a linked
product of the form rij rik ∕rjk . This type of integral, often called a “triangle” integral,
was first discussed by Szasz [29].
While the general development for exponentially correlated wave functions in
principle provided closed analytic formulas for the Hylleraas triangle integrals,
those formulas were often more laborious to evaluate than expansions based on
Laplace-type formulas. For triangle integrals with general spherical harmonics, the
most utilized current approach is probably the Levin u-transformation convergence-
acceleration scheme [30] used by Sims and Hagstrom [9].
The diagrams denoting integrals arising in Hy-CI are the same as those occurring
in its extension E-Hy-CI, the only difference being that each diagram line refers to a
p
factor of type rijij exp(−𝛽ij rij ) instead of simply rij .
Matrix Elements for Explicitly-Correlated Atomic Wave Functions 33

5 Kinetic Energy

Unexpectedly, a bottleneck in Hylleraas-CI computations has been the computation

of kinetic-energy matrix elements. In completely orbital formulations, the deriva-
tives describing the kinetic energy in quantum mechanics simply change the powers
of ri in a matrix element, making it simple to identify the kinetic energy in terms of
potential-energy integrals. However, the presence of factors rij , either as powers or
in exponents, generates new complications.
When no explicit angular factors are present, the kinetic-energy operator T̂ can
be reduced to the form [31]
( )( )
1∑ 1 1 𝜕2 2 𝜕 ∑ 1 ∑ 𝜕2
T̂ = − + + − cos 𝜃ijk . (5)
2 i<j mi mj 𝜕rij2 rij rij i
mi j<k 𝜕rij 𝜕rik
j,k≠i

The indices in Eq. (5) run over all the particles (including any nuclei, whether or not
they are assumed to be of infinite mass), mi is the mass of Particle i, and cos 𝜃ijk is
the cosine of the angle between 𝐫ij and 𝐫ik . It can be evaluated as

rij2 + rik2 − rjk2

cos 𝜃ijk = . (6)
2rij rik

Because Eq. (5) is general, its use yields the kinetic energy even when the particles
are all of finite mass, thereby removing the need for an estimate of the nonphysical
quantity called “mass polarization”.
For wave functions containing explicit angular factors (e.g., spherical harmonics),
the kinetic-energy operator requires additional terms. This topic is discussed in [32–
34].
Evaluation of the kinetic energy for exponentially-correlated wave functions was
examined in 1993 by Rebane [35], who showed how the kinetic-energy matrix ele-
ments could be written in terms of suitable potential-energy contributions. Rebane’s
derivation was later simplified by the author’s research group [36]. Unfortunately
Rebane’s formula involves all the exponents of the exponentially-correlated wave
function and does not apply to the usual .

6 Kinetic Energy in Hylleraas-CI

In the absence of formulas relating Hylleraas kinetic-energy matrix elements to those

of potential energy, expressions based on the Laplace expansion can be differen-
tiated, leading after some complication to a set of kinetic-energy formulas. The
situation is even more cumbersome if, following Ruiz, one uses the interparticle
coordinates directly. In any case, relations involving kinetic-energy integrals would
34 F. E. Harris

be useful in reducing computational effort and/or in providing additional checks for

the numerical work. This section of the present paper summarizes work previously
presented by the present author [37] showing how the use of concepts associated with
vector spherical harmonics [38] can be applied to derive the missing kinetic-energy
formulas for Hy-CI.
Our starting point is to note that after integrating the variables of a kinetic-energy
matrix element that involve orbitals only, there may remain a two- or three-body
integral of which the most difficult are illustrated by the following:
⟨ ⟩
| 1 2|
| |
K2 = r12 𝜙a (1)𝜙b (2) |− ∇1 | r12 𝜙d (1)𝜙e (2) , (7)
| 2 |
⟨ ⟩
| 1 2|
| |
K3 = r13 𝜙a (1)𝜙b (2)𝜙c (3) |− ∇1 | r12 𝜙d (1)𝜙e (2)𝜙f (3) . (8)
| 2 |

Here 𝜙a is a Slater-type orbital (STO) of the form

ga (r) = rna −1 e−𝛼a r .

m
𝜙a = ga (r)Yl a (𝛺) , with (9)
a

For compactness in the final formulas for K2 and K3 we write am l

to denote an orbital
with parameters 𝛼a and na , but with the indicated quantum numbers l, m, which may
differ from the values la , ma of 𝜙a .
For both K2 and K3 we start by evaluating the application of the Laplacian for
Particle 1 to r12 𝜙d (1). Noting that ∇2 Ylm = −l(l + 1)Ylm ∕r2 , we find
[
( )
∇21 [r12 𝜙d (1)] = r12 ∇21 gd (1) + ∇21 r12 gd (1)
]
ld (ld + 1)r12 gd (1) m
− + 2∇1 gd (1) ⋅ ∇1 r12 Yl d (1)
r12 d

m m
+ 2r12 ∇1 gd (1) ⋅ ∇1 Yl d (1) + 2gd (1)∇1 r12 ⋅ ∇1 Yl d (1) . (10)
d d

Equation (10) corrects a misprint in the ld (ld + 1) term of [37].

Most of the quantities in Eq. (10) are easily simplified. Defining 𝐫12 = 𝐫1 − 𝐫2 and
letting an overline circumflex denote a unit vector,
( )
nd (nd − 1) 2𝛼 n
∇21 gd (1) = − d d + 𝛼d2 gd (1) , (11)
r12 r1
2
∇21 r12 = , (12)
r
(12 )
nd − 1
∇1 gd (1) = − 𝛼d gd (1) 𝐫̂ 1 , (13)
r1
∇1 r12 = 𝐫̂ 12 , (14)
Matrix Elements for Explicitly-Correlated Atomic Wave Functions 35

r12 + r12
2
− r22
𝐫̂ 1 ⋅ 𝐫̂ 12 = , (15)
2r1 r12
0 = ∇1 gd (1) ⋅ ∇1 Ylm (1) . (16)

With these substitutions, Eq. (10) becomes

[ ]
n2d − 1 − ld (ld + 1)
(2n + 1)𝛼
∇21 [r12 𝜙d (1)] = − d d
+ 𝛼d2 r12 𝜙d (1)
r12 r1
[ ]
𝛼d r22 (nd − 1)r22 𝜙d (1)
+ nd + 1 − 𝛼d r1 + −
r1 r2 r12
1
m
+ 2gd (1) 𝐫̂ 12 ⋅ ∇1 Yl d (1). (17)
d

Equation (17) corrects a sign error that was present in the corresponding equation of
[37].
One further simplification can now be easily made to the final term of Eq. (17):
m
The orthogonality of 𝐫̂ 1 and ∇1 Yl d (1) permit us to write
d

m 2r2 gd (1) m
2gd (1) 𝐫̂ 12 ⋅ ∇1 Yl d (1) = − 𝐫̂ 2 ⋅ ∇1 Yl d (1) . (18)
d r12 d

As explained in more detail in [37], the properties of vector spherical harmonics

can now be used to complete the evaluation of Eq. (18).
Introducing the complex unit vectors

𝐱̂ + i𝐲̂ 𝐱̂ − i𝐲̂
𝐞̂ 1 = − √ , 𝐞̂ −1 = √ , 𝐞̂ 0 = 𝐳̂ , (19)
2 2

with orthogonality relation

𝐞̂ 𝜇 ⋅ 𝐞̂ −𝜈 = (−1)𝜈 𝛿𝜇𝜈 , (20)

it can be shown that

√ 1
4𝜋 ∑
𝐫̂ 2 = (−1)𝜇 Y1−𝜇 (2) 𝐞̂ 𝜇 , (21)
3 𝜇=−1
1 ( )1∕2
md 1 ∑ ∑ ld (ld + 1)(2ld + 1 − 𝜆
∇1 Yl (1) =
d r1 𝜈=−1 𝜆=±1 2(2ld + 1)
⟨ ⟩
l d + 𝜆 1 ld md −𝜈
× Yl +𝜆 (1) 𝐞̂ 𝜈 , (22)
md −𝜈 𝜈 md d
36 F. E. Harris

and that the final term of Eq. (17) reduces to

√ 1 ( )1∕2
2r g (1) 4𝜋 ∑ ∑ ld (ld + 1)(2ld + 1 − 𝜆
=− 2 d
m
2gd (1) 𝐫̂ 12 ⋅ ∇1 Yl d (1)
d r1 r12 3 𝜈=−1 𝜆=±1 2(2ld + 1)
⟨ ⟩
ld + 𝜆 1 ld md −𝜈
× Yl +𝜆 (1)Y1𝜈 (2).
md −𝜈 𝜈 md d

(23)

In these equations the array in angle brackets is a Clebsch-Gordan coefficient as

defined in the Appendix; the notation we are using for it is not standard but the
author hopes it will become more widely adopted.
The spherical harmonic Y1𝜈 (2) will in overall computations occur multiplied by
the harmonic contained in 𝜙(2); that product can be reduced to a sum of single har-
monics. This reduction takes the form
[ ]
∑ l 1 l + 𝜆′ m +𝜈
𝜈 me
Y1 (2)Yl (2) = e e
Y e (2) , (24)
e
′
me 𝜈 −me−𝜈 le +𝜆′
𝜆 =±1

where the array in brackets is a Gaunt coefficient (also in a nonstandard notation

suggested by the author). There is unfortunately no single widely-accepted defini-
tion for the Gaunt coefficients; we use here that given in the Appendix, which has
the properties of being both analogous to the definition of the Wigner 3-j symbol
[39] and in agreement with the Gaunt-coefficient definition chosen by Pinchon and
Hoggan [40].
Combining Eqs. (17), (18), and (24), we obtain a final expression containing no
differential operators for − 12 ∇21 r12 𝜙d (1)𝜙e (2); it is then straightforward to insert that
expression into Eqs. (7) and (8). We display here the formula for K3 ; that for K2 is
included in [37]:
⟨ ⟩
ld (ld + 1) − n2d + 1 |r r | 𝛼2 ⟨ ⟩
| 12 13 |
K3 = 𝛷abc | 2 | 𝛷def − d 𝛷abc ||r12 r13 || 𝛷def
2 | | 2
| r1 |
⟨ ⟩ ⟨ ⟩
(2nd + 1)𝛼d | r12 r13 | nd + 1 | r13 |
+ 𝛷abc || | | |
2 | 𝛷def − 2
𝛷abc | | 𝛷def
| r1 | | r12 |
⟨ ⟩ ⟨ ⟩
𝛼d | 2
r2 r13 || nd − 1 | 2 |
| r1 r13 | r2 r13 |
+ 𝛷abc | − |𝛷 + 𝛷abc | 2 | 𝛷def
2 | r12 r1 r12 || def 2 |r r |
| | 1 12 |
√ ( )1∕2 1 ⟨ ⟩
4𝜋 ∑ ld (ld +1)(2ld +1−𝜆) ∑
me +𝜈 ld + 𝜆 1 ld
+ (−1)
3 𝜆=±1 2(2𝜆 + 1) md −𝜈 𝜈 md
𝜈=−1
[ ] ⟨ ⟩
∑ l 1 l + 𝜆′ | r r | md −𝜈 me +𝜈 mf
× e e
𝛷abc || 2 13 || dl +𝜆 el +𝜆′ fl . (25)
m e 𝜈 −m e −𝜈 | r1 r12 | d e f
𝜆′ =±1
Matrix Elements for Explicitly-Correlated Atomic Wave Functions 37

This equation is a corrected form of the corresponding formula in [37].

A salient feature of the formula for K3 is that it relates the matrix element for given
angular quantum numbers to those of neighboring angular indices; this behavior is
indeed to be expected because of the properties of the spherical harmonics.
The formula for K2 (not shown but given in [37]) provides an alternative to a
relation identified by Kolos and Roothaan [41]. However, the technique employed
by Kolos and Roothaan does not extend to cover the three-body integral represented
here as K3 .

7 Kinetic Energy in Extended Hylleraas-CI

A procedure similar to that outlined in Sect. 6 can be applied to the kinetic-energy

matrix elements in E-Hy-CI. We summarize here some of the results; a more com-
plete discussion will be published elsewhere [42].
One type of integral relevant here has the form
⟨ ⟩
p′ | 1 | p
r13 e−𝛽 r13 𝜙a (1)𝜙b (2)𝜙c (3) ||− ∇21 || r12 e−𝛽r12 𝜙d (1)𝜙e (2)𝜙f (3) .
′
K3EHCI = (26)
| 2 |

In a process similar to that used for Hy-CI, we start by writing

[
[ ] ( [ p ])
∇21 r12 e−𝛽r12 𝜙d (1) = r12 e−𝛽r12 ∇21 gd (1) + ∇21 r12 e−𝛽r12 gd (1)
p p

]
ld (ld + 1)r12 e−𝛽r12 gd (1)
p
[ p ] m
− + 2∇1 gd (1) ⋅ ∇1 r12 e−𝛽r12 Yl d (1)
r12 d

[ p ]
+ 2r12 e−𝛽r12 ∇1 gd (1) + 2gd (1)∇1 r12 e−𝛽r12 ⋅ ∇1 Yl d (1) ,
p m m
⋅ ∇1 Yl d (1) (27)
d d

which differs from Eq. (10) only by replacement of r12 everywhere it occurs by
r12 e−𝛽r12 . Examination of Eq. (27) shows that we now need to evaluate the new quan-
p

tities
( )
[ ] p(p + 1) 2𝛽(p + 1)
∇21 r12 e−𝛽r12 = + 𝛽 2 r12 e−𝛽r12 ,
p p
2
− (28)
r12 r 12
( )
[ p ] p
∇1 r12 e−𝛽r12 = − 𝛽 r12 e−𝛽r12 𝐫̂ 12 .
p
(29)
r12

Inserting the results from Eqs. (28) and (29) into Eq. (27) and then proceeding as in
Sect. 6, we reach
38 F. E. Harris
⟨ ⟩
ld (ld + 1) − (nd + p)(nd − 1) |f f |
| 12 13 |
K3EHCI= 𝛷abc | 2 | 𝛷def
2 | r |
| 1 |
⟨ ⟩ 2 2
(2nd + p)𝛼d |f f | 𝛼 + 𝛽 ⟨ ⟩
+ 𝛷abc || 12 13 || 𝛷def − d 𝛷abc ||f12 f13 || 𝛷def
2 | 1 |
r 2
⟨ ⟩ ⟨ ⟩
p(nd + p) | | 𝛼d p |f f r 2|
| f12 f13 | | 12 13 1 f12 f13 r2 |
− 𝛷abc | 2 | 𝛷def + 𝛷abc | 2 − |𝛷
2 | r | 2 | r 2 | def
r1 r12
| 12 | | 12 |
⟨ ⟩ ⟨ ⟩
(n − 1)p | 2|
𝛽(nd + 2p + 1) |f f |
| f12 f13 r |
+ d 𝛷abc | 2 2 2 | 𝛷def + 𝛷abc || 12 13 || 𝛷def
2 | r r | 2 | r12 |
| 1 12 |
⟨ ⟩
𝛼d 𝛽 |f f r 2|
| 12 13 12 f12 f13 r1 f12 f13 r2 |
− 𝛷abc | + − |𝛷
2 | r1 r12 r1 r12 || def
|
⟨ ⟩
𝛽(nd − 1) |f f r r22 f12 f13 ||
| 12 13 12
+ 𝛷abc | − 2 | 𝛷def
2 | r2 r1 r12 ||
| 1
√ ( )1∕2 1 ⟨ ⟩
4𝜋 ∑ ld (ld +1)(2ld +1−𝜆) ∑ l + 𝜆 1 ld
+ (−1)me +𝜈 d
3 𝜆=±1 2(2ld + 1) md −𝜈 𝜈 md
𝜈=−1
[ ] [ ⟨ ⟩
∑ l 1 l + 𝜆′ |f f r |
| 12 13 2 | md −𝜈 me +𝜈 mf
× e e
p 𝛷abc | |d e f
me 𝜈 −me −𝜈 | r r2 | ld +𝜆 le +𝜆′ lf
𝜆′ =±1 | 1 12 |
⟨ ⟩]
| f12 f13 r2 | md −𝜈 me +𝜈 mf
−𝛽 𝛷abc | | | .
|d e ′ f
| r1 r12 | ld +𝜆 le +𝜆 lf
(30)

To keep the above formula more compact, we have defined f12 = r12 e−𝛽r12 and f13 =
p

p′ −𝛽 ′ r13
r13 e .
Equation (30) confirms that the kinetic-energy matrix elements in E-Hy-CI reduce
to contiguous potential-energy integrals.

8 Numerical Verification

The formulas for kinetic-energy intergrals in Hy-CI developed here and (in more
detail) in [37] were confirmed by comparing integrals produced using them with
similar integrals computed in other ways by Ruiz [25, 27, 28] and by Sims and
Hagstrom [10]. After making some adjustments needed to achieve consistency (see
[37]), complete agreement with the results of those investigators was obtained.
The errors noted in various equations of [37] arose while transcribing the formu-
las from computer programs and therefore did not affect the numerical verification
process.
Matrix Elements for Explicitly-Correlated Atomic Wave Functions 39

We could not find literature values of E-Hy-CI kinetic-energy integrals for com-
parison with the formula in Sect. 7 of the present contribution and therefore cannot
present data to provide its numerical confirmation.

9 Conclusions

This paper summarizes features of a variety of types of correlated-orbital atomic

calculations and identifies new formulas yielding kinetic-energy matrix elements
for the Hylleeraas-CI method of Sims and Hagstron and of Woźnicki and for the
extended Hylleraas-CI method proposed by the Nakatsuji group.

Acknowledgements Completion of the numerical verifications referred to in this work involved

significant consultations with Drs. María Belén Ruiz and James Sims. The author is pleased and
grateful to acknowledge their assistance.

Appendix. Angular-Momentum Coefficients

The spherical harmonics Ylm (𝜃, 𝜙), alternatively written Ylm (𝛺), can be defined with
the sign convention chosen by Condon and Shortley [17] (Condon-Shortley phase)
by the Rodrigues formula

(−1)m 2 m∕2 d
l+m
Ylm (𝛺) = Nlm (1 − u ) (u2 − 1)l eim𝜙 , (31)
2l l! dul+m
where u = cos 𝜃 and Nlm is the factor
√
(2l + 1)(l − m)!
Nlm = (32)
4𝜋(l + m)!

that makes the Ylm orthonormal. With these definitions,

Ylm (𝛺)∗ = (−1)m Yl−m (𝛺). (33)

A product of spherical harmonics of the same argument 𝛺 can be expanded into

a sum of harmonics of that argument. The coefficients in that expansion are known
as Gaunt coefficients. Unfortunately there is no unanimity as to the definition of the
Gaunt coefficient. Choosing the definition of Pinchon and Hoggan [40], we introduce
a bracket notation that we hope will become adopted:
[ ]
l 1 l2 l3 m m m
= Y 1 (𝛺)Yl 2 (𝛺)Yl 3 (𝛺) d𝛺 . (34)
m1 m2 m3 ∫ l1 2 3
40 F. E. Harris

m m
Expansion of the spherical harmonic product Yl 1 Yl 2 in the orthonormal set YLM ,
1 2
carried out by taking scalar products with (YLM )∗ , leads after use of Eqs. (33) and
(34) to [ ]
∑ l l L
(−1)M 1 2
m m
Yl 1 (𝛺)Yl 2 (𝛺) = Y M (𝛺). (35)
1 2 m1 m2 −M L
LM

Because harmonics with upper indices m1 and m2 form a product all of whose terms
have the same value of M, Eq. (35) can be simplified by dropping the M summation,
setting M = m1 + m2 .
The Gaunt coefficients can be written in terms of Wigner 3-j symbols [39]. Using
the standard notation for that symbol (an array of l and m values in ordinary paren-
theses), the Gaunt coefficients as defined here can be written
[ ] √ ( )( )
l 1 l2 l3 (2l1 + 1)(2l2 + 1)(2l3 + 1) l1 l2 l3 l1 l2 l3
= . (36)
m1 m2 m3 4𝜋 m1 m2 m3 0 0 0

A pair of angular momenta in two independent variables can be coupled to form

a quantity of definite resultant angular momentum by forming a linear combination
of products of the individual angular momenta; the coefficients in that expansion
are called Clebsch-Gordan coefficients. Coupling of the angular-momentum wave
m m
functions 𝜓j 1 (1) and 𝜓j 2 (2) with fixed values of j1 and j2 to form the combined
1 2
function 𝛹JM (1, 2) is described by
⟨ ⟩
∑ j1 j2 J m m
𝛹JM (1, 2) = 𝜓j 1 (1)𝜓j 2 (2), (37)
m1 m2
m1 m2 M 1 2

where the array in angle brackets is our (nonstandard) notation for the Clebsch-
Gordan coefficient. Here all contributing terms must satisfy m1 + m2 = M, so we
can actually reduce Eq. (37) to a single sum over, say, m2 , with m1 set to M − m2 .
The Clebsch-Gordan coefficients can also be written in terms of 3-j symbols:
⟨ ⟩ ( )
j1 j 2 j 3 j1 −j2 +m3
√ j1 j2 j3
= (−1) 2j3 + 1 . (38)
m1 m2 m3 m1 m2 −m3

Well-documented computer programs exist for the evaluation of the 3-j symbols,
making it straightforward to evaluate expressions involving Gaunt or Clebsch-Gordan
coefficients.

References

1. Hylleraas EA (1929) Z Phys 54:347

2. James HM, Coolidge AS (1936) Phys Rev 49:688
Matrix Elements for Explicitly-Correlated Atomic Wave Functions 41

3. Kolos W, Roothaan CCJ (1960) Rev Mod Phys 32:205

4. Larsson S (1968) Phys Rev 169:49
5. Kolos W (1994) J Chem Phys 101:1330
6. Sims JS, Hagstrom SA (1971) J Chem Phys 55:4699
7. Woźnicki W (1971) Theory of electronic shell in atoms and molecules, Jucys A (ed), p 103.
Mintis, Vilnius
8. Pestka G, Woźnicki W (1996) Chem Phys Lett 255:281
9. Sims JS, Hagstrom SA (2004) J Phys B 37:1519
10. Sims JS, Hagstrom SA (2007) J Phys B 40:1575
11. Sims JS, Hagstrom SA (2015) J Phys B 48:175003
12. Bonham RA (1965) J Mol Spectrosc 15:112
13. Bonham RA (1966) J Mol Spectrosc 20:197
14. Rychlewski J, Cencek W, Komasa J (1994) Chem Phys Lett 229:657
15. Fromm DM, Hill RN (1987) Phys Rev A 36:1013
16. Wang C, Mei P, Kurokawa Y, Nakashima H, Nakatsuji H (2012) Phys Rev A 85:042512
17. Condon EU, Shortley G (1951) The theory of atomic spectra, Cambridge University Press
18. Frolov AM, Smith VH (2001) J Chem Phys 115:1187
19. Harris FE (1997) Phys Rev A 55:1820
20. Remiddi E (1991) Phys Rev A 44:5492
21. Pachucki K, Puchalski M, Remiddi E (2004) Phys Rev A 70:032502
22. Harris FE (2009) Phys Rev A 79:032517
23. Arfken GB, Weber HJ, Harris FE (2013) Math Methods Phys, 7th edn. Academic Press, New
York, pp 799–800
24. Jen CK (1933) Phys Rev 43:540
25. Ruiz MB (2009) J Math Chem 46:24
26. Ruiz MB (2009) J Math Chem 46:1322
27. Ruiz MB (2011) J Math Chem 49:2457
28. Ruiz MB (2016) J Math Chem 54:1083
29. Szasz L (1962) Phys Rev 126:169
30. Levin D (1973) Int J Comput Math B 3:371
31. Frost AA (1962) Theor Chim Acta (Berl) 1:36–41
32. Harris FE (2004) Fundamental world of quantum chemistry: a tribute volume to the memory
of Per-Olov Löwdin: Brändas EJ, Kryachko ES (eds), vol 3. Kluwer, Dordrecht, pp 115–128
33. Harris FE (2004) In Advances in Quant Chem: Sabin JR, Brändas EJ (eds), vol 47. Academic
Press, New York, pp 129–155
34. Harris FE (2005) In Advances in Quant Chem: Sabin JR, Brändas EJ (eds), vol 50. Academic
Press, New York, pp 61–75
35. Rebane TK (1993) Opt Spektrosk 75:945, 557
36. Harris FE, Frolov AM, Smith VH (2003) J Chem Phys 119:8833
37. Harris FE (2016) J Chem Phys 144:204110; Erratum (2016) 145:129901
38. Morse PM, Feshbach H (1953) Methods of theoretical physics. Part II. McGraw-Hill, New
York, pp 1898–1901
39. Edmonds AR (1957) Ang Moment Quant Mech. Princeton University Press, Princeton, NJ
40. Pinchon D, Hoggan PE (2007) Int J Quant Chem 107:2186
41. Kolos W, Roothaan CCJ (1960) Rev Mod Phys 32:219
42. Harris FE (to be published)
Effective Bond-Strength Indicators

Gui-Xiang Wang, Yuzhe Stan Chen, Ya-Kun Chen

and Yan Alexander Wang

Abstract To save time and computer resources, we made an attempt to design

reasonable yet simple structural indicators to identify weak chemical bonds, instead
of performing numerous, tedious calculations of individual bond dissociation
energies (BDEs) for all bonds within a molecule. Based on the commonly available
structure-property indicators for bond strength, such as bond length (R), the
Mulliken interatomic electron number (MIEN), the Wiberg bond order (WBO), and
BDE, we have created two new bond-strength indicators, i.e., M = MIEN/R and
K = (WBO × MIEN)/R2, which shall be directly used to efficiently identify almost
all weak bonds with BDE below 350 kJ/mol. If several bonds of the same type
attain the same smallest values of M or K, values of the electron density at the bond
critical points (ρc) alone can almost always pinpoint the weakest bond from the set
of weak bonds, greatly reducing the amount of efforts in carrying out the calcu-
lations of the BDEs of the corresponding bonds.

Keywords Structural indicator ⋅

Weakest bond ⋅
Weak bond
Bond energy ⋅Bond dissociation energy ⋅
Bond order ⋅
Explosives
Trigger bond ⋅Trigger bond indicator ⋅
Bond critical point ⋅
Quantum theory
of atoms in molecules ⋅Population analysis ⋅
Counterpoise correction
Basis set superposition error

G.-X. Wang (✉)

Department of Chemistry, Computation Institute for Molecules and Materials,
Nanjing University of Science and Technology, Nanjing 210094, China
e-mail: [email protected]
G.-X. Wang ⋅ Y. S. Chen ⋅ Y.-K. Chen ⋅ Y. A. Wang (✉)
Department of Chemistry, University of British Columbia, 2036 Main Mall,
Vancouver, BC V6T 1Z1, Canada
e-mail: [email protected]

© Springer International Publishing AG, part of Springer Nature 2018 43

Y. A. Wang et al. (eds.), Concepts, Methods and Applications of Quantum Systems
in Chemistry and Physics, Progress in Theoretical Chemistry and Physics 31,
https://doi.org/10.1007/978-3-319-74582-4_3
44 G.-X. Wang et al.

1 Introduction

Bond dissociation energy (BDE) is important to the understanding of the mecha-

nisms of chemical reactions and crucial to the design, synthesis, and performance
studies of new functional materials. In the field of high-energy density explosives,
the pyrolysis of explosive molecules is considered to be the critical step in the
explosion process. Especially, the BDE of the weakest (trigger) bond plays a sig-
nificant role in the initial reaction of explosion. Hence, identification of the trigger
bond becomes mandatory in the studies of the safety and reliability of explosives.
For a homolytic bond breaking process [1–8],

bond breaking
A − B









! A∙ + B∙ ,

the BDE can be calculated according to the formula [9–11]:

h i
BDEAB = Ẽ ðZP)aAðBÞ
∪b
+ Ẽ ðZP)aðAÞB
∪b
− E(ZP)aAB∪ b
    ð1Þ
+ E(ZP)aA − Ẽ ðZP)aA + E(ZP)bB − Ẽ ðZP)bB .

Here, ZP means all energies (E and Ẽ) are corrected for the zero-point vibrational
effect. The capital letters A and B refer to the corresponding molecular fragments
produced after breaking the bond between A and B, and the small letters a and
b stand for the basis sets attached to the fragments. Before the homolysis reaction
happens, the parent compound optimized within the combined basis set (a ∪ b) has
energy E(ZP)aAB∪ b. After the reaction occurs, A and B radicals optimized each
individually within their own basis sets, a and b, have energies E(ZP)aA and E(ZP)bB,
respectively. To mitigate the inconsistency of the basis sets used before and after
the reaction, the counterpoise correction [10, 11] for the basis-set superposition
error (BSSE) has been adopted in Eq. (1), in which four more single-point energies
with molecular fragments A and B frozen in their geometries within the parent
∪b ∪b
molecule are calculated: Ẽ(ZP)aA(B) with ghost B, Ẽ(ZP)a(A)B with ghost A, Ẽ(ZP)aA,
and Ẽ(ZP)B. For a given ghost structure, the full sets of its basis functions and
b

numerical integration grid points are still present as before, but there are neither
∪b
nuclear charges nor electrons within the ghost. Therefore, both Ẽ(ZP)aA(B) and Ẽ
a∪b
(ZP)(A)B are still calculated within the total combined basis set (a ∪ b).
Equation (1) clearly shows that for every bond breaking occurrence, an accurate
estimate of the BDE involves at least four single-point calculations, two geometry
optimizations, and six vibrational frequency analyses for the ZP correction. Hence,
calculating the BDEs of all bonds of a large molecule to identify the weakest bond
is very labor-intensive and time-consuming. It is thus highly desirable to design a
simpler structural indicator to replace the enormous amount of BDE calculations. In
conventional chemical wisdom, bond length (R), bond order (BO), and bond
energy (BE) are the three commonly used key parameters reflecting the strength of
Effective Bond-Strength Indicators 45

Table 1 Average bond Single Ra BEa BEa/Ra 1/Ra

lengths (Ra) and bond bond (pm) (kJ/mol) (kJ mol−1 Å−1) (Å−1)
energies (BEa) of various
common single bonds [12] H–H 74.14 436 588.1 1.349
O–H 97 464 478.4 1.031
N–H 100 389 389.0 1.000
C–H 110 414 376.4 0.909
Cl–H 127.4 431 338.3 0.785
S–H 132 368 278.8 0.758
Br–H 141.4 364 257.4 0.707
C–O 143 360 251.7 0.699
C–N 147 305 207.5 0.680
C–C 154 347 225.3 0.649
I–H 160.9 297 184.6 0.622
C–Cl 178 339 190.4 0.562
Cl–Cl 199 243 122.1 0.503
Br–Br 228 193 84.6 0.439
I–I 266 151 56.8 0.376
F–H 91.7 565 616.1 1.091
N–N 145 163 112.4 0.690
N–O 136 222 163.2 0.735
O–O 145 142 97.9 0.690
F–F 143 159 111.2 0.699

a chemical bond. Empirically, there exists a good quantitative relationship between

R, BO, and BE: the longer R, the less BE or the lower BO a bond has, the easier the
bond breaks [12]. Thus, R or BO might be directly used to identify the trigger bond,
without relying upon BE.
Along this line of reasoning, the experimental average bond lengths (Ra) and
bond energies (BEa) sampled over many different chemical species containing 20
most common single bonds are gathered in Table 1 [12]. As a whole, the data
collected in Table 1 indeed validate the general observation: the longer R a bond
has, the smaller BE is, i.e., the bond breaks more easily. The linear regression
between BEa/Ra and 1/Ra is drawn in Fig. 1 with a correlation coefficient 0.99.
Despite the fact that specific bonds of the same type come with many different
variations in R and BE due to local chemical environment, it is still very reassuring
in discovering such a strong linear correlation between BEa/Ra (in kJ mol−1 Å−1)
and 1/Ra (in Å−1):
   
BEa 1
= 568.59 − 150.91. ð2Þ
Ra Ra

Accordingly, it might be a good idea to directly employ R to identify the weakest

bond.
46 G.-X. Wang et al.

Fig. 1 Relationship between average bond energies (BEa) and bond length (Ra). The straight line
is a least-square linear fit to the data points denoted by blue squares; data points marked by red
circles are excluded from the linear fit. All data are collected in Table 1

On the other hand, many previous studies [4–7] have primarily used the
Mulliken interatomic electron number (MIEN), calculated via the Mulliken
population analysis [13], to identify the trigger bond. Normally, only the bond with
the smallest MIEN from the same type of bonds is selected for further consideration
[4–7]. However, the bond with the smallest MIEN might not necessarily have the
lowest BDE among the bonds within a molecule. Taking 2,4,6-trinitrotoluene
(TNT) for example, the MIENs of all C–NO2 bonds are among the smallest (see
Table 2), indicating one of the C–NO2 bonds to be the trigger bond. However, to
single out which C–NO2 bond to break first, calculations of the BDEs of these
specific C–NO2 bonds have been carried out. In Table 2, the C3–N11 bond has the
smallest MIEN (0.1468), whereas the C1–N14 or C5–N8 bond has the lowest BDE,
237.75 kJ/mol. Clearly, the smallest MIEN may not correspond to the lowest BDE.
Alternatively, within the quantum theory of atoms in molecules [14, 15], Bader
proposed several parameters to indicate the relative bond strength using the idea
associated with bond critical points (BCPs). These BCPs are located at the inter-
atomic surface between a pair of atoms, at which the electron density reaches
minimum in one dimension, yet reaches maximum in the other two dimensions. As
Bader originally suggested, the value of the electron density at such a bond critical
point between a pair of atoms of a chemical bond, ρc, can be used to measure the
strength of the chemical bond. On the other hand, the BE is an integral of the
electron density over the associated interatomic surface between an atomic pair.
Unfortunately, this BE integral has an unknown system-dependent dimensionless
proportionality pre-factor. As a result, Bader’s ρc indicator can only be used to
compare the same kind of bonds between the same pair of atoms within very similar
local chemical environment and might not be able to evaluate the strengths of
different kinds of bonds even within the same molecule.
Effective Bond-Strength Indicators 47

Table 2 Bond strength indicators of selected explosive molecules

Molecule Bond Ra WBOb MIENc Wd Me Kf ρgc BDEh
TNT C1–N14 1.481 0.9138 0.1504 0.62 0.10 0.06 0.2621 237.75
C3–N11 1.475 0.9214 0.1468 0.62 0.10 0.06 0.2653 265.01
C5–N8 1.481 0.9138 0.1504 0.62 0.10 0.06 0.2621 237.75
C6–C7 1.509 1.0375 0.3819 0.69 0.25 0.17 0.2502 403.65
C7–H19 1.089 0.9062 0.3559 0.83 0.33 0.27 0.2758 360.77
C7–H20 1.089 0.9062 0.3559 0.83 0.33 0.27 0.2758 360.77
C7–H21 1.094 0.8851 0.3611 0.81 0.33 0.27 0.2699 361.26
TNM C1–N16 1.481 0.8880 0.1532 0.60 0.10 0.06 0.2614 232.21
C3–N12 1.478 0.8984 0.1458 0.61 0.10 0.06 0.2625 233.19
C5–N8 1.477 0.9009 0.1447 0.61 0.10 0.06 0.2629 232.24
C2–C15 1.511 1.0278 0.3611 0.68 0.24 0.16 0.2498 404.02
C4–C11 1.511 1.0286 0.3762 0.68 0.25 0.17 0.2495 403.30
C6–C7 1.511 1.0282 0.3709 0.68 0.25 0.17 0.2499 403.41
C7–H19 1.095 0.8910 0.3614 0.81 0.33 0.27 0.2705 365.27
C15–H26 1.089 0.9085 0.3518 0.83 0.32 0.27 0.2764 366.21
PNT C2–N19 1.483 0.8881 0.1484 0.60 0.10 0.06 0.2659 208.11
C3–N16 1.484 0.8904 0.1284 0.60 0.09 0.05 0.2671 200.62
C4–N13 1.484 0.8890 0.1313 0.60 0.09 0.05 0.2671 197.65
C5–N10 1.484 0.8912 0.1263 0.60 0.09 0.05 0.2669 200.63
C6–N7 1.485 0.8856 0.1533 0.60 0.10 0.06 0.2650 208.40
C1–C22 1.510 1.0284 0.3673 0.68 0.24 0.17 0.2493 419.52
C22–H23 1.095 0.8878 0.3600 0.81 0.33 0.27 0.2704 365.39
C22–H24 1.094 0.9039 0.3674 0.83 0.34 0.28 0.2714 365.21
C22–H25 1.088 0.9033 0.3526 0.83 0.32 0.27 0.2778 365.00
TNCr C2–N8 1.479 0.8908 0.1555 0.60 0.11 0.06 0.2623 235.36
C4–N10 1.431 1.0545 0.1707 0.74 0.12 0.09 0.2779 294.13
C6–N12 1.455 0.9805 0.1574 0.67 0.11 0.07 0.2694 250.81
C5–N11 1.337 1.3579 0.3829 1.02 0.29 0.29 0.3413 434.30
C1–C7 1.511 1.0321 0.3656 0.68 0.24 0.17 0.2497 386.34
C3–O9 1.322 1.1728 0.3734 0.89 0.28 0.25 0.3223 453.39
C7–H19 1.094 0.8892 0.3588 0.81 0.33 0.27 0.2699 367.69
C7–H20 1.090 0.9054 0.3655 0.83 0.34 0.28 0.2742 367.31
C7–H21 1.089 0.9044 0.3461 0.83 0.32 0.26 0.2770 367.12
O9–H22 0.997 0.5986 0.2117 0.60 0.21 0.13 0.3106 391.96
N11–H23 1.011 0.7385 0.2746 0.73 0.27 0.20 0.3264 424.61
N11–H24 1.013 0.7360 0.2695 0.73 0.27 0.19 0.3245 405.53
CDNAPY C2–N8 1.466 0.9397 0.1335 0.64 0.09 0.06 0.2695 248.56
C4–N9 1.452 0.9729 0.1237 0.67 0.09 0.06 0.2730 278.67
C5–N10 1.339 1.3124 0.3491 0.98 0.26 0.26 0.3415 455.55
C1–Cl7 1.738 1.0973 0.3046 0.63 0.18 0.11 0.2057 325.22
C3–H15 1.082 0.8624 0.3366 0.80 0.31 0.25 0.2841 470.41
N10–H16 1.011 0.7516 0.2808 0.74 0.28 0.21 0.3255 435.63
N10–H17 1.010 0.7875 0.2942 0.78 0.29 0.23 0.3284 434.71
(continued)
48 G.-X. Wang et al.

Table 2 (continued)
Molecule Bond Ra WBOb MIENc Wd Me Kf ρgc BDEh
PAM C2–N7 1.477 0.9191 0.1475 0.62 0.10 0.06 1.1958 217.54
C4–N9 1.477 0.9191 0.1475 0.62 0.10 0.06 1.1958 217.54
C6–N10 1.474 0.9241 0.1501 0.63 0.10 0.06 1.2627 262.69
C3–O8 1.340 1.0529 0.3040 0.79 0.23 0.18 1.3156 354.45
O8–C17 1.453 0.8424 0.2252 0.58 0.16 0.09 1.0756 263.82
C17–H20 1.090 0.9267 0.3626 0.85 0.33 0.30 0.9198 398.18
C17–H21 1.090 0.9267 0.3626 0.85 0.33 0.28 0.9198 398.18
C17–H22 1.090 0.9305 0.3856 0.85 0.35 0.28 0.9158 398.18
AMNA N2–N4 1.422 0.9807 0.1603 0.69 0.11 0.08 0.3241 107.87
N2–N3 1.397 1.0575 0.2451 0.76 0.18 0.13 0.3306 248.42
C1–N2 1.463 0.9571 0.2703 0.65 0.18 0.12 0.2575 233.11
C1–H7 1.088 0.9131 0.3737 0.84 0.34 0.29 0.2807 376.14
C1–H8 1.099 0.9240 0.3729 0.84 0.34 0.29 0.2718 376.41
C1–H9 1.090 0.9141 0.3713 0.84 0.34 0.29 0.2789 376.59
N3–H10 1.018 0.8192 0.3215 0.81 0.32 0.25 0.3333 315.18
N3–H11 1.024 0.8231 0.3109 0.80 0.30 0.24 0.3274 314.42
AMNFMC N7=N8 1.244 1.4789 0.2938 1.19 0.24 0.28 0.4313 141.55
N5–N11 1.449 0.9167 0.1381 0.63 0.10 0.06 0.3031 124.16
C1–O3 1.361 0.9845 0.2689 0.72 0.20 0.14 0.3014 326.62
C1–N5 1.414 0.9998 0.1857 0.71 0.13 0.09 0.2963 320.40
O3–C4 1.443 0.8441 0.1903 0.58 0.13 0.08 0.2393 264.75
C4–N7 1.454 0.9896 0.2696 0.68 0.19 0.13 0.2683 267.37
C4–H15 1.093 0.9075 0.3719 0.83 0.34 0.28 0.2853 358.88
N5–C6 1.444 0.9583 0.2199 0.66 0.15 0.10 0.2701 330.94
C6–F10 1.378 0.8574 0.2802 0.62 0.20 0.13 0.2465 418.79
NMDACB C1–N2 1.479 0.9456 0.2486 0.64 0.17 0.11 0.2590 196.21
C1–N3 1.479 0.9685 0.3084 0.66 0.21 0.14 0.2704 170.88
C1–H10 1.098 0.9004 0.3552 0.82 0.32 0.27 0.2774 367.45
N2–N6 1.389 0.9756 0.1878 0.70 0.14 0.10 0.3469 157.49
N3–C5 1.455 1.0026 0.3014 0.69 0.21 0.14 0.2704 294.30
C5–H14 1.106 0.9212 0.3525 0.83 0.32 0.27 0.2666 378.11
a
R is the bond length (in Å)
b
WBO is the Wiberg bond order, defined in Refs. [21, 22]
c
MIEN is the Mulliken interatomic electron number, defined in Ref. [13]
d
W = WBO/R (in Å−1), rounded to the second decimal place for optimal sensitivity
e
M = MIEN/R (in Å−1), rounded to the second decimal place for optimal sensitivity
f
K = W × M (in Å−2), rounded to the second decimal place for optimal sensitivity
g
ρc is the electron density at the bond critical point within Bader’s atoms-in-molecules analysis, defined
in Refs. [14, 15]
h
BDE (in kJ/mol) is the bond dissociation energy calculated according to Eq. (1)
Effective Bond-Strength Indicators 49

To sample various single bonds within diverse chemical environment, we have

chosen nine typical explosive molecules (see Fig. 2 for their structural diagrams)
with varied functional groups and geometric shapes, including cyclic 1-nitro-3-
methyl-1,3-diazacyclobutane (NMDACB), branched N-amino-N-methyl-nitramine
(AMNA), linear and branched 1-azide methyl-N-nitro-N-fluoro methyl carbamate
(AMNFMC), and aromatic TNT and its five derivatives: pentanitrotoluene (PNT),
1-methyl-3-hydroxy-6-amino-2,4,6-trinitrobenzene (TNCr), 1-methoxy-2,4,6-
trinitrobenzene (PAM), 2-chloro-3,5-dinitro-6-aminopyridine (CDNAPY), and
1,3,5-trimethyl-2,4,6-trinitrobenzene (TNM).

H20 H20 H 24
21
H H19 21
H H19 23 H H25
O 7 O O 7 O O 22 O

N N N N N N
O 8 6 14 O O 8 6 16 O O 7 1 19 O
5 5 6
1 1 2
4 2 2 3
H 4 H25 O 5 O
3 3 4
11 N 10 16 N
H 15
11 12
H 26 13
N H N H 27 O N O
O O O O O O

TNT TNM PNT

2,4,6-trinitrotoluene 1,3,5-trimethyl-2,4,6-trinitrobenzene pentanitrotoluene

H 22
H20 21 H
19 H H21 20 H
O O H17 17
7 O O8 O
N N N N Cl
7
O 12 1 8 O
16 H 10
5
6 N N
6 1 9
2 O 7
2
3 O
3 2 4
5 O 4 O
24 H 4 3
1 5
N 11 O9 N9 8
N 6

10
H 23 N H22 O H15 O 10
O O N
O O

TNCr CDNAPY PAM

1-methyl-3-hydroxy-6-amino- 2-chloro-3,5-dinitro-6-aminopyridine 1-methoxy-2,4,6-trinitrobenzene
2,4,6-trinitrobenzene

O F10
17
H
H11 H9 N H16
H8 6 O 6
4
N O 5
2
N 2 N
1 11
H10 3
N2 H7 1 O
N
3 1 3
N N O
4 7
N H9 N O
13 H 8 4
5
H10 N9
O O
H14 H15
H15

AMNA NMDACB AMNFMC

N-amino-N-methyl-nitramine 1-nitro-3-methyl-1,3-diazacyclobutane 1-azide methyl-N-nitro-N-fluoro
methyl carbamate

Fig. 2 Molecular structures of selected explosive compounds (some hydrogens omitted for
clarity)
50 G.-X. Wang et al.

Hereafter, the five key bond-strength indicators (i.e., R, MIEN, BO, ρc, and
BDE) of all single bonds of these molecules will be carefully compared to seek
better relationships among them. The end goal is to propose simple, sensitive
structural indicators suitable for identifying the trigger bond (or even better, all
weak bonds), thus saving both computational resources and time.

2 Computational Methods

Many studies [1–8, 16, 17], have already shown that the DFT-B3LYP method [18,
19] in combination with the 6-31G* basis set [20] is able to yield accurate ener-
getics, structures, and other molecular properties. In this paper, the same method
was employed to obtain the fully optimized molecular geometries and electronic
structures, including MIEN and Wiberg bond order (WBO) [21, 22], of the chosen
compounds (Fig. 2) within the Gaussian09 program package [23]. To obtain the
values of ρc, the Bader analysis was performed using AIMAll package [24].
Based on Hess’s law [25], conventional counterpoise correction methods [10,
11] were employed to calculate the BDEs according to Eq. (1). Unfortunately, any
existing counterpoise correction methods [10, 11] cannot address the situation when
a bond within a ring is broken. Instead of trying to contemplate suitable ghost atoms
and their linking strategies for ring opening scenarios, we simply supplemented
extra diffuse polarization basis functions to the existing basis sets for the atoms of
the broken bonds to roughly mimic the effects of ghost atoms. Taking the
four-membered ring of NMDACB for example, when breaking the C1–N2 and
C1–N3 bonds, the BSSE was calculated with additional aug-cc-pV5Z Diffuse (1s,
1p, 1d, 1f, 1g, 1h) basis functions [26] separately placed on the C1, N2, and N3
atoms. Numerical tests confirmed that the magnitude of the resulting correction to
the BDEs from this basis-set local enhancement protocol was in line with the
procedure of existing counterpoise correction methods [10, 11].

3 Results and Discussion

Four commonly available major bond-strength indicators (i.e., R, MIEN, WBO, and
ρc) are listed along with BDEs in Table 2. For the selected compounds, nearly all
the bonds with the longest R or the smallest WBO do not possess the lowest BDE.
The bond with the lowest BDE is the C–NO2 or N–NO2 bond, but, according to
R and WBO, the weakest bond can draw from all sorts of candidates, e.g., C–C, C–H,
C–NO2, C–Cl, N–H, or O–C bond. Similarly, the smallest values of ρc alone never
correspond to the weakest bonds either. This obviously illustrates that R, WBO, or ρc
cannot be used separately to identify the trigger bond.
Particularly, because of AMNFMC containing the azido group (−N7=N8+=N9−),
when the N7=N8 bond is broken to produce the N2 gas, its BDE is the second lowest
Effective Bond-Strength Indicators 51

among all bonds (merely 141.55 kJ/mol). It is comforting that none of the structural
parameters considered here indicates the N7=N8 bond to be the weakest. However,
this also suggests that whenever exotic bonds (not listed in Table 1) are involved in a
bond breaking process, their BDEs should be calculated to verify the weakest bond
for certainty.
Data in Table 2 also show that MIEN alone can correctly single out the type of
bonds to which the trigger bond belongs. For instance, for the selected compounds,
the easiest bond to break is the C–NO2 or N–NO2 bond, but among which MIEN
sometimes incorrectly picks up the weakest bond. For CDNAPY, the C4–N9 bond
has the smallest MIEN (0.1237) and its BDE is 278.67 kJ/mol, but it is the C2–N8
bond that has the lowest BDE (248.56 kJ/mol) even with a bigger MIEN (0.1335).
Therefore, much sensible structural parameters other than the existing four afore-
mentioned must be proposed for trigger bond identification.
In light of the general fact that WBO and MIEN are related positively to BDE
and inversely to R, W = WBO/R and M = MIEN/R were first conceived to be the
better alternatives. Such an idea comes with no surprise, because the numerator of
the 1/Ra term on the right-hand side of Eq. (2) can be interpreted as WBO = 1 for
single bonds. Then, to take local chemical environment into consideration, Eq. (2)
might suggest a general relationship: BDE/R ∝ WBO/R. (Hereafter, numerical
values of bond-strength indicators will be quoted without units unless otherwise
noted.)
Unfortunately, the data shown in Table 2 again proclaim that the minima of
W and BDE still do not align well. For example, in PAM, the O8–C17 bond has the
smallest W, but the C2–N7 or C4–N9 bond has the lowest BDE. In general, the
results of W closely resemble those of WBO.
Fortunately, the bonds with the smallest M enclose the bond with the lowest
BDE (see Table 2). For TNT, the C1–N14, C3–N11, and C5–N8 bonds all have the
same smallest M value (0.10) and both of the C1–N14 and C5–N8 bonds have the
lowest BDE. For PNT, the C3–N16, C4–N13, and C5–N10 bonds have the smallest
M (0.09) and, among them, the C4–N13 bond is the weakest (BDE = 197.65 kJ/
mol). We then considered the product of W and M, K = W × M, as a new indicator.
According to M, for TNCr, the weakest bonds are the C2–N8 and C6–N12 bonds
(M = 0.11), whereas according to K, the weakest bond of TNCr is solely the C2–N8
bond (K = 0.06) in agreement with its lowest BDE value. Overall, K outperforms
M only marginally (see Fig. 3).
Next, we observed an enlightening fact: ρc can almost always single out the
weakest bond among the weaker bonds of the same type within a molecule iden-
tified by M and K (see Table 2). The only exception might be PNT: among the three
most weak bonds (C3–N16, C4–N13, and C5–N10) sorted out by M and K, ρc chooses
the C5–N10 bond (BDE = 200.63 kJ/mol) to be the weakest whereas the C4–N13
bond has the lowest BDE (197.65 kJ/mol). However, this is not a huge failure
because the difference in the BDEs of the three weak bonds of PNT is less than
3 kJ/mol, well within the error margin of the computational methods employed.
Nonetheless, such promising results strongly advocate that M and K, as two new
trigger bond indicators (TBIs), shall be first used to identify the set of most weak
52 G.-X. Wang et al.

Fig. 3 Relative strengths of all single bonds within TNCr measured by bond-strength indicators:
M, K, and BDE

bonds and ρc can subsequently narrow down the candidates whose BDEs can then
be calculated to pinpoint the trigger bond among them. This can really reduce the
amount of work drastically.
In general, the smaller TBI values almost perfectly map to all weak bonds (with
BDE < 350 kJ/mol) sequentially, and both M and K do indeed capture the weakest
bond once these two TBIs reach minimum (see Table 2). In spite of such a general
success of these two TBIs, neither M nor K can always predict the relative order of
bond strength for strong bonds (with BDE > 350 kJ/mol) within a molecule.
Figure 4 showcases this point succinctly. Such a phenomenon is well expected
because the molecular fragments after breaking a strong bond normally undergo

Fig. 4 Comparison of TBIs (K and M) and BDE of all single bonds within TNCr
Effective Bond-Strength Indicators 53

very large structural rearrangement, which cannot be reasonably modeled by

structural indicators of the parent molecule alone.
We thus have successfully designed two new structural indicators, M and K, for
the identification of the trigger bond and weak bonds. If a compound has multiple
weak bonds of the same type, ρc can help determine the weakest bond. The uti-
lization of these two new structural indicators will become utterly efficient and
effective especially for very large systems with hundreds of bonds.

4 Conclusions

In conclusion, ordinary bond-strength indicators, including R, MIEN, WBO, ρc, and

W = WBO/R, are not suitable to be used alone to identify the trigger bond, whereas
M = MIEN/R and the better K = W × M are the indicators of choice for sorting out
weak bonds (with BDE < 350 kJ/mol). When there are several bonds carrying the
same smallest M or K, calculations of their ρc and BDEs must be subsequently
performed to determine the weakest bond. Especially, these two new indicators,
M and K, will become extremely helpful for analyzing bond breaking processes
within large molecular systems by greatly minimizing computational efforts.

Acknowledgements We are grateful to the grant support from the Natural Sciences and Engi-
neering Research Council (NSERC) of Canada, the National Natural Science Foundation of China
(No. 21403110), and the Natural Science Foundation of Jiangsu Province (No. BK20130755).
This work was mainly carried out at UBC during G.X.W.’s one-year visit to UBC from 16 August
2012 to18 August 2013.

References

1. Xu XJ, Xiao HM, Ju XH, Gong XD, Zhu WH (2006) Computational studies on
polynitrohexaazaadmantanes as potential high energy density materials (HEDMs). J Phys
Chem A 110:5929–5933
2. Qiu L, Xiao HM, Gong XD, Ju XH, Zhu WH (2006) Ab initio and molecular dynamics
studies of crystalline TNAD (trans-1,4,5,8-Tetranitro-1,4,5,8-tetraazadecalin). J Phys Chem A
110:3797–3807
3. Qiu LM, Gong XD, Wang GX, Zheng J, Xiao HM (2009) Looking for high energy density
compounds among 1,3-Bishomopentaprismane derivatives with –CN, –NC, and –ONO2
groups. J Phys Chem A 113:2607–2614
4. Xu XJ, Xiao HM, Gong XD, Ju XH, Chen ZX (2005) Theoretical studies on the vibrational
spectra, thermodynamic properties, detonation properties and pyrolysis mechanisms for
polynitroadamantanes. J Phys Chem A 109:11268–11274
5. Wang GX, Shi CH, Gong XD, Xiao HM (2009) Theoretical investigation on structures,
density, detonation properties and pyrolysis mechanism of the derivatives of HNS. J Phys
Chem A 113:1318–1326
54 G.-X. Wang et al.

6. Wang GX, Gong XD, Yan L, Du HC, Xu XJ, Xiao HM (2010) Theoretical studies on the
structures, density, detonation properties, pyrolysis mechanisms and impact sensitivity of
nitro derivatives of toluenes. J Hazard Mater 177:703–710
7. Wang GX, Gong XD, Du HC, Liu Y, Xiao HM (2011) Theoretical prediction of properties of
aliphatic polynitrates. J Phys Chem A 115:795–804
8. Liu Y, Gong XD, Wang LJ, Wang GX, Xiao HM (2011) Substituent effects on the properties
related to detonation performance and sensitivity for 2,2’,4,4’,6,6’-Hexanitroazobenzene
derivatives. J Phys Chem A 115:1754–1762
9. Jensen F (2007) Introduction to computational chemistry, 2nd edn. Wiley, Chichester, West
Sussex, England
10. Boys SF, Bernardi F (1970) The calculation of small molecular interactions by the differences
of separate total energies. Some procedures with reduced errors. Mol Phys 19:553–566
11. Simon S, Duran M, Dannenberg JJ (1996) How does basis set superposition error change the
potential surfaces for hydrogen bonded dimers? J. Chem. Phys. 105:11024–11031
12. Petrucci RH, Herring FG, Madura JD, Bissonette C (2011) General chemistry, 10th edn.
Pearson Publishing, Toronto, Ontario, Canada
13. Mulliken RS (1962) Criteria for the construction of good self-consistent-field molecular
orbital wave functions, and the significance of LCAO-MO population analysis. J. Chem.
Phys. 36:3428–3439
14. Bader RFW, Tang TH, Tal Y, Biegler-Koenig FW (1982) Properties of atoms and bonds in
hydrocarbon molecules. J Am Chem Soc 104:946–952
15. Bader RFW (1994) Atoms in molecules: a quantum theory. Clarendon Press, Oxford, UK
16. Xiao HM, Xu XJ, Qiu L (2008) Theoretical design of high energy density materials. Science
Press, Beijing, China
17. Wang GX, Gong XD, Liu Y, Du HC, Xu XJ, Xiao HM (2011) Looking for high energy
density compounds applicable for propellant among the derivatives of DPO with −N3,
−ONO2, and −NNO2 groups. J Comput Chem 32:943–952
18. Lee C, Yang W, Parr RG (1988) Development of the Colle-Salvetti correlation-energy
formula into a functional of the electron density. Phys. Rev. B 37:785–789
19. Becke AD (1992) Density-functional thermochemistry, II. The effect of the Perdew-Wang
generalized-gradient correlation correction. J. Chem. Phys. 97:9173–9177
20. Hariharan PC, Pople JA (1973) Influence of polarization functions on MO hydrogenation
energies. Theor. Chim. Acta 28:213–222
21. Wiberg KB (1968) Application of the Pople-Santry-Segal CNDO method to the cyclopropy-
lcarbinyl and cyclobutyl cation and to bicyclobutane. Tetrahedron 24:1083–1096
22. Mayer I (1983) Change bond order and valence in the ab initio SCF theory. Chem Phys Lett
97:270–274
23. Frisch MJ, Trucks GW, Schlegel HB, Scuseria GE, Robb MA, Cheeseman JR, Barone V,
Mennucci B, Cossi M, Scalmani G, Petersson GA, et al (2009) Gaussian 09, revision A. 02,
Gaussian, Inc., Wallingford, CT
24. Bader RFW Atoms in molecules, McMaster University, Hamilton, Ontario, Canada. (http://
www.chemistry.mcmaster.ca/bader/aim/ Accessed 3 Jan 2016)
25. Engel T, Reid P, Hehre W (2013) Physical chemistry, 3rd edn. Boston, USA, Pearson
26. Kendall RA, Dunning TH, Harrison RJ (1992) Electron affinities of the first-row atoms
revisited. Systematic basis sets and wave functions. J. Chem. Phys. 96:6796–6806
Advanced Relativistic Energy Approach
in Electron-Collisional Spectroscopy
of Multicharged Ions in Plasmas

Alexander V. Glushkov, Vasily V. Buyadzhi, Andrey A. Svinarenko

and Eugeny V. Ternovsky

Abstract We present the fundamentals of an advanced relativistic approach, based

on the Gell-Mann and Low formalism, to studying spectroscopic characteristics of
the multicharged ions in plasmas, in particular, computing the electron-ion collision
strengths, cross-sections etc. The approach is combined with relativistic many-body
perturbation theory with the Debye shielding model Hamiltonian for electron-nuclear
and electron-electron systems. The optimized one-electron representation in the
perturbation theory zeroth approximation is constructed by means of the correct
treating the gauge dependent multielectron contribution of the lowest perturbation
theory corrections to the radiation widths of atomic levels. The computation results
on the oscillator strengths and energy shifts due to the plasmas environment effect,
the electron-collision strengths, collisional excitation and de-excitation rates for a
number of the Be- and Ne-like ions of argon, nickel and krypton embedded to
different types of plasmas environment (with temperature 0.02–2 keV and electron
density 1016–1024 cm−3) are presented and analyzed.

Keywords Electron-collisional processes ⋅

Multicharged ions
Relativistic energy approach ⋅
Debye plasmas

1 Introduction

An accurate data about spectra, radiative decay widths and probabilities, oscillator
strengths, electron-collision strengths, collisional excitation and de-excitation rates
for atoms and especially ions are of a great interest for different applications,
namely, astrophysical analysis, laboratory, thermonuclear plasmas diagnostics,
fusion research, laser physics etc. [1–30]. It is also very important for studying
energy, spectral and radiative characteristics of a laser-produced hot and dense
plasmas [1, 2, 9, 10]. Above other important factors to studying electron-collisional

A. V. Glushkov (✉) ⋅ V. V. Buyadzhi ⋅ A. A. Svinarenko ⋅ E. V. Ternovsky

Odessa State Environmental University, L’vovskaya str., 15, Odessa 65016, Ukraine
e-mail: [email protected]

© Springer International Publishing AG, part of Springer Nature 2018 55

Y. A. Wang et al. (eds.), Concepts, Methods and Applications of Quantum Systems
in Chemistry and Physics, Progress in Theoretical Chemistry and Physics 31,
https://doi.org/10.1007/978-3-319-74582-4_4
56 A. V. Glushkov et al.

spectroscopy of ions one should mention the X-ray laser problem. It has stimulated
a great number of papers, devoted to modelling the elementary processes in laser,
collisionally pumped plasmas (see [3, 4] and Refs. therein) and construction of the
first VUV and X-ray lasers with using plasmas of Li-, Ne-like ions as an active
medium. Very useful data on the X-lasers problem are collected in the papers by
Ivanova et al. (see [3–6] and Refs. therein). From the other side, studying spectra of
ions in plasmas remains very actual in order to understand the plasmas processes
themselves. In most plasmas environments the properties are determined by the
electrons and the ions, and the interactions between them. The electron-ion colli-
sions play a major role in the energy balance of plasmas. For this reason, modelers
and diagnosticians require absolute cross sections for these processes. The cross
sections for electron-impact excitation of ions are needed to interpret spectroscopic
measurements and for simulations of plasmas using collisional-radiative models. At
present time a considerable interest has been encapsulated to studying elementary
atomic processes in plasmas environments (for example, see [1–30] and Refs.
therein) because of the plasmas screening effect on the plasmas-embedded atomic
systems. In many papers the calculations of various atomic and ionic systems
embedded in the Debye plasmas have been performed [11, 12, 16, 29, 30]. Cal-
culation of emission spectra of the plasmas ions based in the precise theoretical
techniques is practical tool, which may be used instead of very expensive sophis-
ticated experiments. Nevertheless, there are known principal theoretical problems to
be solved in order to receive the correct description of master parameters of the
elementary atomic processes in laser, collisionally pumped plasmas. First of all,
speech is about development of the advanced quantum-mechanical models for the
further accurate computing oscillator strengths, electron-collisional strengths and
rate coefficients for atomic ions in plasmas, including the Debye plasmas. As
usually, a correct accounting of the relativistic, exchange-correlation, a plasmas
environment effects is of a great importance. To say strictly, solving of the whole
problem requires a development of the quantum-electrodynamical approach as the
most consistent one to problem of the Coulomb many-body system.
In this chapter we present the fundamentals of an advanced relativistic energy
approach, based on the Gell-Mann and Low formalism, to studying spectroscopic
characteristics of the multicharged ions in the Debye plasmas, in particular, com-
puting the electron-ion collision strengths, cross-sections etc. The approach is
combined with relativistic many-body perturbation theory (PT) with the Debye
shielding model Hamiltonian for electron-nuclear and electron-electron systems.
The optimized one-electron representation in the PT zeroth approximation is con-
structed by means of the correct treating the gauge dependent multielectron con-
tribution of the lowest PT corrections to the radiation widths of atomic levels. It is
worth to remind that the method of the relativistic many-body PT formalism is
constructed on the base of the same ideas as the well-known PT approach with the
model potential zeroth approximation by Ivanov-Ivanova et al. [31–44]. However
there are a few fundamental differences. For example, in our case the PT zeroth
approximation [51, 54] is in fact the Dirac- Debye-Hückel one. In order to calculate
the radiative and collisional parameters an effective gauge-invariant version of
Advanced Relativistic Energy Approach in Electron-Collisional … 57

relativistic energy approach is used [37, 45–54]. It is important to remind that a

model relativistic energy approach in a case of a multielectron atom has been
developed by Ivanov-Ivanova et al. [33–36]. A generalized gauge-invariant version
of relativistic energy approach in a case of the multielectron atomic systems has
been developed by Glushkov-Ivanov-Ivanova (see Refs. [37–39]). Earlier we have
presented the fundamentals of an advanced generalized energy approach and its
application to many actual problems of modern atomic, nuclear and even molecular
optics and spectroscopy, including, spectroscopy of atoms in a photon vacuum
and an external electromagnetic (laser) field, optics of the cooperative
electron-gamma-nuclear “shake-up” processes (including processes of the NEET
and NEEC: “Nuclear Excitation—Electron Transition”, “Nuclear Excitation—
Electron Capture”), electron-muon-beta-gamma-nuclear spectroscopy, etc. (see
[55–88] and Refs. therein). Below we present and analyze the computation results
on the oscillator strengths and energy shifts due to the plasmas environment effect,
the electron-collision strengths, collisional excitation and de-excitation rates for a
number of the Be- and Ne-like ions of argon, nickel and krypton embedded to
different types of plasmas environment with the temperature 0.02–2 keV and the
electron density 1016–1024 cm−3.

2 Relativistic Many-Body Perturbation Theory

and Relativistic Energy Approach in Scattering Theory

2.1 Formalism of the Relativistic Perturbation Theory

with Dirac-Debye Shielding Model Zeroth
Approximation

Let us start our consideration from formulation relativistic many-body PT with the
Debye shielding model Dirac Hamiltonian for electron-nuclear and
electron-electron systems. Formally, a multielectron atomic systems (multielectron
atom or multicharged ion) is described by the relativistic Dirac Hamiltonian (the
atomic units are used) as follows:


H = ∑ hðri Þ + ∑ V ri rj . ð1Þ
i i>j

Here, h(r) is one-particle Dirac Hamiltonian for electron in a field of a nucleus

and V is potential of the inter-electron interaction.
According to Refs. [35–37] it is useful to determine the interelectron potential
with accounting for the retarding effect and magnetic interaction in the lowest order
on parameter α2 (α is the fine structure constant) as follows:
58 A. V. Glushkov et al.




 1 − αi αj
V ri rj = exp iωij rij ⋅ , ð2Þ
rij

where ωij is the transition frequency; αi,αj are the Dirac matrices.
In order to take into account the plasmas environment effects already in the PT
zeroth approximation we use the known Yukawa-type potential of the following
form:

Vðri , rj Þ = ðZa Zb ̸ jra − rb jÞexpð − μ ⋅ jra − rb jÞ ð3Þ

where ra, rb represent respectively the spatial coordinates of particles, say, A and B
and Za, Zb denote their charges.
The potential (3) is (look, for example, [23–28] and Refs therein) well known,
for example, in the classical Debye-Hückel theory of plasmas. The plasmas envi-
ronment effect is modelled by the shielding parameter μ, which describes a shape of
the long-rang potential. The parameter μ is connected with the plasmas parameters
such as temperature T and the charge density n as follows:
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
μ∼ e2 n ̸ kB T . ð4aÞ

Here e is the electron charge and kB is the Boltzman constant. The density n is
given as a sum of the electron density Ne and the ion density Nk of the k-th ion
species with the nuclear charge qk:

n = Ne + ∑ q2k Nk . ð4bÞ
k

It is very useful to remind the simple estimates for the shielding parameter. For
example, under typical laser plasmas conditions of T ∼ 1 keV and n ∼ 1023 cm−3
the parameter μ is of the order of 0.1 in atomic units. By introducing the
Yukawa-type electron-nuclear attraction and electron-electron repulsion potentials,
the Dirac-Debye shielding model Hamiltonian for electron-nuclear and
electron-electron subsystems is given in atomic units as follows [28]:


1 − αi αj
H = ∑ ½αcp − βmc − Z expð − μri Þ ̸ri  + ∑
2
expð − μrij Þ, ð5Þ
i i>j rij

where c is the velocity of light and Z is a charge of the atomic ion nucleus.
The formalism of the relativistic many-body PT is further constructed in the
same way as the PT formalism in Refs. [31–44]. In the PT zeroth approximation
one should use a mean-field potential, which includes the Yukawa-type potential
(insist of the pure Coulomb one) plus exchange Kohn-Sham potential and addi-
tionally the modified Lundqvist-Gunnarsson correlation potential (with the opti-
mization parameter b) as in Refs. [28–30, 49, 50]. As alternative one could use an
optimized model potential by Ivanova-Ivanov (for Ne-like ions) [31], which is
Advanced Relativistic Energy Approach in Electron-Collisional … 59

calibrated within the special ab initio procedure within the relativistic energy
approach [37, 38].
The most complicated problem of the relativistic PT computing the radiative and
collisional characteristics of the multielectron atomic systems is in an accurate,
precise accounting for the exchange-correlation effects (including polarization and
screening effects, a continuum pressure etc.) as the effects of the PT second and
higher orders. Using the standard Feynman diagram technique one should consider
two kinds of diagrams (the polarization and ladder ones), which describe the
polarization and screening exchange-correlation effects. The polarization diagrams
take into account the quasiparticle (external electrons or vacancies) interaction
through the polarizable core, and the ladder diagrams account for the immediate
quasiparticle interaction. The detailed description of the polarization diagrams and
the corresponding analytical expressions for matrix elements of the polarization
quasiparticles interaction (through the polarizable core) potential are presented in
Refs. [36, 45–50]. An effective approach to accounting for the polarization dia-
grams contributions is in adding the effective two- quasiparticle polarizable operator
into the PT first order matrix elements. In Ref. [36] the corresponding
non-relativistic polarization functional has been derived. More correct relativistic
expression has been presented in the Refs. [45, 46, 49].

2.2 Generalized Relativistic Energy Approach

in a Scattering Problem

In order to calculate different characteristics such as oscillator strengths and energy

shifts due to the plasmas environment effect, the electron-collision strengths, col-
lisional excitation and de-excitation rates etc. we use an advanced generalized
relativistic energy approach combined with the relativistic many-body PT [28, 29,
49, 50, 67–70]. Here we briefly present the key moments of the method.
In the theory of non-relativistic atom a convenient field procedure is known for
calculating the energy shifts ΔE of degenerate states. This procedure is connected
with the secular matrix M diagonalization [33, 37]. In constructing M, the
Gell-Mann and Low adiabatic formula for ΔE is used. The secular matrix elements
are already complex in the PT second order (the first order on the inter-electron
interaction). Their imaginary parts are connected with the radiation decay possi-
bility. It is important to note that the computing the energies and radiative transition
matrix elements is reduced to calculation and the further diagonalization of the
complex matrix M and determination of matrix of the coefficients with eigen state
vectors BIK ie, iv [33–36, 41, 42]. To calculate all necessary matrix elements one must
use the basis of the one-quasiparticle relativistic functions. In many calculations of
the atomic elementary processes characteristics it has been shown that their ade-
quate description requires using the optimized wave functions and an accurate
accounting for the exchange-correlation effects. In Ref. [37] it has been proposed
60 A. V. Glushkov et al.

“ab initio” optimization principle for construction of an effective one-quasiparticle

representation. The minimization of the gauge dependent multielectron contribution
of the lowest QED PT corrections to the radiation widths of atomic levels, which is
determined by the imaginary part of an energy shift ΔE, is used. In the fourth order
of QED PT there appear diagrams, whose contribution into the ImΔE accounts for
the polarization effects. This contribution describes collective effects and it is
dependent upon the electromagnetic potentials gauge (the gauge non-invariant
contribution ΔEninv). This value is considered to be the typical representative of the
electron correlation effects, whose minimization is a reasonable criteria in the
searching for the optimal one-electron basis of the PT. Let us note that this topic is
of a great importance (look, for example, Refs. [49, 50, 89–98], where there are
presented some alternative optimization approaches).
It is worth to remind that E. Davidson had pointed the principal disadvantages of
the traditional representation based on the self-consistent field approach and sug-
gested the optimal “natural orbitals” representation (for example, see Refs. [89,
90]). Our procedure derives an undoubted profit in the routine spectroscopic cal-
culations as it provides the way of the refinement of the atomic characteristics
calculations, based on the “first principles”. The resulting expression looks as the
correction due to the additional nonlocal interaction of the active quasiparticle with
the closed shells. Nevertheless, its calculation is reducible to the solving of the
system of the ordinary differential equations (one-D procedure) [49]. The most
important refinements can be introduced by accounting for the relativistic and the
density gradient corrections to the Tomas- Fermi formula (look details in Refs. [49,
50]). The minimization of the functional Im ΔEninv leads to the integral differential
equation, that is numerically solved. In result one can get the optimal one-electron
representation of the PT, which is further improved within the Dirac-Kohn-Sham-
Sturm approach in order to take into account for the continuum states [49, 50, 60,
61, 96–98].
As some ideas of the energy approach in application to a scattering problem have
been presented in a literature (look, for example, [37, 41, 42], below we concern the
most principal points. Further for definiteness, let us consider a collisional
de-excitation of, say, the Ne-like ion [41]:

ðð2jiv Þ − 1 3jie ½Ji Mi , εin Þ → ðΦo , εsc Þ.

Here Φo is the state of the ion with the closed shells (ground state of the Ne-like
ion); Ji is the total angular moment of the initial target state; indices iv, i.e. are
related to the initial states of a vacancy and an electron; indices εin and εsc are the
incident and scattered energies, respectively to the incident and scattered electrons.
The initial state of the system “atom plus free electron” can be written as

jI > = ain+ ∑ aie+ aiv Φo CmJi ie, M, mi iv ð6aÞ

miv , mie

where CmJi ie, M, mi iv is the Clebsh-Gordan coefficient.

Advanced Relativistic Energy Approach in Electron-Collisional … 61

The final state is as follows:

jF > = asc+ Φo , ð6bÞ

where Φo is the state of an ion with closed electron shells (ground state of Ne-like
ion), |I> represents three-quasiparticle (3QP) state, and |F> represents the
one-quasiparticle (1QP) state.
The justification of the energy approach in the scattering problem is in details
described in Refs. [38, 41–44, 49]. The scattered part of energy shift Im ΔE appears
firstly in the atomic PT second order (the fourth order of the QED PT) in the form
of integral over the scattered electron energy εsc:
Z
dεsc Gðεiv , εie , εin , εsc Þ ̸ðεsc − εiv − εie − εin − i0Þ ð7Þ

ImΔE = π Gðεiv , εie , εin , εsc Þ. ð8aÞ

Here G is a definite squired combination of the two-electron matrix elements (2).

As usually, the value

σ = − 2 ImΔE ð8bÞ

represents the collisional cross-section if the incident electron eigen-function is

normalized by the unit flow condition and the scattered electron eigen-function is
normalized by the energy δ function.
The collisional de-excitation cross section can be further defined as follows:
2
σðIK → 0Þ = 2π ∑ ð2jsc + 1Þf ∑ < 0jjin , jsc jjie , jiv , Ji > BIK
ie, iv g ð9Þ
jin , jsc jie, jiv

The amplitude like combination in (9) has the following form:

pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
< 0jjin , jsc jjie , jiv , Ji > = ð2jie + 1Þð2jiv + 1Þð − 1Þjie + 1 ̸2 × ∑ ð − 1Þλ + Ji ×
λ
" #
jin . . . jsc . . . Ji
× fδλ, Ji ̸ð2Ji + 1ÞQλ ðsc, ie; iv, inÞ + Qλ ðie; in; iv, scÞg
jie . . . jiv . . . ..λ
ð10aÞ
− Yuk
Qλ = QCoul
λ + QBr
λ , ð10bÞ

− Yuk
where QCoul
λ + QBr
λ is the sum of the Coulomb-Yukawa and Breit matrix ele-
− Yuk
ments. The Coulomb part QCoul
λ contains the radial Rλ and angular Sλ integrals
as follows:
62 A. V. Glushkov et al.

− Yuk
QCoul
λ = fRλ ð1243ÞSλ ð1243Þ + Rλ ð1̃243̃ÞSλ ð1̃243̃Þ
ð11Þ
+ Rλ ð12̃4̃3ÞSλ ð12̃4̃3Þ + Rλ ð1̃2̃4̃3ÞS
̃ λ ð1̃2̃4̃3Þg.
̃

Here the tilde designates that the large radial Dirac component f must be
replaced by the small Dirac component g, and instead of li , lĩ = li − 1 should be
taken for ji < li and lĩ = li + 1 for ji > li .
The Breit part can be expressed as follows [39]:

QBr Br Br Br
λ = Qλ, λ − 1 + Qλ, λ + Qλ, λ + 1 ð12Þ

λ = Rλ ð124̃3ÞS
QBr ̃ lλ ð124̃3Þ̃ + Rλ ð1̃2̃43ÞSlλ ð1243Þ +
 ð13Þ
+ Rl ð1̃24̃3ÞSlλ ð1̃24̃3Þ + Rl ð12̃43̃ÞSlλ ð12̃43̃Þ .

The details of their computing can be found in Refs. [36–54]. The modified PC
code ‘Superatom-ISAN” (version-93) has been used in all calculations.

3 Results and Conclusions

Here we present the results of computing the radiative and collisional characteristics
(energy shifts, oscillator strengths, electron-ion cross-sections and collision
strengths) for the Be-, Ne-like ions of Ar, Ni and Kr (Z = 18–36) embedded to the
plasmas environment. Let us remind (see Refs. [11, 12, 16, 28, 39]) that the Be- and
Ne-like ions play an important role in the diagnostics of a wide variety of labo-
ratory, astrophysical, thermonuclear plasmas. Firstly, we list our results on energy
shifts and oscillator strengths for transitions 2s2-2s1/22p1/2,3/2 in spectra of the
Be-like Ni and Kr. The plasmas parameters are as follows: ne = 1022−1024cm−3,
T = 0.5–2 keV (i.e. μ ∼ 0.01–0.3). In Tables 1 and 2 we list the results of

Table 1 Energy shifts ΔE (cm−1) for the 2 s2-[2s1/22p3/2]1 transition in spectra of the Be-like Ni
and Kr ions for different values of the ne (cm−3) and T (in eV) (see explanations in text)
ne 1022 1023 1024 1022 1023 1024
Z kT Li et al. Li et al. Li et al. Our data Our data Our data
NiXXV 500 31.3 292.8 2639.6 33.8 300.4 2655.4
1000 23.4 221.6 2030.6 25.7 229.1 2046.1
2000 18.0 172.0 1597.1 20.1 179.8 1612.5
I-S 8.3 86.6 870.9
KrXXXIII 500 21.3 197.9 2191.9 27.2 215.4 2236.4
1000 15.5 150.5 1659.6 21.3 169.1 1705.1
2000 11.5 113.5 1268.0 16.9 128.3 1303.8
Advanced Relativistic Energy Approach in Electron-Collisional … 63

Table 2 Oscillator strengths gf for the 2 s2-[2s1/22p3/2]1 transition in spectra of the Be-like ion of
Ni for different values of the ne (cm−3) and T (in eV) (gf0–the gf value for free ion)
ne 1022 1023 1024 1022 1023 1024
kT gf0 gf gf gf gfo:our gf: our gf: our gf: our
Li et al. Li et al. Li et al. Li et al. data data data data
500 0.1477 0.1477 0.1478 0.1487 0.1480 0.1480 0.1483 0.1495
1000 0.1477 0.1477 0.1482 0.1480 0.1483 0.1495
2000 0.1477 0.1477 0.1481 0.1479 0.1482 0.1493
I-S 0.1477 0.1477 0.1479

calculation of the energy shifts ΔE (cm−1) for 2s2-[2s1/22p1/2,3/3]1 transitions and

oscillator strengths changes for different plasmas parameters such as the electron
density ne and temperature T.
There are also presented the available theoretical data by Li et al. and
Saha-Frische: the multiconfiguration Dirac-Fock (DF) computation results and
ionic sphere (I-S) model simulation data (from [11, 12, 16] and Refs. therein). The
analysis shows that the presented data are in physically reasonable agreement,
however, some difference can be explained by using different relativistic orbital
basis and different models for accounting of the plasmas screening effect. From the
physical point of view, the behavior of the energy shift is naturally explained, i.e.
by increasing blue shift of the line because of the increasing the plasmas screening
effect.
Further we present the results of computing the electron-collisional
cross-sections and electron-collision strengths for Ne-like ion of Ar (the part of
results has been presented in Refs. [28], but without the plasmas screening effect)
and compare with the known theoretical data: relativistic model potential PT
(RMPPT), relativistic optimized DF PT (ODFPT) [28–30, 41, 42].
In Table 3 we list the electron-collision strengths for Ne-like argon excitation
from the ground state (E = 0.75 keV is the impact electron energy). The corre-
sponding plasmas parameters (θ-pinch plasmas) are as follows: ne = 1016 cm−3,
and Te = 65 eV.
It should be noted that the experimental information about the
electron-collisional cross-sections for high-charged Ne-like ions is very scarce and
is extracted from indirect observations. Such experimental information for a few
collisional excitations of the Ne-like barium ground state has been presented in
Refs. [17–19].
Let us note that the PT first order correction is calculated exactly, the high-order
contributions are taken into account for effectively: polarization interaction of two
above-core quasi-particles and an effect of their mutual screening (correlation
effects). It is interesting to note that here the plasmas effects do not play a critical
quantitative role.
Further we present the results of studying collisional characteristics for the
Ne-like ions in the collisionally pumping plasmas with the parameters Te = 20–
40 eV and density ne = 1019−20 cm−3. This system represents a great interest for
64 A. V. Glushkov et al.

Table 3 The electron-collision strengths for Ne-like Ar excitation from the ground state for
impact electron energy 0.75 keV (numbers in brackets denote the multiplicative powers of ten)
Transition Level J [41] [29] Present data
01–02 2s 2p 0 1,303[−03] 1,415[−03] 1,498[−03]
3 2p3/23s1/2 1 9,017[−03] 9,224[−03] 9,286[−03]
4 2p1/23s1/2 0 2,587[−04] 2,724[−04] 2,783[−04]
5 2p1/23s1/2 1 2,241[−02] 2,342[−02] 2,394[−02]
6 2p3/23p3/2 1 3,456[−03] 3,635[−03] 3,699[−03]
7 2p3/23p3/2 3 2,911[−03] 2,998[−03] 3,065[−03]
8 2p3/23p1/2 2 4,795[−03] 4,922[−03] 4,988[−03]
9 2p3/23p1/2 1 1,033[−03] 1,213[−03] 1,254[−03]
10 2p3/23p3/2 2 6,451[−03] 6,535[−03] 6,597[−03]
11 2p1/23p1/2 1 9,641[−04] 9,993[−04] 1,088[−03]
12 2p1/23p1/2 0 8,794[−04] 8,927[−04] 8,992[−04]
13 2p1/23p3/2 2 7,814[−03] 7,978[−03] 8,113[−03]
14 2p1/23p3/2 1 8,561[−04] 8,723[−04] 9,005[−04]
15 2p3/23p3/2 0 8,670[−02] 8,735[−02] 8,802[−02]
16 2p3/23d3/2 0 1,136[−03] 1,244[−03] 1,296[−03]
17 2p3/23d3/2 1 4,129[−03] 4,327[−03] 4,389[−03]
18 2p3/23d5/2 2 5,227[−03] 5,546[−03] 5,601[−03]
19 2p3/23d5/2 4 3,512[−03] 3,678[−03] 3,714[−03]
20 2p3/23d3/2 3 3,994[−03] 4,133[−03] 4,185[−03]

generation of laser radiation in the short-wave spectral range [3, 4]. Besides, it is
obviously more complicated case in comparison with previous one. Here an
accurate account of the excited, Rydberg, autoionization and continuum states can
play a critical role.
In Table 4 we present the theoretical values of the collisional excitation rates
(CER) and collisional de-excitation rates (CDR) for Ne-like argon transition
between the Rydberg states and from the Rydberg states to the continuum states with

Table 4 The collisional excitation (CER) and de-excitation (CDR) rates (in cm3/s) for Ne-like
argon in plasmas with parameters: ne = 1019−20 cm−3 and electron temperature Te = 20 eV
Parameters ne, cm−3 RMPPT RMPPT RMPPT Present Present Present
results results results
Transition 1 → 2 1 → 3 2 → 3 1 → 2 1 → 3 2 → 3
CDR 1.0 + 19 5.35 − 10 1.64–10 1.13 − 09 5.77 − 10 1.92 − 10 1.28 − 09
(i → i; k)
1.0 + 20 5.51 − 10 1.60 − 10 1.12 − 09 5.94 − 10 1.78 − 10 1.25 − 09
Transition 2 → 1 3 → 1 3 → 2 2 → 1 3 → 1 3 → 2
CER 1.0 + 19 5.43 − 10 5.39 − 12 2.26 − 11 5.79 − 10 1.88 − 12 2.64 − 11
(i → i; k)
1.0 + 20 3.70 − 10 8.32 − 12 2.30 − 11 4.85 − 10 1.13 − 11 2.78 − 11
Advanced Relativistic Energy Approach in Electron-Collisional … 65

Fig. 1 The Rydberg states

zones (Ne-like ion: [Ne,i], nl);
ε0 is the boundary of the
thermalized zone,
neighboring to continuum; ε3
is the ionization potential for
states nl = 3s; εi =
(ε0 +εi+1)/2, i = 1, 2

parameters: ne = 1019−20 cm−3 and electron temperature Te = 20 eV (see details in

Refs. [28, 42]). For comparison there are also listed the data by Ivanov et al.,
obtained within the RMPPT approach (without the shielding effect) [28, 41, 42].
Here we talk about the Rydberg states which converge to the corresponding
lower boundary of continuum −ε0 (see Fig. 1).
As it is indicated in Ref. [42], the parameter −ε0 is the third parameter of the
plasmas environment (together with electron density and temperature). In fact it
defines the thermalized energy zone of the Rydberg and autoionization states which
converge to the ionization threshold for each ion in a plasmas. Usually value ε0 can
be barely estimated from simple relation: ε0 = 0.1 ⋅ Te.
In the consistent theory the final results must not be dependent on the model
parameters, so the concrete value of ε0 is usually chosen in such way that an effect
of its variation in the limits [0.01 ⋅ Te, 0.1 ⋅ Te] (for Ne-like ions) does not influence
on the final results.
In Table 5 we present the theoretical values of the collisional excitation
(CER) and de-excitation (CDR) rates (in cm3/s) for Ne-like argon in plasmas with
the parameters: ne = 1019−20 cm−3 and electron temperature Te = 40 eV. Analysis
of the presented data allows to conclude that the shielding effects play a definite role
for the Debye plasmas. From other side, an account for the highly-lying excited
states is quantitatively important for the adequate description of the collision
cross-sections.

Table 5 The collisional excitation (CER) and de-excitation (CDR) rates (in cm3/s) for Ne-like
argon in plasmas with parameters: ne = 1019−20 cm−3 and electron temperature Te = 40 eV (our
data)
Parameters ne, cm−3 Present results Present results Present results
Transition 1 → 2 1 → 3 2 → 3
CDR (i → i;k) 1.0 + 19 3.18 − 10 8.45 − 11 6.81 − 10
1.0 + 20 5.02 − 10 1.56 − 10 4.99 − 10
Transition 2 → 1 3 → 1 3 → 2
CER (i → i;k) 1.0 + 19 5.33 − 10 5.63 − 10 7.11 − 11
1.0 + 20 7.67 − 10 6.94 − 11 8.93 − 11
66 A. V. Glushkov et al.

The calculations encourage us to believe that using energy approach combined

with the relativistic many-body PT with the optimal one-electron basis is quite
consistent and effective tool from the point of view of the theory correctness and
results exactness. This fact was surely confirmed by other calculations of the
oscillator strengths, radiative widths, hyperfine structure constants for atoms and
multicharged ions (see Refs. [28–30, 49–54]).
To conclude, we have presented an effective quantum approach in
electron-collisional spectroscopy of the multicharged ions in plasmas to compute
the cross sections and other characteristics of the elementary collisional processes.
It is based on the generalized relativistic energy approach and relativistic optimized
many-body PT with the Debye shielding model Hamiltonian for electron-nuclear
and electron-electron systems. The optimized one-electron representation in the PT
zeroth approximation is constructed by means of the correct treating the gauge
dependent multielectron contribution of the lowest PT corrections to the radiation
widths of atomic levels. It is important to note that an approach is universal and,
generally speaking, can be applied to quantum systems of other nature (see, for
example, [57–66] and Refs. therein). Its application is especially perspective when
the experimental information about corresponding properties and systems is very
scarce. We have presented the illustrative results of studying spectra of some
multicharged ions (Be-and Ne-like ions) in plasmas and computing the electron-ion
collision strengths, cross-sections etc. The obtained data can be used in different
applications, namely, astrophysical analysis, laboratory, thermonuclear plasmas
diagnostics, fusion research, laser physics, quantum electronics etc.

Acknowledgements The authors are very much thankful to Prof. Jean Maruani and Prof. Alex
Wang for invitation to make contributions on the QSCP-XXI workshop (Vancouver, Canada). The
useful comments of the anonymous referees are very much acknowledged too.

References

1. Oks E (2010) In: Oks E, Dalimier E, Stamm R, Stehle C, Gonzalez MA (eds) Spectral line
shapes in plasmas and gases. Int J Spectr 1:852581
2. Griem HR (1974) Spectral line broadening by plasmas. Academic Press, New York
3. Ivanova EP (2011) Phys Rev A 84:043829
4. Ivanova EP, Grant IP (1998) J Phys B Mol Opt Phys 31:2871; Ivanova EP, Zinoviev NA
(2001) Phys Lett A 274:239
5. Khetselius OYu (2011) Quantum structure of electroweak interaction in heavy finite
fermi-systems. Astroprint, Odessa
6. Glushkov AV (1991) Opt Spectrosc 70:555
7. Malinovskaya SV, Glushkov AV, Khetselius OYu, Svinarenko AA, Mischenko EV,
Florko TA (2009) Int J Quant Chem 109(14):3325
8. Glushkov AV, Loboda AV, Gurnitskaya EP, Svinarenko AA (2009) Phys Scripta
T135:014022
9. Glushkov AV, Khetselius OYu, Svinarenko AA (2013) Phys Scripta T153:014029
10. Svinarenko AA (2014) J Phys Conf Ser 548:012039
Advanced Relativistic Energy Approach in Electron-Collisional … 67

11. Paul S, Ho YK (2010) J Phys B At Mol Opt Phys 43:065701; Kar S, Ho YK (2011) J Phys B
At Mol Opt Phys 44:015001
12. Glushkov AV (2005) Atom in electromagnetic field. KNT, Kiev
13. Yongqiang L, Jianhua W, Yong H, Jianmin Y (2008) J Phys B At Mol Opt Phys 41:145002
14. Glushkov AV, Mansarliysky VF, Khetselius OYu, Ignatenko AV, Smirnov AV, Prepelitsa GP
(2017) J Phys Conf Ser 810:012034
15. Glushkov AV, Loboda AV (2007) J Appl Spectr 74:305, Springer
16. Saha B, Fritzsche S (2007) J Phys B At Mol Opt Phys 40:259
17. Marrs R, Levine M, Knapp D, Henderson J (1988) Phys Rev Lett 60:1715
18. Zhang H, Sampson D, Clark R, Mann J (1988) Atom Dat Nuc Dat Tabl 37:17
19. Reed KJ (1988) Phys Rev A 37:1791; Talukder MR (2008) Appl Phys Lasers Opt 93:576
20. Smith ACH, Bannister ME, Chung YS, Djuric N, Dunn GH, Wallbank B, Woitke O (1999)
Phys Scripta T80:283
21. Buyadzhi VV (2015) Photoelectronics 24:128
22. Zeng S, Liu L, Wang JG, Janev RK (2008) J Phys B At Mol Opt Phys 41:135202
23. Okutsu H, Sako T, Yamanouchi K, Diercksen GHF (2005) J Phys B At Mol Opt Phys 38:917
24. Nakamura N, Kavanagh AP, Watanabe H, Sakaue HA, Li Y, Kato D, Curell FJ, Ohtani S
(2007) J Phys Conf Ser 88:012066
25. Bannister ME, Djuric N, Woitke O, Dunn GH, Chung Y-S, Smith ACH, Wallbank B,
Berrington KA (1999) Int J Mass Spectr 192:39
26. Badnell NR (2007) J Phys Conf Ser 88:012070; Griffin DC, Balance CP, Mitnik DM,
Berengut JC (2008) J Phys B At Mol Opt Phys 41:215201
27. Glushkov AV, Malinovskaya SV, Prepelitsa G, Ignatenko V (2005) J Phys Conf Ser 11:199
28. Malinovskaya SV, Glushkov AV, Khetselius OYu, Loboda AV, Lopatkin YuM,
Svinarenko AA, Nikola LV, Perelygina TB (2011) Int J Quant Chem 111:288
29. Glushkov AV, Khetselius OYu, Loboda AV, Ignatenko AV, Svinarenko AA,
Korchevsky DA, Lovett L (2008) Spectr Line Shapes. AIP Conf Pro 1058:175
30. Buyadzhi VV, Chernyakova YuG, Antoshkina OA, Tkach TB (2017) Photoelectronics 26:94
31. Ivanov LN, Ivanova EP (1979) Atom Dat Nucl Dat Tabl 24:95
32. Driker MN, Ivanova EP, Ivanov LN, Shestakov AF (1982) J Quant Spectr Rad Transfer
28:531
33. Ivanov LN, Letokhov VS (1985) Com Mod Phys D 4:169
34. Vidolova-Angelova E, Ivanov LN, Ivanova EP, Angelov DA (1986) J Phys B At Mol Opt
Phys 19:2053
35. Ivanov LN, Ivanova EP, Aglitsky EV (1988) Phys Rep 166:315
36. Ivanova EP, Ivanov LN, Glushkov AV, Kramida AE (1985) Phys Scripta 32:513
37. Glushkov AV, Ivanov LN (1992) Phys Lett A 170:33
38. Glushkov AV, Ivanov LN, Ivanova EP (1986) Autoionization phenomena in atoms. Moscow
University Press, Moscow, pp 58–160
39. Ivanova EP, Glushkov AV (1986) J Quant Spectr Rad Transfer 36:127
40. Ivanova EP, Gulov AV (1991) Atom Dat Nuc Dat Tabl 49:1
41. Ivanov LN, Ivanova EP, Knight L (1993) Phys Rev A 48:4365
42. Ivanov LN, Ivanova EP, Knight L, Molchanov AG (1996) Phys Scripta 53:653
43. Glushkov AV, Ivanov LN (1992) Preprint of ISAN. AS N-1 Moscow-Troitsk
44. Glushkov AV, Ivanov LN (1993) J Phys B At Mol Opt Phys 26:L379
45. Glushkov AV (1990) Sov Phys J 33(1):1
46. Glushkov AV (1990) J Str Chem 31(4):529
47. Glushkov AV (1992) JETP Lett 55:97
48. Glushkov A V (2012) Quantum systems in chemistry and physics. In: Nishikawa K,
Maruani J, Brändas E, Delgado-Barrio G, Piecuch P (eds) Progress in theoretical chemistry
and physics, vol 26. Springer, Dordrecht, pp 231–252
49. Glushkov AV (2008) Relativistic quantum theory. Quantum mechanics of atomic systems.
Astroprint, Odessa, p 700
50. Khestelius OYu (2008) Hyperfine structure of atomic spectra. Astroprint, Odessa, p 210
68 A. V. Glushkov et al.

51. Glushkov AV (2013) Advances in quantum methods and applications in chemistry, physics
and biology. In: Hotokka M, Maruani J, Brändas J, Delgado-Barrio G (eds) Progress in
theoretical chemistry and physics, vol 27. Springer, Cham, pp 161–177
52. Glushkov AV, Khetselius OYu, Svinarenko AA, Prepelitsa GP (2010) In: Duarte FJ
(ed) Coherence and ultrashort pulsed emission, InTech, Rijeka, pp 159–186
53. Svinarenko AA, Glushkov AV, Khetselius OYu, Ternovsky VB, Dubrovskaya YuV,
Kuznetsova AA, Buyadzhi VV (2017) In: Orjuela JEA (ed) Rare earth element, InTech,
pp 83–104
54. Glushkov AV, Khetselius OYu, Svinarenko AA, Buyadzhi VV, Ternovsky VB,
Kuznetsova AA, Bashkarev PG (2017) In: Uzunov DI (ed) Recent studies in perturbation
theory, InTech, pp 131–150
55. Buyadzhi VV, Glushkov AV, Lovett L (2014) Photoelectronics 23:38
56. Buyadzhi VV, Glushkov AV, Mansarliysky VF, Ignatenko AV, Svinarenko AA (2015)
Sensor Electr Microsyst Techn 12(4):27
57. Glushkov AV, Khetselius OYu, Malinovskaya SV (2008) Frontiers in quantum systems in
chemistry and physics. In: Wilson S, Grout PJ, Maruani J, Delgado-Barrio G, Piecuch P
(eds) Progress in theoretical chemistry and physics, vol 18. Springer, Dordrecht, pp 525–541
58. Glushkov AV, Khetselius OYu, Malinovskaya SV (2008) Europ Phys J ST 160:195
59. Glushkov AV, Khetselius OYu, Malinovskaya SV (2008) Mol Phys 106:1257
60. Glushkov AV, Khetselius OY, Svinarenko AA (2012) Advances in the theory of quantum
systems in chemistry and physics. In: Hoggan P, Maruani J, Brändas E, Delgado-Barrio G,
Piecuch P (eds) Progress in theoretical chemistry and physics, vol 22. Springer, Dordrecht,
pp 51–68
61. Glushkov AV, Khetselius OYu, Lovett L (2010) Advances in the theory of atomic and
molecular systems. In: Piecuch P, Maruani J, Delgado-Barrio G,Wilson S (eds) Progress in
theoretical chemistry and physics, vol 20. Springer, Dordrecht, pp 125–152
62. Glushkov AV, Khetselius YO, Loboda AV, Svinarenko AA (2008) Frontiers in quantum
systems in chemistry and physics. In: Wilson S, Grout PJ, Maruani J, Delgado-Barrio G,
Piecuch P (eds) Progress in theoretical chemistry and physics, vol 18. Springer, Dordrecht, pp
543–560
63. Glushkov AV, Svinarenko AA, Khetselius OYu, Buyadzhi VV, Florko TA, Shakhman AN
(2015) Frontiers in quantum methods and applications in chemistry and physics. In:
Nascimento M, Maruani J, Brändas E, Delgado-Barrio G (eds) Progress in theoretical
chemistry and physics, vol 29. Springer, Cham, pp 197–217
64. Khetselius OYu, Zaichko PA, Smirnov AV, Buyadzhi VV, Ternovsky VB, Florko TA,
Mansarliysky VF (2017) Quantum systems in physics, chemistry, and biology. In: Tadjer A,
Pavlov R, Maruani J, Brändas E, Delgado-Barrio G (eds) Progress in theoretical chemistry
and physics, Springer, Cham, pp 271–281
65. Glushkov AV, Buyadzhi VV, Kvasikova AS, Ignatenko AV, Kuznetsova AA, Prepelitsa GP,
Ternovsky V B (2017) Quantum systems in physics, chemistry, and biology. In: Tadjer A,
Pavlov R, Maruani J, Brändas E, Delgado-Barrio G (eds) Progress in theoretical chemistry
and physics, vol 30. Springer, Cham, pp 169–180
66. Glushkov AV, Rusov VD, Ambrosov SV, Loboda AV (2003) In: Fazio G, Hanappe F
(eds) New projects and new lines of research in nuclear physics. World Scientific, Singapore,
pp 126–132
67. Glushkov O. Khetselius E Gurnitskaya, Loboda A, Florko VD, Sukharev Lovett L (2008)
Frontiers in quantum systems in chemistry and physics. In: Wilson S, Grout PJ, Maruani J,
Delgado-Barrio G, Piecuch P (eds) Progress in theoretical chemistry and physics, vol 18.
Springer, Dordrecht, pp 507–524
68. Glushkov AV, Ambrosov SV, Loboda AV, Gurnitskaya EP, Khetselius OY (2006) Recent
advances in the theory of chemical and physical systems. In: Julien P, Maruani J, Mayou D,
Wilson S, Delgado-Barrio G (eds). Progress in theoretical chemistry and physics, vol 15.
Springer, Dordrecht, pp 285–299
Advanced Relativistic Energy Approach in Electron-Collisional … 69

69. Glushkov AV, Ambrosov SV, Loboda AV, Chernyakova YuG, Svinarenko AA, Khetselius
OYu (2004) Nucl Phys A Nucl Hadr Phys 734:21
70. Glushkov AV, Malinovskaya SV, Loboda AV, Shpinareva IM, Prepelitsa GP (2006) J Phys
Conf Ser 35:420
71. Glushkov AV, Ambrosov SV, Loboda AV, Gurnitskaya EP, Prepelitsa GP (2005) Int J Quant
Chem 104:562
72. Glushkov A, Malinovskaya S, Loboda A, Shpinareva I, Gurnitskaya E, Korchevsky D (2005)
J Phys Conf Ser 11:188
73. Glushkov AV, Malinovskaya SV, Chernyakova YG, Svinarenko AA (2004) Int J Quant
Chem 99:889
74. Glushkov AV, Ambrosov SV, Ignatenko AV, Korchevsky DA (2004) Int J Quant Chem
99:936
75. Glushkov AV, Malinovskaya SV, Sukharev DE, Khetselius OYu, Loboda AV, Lovett L
(2009) Int J Quant Chem 109:1717
76. Khetselius OYu (2009) Int J Quant Chem 109:3330
77. Khetselius OYu (2009) Phys Scripta T135:014023
78. Khetselius OYu (2012) Quantum systems in chemistry and physics. In: Nishikawa K,
Maruani J, Brändas E, Delgado-Barrio E, Piecuch P (eds) Progress in theoretical chemistry
and physics, vol 26. Springer, Dordrecht, pp 217–229
79. Khetselius YO (2015) Frontiers in quantum methods and applications in chemistry and
physics. In: Nascimento M, Maruani J, Brändas E, Delgado-Barrio E, Piecuch P (eds) Pro-
gress in theoretical chemistry and physics, vol 29. Springer, Cham, pp 55–76
80. Khetselius OYu (2008) Hyperfine structure of atomic spectra. Astroprint, Odessa
81. Khetselius OYu (2005) Hyperfine structure of radium. Photoelectronics. 14:83
82. Khetselius OYu (2012) J Phys Conf Ser 397:012012
83. Khetselius OYu, Florko TA, Svinarenko AA, Tkach TB (2013) Phys Scripta T153:014037
84. Khetselius OYu (2008) Spectral line shape. AIP Conf Proc 1058:363
85. Khetselius OYu, Glushkov AV, Gurnitskaya EP, Loboda AV, Mischenko EV, Florko TA,
Sukharev DE (2008) AIP Conf Proc 1058:231
86. Khetselius OYu (2007) Photoelectronics 16:129
87. Khetselius OYu, Gurnitskaya EP (2006) Sensor Electr Microsyst Techn N3, 35–39
88. Khetselius OYu, Gurnitskaya EP (2006) Sensor Electr and Microsyst Techn Issue 2:25−29
89. Feller D, Davidson ER (1981) J Chem Phys 74:3977
90. Froelich P, Davidson ER, Brändas E (1983) Phys Rev A 28:2641
91. Rittby M, Elander N, Brändas E (1983) Int J Quant Chem 23:865
92. Yan A, Wang C, Yung Y, Ya KC, Chen GH (2011) J Chem Phys 134:241103
93. Maruani J (2016) J Chin Chem Soc 63:33
94. Pavlov R, Mihailov L, Velchev Ch, Dimitrova-Ivanovich M, Stoyanov Zh, Chamel N,
Maruani J (2010) J Phys Conf Ser 253:012075
95. Dietz K, Heβ BA (1989) Phys Scripta 39:682
96. Kohn W, Sham LJ (1964) Phys Rev A 140:1133; Hohenberg P, Kohn W (1964) Phys Rev B
136:864
97. Gidopoulos N, Wilson S (2004) The fundamentals of electron density, density matrix and
density functional theory in atoms, molecules and the solid state. In: Progress in theoretical
chemistry and physics, vol 14. Springer, Berlin
98. Glushkov A V (2006) Relativistic and correlation effects in spectra of atomic systems.
Astroprint, Odessa
Relativistic Quantum Chemistry
and Spectroscopy of Exotic Atomic
Systems with Accounting for Strong
Interaction Effects

O. Yu. Khetselius, A. V. Glushkov, Yu. V. Dubrovskaya,

Yu. G. Chernyakova, A. V. Ignatenko, I. N. Serga
and L. A. Vitavetskaya

Abstract We present the fundamentals of a consistent relativistic theory of spectra

of the exotic pionic atomic systems on the basis of the Klein-Gordon-Fock equation
approach and relativistic many-body perturbation theory (electron subsystem). The
key feature of the theory is simultaneous accounting for the electromagnetic and
strong pion-nuclear interactions by means of using the generalized radiation and
strong pion-nuclear optical potentials. The nuclear and radiative corrections are
effectively taken into account. The modified Uehling-Serber approximation is used
to take into account for the Lamb shift polarization part. In order to take into
account the contribution of the Lamb shift self-energy part we have used the
generalized non-perturbative procedure, which generalizes the Mohr procedure and
radiation model potential method by Flambaum-Ginges. There are presented data of
calculation of the energy and spectral parameters for pionic atoms of the 93Nb,
173
Yb, 181Ta, 197Au, with accounting for the radiation (vacuum polarization),
nuclear (finite size of a nucleus) and the strong pion-nuclear interaction corrections.
The measured values of the Berkley, CERN and Virginia laboratories and alter-
native data based on other versions of the Klein-Gordon-Fock theories with taking
into account for a finite size of the nucleus in the model uniformly charged sphere
and the standard Uehling-Serber radiation correction are listed too.

Keywords Relativistic quantum chemistry ⋅

Spectroscopy of exotic atoms
Relativistic perturbation theory ⋅
Energy approach ⋅
Nuclear and radiative
corrections ⋅ Pionic atomic systems

O. Yu. Khetselius (✉) ⋅ A. V. Glushkov ⋅ Yu. V. Dubrovskaya ⋅ Yu. G. Chernyakova ⋅

A. V. Ignatenko ⋅ I. N. Serga ⋅ L. A. Vitavetskaya
Odessa State Environmental University, L’vovskaya Str., 15, 65016 Odessa, Ukraine
e-mail: [email protected]

© Springer International Publishing AG, part of Springer Nature 2018 71

Y. A. Wang et al. (eds.), Concepts, Methods and Applications of Quantum Systems
in Chemistry and Physics, Progress in Theoretical Chemistry and Physics 31,
https://doi.org/10.1007/978-3-319-74582-4_5
72 O. Yu. Khetselius et al.

1 Introduction

It is well known that studying the energy, spectral, radiation parameters, including
the spectral lines hyperfine structure, for heavy exotic (hadronic, kaonic, pionic)
atomic systems is of a great interest for the further development as atomic and
nuclear theories and quantum chemistry of strongly interacted fermionic systems
(see, for example, Refs. [1–31]). Really, the exotic atoms enable to probe aspects of
atomic and nuclear structure that are quantitatively different from what can be
studied in the electronic (“usual”) atoms. Besides, the corresponding data on the
energy and spectral properties of the hadronic atomic systems can be used as a
powerful tool for the study of particles and fundamental properties.
At present time one of the most sensitive tests for the chiral symmetry breaking
scenario in the modern hadron’s physics is provided by studying the exotic
hadron-atomic systems. Nowadays the transition energies in pionic (kaonic, muonic
etc.) atoms are measured with an unprecedented precision and from studying
spectra of the hadronic atoms it is possible to investigate the strong interaction at
low energies measuring the energy and natural width of the ground level with a
precision of few meV [20–46].
The strong interaction is the reason for a shift in the energies of the low-lying
levels from the purely electromagnetic values and the finite lifetime of the state
corresponds to an increase in the observed level width. For a long time the similar
experimental investigations have been carried out in the laboratories of Berkley,
Virginia (USA), CERN (Switzerland).
The most known theoretical models to treating the hadronic (pionic, kaonic,
muonic, antiprotonic etc.) atomic systems are presented in Refs. [1–46]. The most
difficult aspects of the theoretical modelling are reduced to the correct description of
pion-nuclear strong interaction [10–18] as the electromagnetic part of the problem
can be in principle reasonably accounted for [47–60].
In the present chapter we briefly present the fundamentals of a consistent rela-
tivistic theory of spectra of the exotic pionic atomic systems (with simultaneous
accounting for the electromagnetic and strong pion-nuclear interactions by means of
using the generalized radiation and strong pion-nuclear optical potentials) on the
basis of the Klein-Gordon-Fock. The nuclear and radiative corrections are effec-
tively taken into account. The modified Uehling-Serber approximation is used to
take into account for the Lamb shift polarization part. In order to take into account
the contribution of the Lamb shift self-energy part we have used the generalized
non-perturbative procedure, which generalizes the Mohr procedure and radiation
model potential method by Flambaum-Ginges. There are presented data of calcu-
lation of the energy and spectral parameters for pionic atoms of the 93Nb, 173Yb,
181
Ta, 197Au, with accounting for the radiation (vacuum polarization), nuclear
(finite size of a nucleus) and the strong pion-nuclear interaction corrections. The
measured values of the Berkley, CERN and Virginia laboratories and alternative
data based on other versions of the Klein-Gordon-Fock theories with taking into
Relativistic Quantum Chemistry and Spectroscopy of Exotic … 73

account for a finite size of the nucleus in the model uniformly charged sphere and
the standard Uhling-Serber radiation corrections are listed too.

2 Relativistic Theory of Exotic Quantum Systems

with Accounting of the Electromagnetic and Strong
Interaction Effects

2.1 The Klein-Gordon-Fock Equation. Electromagnetic

Interactions and Nuclear Potential

Here we present a brief description of the key moments of our approach (more
details can be found in Refs. [61–79]). The relativistic electron wave functions are
determined from solution of the Klein-Gordon-Fock equation (pion is the Boson
with spin 0, mass: mπ − = 139.57018 MэB, rπ− = 0.672 ± 0.08 fm) with a general
potential (the latter includes an electric and polarization potentials of a nucleus plus
the strong pion-nuclear interaction potential), which can be written as follows:

1
m2 c2 ΨðxÞ = f ½iℏ∂t + eV0 ðrÞ2 + ℏ2 ∇2 gΨðxÞ ð1Þ
c2

where c is a speed of the light, h is the Planck constant, and Ψ0(x) is the scalar wave
function of the space-temporal coordinates. Usually one considers the central
potential [V0(r), 0] approximation with the stationary solution:

Ψðx) = exp( − iEt ̸ℏÞφðxÞ, ð2Þ

where φðxÞ is the solution of the stationary equation:

1
f ½E + eV0 ðrÞ2 + ℏ2 ∇2 − m2 c2 gφðxÞ = 0 ð3Þ
c2

Here E is the total energy of the system (sum of the mass energy mc2 and
binding energy ε0).
In principle, the central potential V0 is the sum of the following potentials: the
electric potential of a nucleus, vacuum-polarization potential and the strong inter-
action potential. The nuclear potential for the spherically symmetric density ρðrjRÞ
can be presented as follows:

Zr  0   Z∞ 0 0  0  
dr r ρ r R + dr r ρ r R
0 0
Vnucl ðrjRÞ = − ð1 ̸ r Þ 2
ð4Þ
0 r

Further the density can be approximated by the Gaussian function:

74 O. Yu. Khetselius et al.

 ̸2
pffiffiffi  
ρðrjRÞ = 4γ3 ̸ π exp − γr 2 ð5Þ

Z∞
drr 2 ρðrjRÞ = 1,
0

Z∞
drr 3 ρðrjRÞ = R,
0

(here g = 4/R2 and R is the effective nucleus radius) or by the Fermi function:

ρðrÞ = ρ0 ̸f1 + exp½ðr − cÞ ̸ aÞg, ð6Þ

where the parameter a = 0.523 fm, the parameter c is chosen by such a way that it
is true the following condition for average-squared radius:

< r 2 > 1 ̸2 = ð0.836 × A1 ̸3 + 0.5700Þfm. ð7Þ

Further one should use the formulas for the finite size nuclear potential and its
derivatives on the nuclear radius. Here we use the known Ivanov-Ivanova et al.
method of differential equations (look details in Refs. [80–83]). The effective
algorithm for definition of the potential Vnucl ðrjRÞ is used in Refs. [65, 72] and
reduced to solution of the following system of the differential equations (for the
Fermi model):

Zr  0  
0   0 0 
V nuclðr, RÞ = 1 ̸ r 2 dr r 2 ρ r , R ≡ 1 ̸ r 2 yðr, RÞ,
0

y0 ðr, RÞ = r 2 ρðr, RÞ, ð8Þ

ρ′ ðrÞ = ðρ0 ̸ aÞ exp½ðr − cÞ ̸af1 + exp½ðr − cÞ ̸aÞg2

with the corresponding boundary conditions. In a case of the Gaussian model the
corresponding system of differential equations is as follows:

Rr  
V ′ nuclðr, RÞ = ð1 ̸ r 2 Þ dr ′ r ′2 ρ r ′ , R ≡ ð1 ̸r 2 Þyðr, RÞ
0 ð9Þ
y′ ðr, RÞ = r 2 ρðr, RÞ

pffiffiffi   8r
ρ′ ðr, RÞ = − 8γ5 ̸2 r ̸ π exp − γr 2 = − 2γrρðr, RÞ = − 2 ρðr, RÞ
πr

with the boundary conditions:

Relativistic Quantum Chemistry and Spectroscopy of Exotic … 75

Vnucl ð0, RÞ = − 4 ̸ðπr Þ,

yð0, RÞ = 0,
̸2
pffiffiffi
ρð0, RÞ = 4γ3 ̸ π = 32 ̸ R3 ð10Þ

Another, probably, more consistent approach is in using the relativistic

mean-field (RMF) model, which been designed as a renormalizable meson-field
theory for nuclear matter and finite nuclei [47].

2.2 Quantum Electrodynamics Effects in Pionic Atomic

Systems

Consistent and accurate account of the radiation or QED effects is of a great

importance and interest in spectroscopy of the pionic atomic systems. To take into
account the radiation (QED) corrections, namely, the important effect of the vac-
uum polarization one could use the procedure, which is in details described in the
Refs. [41–58, 65, 72–78]. Figure 1 illustrates Feynman diagrams, which describe a

Fig. 1 Feynman diagrams,

which describe a QED effect
of the vacuum polarization:
A1—the Uehling-Serber
term; A2, A3—terms of the
order [αðZαÞn (n = 2, …);
A4—the Källen-Sabry
correction of the order
α2 ðαZ Þ; A5—the
Wichmann-Kroll correction
of order αðZαÞn (n = 3)

(A4)

(A5)
76 O. Yu. Khetselius et al.

QED effect of the vacuum polarization: A1—the Uehling-Serber term; A2, A3–
terms of order члeны пopядкa [αðZαÞn (n = 2, …); A4- the Källen-Sabry cor-
rection of order α2 ðαZ Þ; A5–the Wichmann-Kroll correction of order кa αðZαÞn
(n = 3). An effect of the vacuum polarization is usually taken into account in the
first PT theory order by means of the generalized Uehling-Serber potential with
modification to take into account the high-order radiative corrections. In particular,
the generalized Uehling-Serber potential can be written as follows:

Z∞ pffiffiffiffiffiffiffiffiffiffiffi
2α   t2 − 1 2α
U ðr Þ = − dt expð − 2rt ̸ αZ Þ 1 + 1 ̸2t 2
≡− C ðgÞ, ð11Þ
3πr t 2 πr
1

where g = r ̸ ðαZÞ. More correct and consistent approach is presented in Refs. [42,
43, 52–62]. An accounting of the nuclear finite size effect modifies the potential (7)
as follows:

Z Z∞  pffiffiffiffiffiffiffiffiffiffiffi  
2α2     1 t2 − 1 ρ r′
U ðr Þ = −
FS 3 ′
d r dt exp − 2t r − r ′  ̸αZ × 1+ 2 ,
3π 2t t 2 jr − r ′ j
m
ð12Þ

The Uehling-Serber potential, determined as a quadrature (11), may be

approximated with high precision by a simple analytical function. The use of new
approximation of the Uehling potential permits one to decrease the calculation
errors for this term down to 0.5–1%.
A method for calculation of the self-energy part of the Lamb shift is based on an
idea by Ivanov-Ivanova (see Refs. [80, 81]), which generalizes the known hydro-
gen-like method by Mohr and radiation model potential method by
Flambaum-Ginges (look details in Refs. [41, 52, 61, 62]).
According to Ref. [9], in an atomic system the radiative shift and the relativistic
part of energy are, in principle, defined by one and the same physical field. One
could suppose that there exists some universal function that connects the self -
energy correction and the relativistic energy. The self-energy correction for the
states of a hydrogen-like ion was presented by Mohr [41] as:

Z4
ESE ðHjZ, nljÞ = 0.027148 F ðHjZ, nljÞ ð13Þ
n3

The values of F are given at Z = 10 − 110, nlj = 1s, 2s, 2p1 ̸2 , 2p3 ̸2 .
These results are modified here for the states 1 s2 nlj of the non-H atoms (ions).
It is supposed that for any ion with nlj electron over the core of closed shells the
sought value may be presented in the form [52]:
Relativistic Quantum Chemistry and Spectroscopy of Exotic … 77

ξ4  
ESE ðZ, nljÞ = 0.027148 3
f ðξ, nljÞ cm − 1 ð14Þ
n

The parameter ξ = ðER Þ1 ̸4 , ER is the relativistic part of the bounding energy of

the outer electron; the universal function f ðξ, nljÞ does not depend on the compo-
sition of the closed shells and the actual potential of the nucleus. The procedure of
generalization for a case of the non-H systems with the finite nucleus consists of the
following steps [9]: (1). Calculation of the values ER and ξ for the states nlj of
H-like ions with the point nucleus (in accordance with the Zommerfeld formula);
(2). Construction of an approximating function f ðξ, nljÞ by the found reference
Z and the appropriate F ðHjZ, nljÞ; (3). Calculation of ER and ξ for the states nlj of
Li-like ions with the finite nucleus; (4). Calculation of ESE for the sought states by
the formula (14). The energies of the states of the non-H atoms and ions are
calculated twice: with a conventional constant of the fine structure α = 1 ̸ 137 and
with α̃ = α ̸ 1000. The results of latter calculations were considered as
non-relativistic. This permitted isolation of ER and ξ. A detailed evaluation of their
accuracy may be made only after a complete calculation of ESE n
ðZ, nljÞ. It may be
stated that the above extrapolation method is more justified than using the widely
spread expansions by the parameter αZ. The other details of the theory and com-
putational code can be found in Refs. [61–70, 76–79].

2.3 Strong Pion-Nuclear Interactions in Pionic Atomic

System

The most difficult aspect of the problem is an adequate account for the strong
pion-nuclear interaction in the exotic system. Now it is well known that the most
fundamental and consistent microscopic theory of the strong interactions is pro-
vided by the modern quantum chromodynamics. One should remind that here
speech is about a gauge theory based on the representation of the confined coloured
quarks and gluons. Naturally one could consider the regimes of relatively low and
high energies (asymptotic freedom). In a case of the low energies so called coupling
constant increases to the order 1 and, therefore, this perturbation methods fail to
describe the interaction of strongly interacting hadrons (including pions). Naturally,
to describe the strong pion-nuclear interaction (even at relatively low energies)
microscopically, a different approaches can be developed (look details in Refs. [11–
19, 76–79]).
More simplified and sufficiently popular approach to treating the strong inter-
action in the pionic atomic system is provided by the well known optical potential
model (c.g. [14, 15]). On order to describe the strong π−N interaction we have used
the optical potential model n which the generalized Ericson-Ericson potential is as
follows:
78 O. Yu. Khetselius et al.


4π αðr Þ
Vπ − N = Vopt ðr Þ = − qðr Þ∇ ∇ ,
2m 1 + 4 ̸3πξαðr Þ

mπ   
qð r Þ = 1 + b0 ρðr Þ + b1 ρn ðr Þ − ρp ðr Þ
mN

mπ  
+ 1+ B0 ρ2 ðr Þ + B1 ρðr Þδρðr Þ , ð15Þ
2mN

mπ − 1   
αðr Þ = 1 + c 0 ρð r Þ + c 1 ρn ð r Þ − ρp ð r Þ
mN

mπ − 1  
+ 1+ C0 ρ2 ðr Þ + C1 ρðr Þδρðr Þ .
2mN

Here ρp, n ðr Þ—distribution of a density of the protons and neutrons, respectively,

ξ—parameter (ξ = 0 corresponds to case of “no correlation”, ξ = 1, if anticorrela-
tions between nucleons); respectively isoscalar and isovector parameters b0, c0, B0,
b1, c1, C0 B1, C1—are corresponding to the s-wave and p-wave (repulsive and
attracting potential member) scattering length in the combined spin-isospin space
with taking into account the absorption of pions (with different channels for p-p pair
B0ðppÞ and p-n pair B0ðpnÞ ), the Lorentz-Lorentz effect in the p-wave interaction and
isospin and spin dependence of an amplitude π − N scattering:
 
b0 ρðr Þ → b0 ρðr Þ + b1 ρp ðr Þ − ρn ðr Þ , ð16Þ

The description of numerical values of the potential parameters will be com-

mented below (look details in Refs. [41, 52, 61, 62]).

2.4 Complex Energy of Pionic Atomic System

Further we note that an energy of the hadronic atom can be represented as the
following sum:

E≈EKG + EFS + EVP + EN ; ð17Þ

Here EKG -is the energy of a pion in a nucleus ðZ, AÞ with the point-like charge
(dominative contribution in (17)), EFS is the contribution due to the nucleus finite
size effect, EVP is the radiation correction due to the vacuum-polarization effect, EN
is the energy shift due to the strong interaction VN .
It is easily to note that the strong pion-nucleus interaction contribution into
energy can be directly found from the solution of the Klein-Gordon-Fock equa-
tion with the corresponding pion-nucleon potential, for example, in the optical
potential approximation (8). Since the corresponding optical potential contains the
complex parameters, the relevant energy eigen-values of the Klein-Gordon-Fock
Relativistic Quantum Chemistry and Spectroscopy of Exotic … 79

equation for the definite pionic state ði = nlÞ in an atom are the complex values
too, i.e. [11, 38, 76]:

Ei = Re Ei + iIm Ei = Re Ei − ði ̸ 2ÞΓi , ð18Þ

where the imaginary part determines a width of pionic energy level Gi. The total
width of any level is determined as by the strong pion-nuclear interaction contri-
bution ΓSi (pion absorption) as by the electromagnetic contribution Γrad
i . The latter is
determined by a probability of the electromagnetic radiation transition (including
the Auger process probability ΓAi ) on the lower level. As an example, for the width
of pion 1 s can be written:
 
ΓS1s = Γexp rad S A exp
2p → 1s − Γ2p + Γ2p + Γ2p ≈ Γ2p → 1s , ð19Þ

where Γradi i ΓAi are the radiation and Auger widths respectively. Let us consider
further elements of theory, associated with the implementation of the known rel-
ativistic energy formalism in our theory to calculate the electromagnetic interaction
transition probabilities in spectrum of the pionic atom [80–91]. It is worth to remind
that in relativistic theory of the usual many-electron systems (an energy of any
excited state is a complex quantity) an shift of the total energy level is usually
represented as:

ΔEi = ReΔEi + iImΔEi = ReΔEi − ði ̸2ÞΓrad

i , ð20Þ

where Γradi is a radiation width, and the corresponding radiative transition proba-
bility in the usual atomic system P ∼ Γrad
i . In order to compute the latter we use the
generalized relativistic energy approach.
Let us remind that an initial general energy formalism combined with an
empirical model potential method in a theory of atoms and multicharged ions has
been developed by Ivanov-Ivanova et al. [80–84]; further more general ab initio
gauge-invariant version of relativistic energy approach has been presented by
Glushkov-Ivanov [89]. The imaginary part of the energy shift of an atom is con-
nected with the radiation decay possibility (transition probability). For the α-n
radiation transition ImDE in the lowest order of the PT is determined as:

1
Im ΔE = − ∑ V jωαn j , ð21Þ
4π α > n > f αnαn
½α < n ≤ f 

where ωαn is a frequency of the α-n radiation, (α > n > f) for particle and (α < n <
f) for vacancy. The matrix element V is determined as follows:
80 O. Yu. Khetselius et al.

ZZ
jωj sinjωjr12
Vijkl = dr1 dr2 ψ *i ðr1 Þψ *j ðr2 Þ ð1 − α1 α2 Þψ *k ðr2 Þψ *l ðr1 Þ ð22Þ
r12

The detailed procedure for computing the matrix elements (22) is presented in
Refs. [76–92]. All calculations are performed with using the numeral code
Superatom (version 98).

3 Results and Conclusions

3.1 The Parameters of the Optical Potential

 
In Table 1 we list the values of parameters for the proton and neutron cp , cn Fermi
distribution and nuclear spin I number of some nuclei.
In Table 2 there are presented the concrete values of some optical potential
parameters, which have been used in different calculations [15–39]. Let us explain the
used classification and abbreviations of corresponding sets of the optical potential
parameters: Tauscher, ξ = 0—Tau1; Tauscher, ξ = 1—Tau2; Batty et al.—Bat.;
Seki et al.—Sek; Nagels—Nag; de Laat-Konijin et al.—Laat, Serga-Shakhman—
Serg-Sha, this theory—Odes [16–22, 38, 39, 76–79]. Note that the parameters of the
optical potential in Table 1 (the initial set of parameters) were initially obtained by
calibration of the experimental data on pion-nuclear scattering for the light and
medium nuclei. Further application of the model to the heavy atoms and relatively
low-lying states showed imperfections (in some cases) of these sets of the parameter
values under theoretical studying heavy nuclei. This situation is called pion
anomalies. For example, experimental (the low-energy scattering of pions; LAMPF)
results for relatively low-lying states of pion at levels 3d, 2p, 1 s (in such heavy atoms

Table 1 Parameters for the Nucleus Z A cp, Ф cn, Ф Iπ

proton and neutron cp , cn 20
Ne 10 19.9924 2.963 2.912 0
Fermi distribution and nuclear 24
spin I number of some nuclei Mg 12 23.9850 3.080 3.026 0
93
Mb 41 92.906 4.986 5.127 9/2
133
Cs 55 132.905 5.599 5.774 7/2
165
Ho 67 164.93 6.125 6.329 7/2
173
Yb 70 172.938 6.274 6.570 5/2+
175
Lu 71 174.941 6.274 6.570 7/2
181
Ta 73 180.948 6.347 6.650 7/2+
197
Au 79 196.967 6.454 6.850 3/2+
205
Tl 81 204.974 6.587 6.881 1/2+
208
Pb 82 207.977 6.652 6.892 0
209
Bi 83 208.98 6.688 6.870 9/2−
237
Np 93 237.095 7.001 7.300 5/2+
Relativistic Quantum Chemistry and Spectroscopy of Exotic … 81

Table 2 The values of some optical potential parameters, which have been used in different
calculations (see text)
Tauscher Tauscher Batty Seki Nagels Row78 Laat Odes
ξ=0 ξ=1 ξ=1 ξ=1 ξ=1 ξ=1 ξ=1 ξ = 1 [76,
[16] [16] [17] [18] [22] [19] [38, 39] 79]
b0 −0.0296 −0.0293 −0.017 0.003 −0.013 −0.004 0.007 0.003 mπ− 1
b1 −0.077 −0.078 −0.13 −0.143 −0.092 −0.094 −0.075 −0.094 mπ− 1
ReB0 0 0 −0.0475 −0.15 – – −0.18 −0.15 mπ− 4
ImB0 0.0436 0.0428 0.0475 0.046 – – 0.058 0.046 mπ− 4
c0 0.172 0.227 0.255 0.21 0.209 0.23 0.266 0.21 mπ− 3
c1 0.22 0.18 0.17 0.18 0.177 0.17 0.40 0.18 mπ− 3
ReC0 – – – 0.11 – – 0.07 0.11 mπ− 6
ImC1 – – – – – – −0.34 −0.25 mπ− 6

as Ta, Bi et al.) have shown that the appropriate values of width due to strong
interaction by a factor two and more are less than the values specified within the
optical potential model with using the earliest parameterization [14, 15].
Such relatively significant deviations are typical for strong levels shifts.
A similar anomaly was detected in a case of the strong quadrupole shift of the
pion energy in 3d and 4f states. It should be noted that the results of the hyperfine
structure studying for heavy pionic atoms with pion at low (deep) lying states are
extremely scarce.
Earlier it is indicated a possibility of improving the consistency between theory
and experiment by taking into account of the strong interaction potential isovector
(absorption) terms. In fact, it is a more adequate account of increasing S-wave
repulsion for absorption of pion at 3d-level. In this regard, under the parameteri-
zation of the optical potential authors [33] left without changing the parameters
settings that are the most reliably identified, namely: ReB0, ImB0, c0, c1, ReC0,
ImC0.
At the same time the parameters whose values differ most strongly in different
sets, in particular, b1 (plus parameters, which are usually not included so far in the
basic optical potential parameterizations, i.e. ImB1, ImC1), should be optimized.
This is achieved by receiving preliminary calculated relationships (illustrations are
given below) for the energy shifts and widths (for a number of states fof the
following systems: 20Ne, 24Mg, 93Nb, 133Cs, 175Lu, 181Ta, 197Au, 208Pb) upon the
b1, ImB1, ImC1 parameter values; further, there were chosen such values that satisfy
the smallest standard deviation of the experimental values.
In Tables 3 and 4 we list the dependences of shifts and widths for the 4f, 3d
levels due to the strong pion nuclear interaction upon the parameter ImB1 ImC1
values for the pionic atom of 208Pb (our data).
Further in Tables 5 and 6 we list the analogous dependences of shifts and widths
for the 4f, 3d levels due to the strong pion nuclear interaction upon the parameter
ImB1 ImC1 values for the pionic system 181Ta.
82 O. Yu. Khetselius et al.

Table 3 The dependence of shifts and widths for the 4f, 3d levels due to the strong pion nuclear
interaction upon the parameter ImB1 value for 208Pb
ImB1 ε4f
0 (
208
Pb) Γ4f
0 (
208
Pb) ε3d
0 (
208
Pb) Γ3d
0 (
208
Pb)
0.00 1.596 1.11 18.2 67.7
0.02 1.603 1.09 18.9 63.6
0.04 1.625 1.07 19.7 58.7
0.06 1.633 1.05 20.5 54.8
0.08 1.641 1.03 21.6 50.5
0.10 1.652 0.96 22.8 47.2
0.12 1.723 0.91 23.8 45.3
Exp. 1.68 ± 0.04 0.98 ± 0.05 22.7 ± 2.2 47.1 ± 3.6

Table 4 The dependence of shifts and widths for the 4f, 3d levels due to the strong pion nuclear
interaction upon the parameter ImC1 для ядpa 208Pb
ImC1 ε4f
0 (
208
Pb) Γ4f
0 (
208
Pb) ε3d
0 (
208
Pb) Γ3d
0 (
208
Pb)
0.00 1.610 1.050 18.8 68.3
−0.05 1.622 0.949 19.4 64.4
−0.10 1.633 0.942 19.8 60.3
−0.15 1.642 0.939 20.5 56.5
−0.20 1.653 0.928 21.8 52.3
−0.25 1.665 0.918 22.6 48.2
−0.30 1.676 0.899 23.2 44.5
Exp. 1.68 ± 0.04 0.98 ± 0.05 22.7 ± 2.2 47.1 ± 3.6

Table 5 The dependence of shifts and widths for the 4f, 3d levels due to the strong pion nuclear
interaction upon the parameter ImB1 value for 208Pb 181Ta
ImB1 ε4f
0 (
181
Ta) Γ4f
0 (
181
Ta) ε3d
0 (
181
Ta) Γ3d
0 (
181
Ta)
0.00 0.508 0.41 13.5 34.9
0.02 0.517 0.39 14.1 30.8
0.04 0.525 0.37 14.7 27.5
0.06 0.533 0.35 15.2 24.8
0.08 0.543 0.33 15.8 22.6
0.10 0.554 0.30 16.3 20.3
0.12 0.566 0.28 16.8 18.3
Exp. 0.56 ± 0.04 0.31 ± 0.05 16.2 ± 1.3 20.1 ± 1.5

3.2 Pionic Hydrogen and the Radiation Widths of the 3d, 4f,
5g Levels in Some π − A Systems

In Table 7 we list the calculated (in meV) QED corrections to the energies of the 1s,
2p, 3p, 4p states for pionic hydrogen: data by Indelicato et al. and Serga et al. and
Relativistic Quantum Chemistry and Spectroscopy of Exotic … 83

Table 6 The dependence of shifts and widths for the 4f, 3d levels due to the strong pion nuclear
interaction upon the parameter ImC1 для ядpa 81Ta
ImC1 ε4f 81
0 ( Ta) Γ4f 81
0 ( Ta)
ε3d 81
0 ( Ta) Γ3d 81
0 ( Ta)
0.00 0.486 0.334 12.8 33.7
−0.05 0.502 0.325 13.7 30.8
−0.10 0.519 0.313 14.3 28.5
−0.15 0.534 0.302 14.9 25.7
−0.20 0.547 0.274 15.6 23.4
−0.25 0.559 0.268 16.3 20.6
−0.30 0.571 0.255 16.8 19.9
Exp. 0.56 ± 0.04 0.31 ± 0.05 16.2 ± 1.3 20.1 ± 1.5

our data. The following abbreviations are used: the Uehling-Serber vacuum
polarization correction (VP-US), the Kallen-Sabry correction (VP-KS),
Wichman-Kroll one (PV-WK).
Analysis of data shows that there is physically reasonable agreement between
different theoretical data, namely, data by Schlesser-Indelicato et al. [31, 32, 76–
79]. Naturally, the reason is obvious as the known expansion parameter aZ in the
hydrogen atom is significantly less than one, and the general QED contributions
into the levels energies are not large in comparison with other ones.
In Table 8 we present our calculated (the relativistic Klein-Gordon-Fock theory
combined with energy approach) data on the radiation widths of the 3d, 4f, 5g
levels for a number of the pionic π−A atoms. There are also listed the analogous

Table 7 The calculated (in QED contr. 1s [31, 32] 1s [76, 77] 1s [79]
meV) QED corrections to the
energies of the 1s, 2p, 3p, 4p VP-US −3240.802 −3240.799 −3240.801
states for pionic hydrogen: VP-KS −24.365 −24.363 −24.365
data by Indelicato et al., Serga PV-WK −4.110 −4.113 −4.112
et al. and our data QED contr. 2p 2p 2p
VP-US −35.795 −35.793 −35.794
VP-KS −0.346 −0.343 −0.345
PV-WK −0.008 −0.010 −0.009
QED contr. 3p 3p 3p
VP-US −11.407 −11.405 −11.406
VP-KS −0.108 −0.105 −0.107
PV-WK −0.002 −0.003 −0.002
QED contr. 4p 4p 4p
VP-US −4.921 −4.918 −4.920
VP-KS −0.046 −0.044 −0.045
PV-WK −0.001 −0.002 −0.001
84 O. Yu. Khetselius et al.

Table 8 The radiation widths of the 3d, 4f, 5g levels in some π − A systems
Nucleus Γ1rad Γ2rad (5g) Γ1rad (4f) Γ2rad (4f) [79] Γ1rad (3d) Γ2rad (3d)
(5g) [38, 39] [79] [38, 39] [38, 39] [79]
165
Ho – 15.2 – 56.1 – 228.8
173
Yb – 17.9 – 66.8 – 275.4
175
Lu – 20.7 – 77.5 – 320.3
181
Ta 25.7 23.5 90.9 88.6 369.9 366.1
203
Tl – 37.2 – 136.8 – 557.2
208
Pb 41.5 39.4 146.8 143.2 587.6 583.8
209
Bi 43.7 41.5 156.2 153.1 617.3 613.7

data by Laat-Konijn et al. [38, 39], obtained with using the relativistic
Klein-Gordon-Fock model and Hartree-Fock approximation.

3.3 Transition Energies and Energy Shift and Widths Due

to the Strong Interaction in the Spectra of in Some π − A
Systems

In Figs. 2 and 3 there are presented the parts of the X-ray spectra of the 165Ho
(Fig. 2), 181Ta (Fig. 3) and positions of the hyperfine structure components (5 g-4f
transition; experimental data from [13]).
In Tables 9a and 10 we present theoretical and experimental data (in keV) for the
4f level shift (a) and widths (b) provided by the strong pion-nuclear interaction for a
number of pionic atoms. The shortened designation of the parameter sets for the
strong π−N interaction potential: Tauscher—Tau1; Tauscher, Tau2; Batty et al.—
Bat; Seki et al.—Sek; de Laat-Konijin et al.—Laat, this work—Srg-Sha [16–22, 38,
39, 76–79]. In our parameterization of the strong p−N interaction potential the most
reliably defined (B0, c0, c1, C0) parameters are remained unchanged, and the
parameters whose values differ greatly in different sets, in particular, b1
(b1 = −0.094) plus still not included ones ImB1, ImC1 have been optimized by
computing dependencies of the strong shifts upon the parameters b1, ImB1, ImC1
for π−−20Ne, 24Mg, 93Nb, 133Cs,175Lu, 181Ta, 197Au, 208Pb atoms. Further we have
chosen the values which satisfy the smallest standard deviation of reliable experi-
mental values.
In Table 11 the analogous theoretical and experimental data (in keV) for the 3d
level shift (a) and widths for different pionic atoms are listed [16–22, 38, 39,
76–79].
In Table 12 data on the 4f-3d, 5g-4f transition energies for pionic atoms of the
93
Nb, 173Yb, 181Ta, 197Au are presented. There are also listed the measured values
Relativistic Quantum Chemistry and Spectroscopy of Exotic … 85

Fig. 2 The fragments of the

X-ray spectra of the 165Ho
and positions of the hyperfine
structure components (5-4f
transition; experimental data
from [13])

of the Berkley, CERN and Virginia laboratories and alternative data obtained on the
basis of computing within alternative versions of the Klein-Gordon-Fock
(KGF) theory with taking into account for a finite size of the nucleus in the
model uniformly charged sphere and the standard Uehling-Serber radiation cor-
rection (see Refs. [6, 7, 13, 42, 43, 86, 89]).
The analysis of the presented data indicate on the importance of the correct
accounting for the radiation (vacuum polarization) and the strong pion-nuclear
interaction corrections. The contributions due to the nuclear finite size effect should
be accounted in a precise theory too. More exact knowledge of the electromagnetic
interaction parameters for a pionic atom will make more clear the true values for
parameters of the pion-nuclear potentials. Further it allows to correct a disadvantage
of widely used parameterization of the optical potential. It is especially important if
one takes into account an increasing accuracy of the X-ray pionic atom spec-
troscopy experiments. It is interesting to note that the contributions into transition
energies are about ∼5 keV due to the QED effects, ∼0.2 keV due to the nuclear
86 O. Yu. Khetselius et al.

Fig. 3 The fragments of the

X-ray spectra of the 181Ta and
positions of the hyperfine
structure components (5g-4f
transition; experimental data
from [13])

Table 9 Theoretical and experimental data for the 4f level shift (keV) provided by the strong
pion-nuclear interaction for a number of pionic atoms (see text)
ε4f , Γ4f Exp H-like Tau 1 Tau 2 Ba ξ = 1 Sek ξ = 1 Laat ξ = 1 Srg-Sha Ou r
Func. ξ=0 ξ=1 ξ=1 ξ=1
165
Ho: ε 0.29 ± 0.01 0.21 0.25 0.24 0.24 0.21 0.26 0.29 0.29
0.27 0.26
169
Tm: ε – – – – – – – 0.38 0.38
173
Yb: ε – – – – – – – 0.44 0.44
175
Lu: ε 0.51 ± 0.04 0.36 0.43 0.42 0.41 0.36 0.46 0.50 0.50
181
Ta: ε 0.56 ± 0.04 0.47 0.57 0.54 0.53 0.47 0.60 0.55 0.55
197
Au: ε 1.25 ± 0.07 – 1.21 1.14 1.12 0.98 1.25 1.24 1.24
208
Pb: ε 1.68 ± 0.04 – 1.76 1.62 1.58 1.39 1.68 1.65 1.65
209
Bi: ε 1.78 ± 0.06 – 1.94 1.80 1.78 1.57 1.83 1.77 1.77

finite size effect, and ∼0.07 keV due to the electron screening effect, provided by
the 2[He], 4[Be], 10[Ne] electron shells [79].
To conclude, let us underline that the key factors for the physically reasonable
agreement between experimental and theoretical data on the multi-electron pionic
Relativistic Quantum Chemistry and Spectroscopy of Exotic … 87

Table 10 Theoretical and experimental data for the 4f level widths (keV) provided by the strong
pion-nuclear interaction for a number of pionic atoms (see text)
ε4f , Γ4f Exp H-like Tau 1 Tau 2 Bat Sek Laat Srg-Sha Our
Func. ξ=0 ξ=1 ξ=1 ξ=1 ξ=1 ξ=1 ξ=1
165
Ho: Γ 0.21 ± 0.02 0.08 0.13 0.12 0.13 0.11 0.13 0.20 0.21
169
Tm: Γ – – – – – – – – 0.23
173
Yb: Γ – – – – – – – – 0.26
175
Lu: Γ 0.27 ± 0.07 0.14 0.23 0.22 0.24 0.20 0.24 0.28 0.28
181
Ta: Γ 0.31 ± 0.05 0.16 0.31 0.30 0.31 0.27 0.31 0.30 0.31
197
Au: Γ 0.77 ± 0.04 – 0.73 0.68 0.69 0.58 0.67 0.75 0.77
208
Pb: Γ 0.98 ± 0.05 – 1.18 1.04 1.03 0.86 0.98 0.97 0.99
209
Bi: Γ 1.24 ± 0.09 – 1.35 1.18 1.17 0.99 1.10 1.22 1.25

Table 11 Theoretical and experimental data for the 3d level shifts and widths (keV) provided by
the strong pion-nuclear interaction for a number of pionic atoms (see text)
3d Еxp. Tau 1 Tau 2 Bat Sek Laat Srg-Sha Odes
ξ=0 ξ=1 ξ=1 ξ=1 ξ=1 ξ=1 ξ=1
93
Nb: ε 0.74 ± 0.02 0.66 0.67 0.73 0.66 0.75 0.75 0.73
169
Tm: ε – – – – – – – 11.0
173
Yb: ε – – – – – – – 12.4
175
Lu: ε – – – – – – – 13.9
181
Ta: ε 16.2 ± 1.3 19.6 16.4 10.4 4.4 14.4 16.3 16.1
197
Au: ε 20.6 ± 1.9 27.9 22.5 13.2 5.0 20.3 21.1 20.3
208
Pb: ε 22.7 ± 2.2 34 25 13 3 18 22.8 22.6
209
Bi: ε 20 ± 3 37 27 17 5 19 23 21.1
93
Nb: Γ 0.40 ± 0.02 0.405 0.413 0.459 0.404 0.452 0.42 0.41
169
Tm: Γ – – – – – – – 15.7
173
Yb: Γ – – – – – – – 17.6
175
Lu: Γ – – – – – – – 19.4
181
Ta: Γ 20,1 ± 1,5 40,5 37,5 33,4 26,2 27,6 20.3 20.2
197
Au: Γ 34±3.6 68 62 53 41 42 36.2 35.6
208
Pb: Γ 47.1 ± 3.6 88 78 65 51 51 47.2 47.0
209
Bi: Γ 52 ± 3.6 97 86 72 57 56 53.6 53.4

atoms are provided by a correct consideration of the nuclear, relativistic, radiative

and inter-electron correlation corrections. Using different schemes for accounting of
these correlations explains a difference between calculation results, obtained within
the Klein-Gordon-Fock approach. To reach the further improvement of the com-
puted data one should take into account more correctly the spatial distribution of the
88 O. Yu. Khetselius et al.

Table 12 Transition energies (keV) in the spectra of some heavy pionic atoms (see text)
π−A Trans. Berkley CERN EKGF+EM EKGF-EM EN [6] EN EN [89]
Eexp Eexp [6, 7] [13] [13] a
93
Nb 5g-4f – 307.79 ± – – – – 307.85
0.02
173
Yb 5g-4f – – – – – – 412.26
181
Ta 5g-4f 453.1 ± 453.90 ± 453.06 453.78 – 453.52 453.71
0.4 0.20 453.62
197
Au 5g-4f 532.5 ± 533.16 ± 528.95 – 532.87 531.88 533.08
0.5 0.20
93
Nb 4f-3d – 140.3 ± – – – – 140.81
0.1
173
Yb 4f-3d – – – – – – 838.67
181
Ta 4f-3d – 1008.4 ± – – – 992.75 1008.80
1.3

magnetic moment inside a nucleus (the Bohr-Weisskopf effect), the

nuclear-polarization corrections etc. (for example, within the Woods-Saxon model
or relativistic mean filed theory). In last years this topic has been a subject of
intensive interest.

References

1. PSI Experiment R-98.01. http://www.fz-juelich.de/ikp/exotic-atoms

2. Ericson T, Weise W (1988) Pions and Nuclei. Clarendon, Oxford
3. Deloff A (2003) Fundamentals in Hadronic Atom Theory. World Sci, Singapore
4. Khetselius OYu (2011) Quantum structure of electroweak interaction in heavy finite
fermi-systems. Astroprint, Odessa
5. Deslattes R, Kessler E, Indelicato P, de Billy L, Lindroth E, Anton J (2003) Exotic atoms. Rev
Mod Phys 75:35
6. Backenstoss G (1970) Ann Rev Nucl Sci 20:467
7. Menshikov LI, Evseev MK (2001) Phys Uspekhi 171:150
8. Scherer S (2003) In: Negele JW, Vogt EW (eds) Advances in nuclear physics, vol
27. Springer, Berlin, pp 5–50
9. Schroder H, Badertscher A, Goudsmit P, Janousch M, Leisi H, Matsinos E, Sigg D, Zhao Z,
Chatellard D, Egger J, Gabathuler K, Hauser P, Simons L, El Hassani A (2001) J Phys
C21:473
10. Leon M, Seki R (1974) Phys Rev Lett 32:132
11. Batty CJ, Eckhause M, Gall KP et al. (1989) Phys Rev C 40:2154
12. Chen MY, Asano Y, Cheng SC, Dugan G, Hu E, Lidofsky L, Patton W, Wu CS (1975) Nucl
Phys A 254:413
13. Olaniyi B, Shor A, Cheng S, Dugan G, Wu CS (1982) Nucl Phys A 403:572
14. Erikcson M, Ericson T, Krell M (1969) Phys Rev Lett 22:1189
15. Erikcson M, Ericson T (1966) Ann Phys 36:323
Relativistic Quantum Chemistry and Spectroscopy of Exotic … 89

16. Tauscher L (1971) Analysis of pionic atoms and the p-nucleus optical potential. In:
Proceedings of the international semantic p-Meson nucleus interaction-CNRS-strasbourg,
France, p 45
17. Batty C, Biagi S, Friedman E, Hoath S (1983) Phys Rev Lett 440:931; Batty C J, Friedman E,
Gal A (1978) Nucl Phys A 402:411
18. Seki R, Masutani K, Jazaki K (1983) Phys Rev C 27:1817
19. Rowe G, Salamon M, Landau RH (1978) Phys Rev C 18:584
20. Anagnostopoulos D, Biri S, Boisbourdain V, Demeter M, Borchert G et al. (2003) Nucl Inst
Meth B 205:9
21. Anagnostopoulos D, Gotta D, Indelicato P, Simons LM (2003) arXiv:physics. 0312090v1
22. Nagels MM, de Swart J, Nielsen H et al (1976) Nucl Phys B 109:1
23. Lauss B (2009) Nucl Phys A 827C, 401 PSI experiment R-98.01http://pihydrogen.psi.ch
24. CERN DIRAC Collaboration (2011) Search for long-lived states of p+ p− and p−K atoms,
CERN-SPSLC-2011–001 SPSLC-P-284-ADD p 22
25. Umemoto Y, Hirenzaki S, Kume K, Toki H, Tahihata I (2001) Nucl Phys A 679:549
26. Nose-Togawa N, Hirenzaki S, Kume K (1999) Nucl Phys A 646:467
27. Glushkov AV, Malinovskaya SV, Gurnitskaya EP, Khetselius OYu, Dubrovskaya YuV
(2006) J Phys: Conf Ser 35:425
28. Hatsuda T, Kunihiro T (1994) Phys Rep 247:221
29. Ikeno N, Kimura R, Yamagata-Sekihara J, Nagahiro H, Jido D, Itahashi K et al.
(2011) 1107.5918v1[nucl-th]
30. Kolomeitsev EE, Kaiser N, Weise W (2003) Phys Rev Lett 90:092501
31. Lyubovitskij V, Rusetsky A (2000) Phys Lett B 494:9
32. Schlesser S, Le Bigot E-Q, Indelicato P, Pachucki K (2011) Phys Rev C 84:015211
33. Sigg D, Badertscher A, Bogdan M, Goudsmit P, Leisi H, Schröder H, Zhao Z, Chatellard D,
Egger J, Jeannet E, Aschenauer E, Gabathuler K, Simons L (1996) Rusi El Hassan A. Nucl
Phys A 609:269
34. Gotta D, Amaro F, Anagnostopoulos D, Biri S, Covita D, Gorke H, Gruber A, Hennebach M,
Hirtl A, Ishiwatari T, Indelicato P, Jensen T, Bigot E, Marton J, Nekipelov M, dos Santos J,
Schlesser S, Schmid P, Simons L, Strauch H, Trassinelli M, Veloso J, Zmeskal J ed. Kania Y,
Yamazaki Y (AIP) (2008) CP1037, 162
35. Gotta D, Amaro F, Anagnostopoulos D, Biri S, Covita D, Gorke H, Gruber A, Hennebach M,
Hirtl A, Ishiwatari T, Indelicato P, Jensen T, Bigot E, Marton J, Nekipelov M, dos Santos J,
Schlesser S, Schmid P, Simons L, Strauch H, Trassinelli M, Veloso J, Zmeskal (2008)
Precision physics of simple atoms and molecules. In: Lecture notes in physics, vol 745,
Springer, Berlin, Heidelberg, pp 165–186
36. Taal A, D’Achard van Enschut J, Berkhput J et al (1985) Phys Lett B 156:296
37. Taal A, David P, Hanscheid H, Koch JH, de Laat CT et al (1990) Nucl Phys A 511:573
38. de Laat CT, Taal A, Konijn J et al (1991) Nucl Phys A 523:453
39. de Laat CT, Taal A, Duinker W et al (1987) Phys Lett B 189:7
40. Khetselius OYu, Turin AV, Sukharev DE, Florko TA (2009) Sensor Electr and Microsyst
Techn N1, 30–35
41. Mohr PJ (1993) Atom Dat Nucl Dat Tabl 24:453; (1983) Phys Scripta 46:44
42. Indelicato P (1996) Phys Scripta T65:57; Indelicato P, Trassinelli M (2005) arXiv:
physics.0510126v1
43. Glushkov AV, Malinovskaya SV (2003) In: Fazio G, Hanappe F (eds) New projects and new
lines of research in nuclear physics. World Sci, Singapore, pp 242–250
44. Santos J, Parente F, Boucard S, Indelicato P, Desclaux J (2005) Phys Rev A 71:032501
45. Mitroy J, Ivallov IA (2001) J Phys G Nucl Part Phys 27:1421
46. Strauch T (2009) High-precision measurement of strong-interaction effects in pionic
deuterium, Julich
47. Serot B, Walecka J (1986) Relativistic nuclear many body problem. In: Advances in nuclear
physics, Plenum Press, N-Y
48. Glushkov AV, Ivanov LN (1992) Phys Lett A 170:33
90 O. Yu. Khetselius et al.

49. Glushkov AV, Ivanov LN, Ivanova EP (1986) Autoionization phenomena in atoms. Moscow
University Press, Moscow, pp 58–160
50. Ivanova EP, Glushkov AV (1986) J Quant Spectr Rad Transfer 36:127
51. Glushkov AV, Khetselius OYu, Gurnitskaya EP, Loboda AV, Florko TA, Sukharev DE,
Lovett L (2008) Frontiers in quantum systems in chemistry and physics. In: Wilson S,
Grout PJ, Maruani J, Delgado-Barrio G, Piecuch P (eds) Progress in Theoretical Chemistry
and Physics, vol 1. Springer, Dordrecht, pp 507–524
52. Flambaum VV, Ginges JSM (2005) Phys Rev A 72:052115
53. Safranova UI, Safranova MS, Johnson WR (2005) Phys Rev A 71:052506
54. Glushkov AV, Malinovskaya SV, Khetselius OYu, Loboda AV, Sukharev DE, Lovett L
(2009) Int J Quant Chem 109:1717
55. Gurnitskaya EP, Khetselius OYu, Loboda AV, Vitavetskaya LA (2008) Photoelectronics
17:127
56. Glushkov AV, Ambrosov SV, Loboda AV, Gurnitskaya EP, Prepelitsa GP (2005) Int J Quant
Chem 104:562
57. Johnson W, Sapirstein J, Blundell S (1993) Phys Scripta T 46:184
58. Glushkov AV (2006) Relativistic and correlation effects in spectra of atomic systems,
Astroprint, Odessa
59. Glushkov AV, Khetselius OYu, Malinovskaya SV (2008) Europ Phys Journ ST 160:195
60. Buyadzhi VV, Glushkov AV, Lovett L (2014) Photoelectronics 23:38
61. Glushkov AV, Ambrosov SV, Loboda AV, Chernyakova GYu, Svinarenko AA, Khet-
selius OY (2004) Nucl Phys A Nucl Hadr Phys 734:21
62. Glushkov AV, Ambrosov SV, Loboda AV, Gurnitskaya EP, Khetselius OY (2006) Recent
advances in the theory of chemical and physical systems. In: Julien P, Maruani J, Mayou D,
Wilson S, Delgado-Barrio G (eds) Progress in theoretical chemistry and physics, vol 15.
Springer, Dordrecht, pp 285–299
63. Glushkov AV (2005) AIP Conf Proceedings 796 (1): 206–210
64. Glushkov AV, Lovett L, Khetselius OYu, Gurnitskaya EP, Dubrovskaya YuV, Loboda AV
(2009) Int J Mod Phys A 24:611
65. Glushkov AV, Khetselius OY and Lovett L (2009) Advances in the theory of atomic and
molecular systems. Dynamics, spectroscopy, clusters and nanostructures. In: Piecuch P,
Maruani J, Delgado-Barrio G, Wilson S (eds) Progress in theoretical chemistry and physics,
vol 20. Springer, Dordrecht, pp 125–152
66. Khetselius OY (2012) Quantum systems in chemistry and physics. In: Nishikawa K,
Maruani J, Brändas E, Delgado-Barrio G, Piecuch P (eds) Progress in theoretical chemistry
and physics, vol 26. Springer, Dordrecht, pp 217–229
67. Khetselius OYu (2009) Int J Quant Chem 109:3330
68. Khetselius OYu (2009) Phys Scripta T135:014023
69. Khetselius OYu (2015) Frontiers in quantum methods and applications in chemistry and
physics. In: Nascimento M, Maruani J, Brändas E, Delgado-Barrio G (eds) Progress in
theoretical chemistry and physics, vol 29. Springer, Cham, pp 55–76
70. Khetselius OYu (2008) Hyperfine structure of atomic spectra. Astroprint, Odessa
71. Glushkov AV, Khetselius OYu, Svinarenko AA (2013) Phys Scripta T153:014029
72. Glushkov AV, Khetselius OY, Svinarenko AA (2012) Advances in the theory of quantum
systems in chemistry and physics. In: Hoggan P, Brändas E, Maruani J, Delgado-Barrio G,
Piecuch P (eds) Progress in theoretical chemistry and physics, vol 22. Springer, Dordrecht, pp
51–68
73. Khetselius O, Florko T, Svinarenko A, Tkach T (2013) Phys Scripta T153: 014037
74. Khetselius OYu (2010) AIP Conf Proc 1290:29
75. Khetselius OYu, Florko TA, Nikola LV, Svinarenko AA, Serga IN, Tkach TB,
Mischenko EV (2010) Quantum Theory: reconsideration of foundations (AIP). 1232:243
76. Serga IN, Dubrovskaya YV, Kvasikova AS, Shakhman AN, Sukharev DE (2012) J Phys Conf
Ser 397:012013
77. Serga IN (2013) Photoelectronics 22:71; Shakhman AN (2015) Photoelectronics 24:109
Relativistic Quantum Chemistry and Spectroscopy of Exotic … 91

78. Sukharev DE, Khetselius OYu and Dubrovskaya YuV (2009) Sensor Electr and Microsyst
Techn N3, 16–21
79. Bystryantseva AN, Khetselius OYu, Dubrovskaya YuV, Vitavetskaya LA, Berestenko AG
(2016) Photoelectronics 25:56
80. Glushkov AV (2008) Relativistic quantum theory. Quantum mechanics of atomic systems,
Astroprint, Odessa, 700p
81. Ivanov LN, Letokhov VS (1985) Com Mod Phys D 4:169; Ivanov LN, Ivanova EP,
Aglitsky EV (1988) Phys Rep 166:315
82. Ivanova EP, Ivanov LN, Glushkov AV, Kramida AE (1985) Phys Scripta 32:513
83. Glushkov AV, Kondratenko PA, Buyadgi VV, Kvasikova AS, Sakun TN, Shakhman AN
(2014) J Phys: Conf Ser 548:012025
84. Svinarenko AA, Glushkov AV, Khetselius OYu, Ternovsky VB, Dubrovskaya YuV,
Kuznetsova AA, Buyadzhi VV (2017) In: Rare Earth Element, (ed) Orjuela JEA. InTech,
pp 83–104
85. Glushkov AV, Khetselius OYu, Svinarenko AA, Buyadzhi VV, Ternovsky VB,
Kuznetsova AA, Bashkarev PG (2017) In: Uzunov DI (ed) Recent studies in perturbation
theory. InTech, pp 131–150
86. Baldwin GG, Salem JC, Goldansky VI (1981) Rev Mod Phys 53:687; Goldansky VI,
Letokhov VS (1974) JETP 67:513; Ivanov LN, Letokhov VS (1975) JETP 68:1748
87. Glushkov AV, Khetselius O, Gurnitskaya E, Loboda A, Sukharev D (2009) AIP Conf Proc
1102 (1):168
88. Ivanov LN, Letokhov VS, Glushkov AV (1991) Preprint of Inst. for Spectroscopy of USSR
Acad Sci (ISAN) AS-N5
89. Glushkov AV (2013) Advances in quantum methods and applications in chemistry, physics
and biology. In: Hotokka M, Brändas E, Maruani J, Delgado-Barrio G (eds) Progress in
theoretical chemistry and physics, vol 27. Springer, Cham, pp 161–177
90. Glushkov AV, Ambrosov SV, Loboda AV, Gurnitskaya EP, Khetselius OYu (2006)
Recent advances in theoretical physics and chemistry systems. In: Julien J-P, Maruani J,
Mayou D, Wilson S, Delgado-Barrio G (eds) Progress in theoretical chemistry and physics,
vol 15. Springer, Dordrecht, pp 285–299
91. Glushkov AV (2012) Quantum systems in chemistry and physics. In: Nishikawa K,
Maruani J, Brändas E, Delgado-Barrio G, Piecuch P (eds) Progress in theoretical chemistry
and physics, vol 26. Springer, Dordrecht, pp 231–252
92. Khetselius OYu, Zaichko PA, Smirnov AV, Buyadzhi VV, Ternovsky VB, Florko TA,
Mansarliysky VF (2017) Quantum systems in physics, Chemistry and biology. In: Tadjer A,
Pavlov R, Maruani J, Brändas J, Delgado-Barrio G (eds) Progress in theoretical chemistry and
physics, vol 30. Springer, Cham, pp 271–281
 Part II
Molecular Structure and Dynamics
Difference of Chirality of the Electron
Between Enantiomers of H𝟐 X𝟐

Masato Senami, Ken Inada, Kota Soga, Masahiro Fukuda

and Akitomo Tachibana

Abstract The integrated chirality density of H2 X2 molecules is studied in view-

points of the internal torque for the electron spin. Since the chirality density is pro-
portional to the zeta potential, which is the potential of the zeta force, one of the
torque for the electron spin, the distribution of the chirality density affects the dis-
tribution of the internal torque in molecules. It is seen that the integrated chirality
density is larger for the larger atomic number. It is found that the integrated chiral-
ity density of H2 Te2 has the same sign as the parity-violating energy, while those of
H2 O2 and H2 S2 are opposite to the sign of the parity-violating energy, and the depen-
dence of the integrated chirality density of H2 Se2 on dihedral angle is significantly
different from that of the parity-violating energy.

1 Introduction

Only one form of enantiomeric pair is found in living systems. While both enan-
tiomers of sugar and amino acids are produced in the same quantity in laborato-
ries, only (D)-sugars and (L)-amino acids can be found in the systems. The mecha-
nism how this bias is generated is a long-standing puzzle [1]. On the origin of this
biomolecular homochirality, many hypotheses are proposed, and those are classi-
fied by some features, (1) terrestrial and extra-terrestrial, (2) biotic and abiotic, and
(3) probabilistic and deterministic. We do not know the solution, and however many
researchers believe that the realization of homochirality is deeply related to the par-
ity violation of Nature, which is given by the weak interaction of the standard model
of particle physics. The weak interaction is known to induce nuclear 𝛽 decay. Gauge
theory describes this interaction as well as electromagnetic interaction. Electromag-
netic interaction is formulated as U(1) gauge theory and is mediated by the photon.

M. Senami (✉) ⋅ K. Inada ⋅ K. Soga ⋅ M. Fukuda ⋅ A. Tachibana

Department of Micro Engineering, Kyoto University, Kyoto 615-8540, Japan
e-mail: [email protected]

© Springer International Publishing AG, part of Springer Nature 2018 95

Y. A. Wang et al. (eds.), Concepts, Methods and Applications of Quantum Systems
in Chemistry and Physics, Progress in Theoretical Chemistry and Physics 31,
https://doi.org/10.1007/978-3-319-74582-4_6
96 M. Senami et al.

On the other hand, the weak interaction is formulated as SU(2) gauge theory, and W
and Z bosons are mediator of this interaction. These bosons are coupled to ordinary
fermions, such as electrons, only by the V-A coupling, where V and A is the vec-
tor and axial-vector currents, respectively. Hence parity is violated maximally in the
weak interaction.
It is known that this parity violation makes the energy difference between enan-
tiomers. This energy difference between them, which is called parity-violating energy
shift, is very small [2]. In spite of the smallness, this energy is studied in many
works by computational methods. The parity-violating energy shift is dominantly
induced by virtual Z boson exchanges between nuclei and electrons. Virtual Z boson
exchanges is quantum effect in relation to uncertainty principle. Since Z boson is
very heavy as 90 GeV∕c2 , where c is the speed of light, this particle cannot be pro-
duced energetically, and, however, uncertainty relation allows this particle to affect in
a very restricted region. Single W boson exchange does not contribute to the parity-
violating energy, since if occurs, it is 𝛽 decay. The exchanges between electrons is
known to be subdominant for particularly heavy nuclei [3]. Since nuclei are localized
strongly, the existence of parity-violating energy means the existence of the nonzero
chirality density around nuclei. It is surprising that low energy electrons have polar-
ized chirality, since the electron mass, that is the interaction with Higgs field in the
vacuum, vanishes chirality for free electrons.
The parity-violating energy shift is studied by many researchers as one of solu-
tions for biomolecular homochirality. Due to this energy difference, the amount of
one of enantiomers may be slightly larger than the other through an enhancement
process, such as crystallization. In other viewpoints, some of the authors are inter-
ested in the total integrated chirality density of the electron in a molecule, whose
existence has already been reported [4]. If nonzero integrated chirality density of
the electron exists in enantiomers, its interaction rates through weak interaction are
different between enantiomeric pair. This difference of the reaction rate probably
generates the difference of the number density between enantiomers, which is indis-
pensable for the problem of the biomolecular homochirality. Even though this differ-
ence is not enough, we are interested in the distribution of the chirality density in a
molecule, since the chirality density is known to be proportional to the zeta potential
[5]. The zeta potential is the potential for the zeta force, which is the counter force to
the spin torque, defined only in quantum field theory [6, 7]. Hence chiral molecules
have nonzero zeta force distribution.
In this work, we study the integrated chirality density of the electron in H2 X2
molecules (X = O, S, Se, Te). This structure is one of the simplest chiral molecules
and is chosen in many works [4, 8–10]. The total chirality of the electron of this
structure is reported for H2 Te2 in Reference [4]. In this work, the chirality is investi-
gated also for H2 O2 , H2 S2 , and H2 Se2 in relation to the parity-violating energy shift
and the zeta potential.
Difference of Chirality of the Electron Between Enantiomers of H2 X2 97

2 Theory

First we briefly review the parity-violating energy. The dominant contribution to the
parity-violating energy is the parity odd interaction between electrons and nuclei.
This interaction is given as the coupling of vector and axial-vector currents of elec-
trons and nucleons. For low energy nuclei, nonrelativistic approximation is good,
and then the space-like component of vector current and the time-like component of
axial-vector current vanishes. The internal structure of nuclei is not affected by the
geometry of molecules, and hence the space-like component of axial-vector current
is considered to be negligibly small. Hence, the most important contribution arises
from the coupling of the time-like components of nucleon vector current and electron
axial-vector current. The Hamiltonian of this interaction is given as

∑ GF †
†
HPV = √ QW 𝜓̂ e 𝛾5 𝜓̂ e 𝜓̂ Nn 𝜓̂ Nn
n
(1)
n 2 2

where index n means the species of nuclei, GF = 1.166378 × 10−5 GeV−2 is Fermi
coupling constant [11], 𝜓̂ e and 𝜓̂ Nn are the field operators of electrons and nuclei,
and 𝛾5 ≡ i𝛾 0 𝛾 1 𝛾 2 𝛾 3 . The weak charge of a nucleus QnW is given as QnW = Z n (1 −
4 sin2 𝜃W ) − N n , where Z n and N n are the number of protons and neutrons in a nucleus
n and 𝜃W is the weak-mixing angle, sin2 𝜃W = 0.2313 [11]. The parity-violating
energy is calculated with state vector, |Ψ⟩,

EPV = d3 ⃗r⟨Ψ|HPV |Ψ⟩ (2)

∫

and the parity-violating energy shift is defined as the energy difference between
enantiomers,

ΔEPV = 2|EPV |. (3)

The parity-violating energy can be divided into contributions from each nucleus,
( )
G ∑ n G ∑ n n
EPV = √F QW d3 ⃗r⟨Ψ|𝜓̂ e† 𝛾5 𝜓̂ e 𝜓̂ N† 𝜓̂ Nn |Ψ⟩ = √F QW MPV . (4)
2 2 n ∫ n
2 2 n
n
This MPV is often used for parameterizing the contribution from each nucleus. The
density of nuclei is strongly localized, and hence nucleus density, 𝜓̂ N† 𝜓̂ Nn , can be
n

approximated to 𝜓̂ N† 𝜓̂ Nn = 𝛿 3 (⃗r − ⃗rn ) where ⃗rn is the position of a nucleus n. As

n
a result, the parity-violating energy is well calculated by chirality densities at the
positions of nuclei, ⟨Ψ|𝜓̂ e† (⃗rn )𝛾5 𝜓̂ e (⃗rn )|Ψ⟩.
Chirality density is proportional to the zeta potential, which is the potential of the
zeta force defined in Reference [5]. The definition of the zeta potential is given by
98 M. Senami et al.

ℏc [ † ] ℏc ( † )
𝜙̂ 5 (x) = 𝜓̂ e (x)𝛾5 𝜓̂ e (x) = †
𝜓̂ eR (x)𝜓̂ eR (x) − 𝜓̂ eL (x)𝜓̂ eL (x) , (5)
2 2

where 𝜓̂ eL (x) ≡ 12 (1 − 𝛾5 )𝜓̂ e and 𝜓̂ eR (x) ≡ 12 (1 + 𝛾5 )𝜓̂ e are fields with the left-handed
and right-handed chirality, respectively. The zeta force density operator is defined
with the zeta potential, 𝜙̂ 5 , as

𝜁̂ek (x) = −𝜕k 𝜙̂ 5 . (6)

The zeta force is one of the torque in the equation of motion of the electron spin
defined in quantum field theory. The equation of motion of the spin is derived from
the time-derivative of the spin angular momentum. In quantum field theory, the spin
angular momentum density operator is represented as

1 †
ŝ ke (x) = ℏ𝜓̂ (x)Σk 𝜓̂ e (x), (7)
2 e

where Σk is the Pauli matrix in the four-component representation. The torque den-
sity for the electron spin is derived by the time-derivative of this operator, and the
time-derivative can be reduced by using Dirac equation.

̂ 𝜇 (x)𝜓̂ e (x) = mc𝜓̂ e (x),

iℏ𝛾 𝜇 D (8)

where m is the mass of electron. The covariant derivative D ̂ 𝜇 (x) is defined as D

̂ 𝜇 (x) =
𝜕𝜇 + i ℏc A𝜇 (x), where Ze = −1 is the electric charge of the electron and  𝜇 (x) is the
Ze e ̂

gauge field operator. As a result, we obtain the equation of motion of the spin,

𝜕̂ske (x)
= ̂tek (x) + 𝜁̂ek (x), (9)
𝜕t

where the first term, ̂tek (x), is the spin torque density. The spin torque term is defined
so that this term matches the well-known spin torque term in relativistic quantum
mechanics. The Heisenberg equation of the spin angular momentum in quantum
mechanics is given by d⃗ŝ e ∕dt = −c𝜋⃗̂ × 𝛼⃗ [12]. The spin torque density operator is
defined with the relativistic stress tensor density, 𝜏̂eΠln (x), as

̂tek (x) = −𝜖lnk 𝜏̂eΠln (x), (10)

where 𝜖lnk is the Levi-Civita tensor. The relativistic stress tensor operator is given by
[5],
[ ( ) ]
𝜏̂eΠln (x) =
iℏc † ̂ l (x)𝜓̂ e (x) − D
𝜓̂ e (x)𝛾 0 𝛾 n D ̂ l (x)𝜓̂ e (x) † 𝛾 0 𝛾 n 𝜓̂ e (x) . (11)
2
Difference of Chirality of the Electron Between Enantiomers of H2 X2 99

The stress tensor is known to classify the chemical bond [13, 14]. In the following,
we call only ̂tek (x) the spin torque, and the sum of the terms in the right-hand side is
called the torque for the spin. Our equation of motion for the spin is recently shown
to be derived from the spin vorticity principle in a sophisticated way [15].
One may wonder whether this new contribution disturbs the consistency between
experimental observations and a prediction by quantum mechanics. The expecta-
tion value of the zeta force is zero after the integration over the whole space, since
the zeta force density operator is given as the gradient form of the zeta potential.
Hence, contributions from the zeta force are considered to be negligible in past exper-
iments. However, the contribution from the zeta force give a nonzero effect in a local
region, even after the integration over a restricted local region. Hence, the zeta force
is observable quantity if an experimental setup is carefully designed for this purpose.
In addition, the equation of motion from quantum field theory has another advan-
tage over that from quantum mechanics. In a time-independent stationary state of
the electron spin. the spin torque and zeta force are canceled out with each other and
the torque for the spin is zero at any point. Hence in quantum field theory a local
picture of the spin dynamics can be correctly described. In quantum mechanics, any
local spin dynamics cannot be predicted. The Heisenberg equation of the spin gives
generically nonzero torque for a local region even for a spin stationary state. This
is because the expectation value of quantum mechanics is defined as the integration
over the whole space and hence local description is theoretically out of scope.

3 Computational Details

Our equation is defined in quantum field theory, and hence a state should also be
prepared in the theory. However, a generic state based on quantum field theory is
not available for our purpose, since most computation code is based on quantum
mechanics. Hence in this work we use wave functions derived from ordinary elec-
tronic structure computations as a substitution. With the usage of these wave func-
tions, computations of physical quantities, parity-violating energy, zeta potential,
and so on, are performed by QEDynamics program package [16–18].
For ordinary electronic structure computations, we use DIRAC14 program pack-
age [19]. Four-component wave function by relativistic quantum mechanics can be
computed by this code, which is indispensable for the study of spin. The dyall.ae2z
basis set [20] is used for large components of all atoms. The small component
basis set is generated by restricted kinetic balance in the code. The effect of three-
component vector potential is ignored in our calculations, since it is quantitatively
small for states by quantum mechanics computations [21]. The structure of H2 X2
molecules are determined as follows. First, the geometrical optimization computa-
tions are performed by Hartree-Fock computations with Dirac-Coulomb Hamilto-
nian. The internuclear lengths between heavy atoms X, 1.390 Å for oxygen atoms,
2.058 Å for sulfur atoms, 2.333 Å for selenium atoms, and 2.729 Å for tellurium
atoms, while the internuclear lengths between X and H atoms, 0.9439 Å for oxygen
100 M. Senami et al.

Fig. 1 Geometry of H2 X2
molecule and the definition
of the dihedral angle

atom, 1.332 Å for sulfur atom, 1.455 Å for selenium atom, and 1.649 Å for tellurium
atom. In the following, these internuclear lengths are adopted for all dihedral angles.
For wave functions derived by computations as above, parity-violating energy,
n
MPV , total chirality, spin torque density, zeta force, and zeta potential are computed
with special attention to dihedral angle of H2 X2 . Since H2 X2 is a chiral molecule,
two choices of dihedral angle exist. Our definition of the dihedral angle is shown in
Fig. 1, which is the opposite to Reference [8, 9]. Results can easily be compared by
replacing the angle 360◦ − 𝜙. Note that parity-violating energy is known to be heav-
ily dependent on computational methods, geometry of molecules, basis sets and so
on. Our basis set is smaller than previous works, and hence there are some differences
between our results and previous works [4, 8–10] as discussed later.

4 Result and Discussion

For the purpose of the check of our electronic structure, our results of parity-violating
energy and contributions from heavy atoms are compared to previous works. In
X
Fig. 2, the contributions from heavy atoms to parity-violating energy, MPV , are shown
as a function of the dihedral angle for H2 X2 (X = O, S, Se, Te) molecules. All
X
molecules have similar pattern and it is seen that heavier X atoms give larger MPV
due to larger relativistic effects. The tendencies of these curves are consistent with
previous works [4, 8–10] qualitatively. The contribution of hydrogen atoms to EPV
is known to be much smaller than that of heavier atoms. Our values have some devi-
ation from previous works. This deviation is speculated to be due to the smallness
of our basis set. Parity-violating energy of H2 X2 and contributions from heavy atom
X are summarized in Table 1. The dihedral angle is chosen to be 45◦ or −45◦ as
the value reported in references, which have the same value with opposite sign. The
abbreviation, HF, means Hartree-Fock with Dirac-Coulomb Hamiltonian, CCSD is
coupled-cluster singles-and-doubles, and CISD is configuration interaction includ-
ing single and double excitations. As seen from this table, larger basis sets as triple
or quadruple zeta function are critically important, and results of double zeta basis
sets are much smaller than that of larger ones. Larger basis set is speculated to be
Difference of Chirality of the Electron Between Enantiomers of H2 X2 101

-4
1.0×10
4.0×10-6

5.0×10-5
2.0×10-6
MPVO [a.u.]

MPVS [a.u.]
0.0×100 0.0×100

-2.0×10-6 -5
-5.0×10

-6
-4.0×10
-4
-1.0×10
0 50 100 150 200 250 300 350 0 50 100 150 200 250 300 350
dihedral angle [degrees] dihedral angle [degrees]

(a) H2 O2 (b) H2 S2

4.0×10-3 4.0×10-2

2.0×10-3 2.0×10-2
[a.u.]

[a.u.]
0.0×100 0.0×100
Se

Te
MPV

MPV

-3 -2
-2.0×10 -2.0×10

-4.0×10-3 -4.0×10-2

0 50 100 150 200 250 300 350 0 50 100 150 200 250 300 350

dihedral angle [degrees] dihedral angle [degrees]

(c) H2 Se2 (d) H2 Te2

Fig. 2 Contribution from heavy atom to parity-violating energy of H2 X2 molecules as a function

of the dihedral angle

required for the description of the cusp structure near nuclei. Since our triple zeta
result is well consistent with other triple zeta results, most deviation of our results
from others comes from the smallness of the basis set. In addition, the effect of post
Hartree-Fock computation is seen to be about 10%. Due to the limit of computational
resources, the computations of the latter part of this work is restricted to the basis
set, dyall.ae2z. Nevertheless, our wave functions are seen to be reasonable within
our computational methods from this table.
In addition, we investigate the spin torque and the zeta force of H2 O2 . Figure 3
shows the distributions of the spin torque, the zeta force, and their sum for H2 O2
with 𝜙 = 45◦ . It can be seen that the sum of the spin torque and the zeta force is
much smaller than the spin torque and the zeta force itself in the whole region. This
result is consistent with the fact that the nonzero spin torque is in balance with the
zeta force for the spin stationary state. Hence, although our computational result is
derived from wave functions of quantum mechanics, it is considered that we can use
these wave functions. The values of the norm of the spin torque and zeta force in the
vicinity of oxygen nuclei amount to 10−3 [a.u.], which is much large than that in the
vicinity of hydrogen nuclei. The smallness of the spin torque around hydrogen nuclei
102 M. Senami et al.

Table 1 Parity-violating energy of H2 X2 and contributions from heavy atom X are summarized.
The dihedral angle is chosen to be 45◦ or −45◦ , which have the same value with opposite sign. HF
means Hartree-Fock with Dirac-Coulomb Hamiltonian
Molecule Method Basis set for X |MPVX
| [a.u.] |EPV | [a.u.] Reference
H2 O2 HF dyall.ae2z 3.3401 × 10−6 3.8853 × 10−19 This study
dyall.ae3z 5.1839 × 10−6 6.0300 × 10−19 This study
dyall.ae3z − 6.051 × 10−19 [10]
dyall.ae4z − 6.376 × 10−19 [10]
cc-pVDZ+3p 5.801 × 10−6 − [9]
[22]
25s25p5d [8] 6.057 × 10−6 − [9]
aug-cc-pVDZ 3.729 × 10−6 − [8]
[22]
aug-cc-pVTZ 4.239 × 10−6 − [8]
[22]
aug-cc-pVQZ 4.687 × 10−6 − [8]
[22]
25s25p5d 6.057 × 10−6 − [8]
CISD cc-pVDZ+3p 5.410 × 10−6 − [9]
CCSD dyall.ae3z − 5.323 × 10−19 [10]
dyall.ae4z − 5.583 × 10−19 [10]
cc-pVDZ+3p 5.299 × 10−6 − [9]
H2 S2 HF dyall.ae2z 7.0563 × 10−5 1.6416 × 10−17 This study
cc-pCVTZ [22] − 1.825826 × [10]
10−17
cc-pVDZ+3p 8.916 × 10−5 − [9]
25s25p5d 9.581 × 10−5 − [8]
CCSD cc-pCVTZ [22] − 1.82103 × 10−17 [10]
cc-pVDZ+3p 9.283 × 10−5 − [9]
H2 Se2 HF dyall.ae2z 2.3917 × 10−3 1.6334 × 10−15 This study
25s25p5d 3.586 × 10−3 − [8]
CCSD dyall.cv3z [23] − 2.115 × 10−15 [10]
H2 Te2 HF dyall.ae2z 2.3842 × 10−2 2.7769 × 10−14 This study
25s25p5d 3.149 × 10−2 − [8]
CCSD dyall.cv3z − 3.289 × 10−14 [10]

O
is consistent with our previous results [6]. The nonzero MPV means the existence of
the zeta force around oxygen nuclei, since the chirality density is proportional to the
zeta potential and the zeta potential is the potential for the zeta force. Hence this
result is consistent with the existence of the parity-violating energy.
In Fig. 4, the integrated chirality density as a function of the dihedral angle is
shown for (a) H2 O2 , (b) H2 S2 , (c) H2 Se2 , and (d) H2 Te2 molecules. Our result of
Difference of Chirality of the Electron Between Enantiomers of H2 X2 103

(a) Spin torque distribution (b) Zeta force distribution

(c) Sum of spin torque and zeta force

Fig. 3 The distributions of the magnitude and the direction of a the spin torque distribution,
b the zeta force, and c the sum of the spin torque and zeta force are shown for H2 O2 molecule.
The dihedral angle is 45◦

H2 Te2 is consistent with the result reported in Reference [4], and it is confirmed that
the electron chirality is nonzero in chiral molecules. The integrated chirality density
of H2 Te2 has almost the same dependence on the dihedral angle. However, the inte-
grated chirality density of H2 O2 and H2 S2 are almost opposite to the parity-violating
X
energy, which is determined dominantly from MPV shown in Fig. 2. Moreover, the
integrated chirality density of H2 Se2 has different oscillation pattern from the parity-
violating energy. Although we guessed that the integrated chirality density and the
parity-violating energy of H2 X2 have some correlation as in Reference [4], our inte-
grated chirality density is not inconsistent with the parity-violating energy, since the
parity-violating energy is determined dominantly only by the chirality density nearby
heavy nuclei. Nevertheless, we should improve our computations with larger basis
set and perform post Hartree-Fock computations in order to check our results.
In Fig. 5, the distributions of zeta potential around one Te atom of H2 Te2 at the
dihedral angle, (a) 15◦ , (b) 45◦ and (c) 90◦ , are shown on the xy-plane for the z coordi-
nate on Te atoms. Our results are well consistent with those reported in Reference [4].
We have shown only the results for 𝜙 = 15◦ , 45◦ , 90◦ , while at other dihedral angles,
our results are consistent with Reference [4]. The distribution pattern of zeta poten-
tial agrees with their results well. The difference of the values arises from the factor
ℏc∕2, which is the coefficient of the zeta potential over the chirality density. For
comparison, the same figure for H2 O2 is shown in Fig. 6 at the dihedral angle, (a)
15◦ , (b) 45◦ and (c) 90◦ . The results are shown on the xy-plane for the z coordinate
104 M. Senami et al.

1.0 ×10-7
4.0 ×10-7
chirality density [a.u.]

chirality density [a.u.]

-8
5.0 ×10
2.0 ×10-7

0.0 ×100 0.0 ×100

-7
-2.0× 10
-5.0 ×10-8

-4.0× 10-7
-1.0 ×10-7
0 50 100 150 200 250 300 350 0 50 100 150 200 250 300 350
dihedral angle [degrees] dihedral angle [degrees]
(a) H2 O2 (b) H2 S2
-6 -5
1.0 ×10 2.0 ×10
-5
1.5 ×10

chirality density [a.u.]

chirality density [a.u.]

5.0 ×10-7 1.0 ×10-5

-6
5.0 ×10
0
0.0 ×10 0.0 ×100
-6
-5.0× 10
-7 -5
-5.0 ×10 -1.0 ×10
-5
-1.5 ×10
-6
-1.0 ×10 -2.0 ×10-5
0 50 100 150 200 250 300 350 0 50 100 150 200 250 300 350
dihedral angle [degrees] dihedral angle [degrees]
(c) H2 Se2 (d) H2 Te2

Fig. 4 The integrated chirality density as a function of the dihedral angle for a H2 O2 , b H2 S2 ,
c H2 Se2 , and d H2 Te2 molecules

(a) φ = 15◦ (b) φ = 45◦ (c) φ = 90◦

Fig. 5 The distribution of zeta potential of H2 Te2 at the dihedral angle, a 15◦ , b 45◦ and c 90◦ .
The result is shown on the xy-plane for the z coordinate on Te atoms

at one O atom. The localization of the innermost core electrons are strongly different
between O and Te atoms, and hence the distribution pattern of the zeta potential is
largely extended. The sign of the zeta potential at the position of a nucleus is the
same for 15–45◦ , and opposite for 90◦ , this corresponds to the dependence of MPV X

on the dihedral angle.

Difference of Chirality of the Electron Between Enantiomers of H2 X2 105

(a) φ = 15◦ (b) φ = 45◦ (c) φ = 90◦

Fig. 6 The distribution of zeta potential of H2 O2 at the dihedral angle, a 15◦ , b 45◦ and c 90◦ . The
result is shown on the xy-plane for the z coordinate on O atoms

5 Conclusion

In this work, we have studied H2 X2 molecules in viewpoints of spin related local

physical values. Since the chirality density is proportional to the zeta potential, which
is the potential of the zeta force, one of the torque for the electron spin, the distribu-
tion of the chirality density affects the distribution of the internal torque in molecules.
Our quantum states are well consistent with those in previous works within the choice
of basis set, and it has been confirmed that the spin torque and the zeta force are in bal-
ance with each other. We have found that the integrated chirality density is larger for
the larger atomic number as speculated from the trend of the parity violating energy.
The dependence of the integrated chirality density of H2 Te2 on dihedral angle is con-
sistent with the previous work. We have found that the integrated chirality density
of H2 Te2 has the same sign as the parity violating energy, while those of H2 O2 and
H2 S2 are opposite to the sign of the parity-violating energy, and moreover the depen-
dence of the integrated chirality density of H2 Se2 on dihedral angle is significantly
different from that of the parity-violating energy.
In our future work, we should check the dependences of the integrated chiral-
ity density of H2 O2 , H2 S2 , and H2 Se2 on dihedral angle, which are different from
that of H2 Te2 . For this purpose, larger basis set and post Hartree-Fock computa-
tions are used. In addition, we investigate the relation between the distribution of the
spin torque and the zeta force and the dihedral angle.

References

1. Meyerhenrich U (2008) Amino acids and the asymmetry of life. Springer, Heidelberg
2. Mason SF (1984) Nature 311:19
3. Hegstrom RA, Rein DW, Sandars PGH (1980) J Chem Phys 73:1
4. Bast R, Koers A, Gomes ASP, Iliaš M, Visscher L, Schwerdtfeger P, Saue T (2011) Phys Chem
Chem Phys 13:864
5. Tachibana A (2001) J Chem Phys 115:3497; J Mol Model 11:301 (2005); J Mol Struct:
(THEOCHEM), 943:138 (2010). see also Tachibana A (2017) New aspects of quantum elec-
trodynamics. Springer
106 M. Senami et al.

6. Fukuda M, Senami M, Tachibana A (2013) Prog Theor Chem Phys 27:131; Fukuda M, Soga
K, Senami M, Tachibana A (2016) Int J Quant Chem 116:920
7. Senami M, Nishikawa J, Hara T, Tachibana A (2010) J Phys Soc Jpn 79:084302; Hara T,
Senami M, Tachibana A (2012) Phys Lett A 376:1434; Fukuda M, Soga K, Senami M,
Tachibana A (2013) Phys Rev A 93:012518; Fukuda M, Ichikawa K, Senami M, Tachibana
A (2016) AIP Adv 6:025108
8. Laerdahrl JK, Schwerdtfeger P (1999) Phys Rev A 60:4439
9. Thyssen J, Laerdahrl JK, Schwerdtfeger P (2000) Phys Rev Lett 85:3105
10. Shee A, Visscher L, Saue T (2016) J Chem Phys 145:184107
11. Patrignani C et al (2016) Particle data group. Chin Phys C 40:100001
12. Sakurai JJ (1976) Advanced quantum mechanics. (Addison-Wesley, Reading, Mass)
13. Tachibana A (2004) Int J Quantum Chem 100:981
14. Ichikawa K, Nozaki H, Komazawa N, Tachibana A (2012) AIP Adv 2:042195
15. Tachibana A (2012) Math J Chem 50(669); In: (2013) Ghosh SK, Chattaraj PK (eds) Concepts
and methods in modern theoretical chemistry. CRC, Boca Raton, FL, pp 235–251
16. Senami M, Ichikawa K, Tachibana A, QEDynamics http://www.tachibana.kues.kyoto-u.ac.jp/
qed
17. Ichikawa K, Fukuda M, Tachibana A (2013) Int J Quantum Chem 113:190
18. Senami M, Miyazato T, Takada S, Ikeda Y, Tachibana A (2013) J Phys Conf Ser 454:012052;
Senami M, Ogiso Y, Miyazato T, Yoshino F, Ikeda Y, Tachibana A (2013) Trans Mat Res Soc
Jpn 38(4):535
19. Saue T, Visscher L, Jensen HJAa, Bast R, Bakken V, Dyall KG, Dubillard S, Ekström U, Eliav
E, Enevoldsen T, Faßhauer E, Fleig T, Fossgaard O, Gomes ASP, Helgaker T, Lærdahl JK, Lee
YS, Henriksson J, Iliaš M, Jacob ChR, Knecht S, Komorovský S, Kullie O, Larsen CV, Nataraj
HS, Norman, P Olejniczak G, Olsen J, Park YC, Pedersen JK, Pernpointner M, di Remigio R,
Ruud K, Sałek P, Schimmelpfennig B, Sikkema J, Thorvaldsen AJ, Thyssen J, van Stralen J,
Villaume S, Visser O, Winther T, Yamamoto S (2014) DIRAC a relativistic ab initio electronic
structure program. http://www.diracprogram.org
20. Dyall KG (2012) Theor Chem Acc 99:366 (1998); Theor Chem Acc 115:441 (2006); Theor
Chem Acc 117:483 (2007); J Phys Chem A 113:12638 (2009); Theor Chem Acc 131:1217
21. Senami M, Ikeda Y, Fukushima A, Tachibana A (2010) Jpn J Appl Phys 49:115002
22. Dunning TH Jr (1989) J Chem Phys 90:1007; Kendall RA, Dunning TH Jr, Harrison RJ, J
Chem Phys 96:6796 (1992); Woon DE, Dunning TH Jr, J Chem Phys 98:1358 (1993)
23. Dyall KG (2009) J Phys Chem A 113:12638; Theor Chem Acc 108:335 (2002); Chem Acc
115:441 (2006); Theor Chem Acc 117:483 (2007)
A Crystallographic Review of Alkali
Borate Salts and Ab Initio Study
of Borate Ions/Molecules

Cory C. Pye

Abstract The crystal structures of alkali metal borate salts are reviewed. A wide
diversity of structures is noted. Structures with discrete ions containing two or more
boron atoms are targeted for further study with ab initio methods (HF, B3LYP, and
MP2) using modest basis sets (6-31G*, 6-31+G* and 6-311+G*). The ions iden-
tified for study are: [B3O6]3−, [B3O5(OH)2]3−, [B3O4(OH)4]3−, [B3O3(OH)4]−,
[B4O5(OH)4]2−, [B5O6(OH)4]−, [B2O5]4−, and [B4O9]6−. Some structurally related
ions are examined, and an investigation of the diborates launched, including [B2O
(OH)6]2−, observed in the magnesium salt, and [B2(OH)7]−, postulated as the
intermediate responsible for signal exchange between borate anion and boric acid in
11
B NMR. The B-O bond distances and general structures are in good agreement
with both crystallographic data and previous ab initio calculations.

Keywords Crystal structure ⋅ Ab initio study ⋅ Alkali borate ⋅ Polyborate

1 Introduction

In the preceding paper, the importance of the speciation of boron (III) to geo-
chemistry, oceanography, and the nuclear industry, was discussed [1]. In addition,
the crystal structures of boron oxide, boric acid, and monomeric forms of sodium
and lithium borates were presented and discussed. Molecular forms exist for the
orthoboric acid [B(OH)3] and metaboric acid [B3O3(OH)3], and certain lithium and
sodium borates contain discrete [B(OH)4]−, [BO2(OH)]2−, and [BO3]3− ions.
The structure, energy, and vibrational frequencies of the crystallographically

Electronic supplementary material The online version of this chapter (https://doi.org/10.1007/

978-3-319-74582-4_7) contains supplementary material, which is available to authorized users.

C. C. Pye (✉)
Department of Chemistry, Saint Mary’s University, Halifax, NS B3H 3C3, Canada
e-mail: [email protected]

© Springer International Publishing AG, part of Springer Nature 2018 107

Y. A. Wang et al. (eds.), Concepts, Methods and Applications of Quantum Systems
in Chemistry and Physics, Progress in Theoretical Chemistry and Physics 31,
https://doi.org/10.1007/978-3-319-74582-4_7
108 C. C. Pye

characterized ortho- and metaboric acid, tetrahydroxoborate, trioxoborate, and

dioxomonohydroxoborate were calculated ab initio, in addition to the postulated
conjugate base of boric acid, oxodihydroxoborate [BO(OH)2]−, as well as the
transition structure leading to it. The effect of additional water was also calculated.
In this paper, the focus is on the polyborates.
The crystal structures of numerous sodium borates are given in Table 1. The
notation (l:m:n) is shorthand for a substance of oxide formula lM2O:mB2O3:nH2O,
where M is an alkali metal. Anhydrous (3:1:0) consists of discrete BO33− ions [2].
The hydrated sodium borate (2:1:1) consists of discrete [BO2(OH)]2− ions con-
necting sheets of edge-shared NaO polyhedral [3]. There are several hydrates of
sodium metaborate (1:1:x). The octahydrate [4] consists of the tetrahedral [B
(OH)4]− anion, as does the tetrahydrate [5]. (3:3:4) consists of the [B3O4(OH)4]3−
anion with two tetrahedral and one trigonal boron atoms [6]. (3:3:2) consists of
isolated [B3O5(OH)2]3− ions with two trigonal and one tetrahedral ion [7]. The
anhydrous sodium metaborate (1:1:0) contains discrete B3O63− anions in which all
boron atoms are trigonal [8]. (2:3:1) consists of the unique [B12O20(OH)4]8− anion,
which consists of 6 6-membered rings spirofused together in a cycle. There are 6
tetrahedral and 6 trigonal boron atoms. On the outer periphery of the ion, four of the
trigonal boron atoms have hydroxyl groups and two have oxo groups [9]. Both
(3:5:4) [10] and (3:5:2) [11] consists of layers of B5O6 polyhedra containing three
tetrahedral and two trigonal boron atoms.
Borax (1:2:10) [12], also known as sodium tetraborate decahydrate, consist of
chains of edge-shared hexaaquasodium octahedra alternating with chains of discrete
[B4O5(OH)4]2− ions hydrogen-bonded to each other [13]. The tetraborate anion
consists of an oxygen-boron core with two tetrahedral and two trigonal boron atoms
reminiscent of bicyclo [3.3.1] nonane, where the bridgeheads correspond to the
tetrahedral boron atoms. The structure can therefore be reformulated as
Na2[B4O5(OH)4] ⋅ 8H2O. The structure was refined by neutron diffraction [14].
The determination of the hydrogen atom positions indicate that the ion has C2
symmetry [15]. The oxo atoms of the anion accept hydrogen bonds either from the
aquasodium chains (cross-chain) or from the hydroxyl groups (in-chain), and the
hydroxyl group donate cross-chain hydrogen-bonds to the water oxygen atoms.
Sodium tetraborate “pentahydrate” [16] (the mineral tincalconite, 1:2:4.667) is a
reversible dehydration product of the decahydrate. The [B4O5(OH)4]2− ions persist
in this structure, however, the hydroxyls of the tetrahedral boron atoms are now
directly coordinated to the sodium atoms. The structure can therefore be reformu-
lated as Na2[B4O5(OH)4] ⋅ 3H2O. The hydroxyls of the trigonal boron hydrogen
bond to the oxo group between the two tetrahedral boron atoms. Upon refinement,
incomplete occupancy of the water sites was found, corresponding to the formula
Na2 [B4O5(OH)4] ⋅ 2.667H2O [17]. The disorder was also confirmed at room
temperature [18]. It was suggested that sodium tetraborate tetrahydrate (the mineral
kernite, 1:2:4) was similar to borax in structure [19]; however this was shown to be
incorrect, as the boron units formed an infinite chain of six-membered rings
comprised of [B4O6(OH)2]2n− n [20, 21]. A later redetermination also included an
electron population analysis [22]. The monohydrate consists of chains of
Table 1 Crystal structures of sodium borates
Stoichiometry Name Synth T(K) Space group Unit cell X-ray T(K) B-O (Å) Comments
NalBmHnO(l+3m+n)/ (Å or deg.) ranges
2

l m n Tetrahedral Trigonal Refs.

3 1 0 748 P21 ̸ c 5.687,7.530,9.993 n/a 1.377–1.409 [BO3]3− ions [2]
127.15
2 1 1 523 Pnma 8.627,3.512,9.863 n/a 1.351–1.439 [BO2(OH)]2−
ions [3]
1 1 8 298 P1̄ 6.126,8.1890,6.068 1.474–1.479 n/a [B(OH)4]− ions [4]
67.92,110.58,101.85
1 1 4 293 P21 ̸ a 5.886,10.566,6.146 1.463–1.483 n/a [B(OH)4]− ions [5]
111.60
3 3 4 383 Cc 12.8274,7.7276,6.9690 293 1.445–1.497 1.340–1.399 [B3O4(OH)4]3−
98.161 ions [6]
3 3 2 423 Pnma 8.923,7.152,9.548 1.464–1.488 1.318–1.432 [B3O5(OH)2]3− ions [7]
1 1 0 Sodium metaborate R3̄c 11.925,6.439 (hex) n/a 1.280,1.433 [B3O6]3- ions [8]
7.212, 111.54 (rhomb)
2 3 1 523 P21 ̸ c 8.709,11.917,9.468 1.443–1.512 1.334–1.396 [B12O20(OH)4]8− ion [9]
A Crystallographic Review of Alkali Borate Salts …

96.02
3 5 4 423 Pbca 8.804,18.371,10.924 1.425–1.531 1.351–1.398 Layers of B5O6
polyhedra [10]
3 5 2 523 Pca21 11.2373,6.0441,11.1336 291 1.440–1.510 1.367–1.377 Layers of B5O6
polyhedra [11]
1 2 10 Borax (sodium C2/c 11.89, 10.74, 12.19 [12]
tetraborate decahydrate)
C2/c 11.858,10.674,12.197 298 1.46–1.54 1.32–1.40 Chains of [B4O5(OH)4]2−
106.68 ions [13]
298 C2/c 11.885,10.654,12.206 297 1.464–1.500 1.363–1.374 Neutron diffraction [14]
(continued)
109
Table 1 (continued)
110

Stoichiometry Name Synth T(K) Space group Unit cell X-ray T(K) B-O (Å) Comments
NalBmHnO(l+3m+n)/ (Å or deg.) ranges
2

l m n Tetrahedral Trigonal Refs.

106.623
C2/c 11.8843,10.6026,12.2111 145 1.4451–1.5075 1.3655–1.3784 [15]
106.790
3 6 14 Tincalconite 353 R32 11.09,21.07 1.454–1.507 1.330–1.386 Assumed n = 15 [16]
(“pentahydrate”)
348 R32 11.097,21.114 108 1.442–1.495 1.361–1.374 Incomplete
occupancy [17]
R32 11.1402,21.207 291 1.445–1.493 1.364–1.370 [18]
1 2 4 Kernite (tetrahydrate) P21 ̸ c 7.022,9.151,15.676, n/a n/a Assumed similar to
108.83 borax [19]
P21 ̸ c 7.0172,9.1582,15.6774 1.48 1.37 Chain of 6-membered
108.861 rings [20]
P21 ̸ c 7.022,9.151,15.676, 1.428–1.521 1.353–1.384 [21]
108.8
P21 ̸ c 7.0172,9.1582,15.6774 1.436–1.523 1.350–1.383 [22]
108.861
1 2 1 (Monohydrate) 423 Pbca 8.540,10.263,14.547 1.452–1.503 1.356–1.392 [23]
1 2 0 973 P1̄ 6.5445,8.6205,10.4855 295 1.434–1.492 1.295–1.453 6- and 6,6-spiro fused
93.279,94.870,90.843 rings [24]
2 5 7 Ezcurrite P1̄ 8.598,9.570,6.576 1.431–1.494 1.322–1.385 Chains of cond.
102.75,107.5,71.52 [B5O6(OH)4]− [25]
2 5 5 Nasinite 423 Pna21 12.015,6.518,11.173 1.440–1.496 1.345–1.399 Sheets of cond.
[B5O6(OH)4]− [26]
453 Pna21 11.967,6.5320,11.126 153 1.4746 1.3685 [27]
443 Pc 11.323,6.5621,12.244 298 1.444–1.534 1.351–1.434 [28]
91.050
2 5 3 Biringuccite 523 P21 ̸ c 11.1955,6.5607,20.7566 1.421–1.513 1.343–1.413 Condensed
93.891 Spiro-B5O6 [29]
1 3 4 Ameghinite C2/c 18.428,9.882,6.326 1.443–1.506 1.342–1.398 [B3O3(OH)4]− [30]
(continued)
C. C. Pye
Table 1 (continued)
Stoichiometry Name Synth T(K) Space group Unit cell X-ray T(K) B-O (Å) Comments
NalBmHnO(l+3m+n)/ (Å or deg.) ranges
2

l m n Tetrahedral Trigonal Refs.

104.38
1 3 0 α-sodium triborate 998 P21 ̸ c 10.085,11.363,10.845 295 1.443–1.512 1.343–1.399 Spiro-B5O6 and BO4 [31]
104.48
β-sodium triborate 1003-1013 P21 ̸ c 8.990,11.033,12.107 295 1.431–1.504 1.338–1.416 Spiro-B5O6, B3O3,
90.50 and BO4 [32]
1 4 0 Sodium tetraborate Melt P21 ̸ a 6.507,17.796,8.377 1.448–1.511 1.331–1.396 Spiro-B5O6 and B3O3 [33]
96.57
1 5 10 Sborgite C2/c 11.119,16.474,13.576 1.456–1.482 1.339–1.412 [B5O6(OH)4]− ions [34]
112.83
1 5 4 Sodium pentaborate 423 P21 ̸ c 8.701,8.067,12.977 1.451–1.489 1.342–1.394 [B5O6(OH)4]− ions [35]
106.77
A Crystallographic Review of Alkali Borate Salts …
111
112 C. C. Pye

[B4O6(OH)2]2− in which the trigonal hydroxyl of a [B4O5(OH)4]2− ion has con-

densed with a tetrahedral hydroxyl of another to eliminate water [23]. The anhy-
drous sodium diborate (1:2:0) consists of single and spiro-fused double
6-membered rings [24]. The single rings are linked to double rings, which are
linked together with oxide bridges.
Ezcurrite (2:5:7) consists of chains of [B5O6(OH)4]− ions which have condensed
to give two tetrahedral and three trigonal boron atoms [25]. Nasinite (2:5:5) consists
of sheets of [B5O6(OH)4]− ions which have condensed to give two tetrahedral and
three trigonal boron atoms [26]. The structure was refined further [27] and a very
similar structure identified [28]. Biringuccite (2:5:3) consists of sheets of condensed
spiro-B5O6 units containing three trigonal and two tetrahedral boron atoms [29].
Ameghinite (1:3:4) consists of monomeric [B3O3(OH)4]− ions [30]. Anhydrous
α-sodium triborate (1:3:0) contains both a spiro-fused double 6-membered ring and
a tetraborate core, which link to each other by oxo groups [31]. The β-modification
consists of the spiro-fused double 6-membered ring, a simple 6-membered ring, and
a standalone BO4 tetrahedra [32]. Anhydrous sodium tetraborate (1:4:0) consists of
layers of single and spiro-fused double 6-membered rings [33]. The single rings are
linked to the double rings, which are linked together with oxide bridges. Both
sborgite, NaB5O8 ⋅ 5H2O (1:5:10) [34], and its dehydrated form of sodium
pentaborate (1:5:4) [35], consist of discrete [B5O6(OH)4]− ions containing one
tetrahedral and four trigonal boron atoms.
To summarize these results, the following discrete polyborate ions (as a sodium
salt) are known in the solid phase. The trinuclear [B3O3(O)(OH)4]3−,
[B3O3(O)2(OH)2]3−, [B3O3(O)3]3−, and [B3O3(OH)4]− all contain the B3O3 core.
The tetranuclear [B4O5(OH)4]2− contains the B4O5 core. The pentanuclear
[B5O6(OH)4]− contains the B5O6 core (two B3O3 cores spiro fused at a boron). The
dodecanuclear [B12O20(OH)4]8− containing the B12O18 core (six B3O3 cores spiro
fused into a macrocycle). Upon dehydration, typically chains or sheets of con-
densed units form (sometimes more than one unit is present). Some plausible
species not observed to date in sodium salts are: BO(OH)− 2 , any dinuclear species;
[B3O3(OH)6]3−, or [B3O3(OH)5]2−; [B4O5(OH)6]3−, or [B4O5(OH)5]2−;
[B5O6(OH)6]3−, or [B5O6(OH)5]2−.
The crystal structures of numerous potassium borates are given in Table 2. The
previously described discrete ions [B3O6]3−, [B4O5(OH)4]2−, [B3O3(OH)4]−, and
[B5O6(OH)4]−, are all represented. The only new borate anion appearing is
[B12O16(OH)8]4-. For lithium borates (Table 3), only the previously-mentioned
discrete ion [B(OH)4]− is found, but two new ions, B2O54− and B4O96−, appear. For
rubidium borates (Table 4), the previously mentioned discrete ions [B3O4(OH)4]3−,
[B3O6]3−, [B4O5(OH)4]2−, and [B5O6(OH)4]− are represented, and the new ion
[B7O9(OH)5]2− ion appears. For cesium borates (Table 5), the previously discussed
[B3O6]3−, [B4O5(OH)4]2− and [B5O6(OH)4]− ions are represented, but no new ions
have been found. Condensation of the triborate, tetraborate, and pentaborate into
chains and sheets is quite common as the water content of the borates decrease.
Conversely, hydration/dissolution of the water-poor solids might be expected
Table 2 Crystal structures of Potassium Borates
Stoichiometry Name Synth T(K) Space group Unit cell (Å or deg.) X-ray T(K) B-O (Å) ranges Comments
KlBmHnO(l+3m+n)/
2

l m n Tetrahedral Trigonal Refs.

1 1 0 Potassium metaborate R3̄c 7.76, 110.6 n/a 1.33–1.38 [B3O6]3− ions [36]
R3̄c n/a 1.331–1.398 [37]
1 2 4 Potassium tetraborate tetrahydrate 298 P21 21 21 12.899, 11.774, 6.859 1.447–1.511 1.312–1.401 [B4O5(OH)4]2− ions [38]
1 2 0 Potassium diborate 1073 6.46, 9.56,10.38, [39]
90.7, 102.6, 101.2
973 P1̄ 6.484, 9.604,10.413, 1.435–1.508 1.357–1.381 Condensed B4O5, B3O3
89.28, 102.75,101.25 and BO3 units [40]
2 5 5 Pna21 12.566,6.671,11.587 1.454–1.496 1.346–1.412 [B5O8(OH)] unit [41]
2 5 2 443 C2/c 18.077,6.857,13.266 298 1.449–1.513 1.350–1.413 B3O3 and B4O5 units [42]
95.271
1 3 6 298 C2/c 15.540,6.821,14.273, 163 1.462–1.489 1.347–1.393 [B3O3(OH)4]− ions [43]
104.44
A Crystallographic Review of Alkali Borate Salts …

1 3 2 428 P4 ̸ncc 11.3482,15.9169 1.426–1.471 1.270–1.468 [B12O16(OH)8]4- ions [44]

1 3 1 453 P21 ̸n 9.036,6.6052,15.997, 298 1.451–1.482 1.337–1.398 B3O3 units [45]
91.862
1 3 0 Potassium triborate 1073 10.63, 10.64, 13.01, [39]
90.7, 95.2, 114.2
1073 P21 ̸c 9.319, 6.648,21.094 1.455–1.492 1.332–1.406 B3O3 units [46]
94.38
5 19 0 γ-potassium pentaborate 1073 11.59, 17.80, 12.99, [39]
95.8
Pentapotassium 1023 C2/c 17.888,11.479,12.973, 295 1.459–1.494 1.336–1.404 B5O6, B3O3, BO3,
enneakaidekaborate 95.52 and BO4 groups [47]
1 5 8 Potassium pentaborate Aea2 11.08,11.11,8.97 1.53 1.28–1.42 [48]
hydrate
Aea2 11.065, 11.171, 9.054 [49]
113

(continued)
Table 2 (continued)
114

Stoichiometry Name Synth T(K) Space group Unit cell (Å or deg.) X-ray T(K) B-O (Å) ranges Comments
KlBmHnO(l+3m+n)/
2

l m n Tetrahedral Trigonal Refs.

Aea2 11.062, 11.175, 9.041 1.474–1.481 1.345–1.380 [B5O6(OH)4]− ions [50]
1 5 4 443 P21 ̸c 9.4824,7.5180,11.4154, 298 1.457–1.479 1.342–1.399 B5O6 units [42]
97.277
1 5 3 473 P1̄ 7.5612,9.2236,11.7298, 100 1.468–1.483 1.340–1.396 Condensed B5O6 units [51]
99.038,106.595,91.314
1 5 2 P21 ̸c 7.6690,9.0445,12.2304 100 1.464–1.482 1.339–1.391 Condensed B5O6 units [52]
119.132
1 5 0 α-potassium pentaborate 1033 Pbca 8.52,8.53,21.74 [53]
1033 Pbca 8.383,8.418,21.540 1.469–1.477 1.319–1.405 Condensed B5O6 units [54]
β-potassium pentaborate 973 Pbca 7.42, 11.69,14.72 1.35–1.57 1.07–1.51 [55]
Pbca 7.418,11.702,14.745 1.40–1.49 1.33–1.41 Condensed B5O6 units [56]
C. C. Pye
Table 3 Crystal structures of Lithium Borates
Stoichiometry Name Synth T(K) Space group Unit cell (Å or deg.) X-ray T(K) B-O (Å) ranges Comments
LilBmHnO(l+3m+n)/
2

l m n Tetrahedral Trigonal Refs.

2 1 0 β 903 I2 ̸a 10.2269,4.6988,8.7862, 153 n/a 1.345–1.444 B2O54− ions [57]
93.562
α 908 Pca21 10.1497,4.7365,17.5880 153 n/a 1.340–1.454 B2O54− ions [57]
3 2 0 873 P21 ̸ n 3.31913,23.361,9.1582, 293 n/a n/r, 1.316–1.436 cif Planar B4O96−
92.650 ions [58]
1 1 16 P3 6.5534,6.1740 1.472–1.473 n/a [Li(H2O)4]+ and [B(OH)4]−
ions [59]
P3 6.555,6.177 [60]
1 1 4 Pbca 9.16,7.95,8.54 [B(OH)4]− ions [61]
Pbca 7.9362,8.5220,9.1762 120 1.4644–1.4989 n/a [62]
1 1 0 Lithium Melt P21 ̸ c 5.838,4.348,6.449, n/a 1.330–1.400 BO3 chains [60]
metaborate 115.12
A Crystallographic Review of Alkali Borate Salts …

γ-LiBO2 1223, I4̄2d 4.1961,6.5112 1.483 n/a BO4 units [63]

15 kbar
1 2 3 Pnna 9.7984,8.2759,9.6138 powder XRD,
spiro-linked B3O3 chains [64]
1 2 0 Dilithium 1173 I41 cd 9.47,10.26 1.43–1.47 1.37–1.42 Condensed B4O5 units [65]
tetraborate
I41 cd 9.479,10.280 1.448–1.507 1.349–1.386 [66]
I41 cd 9.477,10.286 1.453–1.505 1.347–1.375 [67]
I41 cd 9.479,10.290 298 1.453–1.506 1.355–1.374 [68]
3 7 0 “2:5” 1113 P1̄ 6.487,7.840,8.510, 1.443–1.502 1.347–1.392 B3O3 and BO3 units [69]
92.11,104.85,99.47
(continued)
115
Table 3 (continued)
116

Stoichiometry Name Synth T(K) Space group Unit cell (Å or deg.) X-ray T(K) B-O (Å) ranges Comments
LilBmHnO(l+3m+n)/
2

l m n Tetrahedral Trigonal Refs.

1 3 0 Lithium melt Pna21 8.447,7.3789,5.1408 298 1.4610–1.4872 1.3481–1.3967 B3O3 unit [70], [71]
triborate
1107 Pna21 8.444,7.378,5.1416 293 1.4614–1.4847 [72]
3 11 0 “Li2B8O13” 873 P21 ̸ c 17.7607,7.7737,9.6731 293 n/r n/r B3O3 and B5O6 units [58]
100.906
1273 P21 ̸ a 9.766, 7.849, 17.899 1.464–1.496 1.345–1.412 [73]
1 5 2 573, P21 ̸ a 13.576,9.077,5.543, 293 1.445–1.492 1.346–1.382 B5O6 unit [74]
10 MPa 91.47
C. C. Pye
Table 4 Crystal structures of Rubidium Borates
Stoichiometry Name Synth T(K) Space group Unit cell (Å or deg.) X-ray T(K) B-O (Å) ranges Comments
RblBmHnO(l+3m+n)/
2

l m n Tetrahedral Trigonal Refs.

3 3 8 Pna21 7.90,13.95,9.28 1.44–1.50 1.37–1.45 [B3O4(OH)4]3−
ions [75]
1 1 0 R3̄c 13.21, 7.78 [37]
Rubidium 1173 R3̄c 13.1572,7.7434 293 1.315–1.407 [B3O6]3− ions [76]
metaborate
1 2 5.6 298 Pbcn 11.276,13.097,16.751 298 1.416–1.524 1.352–1.391 [B4O5(OH)4]2−
ions [77]
1 2 0 980 P1̄ 9.860,10.653,6.649, 1.41–1.50 1.33–1.40 B4O5, B3O3 and BO3
103.4,101.4,89.1 units [78]
3 7 0 913 P1̄ 6.603,6.632,30.085, 1.42–1.53 1.31–1.41 B5O6, B3O3 and BO4
A Crystallographic Review of Alkali Borate Salts …

91.183,91.781,119.293 units [79]

1 3 0 α-rubidium 873 P21 21 21 8.209,10.092,5.382 1.44–1.58 1.31–1.40 powder XRD, B3O3
triborate units [80]
β-rubidium 973 P21 21 21 8.438,8.719,6.240 293 1.45–1.49 1.34–1.40 B3O3 units [81]
triborate
2 7 5 443 P21 ̸ n 8.047,11.785,13.952, 298 1.420–1.500 1.333–1.395 [B7O9(OH)5]2−
102.974 ions [82]
1 5 8 298 Aea2 11.302,10.962,9.335 298 1.473–1.474 1.344–1.375 [B5O6(OH)4]−
ions [83]
1 5 0 β-rubidium 973 Pbca 7.50, 11.84, 14.80 B5O6 units [55]
pentaborate
1003 Pbca 7.553, 11.857, 14.813 298 1.464–1.481 1.339–1.391 [84]
117
 118

Table 5 Crystal structures of Cesium Borates

Stoichiometry Name Synth T(K) Space group Unit cell (Å or deg.) X-ray T(K) B-O (Å) ranges Comments
CslBmHnO(l+3m+n)/
2

l m n Tetrahedral Trigonal Refs.

1 1 0 Cesium metaborate R3̄c 13.68, 8.36 [37]
R3̄c 13.637, 8.365 1.301–1.418 [B3O6]3− ions [85]
1 2 5 298 P21 ̸ c 8.424,11.378,13.160, 298 1.439–1.516 1.361–1.379 [B4O5(OH)4]2−ions [86]
92.06
3 7 0 993 C21 ̸ c 59.911, 11.520.34.525, 293 1.42–1.52 1.32–1.42 3-layers, 9xsupercell [87]
101.052
1 3 0 Cesium triborate 1073 P21 21 21 6.18,8.48,9.17 1.45–1.50 1.34–1.44 B3O3 unit [88]
P21 21 21 6.213,8.521,9.170 295 1.458–1.485 1.336–1.399 [89]
3 13 0 1003 C2 ̸c 23.064,13.367, 24.464, 298 1.39–1.56 1.33–1.43 B3O3 and B5O6
90.281 units [90]
1 5 8 298 P21 ̸ c 11.584,7.174,13.959 1.46–1.49 1.33–1.40 [B5O6(OH)4]−
94.61 ions [91]
1 5 0 α-cesium 823 P21 ̸ n 7.117,9.634,10.391 298 1.463–1.495 1.334–1.409 B5O6 units [84]
pentaborate 101.160
α-HT 923 P21 ̸ c 7.122,9.640,11.411 1.466–1.497 1.346–1.387 B5O6 units [92]
116.64
γ-cesium 733 Pbca 8.697, 8.431,21.410 298 1.46–1.48 1.30–1.45 B5O6 units [84]
pentaborate
1 9 0 Cesium 853 P2221 8.768,15.790 1.43–1.46 1.23–1.44 B3O3 units [93]
enneaborate
C. C. Pye
A Crystallographic Review of Alkali Borate Salts … 119

initially to result in partial hydrolysis into discrete structural motifs already present
in the solid state, followed by equilibration in solution.
Ab initio calculations have been carried out on some of the discrete ions seen in
the crystal structures. Gupta and Tossell fixed the symmetry of B2O(OH)4 to C2v
(B-O fixed at 1.353 Å (HF/STO-3G) and 1.375 Å (4-31G), and found that the B-O-B
linkage was bent [94]. They also found B-O distances of 1.430 Å and 1.279 Å for
B3O63− at HF/STO-3G, which compares well with the crystal structure. Zhang et al.
[95] completely optimized B2O(OH)4 with symmetry C2v (HF/STO-3G, 6-31G*), Cs
(HF/6-31G) and C2 (HF/STO-3G, 3-21G*, 4-31G, 6-31G, 6-31G*). They also
calculated the B2O(OH)2− 6 (C2v), B2O(OH)− −
5 (Cs), B3O3(OH)4 (C2v) and B3O3
(OH)5 (Cs) ions at HF/STO-3G. Oi calculated B2O(OH)4 (C2), B2O(OH)−
2−
5 (C1), and
−
B2O(OH)2− 6 (C2) at HF/6-31G* [96]. In addition, Oi also calculated B3O3(OH)4
2− 2− −
(C2), B3O3(OH)5 (C1), B4O5(OH)4 (C2), and B5O6(OH)4 (S4) at HF/6-31G* [97].
A combined Raman and DFT (B3LYP/aug-cc-pVDZ) investigation of B2O(OH)4,
B2O(OH)− 2− − 2− 3−
5 , B2O(OH)6 , B3O3(OH)4 , B3O3(OH)5 , B3O3(OH)6 , B4O5(OH)4 ,
2−
−
and B5O6(OH)4 was presented by Zhou et al [98]. In addition to these ions, two
heptamers B7O9(OH)2− 5 were calculated by Beckett et al. at B3LYP/6-311++G(d, p)
[99].

2 Methods

Calculations were performed using Gaussian 03 [100]. The MP2 calculations use
the frozen core approximation. The geometries were optimized using a stepping
stone approach, in which geometries at the levels HF/6-31G*, HF/6-31+G*, HF/
6-311+G*, B3LYP/6-31G*, B3LYP/6-31+G*, B3LYP/6-311+G*, MP2/6-31G*,
MP2/6-31+G* and MP2/6-311+G* were sequentially optimized, with the geom-
etry and molecular orbitals reused for the subsequent level. Default optimization
specifications were normally used. After each level, where possible, a frequency
calculation was performed at the same level and the resulting Hessian was used in
the following optimization. Z-matrix coordinates constrained to the appropriate
symmetry were used as required to speed up the optimizations. Because frequency
calculations are done at each level, any problems with the Z-matrix coordinates
would manifest themselves by giving imaginary frequencies corresponding to
modes orthogonal to the spanned Z-matrix space. The Hessian was evaluated at the
first geometry (Opt = CalcFC) for the first level in a series in order to aid geometry
convergence.
120 C. C. Pye

3 Results and Discussion

In our previous work, the structure, energy, and vibrational spectra of orthoboric
acid, metaboric acid, and tetrahydroxoborate was thoroughly discussed. We now
focus on the polyborate ions. First we discuss those observed in the crystal structure
of alkali metal borates, followed by structurally related ions. The more
highly-charged ions might be observed in low-water, high ionic-strength environ-
ments. We also discuss other possibilities.

3.1 Triborate Species

3.1.1 [B3O3(OH)4]−

The structure of the crystallographically-observed triborate ion, [B3O3(OH)4]−, is

given in Fig. 1. The ion, present in the sodium borate mineral ameghinite, could be
formed by the addition of hydroxide ion to metaboric acid. Firstly, four different
C2v structures were optimized. None of these was an energy minimum. All possess
an imaginary A2 mode, suggesting desymmetrization to C2. Structures 3 and 4
possess also an imaginary B2 mode, and for some MP2 levels, an imaginary B1
mode, suggesting desymmetrization to Cs. Most of the C2 forms are minima at
some levels. None of the three Cs forms are minima, desymmetrizing instead to C1
#5–7, respectively. C2 #1 was not a minimum at HF/6-31+G* and MP2/6-31+G*,
coalescing into C1 #6. C2 #2 was only a minimum ay B3LYP/6-31G* and MP2/
6-31G*, morphing into C2 #4 at HF/6-31+G* and HF/6-311+G* levels, or
desymmetrizing to C1 #7 at the other levels. C2 #3 was not a minimum at MP2/6-31
+G*, becoming C1 #4. C2 #4 was not stable at the MP2 levels, becoming C1 #2
instead. The order of stability was C1 #6 (0.0 kJ/mol) < C2 #3 (−0.4 to 1.4 kJ/mol)
< C1 #5 (2.8–3.6 kJ/mol) < C2 #1 (4.7–9.8 kJ/mol) < C1 #7 (12–14 kJ/mol) < C2
#4 (11–23 kJ/mol) < C2 #2 (17–24 kJ/mol).

3.1.2 [B3O3(OH)5]2−

The structure of the triborate ion [B3O3(OH)5]2− is given in Fig. 1. Neither of the
two Cs forms is stable and both desymmetrize to the corresponding C1 forms. Of
these, structure #2 is 2.1–3.4 kJ/mol more stable than structure #1.

3.1.3 [B3O3(OH)6]3−

The structure of the triborate ion [B3O3(OH)6]3− is given in Fig. 1. First, two D3h
structures were tried. Neither were minima, and the number of imaginary
A Crystallographic Review of Alkali Borate Salts … 121

C2v #1 C2v #2 C2v #3 C2v #4 Cs #1

C2 #1 C2 #2 C2 #3/C1 #4 C2 #4/C1 #2 Cs #2

C1 #5 C1 #6 C1 #7 Cs #3

Cs #1 Cs #2 C1 #1 C1 #2

D3h #1 D3h #2 D3 C3h C2v #1

Cs #1 Cs #2 C3 #1 C3 #2 C2v #2

C2 #1 C1 #1 C1 #2 C1 #3

Fig. 1 Structure of Triborate Ion, [B3O3(OH)3+n]n−, n = 1 − 3. A bold symmetry label indicates

a minimum energy structure
122 C. C. Pye

frequencies suggested many possible structures of lower symmetry. The two pos-
sible D3 structures (via A1”) coalesced into one, but it was not a minimum. The two
possible C3h structures (via A2’) coalesced into one, and it too is not a minimum.
While desymmetrization along the E” and E’ modes should lead to a C2 and Cs
structure, respectively, it was thought that it would be advantageous to first
desymmetrize to the corresponding C2v structures (which might ascend in symmetry
back to the corresponding D3h structures). Neither of these was a minimum, as
expected, and C2v #1 did indeed ascend to D3h #1 at the HF levels. The D3 and C3h
forms desymmetrize into the stable C3 #1 and #2 forms, respectively, along the A2
and A” modes, respectively. The two possible C2v structures could desymmetrize
into either two C2 structures, or into one of four possible Cs structures. The two C2
structures coalesce into a structure that is only stable at MP2/6-31G*. Cs #3 (from
C2v #1) and Cs #4 (from C2v #2) ascend in symmetry to either C3h or C2v #1.
Neither of the Cs structures are stable. The C2 structure desymmetrizes into C1 #1 at
MP2/6-31+G* and MP2/6-311+G*, and ascends in symmetry to C3 #1 at the HF
and B3LYP levels. The Cs structures desymmetrize to the stable C1 #2 and 3 at
most levels. The C3 #1 structure is the most stable.

3.1.4 [B3O4(OH)4]3−

The structure of the triborate ion [B3O4(OH)4]3− is given in Fig. 2. The C2v
structure has at least three imaginary frequencies of irreducible representation A2,
B1 and B2. These suggest desymmetrization to a C2 or to two different Cs structures,
respectively. None of these are energy minima (except Cs #1 at B3LYP/6-31G*)
and all desymmetrize to C1 structures #1–3, respectively. C1 #1 is the most stable.
This ion has been characterized crystallographically as the sodium and rubidium
salt (see Tables 1 and 4).

3.1.5 [B3O5(OH)2]3−

The structure of the triborate ion [B3O5(OH)2]3− is given in Fig. 2. The two C2v
structures are unstable and all contain an imaginary A2 mode, leading to two C2
structures that coalesce into the most stable minimum. The higher energy C2v #2
structure also contains a B2 mode. The lower energy C2v #1 structure also contains
either a B2 mode (B3LYP/6-31G*) or B1 mode (MP2/6-31+G* and MP2/6-311
+G*). The Cs #1 structure, derived from C2v #2, is only stable at B3LYP/6-31G*.
Cs #2, derived from C2v #1, coalesces into Cs #1. Cs #3, also derived from C2v #1, is
not stable. The C1 #1 structure, derived from Cs #1, coalesces into C2, except at the
stable B3LYP/6-31+G* and B3LYP/6-311+G* levels, whereas the C1 #2 structure,
derived from Cs #3, coalesces into C2 at both levels. This ion has been characterized
as the sodium salt (see Table 1).
A Crystallographic Review of Alkali Borate Salts … 123

C2v C2 Cs #1 Cs #2 C2v #1

D3h C1 #1 C1 #2 C1 #3 C2v #2

C2 Cs #1 Cs #3 C1 #1

Fig. 2 Structure of triborate ion, [B3O3+n(OH)6-2n]3−, n = 1 − 3

3.1.6 [B3O6]3−

The only structure of the triborate ion [B3O6]3−, of D3h symmetry, is given in Fig. 2.
This ion is characterized crystallographically in sodium, potassium, rubidium, and
cesium metaborate (see Tables 1, 2, 3, 4 and 5). There does not appear to be any
systematic trend in the range of the B-O distances upon varying the alkali metal.

3.2 Tetraborate Species

3.2.1 [B4O5(OH)4]2−

The structure of the tetraborate ion, [B4O5(OH)4]2−, is given in Fig. 3. Initially, four
structures of C2v symmetry were tried. None of these were energy minima. All of
these had both an A2 and a B2 imaginary mode, and, at some levels, some have a B1
imaginary mode. The four structures of C2 symmetry derived by desymmetrizing
along the A2 mode were all minima, but only three unique structures were found, as
some coalesced. The four structures of Cs symmetry derived by desymmetrizing
along the B2 mode (#1–#4) coalesced into two different structures, but only Cs #2
was a minima, at all levels except B3LYP/6-31G* and MP2/6-31G*. The four
structures of Cs symmetry derived by desymmetrizing along the B1 mode (#5–#8)
did not give minimum energy structures, and Cs #8 ascended in symmetry to C2v
124 C. C. Pye

#3. Cs #1 desymmetrized to the stable C1 #1. The C2 #2, C2 #3, and C1 #1 are quite
close in energy (5 kJ/mol). This ion has been characterized as a salt of all the alkali
metal ions (except lithium), and the sodium salt is commonly known as borax.
An alternative structure might exist where two of the hydroxyls connect to the
non-bridgehead boron atoms. Initially, four structures of C2v symmetry (#5–#8)
were tried. One of these (C2v #6) was an energy minima at all levels except B3LYP/
6-31G*. For this structure at this level, and for the other three structures, there exists
A2 and B1 imaginary modes. This suggests desymmetrization to C2 #5–#8 and Cs
#9–#12, respectively. The C2 #6 (B3LYP/6-31G* only) and #7 structures are stable.
All other attempts at structures of C2 symmetry revert to either C2 #7 or C2v #6. The
Cs #10 (B3LYP/6-31G*) and Cs #11 is stable. All other attempts at structures of Cs
symmetry revert to either Cs #11 or C2v #6.

3.2.2 [B4O9]6−

Another tetraborate ion, observed in the lithium salt (Table 3), is shown in Fig. 3.
Initially we considered structures of C2h and C2v symmetry, of which there are two
each. None of these is an energy minimum. For C2h #1, there are imaginary modes
of irreducible representation Au (desymmetrization to C2), Bg (desymmetrization to
Ci), and, at MP2/6-311+G*, Bu (desymmetrization to Cs #2). For C2h #2, there are
Au imaginary modes, and at MP2/6-31+G* and MP2/6-311+G*, Bu imaginary
modes (desymmetrization to Cs #3). For C2v #1, there are imaginary modes of
irreducible representation A2 (desymmetrization to C2) and B2 (desymmetrization to
Cs), whereas for C2v #2, there is only an A2 imaginary mode, giving rise to the
stable C2 #4. Desymmetrization of C2h #1 to a Ci structure results in ascent in
symmetry to C2h #2. Desymmetrization of C2v #1 to a Cs structure usually results in
ascent in symmetry to C2h #2, except for B3LYP/6-311+G* and MP2/6-311+G*,
where the Cs #1 structure remains. None of the three Cs structures are energy
minima. The Cs #1 structure desymmetrizes to the stable C1 #1, whereas attempts to
desymmetrize Cs #2 and Cs #3 result in ascent in symmetry to C2 #4.

3.3 Pentaborate Species

3.3.1 [B5O6(OH)4]−

The pentaborate ion, [B5O6(OH)4]−, observed as a sodium, potassium, rubidium,

and cesium salt (Tables 1, 2, 4, and 5), is given in Fig. 4. Initially, two structures of
D2d symmetry were tried, and these were indeed minima at most levels. In addition,
other structures formed from rotating hydroxyls by 180 degrees were also minima
at most if not all levels. These are labelled C2v #1, C2 #1, Cs #1, and Cs #2. The only
possible levels where these structures are not minima are MP2/6-31+G* and MP2/
6-311+G*. Structure D2d #1 has an imaginary E mode at MP2/6-31+G*, whereas
A Crystallographic Review of Alkali Borate Salts … 125

C2v #1 C2v #2 C2v #3 C2v #4

C2 #1/C1 #4 C2 #2 C2 #3 Cs #1 Cs #2/C1 #2

Cs #5 Cs #6 Cs #7 C1 #1 C2v #5

C2v/2 #6/
C2v #7 C2v #8 C2 #7 Cs #11
Cs #10

C2h #1 C2h #2 C2v #1 C2v #2

C2 #4 Cs #1 Cs #2 Cs #3 C1 #1

Fig. 3 Structure of tetraborate ions, [B4O5(OH)4]2− and [B4O9]6–

structure D2d #2 has an imaginary E mode at both MP2/6-31+G* and MP2/6-311

+G*. These structures could either desymmetrize to C2 #2/3 (stable), or Cs #3/4
(unstable), both of which have slightly puckered ring(s). Desymmetrization of these
Cs #3 and #4 structures via C1 #3/4 results in ascent in symmetry to C2 #2/3. C2 #1
is stable at all levels. C2v #1 has an imaginary B1 mode at both MP2/6-31+G* and
MP2/6-311+G*, and an imaginary B2 mode at MP2/6-31+G*. These lead to Cs #5
and 6 respectively. Cs #5 is stable at MP2/6-311+G*. Both Cs structures are
unstable at MP2/6-31+G*, leading to stable C1 #5 and 6 structures. Cs #1 and #2
are unstable at MP2/6-31+G*, converting into C1 #1 and #2. The energy ordering is
D2d #1 < Cs #1 < C2 #1 < C2v #1 < Cs #2 < D2d #2.
126 C. C. Pye

3.3.2 [B5O6(OH)6]3−

The pentaborate ion, [B5O6(OH)6]3−, is given in Fig. 4. We initially started with two
C2v structures and two C2 structures. Neither C2v structure is stable. Both have
imaginary A2, B1, and B2 modes at most levels. Both C2 structures are stable at most
levels, with the exception of B3LYP/6-311+G* (C2 #1) and B3LYP/6-31+G* (C2
#2), which desymmetrize to stable C1 #1 and #2, respectively. The C2 #3 and #4
structures, derived from C2v #1 and #2, are unstable at all levels and desymmetrize to
the stable C1 #3 and #4. At B3LYP/6-31G* and MP2/6-31G*, one of the central B-O
bonds in C1 #3 breaks. Attempts to locate the Cs #1 and #2 structures also result in
breaking of one of the central B-O bonds. The Cs #3 and #4 structures are not stable,

C2v #1/Cs #5/

D2d #1/C2 #2 D2d #2/C2 #3 C2 #1 Cs/1 #1
C1 #5/6

Cs/1 #2 C2/1 #1 C2/1 #2

C2v #1 C2v #2 C2 #3 C2 #4 Cs #3

Cs #4 C1 #3 C1 #4 C1 #5 C1 #6

D2d S4 D2 Cs

C1 #1 C2 #2 C2 #3 C1 #2

Fig. 4 Structure of pentaborate ions, [B5O6(OH)4+2n]−1−2n, n = 0 − 2

A Crystallographic Review of Alkali Borate Salts … 127

and desymmetrize to C1 #5 and #6, respectively. Structure C1 #5 fragments at all levels

except HF/6-31+G*, but structure C1 #6 is either stable (B3LYP/6-31+G*, MP2/6-31
+G*, and MP2/6-311+G*) or coalesces into C1 #4.

3.3.3 [B5O6(OH)8]5−

The pentaborate ion, [B5O6(OH)8]5−, is given in Fig. 4. We initially started with a

D2d structure, but this was not stable and possessed imaginary B1 (all levels), A2
(HF and MP2), and E (MP2/6-31G* and MP2/6-311+G*) modes, to give D2, S4,
and Cs structures, respectively. Both the D2 and S4 structures are stable at most
levels examined. The D2 is unstable at HF/6-31+G*, reverting via C2 #1 to S4, and
at MP2/6-311+G* giving the stable C2 #2 or C2 #3. The S4 structure is unstable at
MP2/6-31G* and MP2/6-311+G*, desymmetrizing to the stable C1 #1. The Cs
structure is unstable at both levels and desymmetrizes to the stable C1 #2.

3.4 Diborate Species

3.4.1 [B2O5]4−

The diborate ion, [B2O5]4−, crystallographically observed as the lithium salt

(Table 3), is given in Fig. 5. Initially, structures of D2d and D2h symmetry were

D2d C2 #1 Cs #1 D2h

C2v #1 C2v #2 C2h

C2v #1 C2 #1 C2 #2 Cs #1 Cs #2

C2v #2 C1 #1 C1 #2 C1 #3 C1 #4

Fig. 5 Structure of diborate ions, [B2O5]4− and [B2(OH)7]−

128 C. C. Pye

tried. The D2d structure was stable at B3LYP/6-31G* and MP2/6-31G*, but had an
imaginary E mode at the other levels, suggesting desymmetrization to either a C2 #1
or Cs #1. The C2 #1 is stable at all levels except HF/6-31+G* and HF/6-31+G*, for
which the Cs #1 is stable. The D2h structure is unstable at all levels. Desym-
metrization along an imaginary Au mode at all levels results in ascension in sym-
metry to the D2d structure. Desymmetrization along the B2u, B3u, and B3g modes
give unstable C2v #1 (all levels except B3LYP/6-31G*), C2v #2 (MP2/6-31+G* and
MP2/6-311+G*), and C2h (B3LYP/6-31+G* and B3LYP/6-311+G*) structures.
Desymmetrization of these structures leads to structures already observed.

3.4.2 [B2(OH)7]−

In the polyborate structures discussed thus far, any oxygen that is bound to two
boron atoms is not bound to a hydrogen. However, the diborate species [B2(OH)7]−
with a bridging hydroxide was postulated as the intermediate to explain the boric
acid–borate interchange in aqueous solution as studied by 11B NMR [101]. Such an
intermediate may lie on the reaction path to condensation to form the other
polyborates. It is expected that the hydrogen of the bridging oxygen would be quite
acidic and easily lost.
The diborate ion, [B2(OH)7]−, is given in Fig. 5. Initially, two C2v structures
were tried. These had numerous imaginary A2, B1, and B2 frequencies, which
suggested desymmetrization to two C2 and four Cs structures. Neither C2 structure
was stable, possessing an imaginary B mode. The Cs #1 and Cs #2 structures,
derived from C2v #1, were not stable, containing A” imaginary frequencies. The Cs
#3 and Cs #4 structures, derived from C2v #2, either dissociated into a hydrogen
bonded [B(OH)4]−…B(OH)3 hydrogen bonded complex (Cs #3) or coalesced into
Cs #2. The C2 structures desymmetrized into the corresponding stable C1 #1 and #2
structures. The Cs structures desymmetrized into the corresponding stable C1 #3 and
#4 structures. The four stable structures are quite close in energy, and the gas-phase
association energy of boric acid and borate is in the range −47.9 to 108.5 kJ/mol.

3.4.3 [B2(OH)6]0

We were curious to see whether a dimer of boric acid could exist possessing a
single hydroxyl bridge. The bridge necessarily converts one of the boron atoms into
tetrahedral, and under the constraint of Cs symmetry, results in eight possible
structures. None of these are stable at the HF/6-31G* level, resulting in fragmen-
tation of the B-O bond to give hydrogen bonded structures. In addition, structure Cs
#1 is unstable at all levels investigated. These results suggest that such a structure
does not exist.
Next, a dimer of boric acid containing two bridging hydroxyls was investigated.
The optimized structures are shown in Fig. 6. Initially two structures of D2h
symmetry were considered. Neither of these structures were stable, possessing
A Crystallographic Review of Alkali Borate Salts … 129

D2h #1 D2 #1 C2h #4 C2h #3 D2h #2

C2h #1 C2v #1 D2 #2 C2v #2

Ci #1 C2 #1 Cs #10 C2 #5 Cs #9

C2v/2 #4 C2v #3 C1 #2

C2v #1 C2v #2 C2v #3 Cs/1 #1 Cs/1 #2

C2 #1 C2 #2 C2 #3 Cs/1 #3 Cs/1 #4

Cs #7 Cs #8 Cs #9 Cs/1 #5 Cs/1 #6

Fig. 6 Structure of diboric acid, [B2(OH)6]0, and [B2(OH)5]+

imaginary B3g and B2u modes at all levels, suggesting desymmetrization to the C2h
#1 and #3, and C2v #1 and #2 structures. The D2h #2 structure possessed an Au
mode at all levels (as did D2h #1 at MP2/6-311+G*), suggesting desymmetrization
to D2 #2 (D2 #1). The D2h structures also contained a B2g imaginary mode at some
levels suggesting desymmetrization (D2h #1: B3LYP/6-311+G* to C2h #2; D2h #2:
HF/6-311+G* to C2h #4).
To our surprise, the C2h #1 structure was stable at all B3LYP levels and at MP2/
6-31+G*, but possessed a Bg imaginary mode at the other levels, suggesting
130 C. C. Pye

desymmetrization to Ci #1. The C2h #3 structure was stable at all levels! The C2h #2
structure (B3LYP/6-311+G*) breaks apart into two boric acid molecules. The C2v
#1 structure has an A2 imaginary mode at all levels leading to C2 #1. The D2 #1
structure (MP2/6-311+G*) has an imaginary B3 and B2 mode, leading to C2 #2 and
C2 #3. The C2v #2 structure has an A2 and B1 imaginary mode at all levels leading to
C2 #4 and Cs #9. The D2 #2 structure has an imaginary B3 and B2 modes at all
levels, leading to C2 #5 and C2 #6. The C2h #4 structure (HF/6-311+G*) is
unstable, possessing both a Bg and Au mode, suggesting desymmetrization to Ci #2
and C2 #7.
The Ci #1 structure is stable at all levels except HF/6-31G* and HF/6-311+G*,
where it dissociated. The C2 #1 structure is only stable at HF/6-31+G*. It disso-
ciates at HF/6-31G*, HF/6-311+G*, and B3LYP/6-311+G*. At the other levels, it
ascends in symmetry via C1 #1 to the new Cs #10 structure, which exists at all
levels. The C2 #2 structure ascends in symmetry to C2h #1. The C2 #3 structure
coalesces to C2 #1. The C2 #4 structure ascends in symmetry to the new C2v #3. The
unstable Cs #9 structure only exists at MP2/6-31G*, ascending in symmetry to C2v
#3 otherwise. It desymmetrizes to the stable C1 #2 at MP2/6-31G*. The C2 #5
structure is stable at all levels, but the C2 #6 structure ascends in symmetry to C2v
#3. The Ci #2 and C2 #7 ascend in symmetry to C2h #3 and C2v #3, respectively.
To summarize these results, there are at least four stable structures at all levels.
Their energy ordering is as follows: C2v #3 (0.0 kJ/mol) < C2h #3 (5.7–9.8 kJ/mol)
< Cs #10 (8.1–13.1 kJ/mol) < C2 #5 (16.4–21.6 kJ/mol). The endothermic
gas-phase dimerization energy of boric acid lies in the range 35.8–103.8 kJ/mol.

3.4.4 [B2(OH)5]+

Another structure of interest is the cationic [B2(OH)5]+ (Fig. 6). Initially we tried
four structures of C2v symmetry. None of these was a minimum except for C2v #4 at
all levels except MP2/6-311+G*. All non-minima had imaginary A2 modes, and in
some cases, B1 modes. Deymmetrization along the A2 modes led to the stable C2
#1–#4 structures. Deymmetrization along the B1 modes led to the unstable Cs #7–
#9 structures. In addition, there are six other Cs structures obtainable by flipping the
hydrogen atoms. The Cs #1 and #3 structures are unstable at all levels. The Cs #2
structure is only stable at HF/6-31G*, HF/6-31+G*, and B3LYP/6-31+G*. The Cs
#4, #5, and #6 structures are stable at all levels except MP2/6-311+G*. These Cs
structures desymmetrize into the corresponding C1 #1–#9 structures. The C1 #1–#6
structures are stable, but the C1 #7–#9 structures convert to other structures already
obtained. Of these structures, Cs #4 is the most stable, followed by C2v #4 (8.0–
10.2 kJ/mol). The structure (with hydroxide) is thermodynamically unstable rela-
tive to two boric acid molecules (880–1020 kJ/mol).
A Crystallographic Review of Alkali Borate Salts … 131

3.4.5 [B2O(OH)6]2−

The diborate ion [B2O(OH)6]2− has been observed as the magnesium salt in the
mineral pinnoite. Stadler found that the space group of pinnoite was P42 or P42/m,
with a = 7.617(2) and c = 8.190(2), and proposed that the ion was [B2O(OH)6]2−
[102]. The space group was confirmed as P42 (a = 7.62(1), c = 8.19(1) by Paton
and MacDonald and the structure of the ion confirmed [103]. The structure was
further refined by Krogh-Moe [104]. Its structure is shown in Fig. 7. We initially
tried two C2v structures. Both were unstable and possessed A2 and B2 imaginary
modes. In addition C2v #1 possessed an imaginary B1 mode. The two C2 structures
thus derived were stable (with C2 #2 coalescing into C2 #1 at B3LYP/6-31G* and
MP2/6-31G*). None of the three Cs structures were stable, and desymmetrized to
the corresponding C1 structures. C1 #3 coalesced into C1 #1 at B3LYP/6-31G* and
MP2/6-31G*. Other possibilities include two Cs structures (#4 and #5), related to
the C2v structures by rotation of one of the hydroxyls. Neither of these are stable,
and desymmetrize to the corresponding C1 structures.

3.4.6 [B2O(OH)5]−

The diborate ion [B2O(OH)5]− was initially considered to contain a only a single
oxo bridge (Fig. 8). Eight such structures of Cs symmetry were considered, with the
BO3 unit in the plane of symmetry. None of these was stable and led to the
corresponding C1 structures. In some cases, these coalesced. It could also poten-
tially exist as a (μ-O)(μ-OH) doubly bridged dimer of C2v symmetry. This structure
is unstable, and has imaginary A2, B1, and B2 modes leading to C2, Cs #9, and Cs
#10 structures. In Cs #10, one of the B-O(H) bonds has broken. All of these
structures also have imaginary modes, leading potentially to C1 #9−#11,

C2v #1 C2 #1 C2 #2 C2v #2

Cs #1 Cs #2 Cs #3 Cs #4 Cs #5

C1 #1 C1 #2 C1 #3 C1 #4 C1 #5

Fig. 7 Structure of diborate ion, [B2O(OH)6]2−

132 C. C. Pye

Cs #1 Cs #2 Cs #3 Cs #4 Cs #5

Cs #6 Cs #7 Cs #8 C1 #1 C1 #2

C1 #3 C1 #4 C1 #5 C1 #6 C1 #7

C1 #8 C2v #1 C2 #1 Cs #9 Cs #10

Cs #11 Cs #12 Cs #13 Cs #14 Cs #15

C1 #9 C1 #12 C1 #13 C1 #14 C1 #15

Fig. 8 Structure of diborate ion, [B2O(OH)5]−

respectively. Only C1 #9 is stable, with C1 #10 and #11 coalescing into C1 #6 and
#10, respectively. The bond breaking in Cs #10 suggests that we try three additional
Cs structures (#11-#13) in which the in plane OH hydroxyls have rotated 180
degrees. To our surprise, Cs #11 and #13 resulted in inversion about the bridging
oxygen. These structures, in turn, suggested trying two other Cs structures (#14 and
#15), keeping the inverted oxygen but placing the hydrogen in its original position.
None of these Cs structures were stable. Desymmetrization of Cs #11–15 to C1 #12–
16 gives stable structures at some levels, or coalescence to others.

3.4.7 [B2O(OH)4]0

A dehydrated form of boric acid dimer with a bridging oxo group was considered
(Fig. 9). The high-symmetry structures D2h #1 and #2 has B2u, Au, and B3u
A Crystallographic Review of Alkali Borate Salts … 133

D2h #1 D2d #1 Cs #12 D2d #2 D2h #2

C2v #1/Cs #7 C2v #5 Cs #13 C2v #6 C2v #4/Cs #9

C2 #1 C2v #2/Cs #8 C2v #3 C2 #4

C2 #2 C2 #3 Cs/1 #6

Cs/1 #1 Cs/1 #2 Cs/1 #3 Cs/1 #4 Cs/1 #5

Fig. 9 Structure of diboric acid, [B2O(OH)4]0

imaginary frequencies, suggesting desymmetrization to C2v #1/4, D2 #1/2, and C2v

#5/6, respectively. The D2d #1 and #2 had an imaginary E mode, suggesting
desymmetrization to C2 #7/8 or Cs #12/13. Two additional C2v structures were also
tried (#2, #3). The D2 structures ascended in symmetry to the corresponding D2d
structures. None of the C2v structures #1-#4 was stable. All had an imaginary A2
mode, which led to the stable C2 structures #1–#4. In some cases at some levels,
there was also an imaginary B1 mode leading to a Cs structure (#7–#9) which were
unstable and desymmetrized, via the corresponding C1 structure, to one of the
observed C2 structures. The C2v #5 and #6 structures had imaginary B1 and A2
modes. The putative C2 #5 and #6 derived along the A2 mode coalesced into C2 #1
and #4, whereas the putative Cs #10 and #11 derived along the B1 mode ascended in
symmetry or coalesced to C2v #1/Cs #7 and C2v #4. Six additional Cs structures
were optimized (#1–#6). Most (#2, #4, #5, and #6) were stable at all levels except
MP2/6-311+G*. Unstable Cs structures desymmetrized to the corresponding C1
structures (#1–#6). The Cs #12/13 structures derived from the D2d #1 and #2
structures were not stable. When these were desymmetrized (C1 #10, #11), they
ascended in symmetry to C2 #1 and C2 #4, respectively. The C2 #7/8 structures
coalesced into C2 #1/4 as well.
134 C. C. Pye

3.4.8 [B2O2(OH)4]2–

Structures containing two oxo bridges between the two boron atoms are now
considered. The first such structure, [B2O2(OH)4]2–, can be considered to be the
dimer of the hypothetical oxodihydroxoborate ion, [BO(OH)2]– (Fig. 10). First, the
high symmetry D2h #1 and #2 structures were considered. These had imaginary Au,
B3g, B2g, and (#2) B1u modes, leading to the potential D2 #1/2, C2h #1/3, C2h #2/4,
and C2v structures.
The D2 #1 structure is stable at all levels, whereas D2 #2 is only stable at
B3LYP/6-31G* and MP2/6-31G*. It has imaginary B2 and B1 modes at the other
levels, giving stable C2 #3 and #4 structures. The C2v structure, derived from D2h
#2, ascends in symmetry to the D2h #1 structure. The C2h #1 structure is stable at all
levels except B3LYP/6-311+G* and the HF levels, for which it has an Au mode to
desymmetrize to the putative C2 #1 structure, which ascends in symmetry to D2 #1.
The C2h #2 structure, which exists at all levels except HF/6-31+G* and HF/6-311
+G*, has an Au and Bg mode at all levels, suggesting desymmetrization to C2 #2
and Ci #1, respectively. These ascend in symmetry to D2 #1 and C2h #1, respec-
tively. The C2h #3 structure, which exists for calculations with diffuse basis sets, has
imaginary Bg and Bu modes, desymmetrizing to Ci #2 and Cs #1, respectively,
which are stable at all levels, with the exception of Cs #1 at HF/6-31+G*, which
ascends in symmetry via C1 #1 to C2 #3. The C2h #4 structure coalesced with either
the C2h #2 or D2h #1 structure.

D2h #1 D2 #1 C2h #3 D2h #2

C2h #1 C2h #2 Cs #1 Ci #2 D2 #2

C2h C2v C2 #3 C2 #4

C2v #1 Cs #2 C1 #4 C2 #2 C2v #2

C2 #1 Cs #1 C1 #1 C1 #5 Cs #4

Fig. 10 Structure of diborate ions, [B2O2(OH)4-n]−2+n, n = 0 − 2

A Crystallographic Review of Alkali Borate Salts … 135

3.4.9 [B2O2(OH)4]−

The diborate ion, [B2O2(OH)3]−, is shown in Fig. 10. Initially, two C2v structures
were tried. Both had imaginary B1, B2, and A2 modes. These lead to Cs #1/3, Cs #2/
4, and C2 #1/2, respectively. The structure Cs #3 coalesced into Cs #1. None of
these structures were stable. The structure Cs #1 desymmetrized to C1 #1. The
structure Cs #2, desymmetrized via C1 #2 to coalesce with C1 #1. The structure C2
#1 also desymmetrized via C1 #3 to coalesce with C1 #1. The structure C2 #2
desymmetrized to C1 #4. The structure Cs #4 desymmetrized via C1 #5 to coalesce
to C1 #4 (except at MP2/6-311+G*).

3.4.10 [B2O2(OH)4]0

The diboric acid, [B2O2(OH)2]0, which is an isomer of metaboric acid, is also given
in Fig. 10. The two forms examined, C2h and C2v, were both minima, with C2h
slightly lower in energy.

3.5 Structural Comparisons

A comparison of the calculated ab initio with the crystallographically observed

boron-oxygen bond lengths is given in Table 6. For ease of reporting, we take
high-symmetry structures in some cases and average the crystal structures
according to the approximate symmetry mentioned. Inspection of Table 6 reveals
that in most instances, calculations with the 6-31+G* and 6-311+G* basis sets are
within 0.01 Å of each other, the exception being [B4O5(OH)4]2−, where the con-
formation of the hydroxyl on the tetrahedral boron is significantly different. For the
singly and doubly charged anions, the 6-31G* results are, with one exception,
within 0.01 Å of the 6-31+G* results, but the more highly charged anions show
more deviation. The B3LYP and MP2 calculations are usually within 0.01 Å of
each other, except for the highly charged [B4O9]6− anion. The Hartree-Fock cal-
culations tend to be slightly shorter. There seems to be mostly good agreement
between the crystal structures and the correlated calculations with a diffuse basis
set, with the experiments usually being slightly shorter. The agreement between
experiment and theory here is rather remarkable in the sense that the calculations
correspond to a gas-phase ion (no medium effects were included in the calculation),
whereas the experiment corresponds to the solid-state, surrounded by counterions.
We may compare our results to previous ab initio calculations (see Introduction).
Although Zhang et al. [95] completely optimized B2O(OH)4 with symmetry C2v
(our #3), Cs, and C2 (our #3), our lowest energy structure corresponds to a planar Cs
structure with an internal hydrogen bond. However, in most cases, the previous
calculations, if they exist, are comparable to ours.
Table 6 Theoretical and literature experimental B-O distances (Å) in crystallographically observed Polyborate Ions
136

Species X-ray HF B3LYP MP2

6-31G* 6-31+G* 6-311+G* 6-31G* 6-31+G* 6-311+G* 6-31G* 6-31+G* 6-311+G*
4−
B2O5 D2d [57] 1.340 1.376 1.386 1.384 1.39 1.399 1.39 1.398 1.415 1.407
1.454 1.485 1.438 1.442 21.526 1.440 51.445 1.528 1.443 1.445
B2O(OH)2−
6 C2v #1 [103] 1.42,1.46, 1.431 1.431 1.429 1.444 1.445 1.443 1.447 1.449 1.443
1.54,1.56 1.448 1.449 1.448 1.461 1.464 1.462 1.466 1.469 1.463
1.569 1.565 1.565 1.579 1.576 1.578 1.574 1.568 1.563
B3O63− D3h [8] 1.280 1.305 1.314 1.311 1.320 1.332 1.328 1.326 1.342 1.335
1.433 1.421 1.417 1.416 1.440 1.433 1.433 1.445 1.440 1.435
B3O5(OH)3−
2 C2 #1 [7] 1.326 1.307 1.318 1.315 1.320 1.335 1.331 1.327 1.345 1.339
1.39 1.417 1.410 1.409 1.439 1.427 1.426 1.443 1.432 1.427
1.42 1.422 1.419 1.418 1.440 1.435 1.434 1.446 1.442 1.437
1.47 1.446 1.447 1.446 1.461 1.461 1.460 1.467 1.468 1.463
1.488 1.528 1.523 1.522 1.544 1.541 1.541 1.542 1.539 1.532
B3O4(OH)3−
4 C2v [6] 1.340 1.309 1.323 1.319 1.322 1.340 1.336 1.329 1.351 1.344
1.397 1.418 1.411 1.410 1.438 1.428 1.427 1.442 1.433 1.428
1.445 1.441 1.438 1.438 1.455 1.450 1.449 1.459 1.455 1.452
1.487 1.447 1.453 1.453 1.458 1.468 1.468 1.465 1.477 1.470
1.493 1.528 1.523 1.523 1.546 1.541 1.541 1.543 1.540 1.533
B3O3(OH)−
4 C2 #3 [30] 1.344 1.327 1.329 1.326 1.341 1.343 1.341 1.326 1.350 1.344
1.362 1.373 1.373 1.371 1.388 1.388 1.386 1.393 1.394 1.389
1.396 1.381 1.380 1.377 1.396 1.396 1.393 1.400 1.400 1.394
1.452 1.452 1.454 1.452 1.464 1.468 1.466 1.466 1.471 1.464
1.503 1.484 1.482 1.483 1.498 1.496 1.497 1.500 1.499 1.495
(continued)
C. C. Pye
Table 6 (continued)
Species X-ray HF B3LYP MP2
6-31G* 6-31+G* 6-311+G* 6-31G* 6-31+G* 6-311+G* 6-31G* 6-31+G* 6-311+G*
6−
B4O9 C2h #1 [58] 1.321 1.323 1.331 1.328 1.339 1.345 1.339 1.344 1.354 1.346
1.344 1.359 1.373 1.370 1.373 1.374 1.368 1.378 1.394 1.385
1.349 1.366 1.387 1.385 1.377 1.394 1.387 1.384 1.429 1.424
1.390 1.444 1.442 1.443 1.467 1.448 1.452 1.472 1.459 1.461
1.394 1.478 1.447 1.451 1.513 1.453 1.452 1.517 1.476 1.474
1.434 1.548 1.481 1.484 1.607 1.473 1.469 1.608 1.485 1.485
B4O5(OH)2−
4 C2 #1 [15] 1.3655 1.334 1.350 1.335 1.351 1.350 1.347 1.354 1.357 1.353
1.3737 1.348 1.415 1.344 1.362 1.366 1.364 1.370 1.373 1.364
1.3784 1.418 1.416 1.415 1.435 1.434 1.431 1.438 1.438 1.431
1.4451 1.436 1.438 1.419 1.452 1.454 1.452 1.454 1.459 1.432
1.4657 1.463 1.461 1.456 1.476 1.476 1.473 1.477 1.478 1.468
A Crystallographic Review of Alkali Borate Salts …

1.4902 1.488 1.485 1.515 1.509 1.501 1.499 1.505 1.503 1.528
1.5075 1.515 1.514 1.518 1.520 1.527 1.534 1.529 1.530 1.533
B5O6(OH)−
4 D2d #1 [34] 1.356 1.332 1.333 1.331 1.347 1.348 1.346 1.352 1.355 1.350
1.367 1.371 1.371 1.370 1.386 1.386 1.384 1.391 1.393 1.387
1.383 1.375 1.374 1.372 1.389 1.390 1.387 1.393 1.394 1.388
1.468 1.464 1.464 1.464 1.476 1.477 1.478 1.478 1.479 1.476
137
138 C. C. Pye

4 Conclusions

Boron, when surrounded by oxygen, can be either trigonal planar or tetrahedral. An

examination of existing crystal structures of the alkali metal salts reveal a wide
range of possible ways in which these may condense together to form B2O,
cyclo-B3O3, bicyclo-B4O5, and spirobicyclo-B5O6 units. In some cases, discrete
ions are formed. The structures of these ions have been calculated using
Hartree-Fock, density functional, and Moller Plesset theory with modest basis sets
and the predicted gas-phase geometries are shown to agree quite well with the
crystal structures. In addition, several other structurally-related ions and neutral
molecules have been shown to be minima on the potential energy surface and, while
as yet unobserved, could potentially exist.

Acknowledgements CCP acknowledges ACENet for computational facilities.

References

1. Pye CC (2018) An ab initio study of boric acid, borate, and their interconversion. Prog Theor
Chem Phys 31:143–177
2. Konig H, Hoppe R (1977) Zur Kenntnis von Na3BO3. Z Anorg Allg Chem 434:225–232 (in
German)
3. Menchetti S, Sabelli C (1982) Structure of hydrated sodium borate Na2[BO2(OH)]. Acta
Crystallogr B 38:1282–1284
4. Block S, Perloff A (1963) The direct determination of the crystal structure of NaB
(OH)4.2H2O. Acta Crystallogr 16:1233–1238
5. Csetenyi LJ, Glasser FP, Howie RA (1993) Structure of sodium tetrahydroxyborate. Acta
Crystallogr C 49:1039–1041
6. Andrieux J, Goutaudier C, Laversenne L, Jeanneau E, Miele P (2010) Synthesis,
characterization, and crystal structure of a new trisodium triborate, Na3[B3O4(OH)4]. Inorg
Chem 49:4830–4835
7. Corazza E, Menchetti S, Sabelli C (1975) The crystal structure of Na3[B3O5(OH)2]. Acta
Crystallogr B 31:1993–1997
8. Marezio M, Plettinger HA, Zachariasen WH (1963) The bond lengths in the sodium
metaborate structure. Acta Crystallogr 16:594–595
9. Menchetti S, Sabelli C (1979) A new borate polyanion in the structure of Na8[B12O20(OH)4].
Acta Crystallogr B 35:2488–2493
10. Menchetti S, Sabelli C (1977) The crystal structure of synthetic sodium pentaborate
monohydrate. Acta Crystallogr B 33:3730–3733
11. Menchetti S, Sabelli C, Trosti-Ferroni R (1982) The structure of sodium borate Na3[B5O9].
H2O. Acta Crystallogr B 38:2987–2991
12. Font Tullot JM (1947) El borax como estructura cristalina en cadenas. Estud Geol 7:13–20
(in Spanish)
13. Norimoto N (1956) The crystal structure of borax. Mineral J 2:1–18
14. Levy HA, Lisensky GC (1978) Crystal structures of sodium sulfate decahydrate (Glauber’s
Salt) and sodium tetraborate decahydrate (Borax). Redetermination by neutron diffraction.
Acta Crystallogr B 34:3502–3510
15. Gainsford GJ, Kemmitt T, Higham C (2008) Redetermination of the borax structure from
laboratory X-ray data at 145 K. Acta Crystallogr E 64:i24–i25
A Crystallographic Review of Alkali Borate Salts … 139

16. Giacovazzo C, Menchetti S, Scordari F (1973) The crystal structure of tincalconite. Am

Mineral 58:523–530
17. Powell DR, Gaines DF, Zerella PJ, Smith RA (1991) Refinement of the structure of
tincalconite. Acta Crystallogr C 47:2279–2282
18. Luck RL, Wang G (2002) On the nature of tincalconite. Am Mineral 87:350–354
19. Ross V, Edwards JO (1959) On the crystal structure of kernite, Na2B4O7.4H2O. Acta
Crystallogr 12:258
20. Giese RF Jr (1966) Crystal structure of kernite, Na2B4O6(OH)2•3H2O. Science 154:1453–
1454
21. Cialdi G, Corazza E, Sabelli C (1967) La struttura cristallina della kernite, Na2B4O6(OH)2
3H2O. Rend. Lincei Sc Fis Mat Nat Ser 8, 42:236–251 (in Italian)
22. Cooper WF, Larsen FK, Coppens P, Giese RF (1973) Electron population analysis of
accurate diffraction data. V. Structure and one-center charge refinement of the light-atom
mineral kernite, Na2B4O6(OH)2•3H2O. Am Mineral 58:21–31
23. Menchetti S, Sabelli C (1978) The crystal structure of Na2[B4O6(OH)2]. Acta Crystallogr B
34:1080–1084
24. Krogh-Moe J (1974) The crystal structure of sodium diborate, Na2O.2B2O3. Acta
Crystallogr B 30:578–582
25. Cannillo E, Dal Negro A, Ungaretti L (1973) The crystal structure of ezcurrite. Am Mineral
58:110–115
26. Corazza E, Menchetti S, Sabelli C (1975) The crystal structure of nasinite,
Na2[B5O8(OH)].2H2O. Acta Crystallogr B 31:2405–2410
27. Wang Y, Pan S, Tian X, Zhou Z, Liu G, Wang J, Jia D (2009) Synthesis, structure, and
properties of the noncentrosymmetric hydrated borate Na2B5O8(OH)•2H2O. Inorg Chem
48:7800–7804
28. Liu Z-H, Li L-Q, Wang M-Z (2006) Synthesis, crystal structure and thermal behavior of
Na4[B10O16(OH)2]•4H2O. J Alloys Compd 407:334–339
29. Corazza E, Menchetti S, Sabelli C (1974) The crystal structure of biringuccite,
Na4[B10O16(OH)2] •2H2O. Am Mineral 59:1005–1015
30. Dal Negro A, Martin Pozas JM, Ungaretti L (1975) The crystal structure of ameghinite. Am
Mineral 60:879–883
31. Krogh-Moe J (1974) The crystal structure of α sodium triborate, α-Na2O.3B2O3. Acta
Crystallogr B 30:747–752
32. Krogh-Moe J (1972) The crystal structure of a sodium triborate modification,
β-Na2O.3B2O3. Acta Crystallogr B 28:1571–1576
33. Hyman A, Perloff A, Mauer F, Block S (1967) The crystal structure of sodium tetraborate.
Acta Crystallogr 22:815–821
34. Merlino S, Sartori F (1972) The crystal structure of sborgite, NaB5O6(OH)4.3H2O. Acta
Crystallogr B 28:3559–3567
35. Menchetti S, Sabelli C (1978) The crystal structure of NaB5O6(OH)4. Acta Crystallogr B
34:45–49
36. Zachariasen WH (1937) The crystal structure of potassium metaborate, K3(B3O6). J Chem
Phys 5:919–922
37. Schneider W, Carpenter GB (1970) Bond lengths and thermal parameters of potassium
metaborate, K3B3O6. Acta Crystallogr B 26:1189–1191
38. Marezio M, Plettinger HA, Zachariasen WH (1963) The crystal structure of potassium
tetraborate tetrahydrate. Acta Crystallogr 16:975–980
39. Krogh-Moe J (1961) Unit-cell data for some anhydrous potassium borates. Acta Crystallogr
14:68
40. Krogh-Moe J (1972) The crystal structure of potassium diborate, K2O.2B2O3. Acta
Crystallogr B 28:3089–3093
41. Marezio M (1969) The crystal structure of K2[B5O8(OH)].2H2O. Acta Crystallogr B
25:1787–1795
140 C. C. Pye

42. Zhang H-X, Zhang J, Zheng S-T, Yang G-Y (2005) Two new potassium borates,
K4B10O15(OH)4 with stepped chain and KB5O7(OH)2•H2O with double helical chain. Cryst
Growth Des 5:157–161
43. Salentine CG (1987) Synthesis, characterization, and crystal structure of a new potassium
borate, KB3O5•3H2O. Inorg Chem 26:128–132
44. Wang G-M, Sun Y-Q, Zheng S-T, Yang G-Y (2006) Synthesis and crystal structure of a
novel potassium borate with an unprecedented [B12O16(OH)8]4− anion. Z Anorg Allg Chem
632:1586–1590
45. Li H-J, Liu Z-H, Sun L-M (2007) Synthesis, crystal structure, vibrational spectroscopy and
thermal behavior of the first alkali metal hydrated hexaborate: K2[B6O9(OH)2]. Chin J Chem
25:1131–1134
46. Bubnova RS, Fundamenskii VS, Filatov SK, Polyakova IG (2004 ) Crystal structure and
thermal behavior of KB3O5. Doklady Phys Chem 398:249–253
47. Krogh-Moe J (1974) The crystal structure of pentapotassium enneakaidekaborate,
5K2O.19B2O3. Acta Crystallogr B 30:1827–1832
48. Zachariasen WH (1937) The crystal structure of potassium acid dihydronium pentaborate
KH2(H3O)2B5O10, (Potassium Pentaborate Tetrahydrate). Z Kristallogr 98:266–274
49. Cook WR Jr, Jaffe H (1957) The crystallographic, elastic, and piezoelectric properties of
ammonium pentaborate and potassium pentaborate. Acta Crystallogr 10:705–707
50. Zachariasen WH, Plettinger HA (1963) Refinement of the structure of potassium pentaborate
tetrahydrate. Acta Crystallogr 16:376–379
51. Gao Y-H (2011) Potassium decaborate monohydrate. Acta Crystallogr E 67:i57
52. Wu Q (2011) Potassium pentaborate. Acta Crystallogr E 67:i67
53. Krogh-Moe J (1959) On the structural relationship of vitreous potassium pentaborate to the
crystalline modifications. Arkiv Kemi 14:567–572
54. Krogh-Moe J (1972) The crystal structure of the high-temperature modification of potassium
pentaborate. Acta Crystallogr B 28:168–172
55. Krogh-Moe J (1959) The crystal structures of potassium pentaborate, K2O.5B2O3, and the
isomorphous rubidium compound. Arkiv Kemi 14:439–449
56. Krogh-Moe J (1965) Least-squares refinement of the crystal structure of potassium
pentaborate. Acta Crystallogr 18:1088–1089
57. He M, Okudera H, Simon A, Kohler J, Jin S, Chen X (2013) Structure of Li4B2O5:
high-temperature monoclinic and low-temperature orthorhombic forms. J Solid State Chem
197:466–470
58. Rousse G, Baptiste B, Lelong G (2014) Crystal structures of Li6B4O9 and Li3B11O18 and
application of the dimensional reduction formalism to lithium borates. Inorg Chem 53:6034–
6041
59. Touboul M, Betourne E, Nowogrocki G (1995) Crystal structure and dehydration process of
Li(H2O)4B(OH)4•2H2O. J Solid State Chem 115:549–553
60. Zachariasen WH (1964) The crystal structure of lithium metaborate. Acta Crystallogr
17:749–751
61. Hohne E (1964) Die Kristallstruktur des LiB(OH)4. Z Chem 4:431–432 (in German)
62. Fronczek FR, Aubry DA, Stanley GG (2001) Refinement of lithium tetrahydroxoborate with
low-temperature CCD data. Acta Crystallogr E 57:i62–i63
63. Marezio M, Remeika JP (1966) Polymorphism of LiMO2 compounds and high-pressure
single-crystal synthesis of LiBO2. J Chem Phys 44:3348–3353
64. Louer D, Louer M, Touboul M (1992) Crystal structure determination of lithium diborate
hydrate, LiB2O3(OH).H2O, from X-ray powder diffraction data collected with a curved
position-sensitive detector. J Appl Crystallogr 25:617–623
65. Krogh-Moe J (1962) The crystal structure of lithium diborate, Li2O.2B2O3. Acta Crystallogr
15:190–193
66. Krogh-Moe J (1968) Refinement of the crystal structure of lithium diborate, Li2O.2B2O3.
Acta Crystallogr B 24:179–181
A Crystallographic Review of Alkali Borate Salts … 141

67. Natarajan M, Faggiani R, Brown ID (1979) Dilithium tetraborate, Li2B4O7. Cryst Struct
Commun 8:367–370
68. Radaev SF, Muradyan LA, Malakhova LF, Burak YV, Simonov VI (1989) Atomic Structure
and electron density of lithium tetraborate Li2B4O7. Sov Phys Crystallogr 34:842–845
69. Jiang A, Lei S, Huang Q, Chen T, Ke D (1990) Structure of lithium heptaborate, Li3B7O12.
Acta Crystallogr C 46:1999–2001
70. Radaev SF, Genkina EA, Lomonov VA, Maksimov BA, Pisarevskii YV, Chelokov MN,
Simonov VI (1991) Distribution of deformation electron density in lithium triborate LiB3O5.
Sov Phys Crystallogr 36:803–807
71. Radaev SF, Maximov BA, Simonov VI, Andreev BV, D’Yakov VA (1992) Deformation
density in lithium triborate, LiB3O5. Acta Crystallogr B 48:154–160
72. Le Henaff C, Hansen NK, Protas J, Marnier G (1997) Electron density distribution in LiB3O5
at 293 K. Acta Crystallogr B 53:870–879
73. Sennova N, Albert B, Bubnova R, Krzhizhanovskaya M, Filatov S (2014) Anhydrous
lithium borate, Li3B11O18, crystal structure, phase transition and thermal expansion.
Z Kristallogr 229(7):497–504
74. Cardenas A, Solans J, Byrappa K, Shekar KVK (1993) Structure of lithium catena-Poly
[3,4-dihydroxopentaborate-1:5-mu-oxo]. Acta Crystallogr C 49:645–647
75. Zviedre I, Ievins A (1974) Crystalline structure of rubidium monoborate RbBO2•4/3 H2O.
Latvijas PSR Zinatnu Akademijas Vestis Kimijas Seria 4:395–400 (in Russian)
76. Schmid S, Schnick W (2004) Rubidium metaborate, Rb3B3O6. Acta Crystallogr C 60:i69–i70
77. Touboul M, Penin N, Nowogrocki G (2000) Crystal structure and thermal behavior of
Rb2[B4O5(OH)4]•3.6H2O. J Solid State Chem 149:197–202
78. Krzhizhanovskaya MG, Bubnova RS, Bannova II, Filatov SK (1997) Crystal structure of
Rb2B4O7. Crystallogr Rep 42:226–231
79. Bubnova RS, Krivovichev SV, Shakhverdova IP, Filatov SK, Burns PC,
Krzhizhanovskaya MG, Polyakova IG (2002) Synthesis, crystal structure and thermal
behavior of Rb3B7O12, a new compound. Solid State Sci 4:985–992
80. Krzhizhanovskaya MG, Kabalov YK, Bubnova RS, Sokolova EV, Filatov SK (2000)
Crystal structure of the low-temperature modification of α-RbB3O5. Crystallogr Rep 45:572–
577
81. Krzhizhanovskaya MG, Bubnova RS, Fundamentskii VS, Bannova II, Polyakova IG,
Filatov SK (1998) Crystal structure and thermal expansion of high-temperature β-RbB3O5
modification. Crystallogr Rep 43:21–25
82. Liu Z-H, Li L-Q, Zhang W-J (2006) Two new borates containing the first examples of large
isolated polyborate anions: chain [B7O9(OH)5]2− and ring [B14O20(OH)6]4−. Inorg Chem
45:1430–1432
83. Behm H (1984) Rubidium pentaborate tetrahydrate, Rb[B5O6(OH)4].2H2O. Acta Crystal-
logr C 40:217–220
84. Penin N, Seguin L, Touboul M, Nowogrocki G (2001) Crystal structures of three MB5O8
(M = Cs, Rb) borates (α-CsB5O8, γ-CsB5O8, and β-RbB5O8). J Solid State Chem 161:205–213
85. Schlager M, Hoppe R (1994) Darstellung und Kristallstruktur von CsBO2. Z Anorg Allg
Chem 620:1867–1871 (in German)
86. Touboul M, Penin N, Nowogrocki G (1999) Crystal structure and thermal behavior of
Cs2[B4O5(OH)4] 3H2O. J Solid State Chem 143:260–265
87. Nowogrocki G, Penin N, Touboul M (2003) Crystal structure of Cs3B7O12 containing a new
large polyanion with 63 boron atoms. Solid State Sci 5:795–803
88. Krogh-Moe J (1960) The crystal structure of cesium triborate, Cs2O.3B2O3. Acta Crystallogr
13:889–892
89. Krogh-Moe J (1974) Refinement of the crystal structure of caesium triborate, Cs2O.3B2O3.
Acta Crystallogr B 30:1178–1180
90. Penin N, Seguin L, Touboul M, Nowogrocki G (2002) A new cesium borate Cs3B13O21.
Solid State Sci 4:67–76
142 C. C. Pye

91. Behm H (1984) Structure determination on a twinned crystal of cesium pentaborate

tetrahydrate, Cs[B5O6(OH)4].2H2O. Acta Crystallogr C 40:1114–1116
92. Bubnova RS, Fundamentsky VS, Anderson JE, Filatov SK (2002) New layered polyanion in
α-CsB5O8 high-temperature modification. Solid State Sci 4:87–91
93. Krogh-Moe J, Ihara M (1967) The crystal structure of caesium enneaborate, Cs2O.9B2O3.
Acta Crystallogr 23:427–430
94. Gupta A, Tossell JA (1983) Quantum mechanical studies of distortions and polymerization
of borate polyhedra. Am Mineral 68:989–995
95. Zhang ZG, Boisen MB Jr, Finger LW, Gibbs GV (1985) Molecular mimicry of the geometry
and charge density distribution of polyanions in borate minerals. Am Mineral 70:1238–1247
96. Oi T (2000) Calculations of reducted partition function ratios of monomeric and dimeric
boric acids and borates by the ab initio molecular orbital theory. J Nucl Sci Tech 37(2):
166–172
97. Oi T (2000) Ab initio molecular orbital calculations of reduced partition function ratios of
polyboric acids and polyborate anions. Z Naturforsch A 55:623–628
98. Zhou Y, Fang C, Fang Y, Zhu F (2011) Polyborates in aqueous borate solution: a Raman
and DFT theory investigation. Spectrochim Acta A 83:82–87
99. Beckett MA, Davies RA, Thomas CD (2014) Computational studies on gas phase polyborate
anions. Comput Theor Chem 1044:74–79
100. Frisch MJ et al (2004) Gaussian 03, revision D.02. Gaussian Inc, Wallingford, CT
101. Ishihara K, Nagasawa A, Umemoto K, Ito H, Saito K (1994) Kinetic study of boric
acid-borate interchange in aqueous solution by 11B NMR spectroscopy. Inorg Chem
33:3811–3816
102. Stadler HP (1947) The cell dimensions and space-group of pinnoite. Mineral Mag 28:26–28
103. Paton F, MacDonald SGG (1957) The crystal structure of pinnoite. Acta Crystallogr
10:653–656
104. Krogh-Moe J (1967) A note on the structure of pinnoite. Acta Crystallogr 23:500–501
An Ab Initio Study of Boric Acid,
Borate, and their Interconversion

Cory C. Pye

Abstract The chemistry of boric acid and monomeric borates is reviewed. Fol-
lowing a discussion of the crystal structures and nuclear magnetic resonance
studies, ab initio results are presented of molecular ortho- and metaboric acid,
(tetrahydroxo)borate, and the hydrates of orthoboric acid and borate. The structures
and vibrational frequencies are compared with experiment. Attempts to study their
interconversion lead us to a discussion of oxodihydroxoborate (the conjugate base
of boric acid), and of the hydroxide-boric acid complex. It is hypothesized that the
conversion of boric acid into borate proceeds via the oxodihydroxoborate inter-
mediate. Finally, the calculated structures of hydroxodioxo- and trioxoborate are
compared with experiment.

Keywords Boric acid ⋅ Borate ⋅ Ab initio

1 Introduction

The nature of boron(III) in aqueous solution has been of longstanding interest to

chemists as numerous hydroxooxoborates can exist as anions in metal ion salts [1].
The isotope 10B has important applications in nuclear science because of its large
thermal neutron absorption cross section, and the separation of 10B (natural
abundance 19.58%) from the remaining 11B (natural abundance 80.42%) is an
important technological problem requiring knowledge of the reduced partition
function ratio for isotope exchange [2–4]. Many borate minerals exist in nature and
the pure compounds can sometimes be prepared in the laboratory. The boron

Electronic supplementary material The online version of this chapter (https://doi.org/10.1007/

978-3-319-74582-4_8) contains supplementary material, which is available to authorized users.

C. C. Pye (✉)
Department of Chemistry, Saint Mary’s University, Halifax, NS B3H 3C3, Canada
e-mail: [email protected]

© Springer International Publishing AG, part of Springer Nature 2018 143

Y. A. Wang et al. (eds.), Concepts, Methods and Applications of Quantum Systems
in Chemistry and Physics, Progress in Theoretical Chemistry and Physics 31,
https://doi.org/10.1007/978-3-319-74582-4_8
144 C. C. Pye

isotope ratio is also an important consideration in geochemistry because, for marine

carbonates, it is used as a paleo-pH recorder for ancient seawater [5–9]. It is also
known that sound absorption in the ocean (∼1 kHz) involves chemical equilibria
with relaxation rates that correspond to the boric acid-borate equilibria [10].
Important information on the structure of these ions can be obtained by solubility
and pH measurements on solutions, X-ray diffraction study of crystals, and nuclear
magnetic resonance.
The solubilities of the simpler water-boron oxide system was investigated by
Kracek et al. [11]. They identified the solid compounds ice, H3BO3, three modi-
fications of HBO2, and crystalline B2O3. The solubility curves of boric acid and
several sodium borates were reported by Blasdale and Slansky soon thereafter [12].
The solid phases investigated were (ortho)boric acid, H3BO3; three hydrated forms
of sodium tetraborate, Na2B4O7 ⋅ nH2O, with n = 4 (the mineral kernite), 5, 10;
sodium pentaborate pentahydrate, NaB5O8 ⋅ 5H2O; and two hydrated forms of
sodium metaborate, NaBO2 ⋅ nH2O, with n = 2, 4. An alternate way to denote a
series of stoichiometries, common in geochemistry, is with respect to the ratio of
oxides lNa2O: mB2O3: nH2O, or shortened as (l:m:n).
The crystal structures of boron oxide and its hydrates ortho- and metaboric acid
are given in Table 1. Trigonal boron oxide, originally thought to contain tetrahedral
boron [13], actually consists of a 3D-network of corner-linked BO3 units [14–16].
The high-pressure orthorhombic form does consist of fused 6-membered rings of
BO4 tetrahedra [17]. Orthoboric acid was originally assumed to have the hydrogen
atoms halfway between the oxygens [18]. It was also shown that there could be
disorder in the layering by electron diffraction [19]. Refinement shows that both
polytypes (AB [20] or ABC [21] stacking) of orthoboric acid (0:1:3) consist of
stacks of approximately planar layers of hydrogen-bonded B(OH)3 molecules with
approximate C3h symmetry. Metaboric acid (0:1:1) exists in at least three forms,
two of which were discovered by Tazaki [22]. The orthorhombic α-form consists of
sheets [23] of B3O3(OH)3 with approximately Cs symmetry, held together by
hydrogen bonds [24]. The monoclinic β-form was shown to contain BO4 tetrahedra
and planar B2O5 groups [25], and further refinement showed that it consists of
endless zigzag chains of [B3O4(OH)(OH2)] [26, 27]. In essence, the H3B3O6
molecules have condensed together and the water molecule produced has bonded to
one of the two boron condensation sites. The cubic γ-form contains only tetrahedral
boron atoms [27, 28].
The crystal structures of some sodium and lithium borates are given in Table 2.
Anhydrous sodium borate (3:1:0) consists of discrete BO33− ions [29]. The
hydrated sodium borate (2:1:1) consists of discrete [BO2(OH)]2− ions connecting
sheets of edge-shared NaO polyhedra [30]. There are several hydrates of sodium
metaborate (1:1:n), but only two of them consist of the tetrahedral [B(OH)4]− anion,
the octahydrate [31] and the tetrahydrate [32]. There are also two monomeric
lithium salts (n = 16,8) [33–36]. The crystal structure of polyborates will be dis-
cussed in a separate paper.
Table 1 Crystal structures of boron oxide and boric acid
Stoichiometry Name Synth Space Unit cell X-ray B-O (Å) ranges Comments
BmHnO(3m+n)/2 T(K) group (Å or deg.) T(K)
m n Tetrahedral Trigonal References
1 0 Boron oxide-I 513 C31 4.334, 8.334 1.31–2.145 n/a Powder, BO4
tetrahedra [13]
P31 4.33, 8.39 293 n/a 1.38 Powder XRD and
ED, BO3 [14]
573 P31 4.336,8.340 n/a 1.336–1.404 Chain of BO3 [15]
P31 21 4.3358,8.3397 n/a 1.356–1.375 [16]
Boron oxide-II 1373 Ccm21 4.613, 1.373-1.512 n/a BO4 tetrahedra
65 kb 7.803,4.129 [17]
1 3 Orthoboric acid P1 7.04,7.04,6.56 n/a 1.36 Assumed H in
(sassolite) 92.50, middle of Os [18]
101.16,120.0
An Ab Initio Study of Boric Acid, Borate …

1.36 Electron diff.,

small xtal [19]
P1 7.039, 298 n/a 1.353–1.365 AB stacked layers
7.053,6.578, [20]
92.58,101.17,
119.83
Orthoboric 298 P32 7.0453,9.5608 297 n/a 1.349–1.377 ABC stacked
acid-T layers [21]
1 1 Metaboric acid 423 Pnma or 8.015,9.679, n/a 1.37 [22, 23]
(orthorhombic α) Pa 6.244
443 Pbnm 8.019, 143 n/a 1.347–1.391 B3O3(OH)3 layers
9.703,6.13 [24]
Metaboric acid 433 P21 ̸ c 6.76,8.80,7.15 [22]
(monoclinic β) 92.67
P21 ̸ a 7.132,8.852, 1.42–1.51 1.29–1.41 [25]
6.772
93.253
P21 ̸ a 7.122,8.842, 1.433–1.553 1.345–1.386 B3O4(OH)(OH2)
6.771 93.26 chains [26]
145

(continued)
Table 1 (continued)
146

Stoichiometry Name Synth Space Unit cell X-ray B-O (Å) ranges Comments
BmHnO(3m+n)/2 T(K) group (Å or deg.) T(K)
m n Tetrahedral Trigonal References
423 P21 ̸ c 6.758,8.844, 183 1.3284– 1.3485– [27]
7.075 93.5 1.5612 1.3812
Metaboric acid P43n 8.886 1.436–1.505 n/a [28]
(cubic γ)
453 P43n 8.8811 183 1.4428– n/a [27]
1.5094
C. C. Pye
Table 2 Crystal structures of monomeric sodium and lithium borates
Stoichiometry Synth Space Unit cell X-ray B-O (Å) ranges Comments
MlBmHnO(l+3m+n)/2 T(K) group (Å or deg.) T(K)
M l m n Tetrahedral Trigonal Ref.
Na 3 1 0 748 P21 ̸ c 5.687,7.530,9.993 n/a 1.377–1.409 [BO3]3− ions [29]
127.15
Na 2 1 1 523 Pnma 8.627,3.512,9.863 n/a 1.351–1.439 [BO2(OH)]2− ions [30]
An Ab Initio Study of Boric Acid, Borate …

Na 1 1 8 298 P1 6.126,8.1890,6.068 1.474–1.479 n/a [B(OH)4]− ions [31]

67.92,110.58,101.85
Na 1 1 4 293 P21 ̸ a 5.886,10.566,6.146 1.463–1.483 n/a [B(OH)4]− ions [32]
111.60
Li 1 1 16 P3 6.5534,6.1740 1.472–1.473 [Li(H2O)4]+ and [B(OH)4]−
ions [33]
C3 6.555,6.177 [34]
Li 1 1 4 Pbca 9.16,7.95,8.54 [B(OH)4]− ions [35]
Pbca 7.9362,8.5220,9.1762 120 1.4644–1.4989 [36]
147
148 C. C. Pye

Early work on the nuclear magnetic resonance (NMR) spectra of boron com-
pounds (11B-12.83 MHz) was done and showed a range of chemical shifts and
some 1-bond 11B-1H and 11B-2H coupling constants, using BF3 ⋅ Et2O as a ref-
erence [37]. Of interest in this study are the 11B chemical shifts of NaBO2(aq)
(−1.3 ± 0.5 ppm, due to [B(OH)4]−), NaB5O8(aq) (−1.3 and −14.4 ± 1.0 ppm),
K2B4O7 (−7.5 ± 1.0 ppm), Na2B4O7 (−8.0 ± 0.5 ppm), (NH4)2B4O7 (−10.3 ±
0.5 ppm), KB5O8(aq) (−13.0 ± 0.5 ppm), and B(OH)3(aq) (−18.8 ± 1.0 ppm).
These were interpreted as being due to a dynamic equilibrium between B(OH)4−
and B(OH)3. A later study by Momii and Nachtrieb (sat. B(OH)3 (aq) reference,
11
B-14 MHz) reexamined these results and gave 0.090–0.900 M NaBO2(aq) at
17.4 ± 0.5 ppm and 0.090–0.900 M KBO2 at 15.5 ± 0.5 ppm [38]. These were
interpreted as due to [B(OH)4]−. For sodium pentaborate solutions, the peak at
15.0 ppm was assigned to [B5O6(OH)4]−, and the peak at 1.1 ppm assigned to a
rapid equilibrium between B(OH)3, B(OH)4− and B3O3(OH)4−. For the tetrabo-
rates, a single peak is observed whose chemical shift increases with concentration
from 8 to 11 ppm, and this was assigned to a rapid equilibrium between B(OH)3,
B(OH)4− and at least two other ions. How and coworkers showed that the chemical
shift of a 50 g/L solution at 33 °C varied from −2 to −20 ppm between pH 12–2
respectively (11B, 12.83 MHz, BF3 ⋅ Me2O ref.) [39]. Smith and Wiersema
(11B-80 MHz) noted that one NMR peak in all borate solutions was linearly related
to the sodium to boron ratio and could this be interpreted as the peak of rapidly
exchanging B(OH)3 and [B(OH)4]− [40]. For tetraborate solutions, three peaks
could be observed, with the 5.0 ppm peak assigned to [B3O3(OH)4]−. Pentaborate
solutions also showed three peaks, with the 5 ppm peak assigned to [B3O3(OH)4]−,
and the peak at 18 ppm assigned to [B5O6(OH)4]−. Covington and Newman
examined the 11B spectra (28.87 MHz, rel. to infinite dilution [B(OH)4]−) of
sodium and potassium borate in water and in ∼0.1 mol/L added [OH−] in an effort
to determine the pKb of borate [41]. Henderson et al., in their study of the com-
plexation of borate with diols, showed the 11B NMR (12.83 MHz) of borax, boric
acid, and sodium metaborate from pH 2–12, along with the line width at half height
[42]. Janda and Heller examined the 11B spectra (60 MHz) of sodium, potassium,
and ammonium polyborates as a function of concentration and pH (0.5–13.8) and
either one or two lines were observed [43]. Epperlein et al. examined the 10B
spectra (1.807T, 8.267 MHz) of some boron species and found B(OH)3 at 0 ppm
(reference), B5O6(OH)4− at 17 ppm, B4O5(OH)42− between 70–85 ppm, and B
(OH)4− at around 140 ppm [44]. Salentine confirmed earlier results (11B, 127 and
160 MHz, external reference BF3 ⋅ Et2O) on the pentaborate (18, 13, 1 ppm) and
tetraborate (12, 8, 1 ppm) [45]. It was proposed that the resonance at 13 ppm, due
to triborate ion, and at 1 ppm, due to pentaborate ion, are due to the tetrahedral
boron atoms, and the trigonal boron atoms are not observed because of quadrupolar
relaxation.
We have reviewed the crystallography of boric acid and monomeric borates, and
the boron NMR of boric-acid/borate containing solutions. Our remaining goals are
to compare the ab initio energy, structure, and vibrational spectra to experiment
(where known) of orthoboric acid, metaboric acid, and tetrahydroxoborate, and to
An Ab Initio Study of Boric Acid, Borate … 149

determine the effect of hydration where appropriate. We also examine the mecha-
nism of interconversion of orthoboric acid and tetrahydroxoborate, which leads to a
discussion of the crystallography unobserved oxodihydroxoborate anion. We then
discuss the related crystallographically observed trioxoborate and dioxohydroxob-
orate anions.

2 Methods

Calculations were performed using Gaussian 03 [46]. The MP2 calculations utilize
the frozen core approximation, which is valid in nearly all cases except where
excessive core/valence mixing occurs (denoted c/v). The geometries were opti-
mized using a stepping stone approach, in which the geometries at the levels HF/
6-31G*, HF/6-31+G*, HF/6-311+G*, B3LYP/6-31G*, B3LYP/6-31+G*, B3LYP/
6-311+G*, MP2/6-31G*, MP2/6-31+G*and MP2/6-311+G* were sequentially
optimized, with the geometry and molecular orbital reused for the subsequent level.
Default optimization specifications were normally used. After each level, where
possible, a frequency calculation was performed at the same level and the resulting
Hessian was used in the following optimization. Z-matrix coordinates constrained
to the appropriate symmetry were used as required to speed up the optimizations.
Because frequency calculations are done at each level, any problems with the
Z-matrix coordinates would manifest themselves by giving imaginary frequencies
corresponding to modes orthogonal to the spanned Z-matrix space. The Hessian
was evaluated at the first geometry (Opt = CalcFC) for the first level in a series in
order to aid geometry convergence. To facilitate comparison with results from
gas-phase, solution, and solid phase measurements, no solvent corrections were
applied except via the supermolecule approach (explicit water molecules).

3 Results and Discussion

3.1 Orthoboric Acid, H3BO3

Six forms of (ortho)boric acid were investigated (Fig. 1). Two of these (C3h and Cs #2)
were minima, with the C3h structure being lower in energy by 22–26 kJ/mol
(Table 3). The C3v structure, a third-order saddle point, was much higher in energy
(139–152 kJ/mol). The other three structures were transition states linking the minima.
Both the Cs #1 and Cs #3 linked the Cs #2 structure to itself, whereas the C1 structure
linked the C3h and Cs #2 structures. The C1 structure was 38–43 kJ/mol higher than
the C3h structure, whereas the Cs #1 and Cs #3 structures were 34–38 and
20–22 kJ/mol, respectively, higher than the Cs #2.
150 C. C. Pye

C3h C3v Cs #1

Cs #2 Cs #3 C1 #1

Fig. 1 Structure of boric acid, [B(OH)3]. A bold symmetry label indicates a minimum energy
structure

Table 3 Relative Energies of B(OH)3 (kJ/mol)

C3h Cs #2 Cs #1 Cs #3 C1 #1 C3v
HF/6-31G* 0.0 24.5 60.3 46.2 41.3 146.4
HF/6-31+G* 0.0 24.6 60.3 46.1 41.3 142.1
HF/6-311+G* 0.0 26.3 62.7 47.8 42.5 145.7
B3LYP/6-31G* 0.0 22.1 56.8 42.7 38.8 145.2
B3LYP/6-31+G* 0.0 22.8 57.6 43.4 39.4 139.6
B3LYP/6-311+G* 0.0 24.3 59.4 44.7 40.2 141.4
MP2/6-31G* 0.0 23.8 61.1 45.8 41.6 155.7
MP2/6-31+G* 0.0 24.3 61.5 46.0 41.8 149.4
MP2/6-311+G* 0.0 25.6 63.5 47.5 42.9 152.6

Of these structures, the C3h structure is observed in the crystal structure of both
polymorphs of boric acid (Table 1). Our calculated B-O bond lengths range from
1.3553 to 1.3780 Å, with HF < B3LYP < MP2 (Table 4). These are in good
agreement with the experimentally determined (X-ray) average bond length of
1.36 Å and with previous literature values (Tables 1 and 4).
The vibrational spectra (unscaled) of the lowest-energy form of boric acid, as
well as some experimental vibrational frequencies from the literature, is given in
Table 5. Undistorted boric acid, of C3h symmetry, has 15 modes of internal
vibration and spans the vibrational representation

Γvib = 3A′ ðR, pÞ + 2A′′ ðIR; R, dpÞ + 4E ′ ðIRÞ + E ′′ ðR, dpÞ.

There is little difference in vibrational frequency between the A′ and E′ modes of

the OH stretch and HOB deformation. The computations are in reasonable
An Ab Initio Study of Boric Acid, Borate … 151

Table 4 Geometrical Parameters of the C3h form of B(OH)3. n/r = not reported
Level B-O (Å) O-H (Å) B-O-H angle (deg.)
HF/6-31G* 1.3581 0.9466 112.56
HF/6-31+G* 1.3584 0.9471 113.69
HF/6-311+G* 1.3553 0.9401 113.99
B3LYP/6-31G* 1.3721 0.9675 110.90
B3LYP/6-31+G* 1.3729 0.9683 112.46
B3LYP/6-311+G* 1.3691 0.9630 113.01
MP2/6-31G* 1.3762 0.9700 110.38
MP2/6-31+G* 1.3780 0.9722 111.77
MP2/6-311+G* 1.3709 0.9614 112.25
Literature
HF/STO-3G [2, 47] 1.389 n/r 114 (fixed)
HF/STO-3G [48] 1.364 0.98 110
HF/4-31G [48] 1.364 0.95 121
HF/3-21G* [49] 1.377 0.962 n/r
HF/6-31G [49] 1.370 0.947 n/r
HF/6-31G* [49] 1.358 0.947 n/r
HF/6-31G* [2, 49] 1.358 0.947 112.6
MP2/6-31G** [50] 1.357 0.942 113.0
B3LYP/6-311++G** [51] 1.380 0.971 112.6
B3LYP/6-311++G** [52] 1.370 0.962 112.8
MP2/6-311++G** [52] 1.373 0.961 110.7
B3LYP/aug-cc-pVQZ [53] 1.369 0.960 113.1
MP2/aug-cc-pVTZ [53] 1.374 0.962 111.5
MP2/aug-cc-pVQZ [53] 1.370 0.959 111.8
QCISD/6-311++G** [53] 1.371 0.959 111.1
B3LYP/aug-cc-pVDZ [54] 1.376 n/r n/r

agreement with the IR spectra measured in an argon matrix, and with most modes
of solid and aqueous boric acid. However, the in-plane BO3 deformation, HOB
deformation, and OH stretching frequencies differ, which would be expected,
because in solid boric acid, the molecules are held together by a network of
hydrogen bonds. It might be expected that aqueous solutions may exhibit similar
behavior to the solid. The vibrational spectra of the higher-energy Cs #2 confor-
mation is also given in Table 6. The E modes correlate with 2A modes of the same
reflection symmetry. The main differences in the vibrational frequencies of the Cs
#2 conformer are that the BOH torsion is much lower, the in plane BO3 deformation
is somewhat higher in frequency, the out of plane BO3 deformation is slightly
lower, the symmetric BO stretch is slightly higher, one component of the HOB
deformation and asymmetric B-O stretch is somewhat lower, and the OH fre-
quencies slightly higher.
Table 5 Theoretical and literature experimental vibrational frequencies (cm−1) of the C3h form of boric acid, B(OH)3
152

BO3 def BOH BOH tors BO3 oop def BO str HOB HOB BO str OH str OH str
tors def def
E′ A″ E″ A″ A′ E′ A′ E′ E′ A′
Theory, C3h
HF/6-31G* 458 476 567 736 927 1110 1116 1561 4135 4138
HF/6-31+G* 452 484 572 741 924 1086 1092 1540 4129 4132
HF/6-311+G* 454 491 577 747 919 1098 1112 1533 4188 4191
B3LYP/6-31G* 428 428 535 665 877 1038 1045 1478 3800 3800
B3LYP/6-31+G* 420 443 543 669 871 1010 1015 1448 3796 3796
B3LYP/6-311+G* 420 446 544 676 869 1018 1029 1441 3829 3830
MP2/6-31G* 425 445 551 675 880 1052 1056 1494 3823 3823
MP2/6-31+G* 419 450 551 677 872 1021 1024 1458 3793 3794
MP2/6-311+G* 426 447 548 680 873 1031 1038 1459 3872 3873
Theory (Literature)
MP2/6-31G** [50] 428 445 547 682 886 1044 1045 1509 3949 3948
B3LYP/6-311++G** 427 473 669 1015 1442 3870
[51]
Experiments
Raman, sat.aq. [55] 875
Raman, powder [56] 880 3172 3256
Raman, solution [57] 506 875 1130? 1060 1420
Raman, powder 503 1155 1065
Raman, sol. 70 °C [58] 882 3195 3290
Raman, powder 515 872 986
Raman, powder [59] 883 1169 1032 1391 3170 3248
(continued)
C. C. Pye
Table 5 (continued)
BO3 def BOH BOH tors BO3 oop def BO str HOB HOB BO str OH str OH str
tors def def
E′ A″ E″ A″ A′ E′ A′ E′ E′ A′
IR, mull [60] 885 1195 1450 3270
IR, mull and xtal. [61] 540 648 882 1197 1450 3200
IR, KBr pellets [62] 544 639 882? 1183 1428 3150
IR, 10B(OH)3 545 668 882? 1195 1490
IR, B(OD)3 540 654 1428 2380
Raman, sol 500 880 1430 3200
R, pH = 6.1–10 [63] 875
R, pH = 6.38–10.60 [64] 872
An Ab Initio Study of Boric Acid, Borate …

R, powder [65] 500 881 1168 1375

R, pH = 0.0–9.3 492–498 877
N2 matrix [66] 448.9 513.8 675.0 1009.9 1426.2 3668.5
10
B(OH)3 450.5 700.9 1478.0
B(18OH)3 430.7 668.1 999.5 1409.7 3657.6
10 18
B( OH)3 431.9 694.5 1461.2
B(OD)3 404.7 376.0 2704.6
10
B(OD)3 405.2 825.5 1452.7
IR+R (g), 150 °C [67] 866 1017 1020 1429 3706 3705
H10
3 BO3 700 866 1017 1020 1477 3706 3705
D11
3 BO3 887 810 706 1405 2726
D10
3 BO3 1456
Ar matrix [68] 432.1 436.0 (520) ± 5 666.4 992.4 1414.9 3688.6
692.2 1471.9
R,NaB5O8,1.39 M [54] 875
153
154 C. C. Pye

Table 6 Theoretical vibrational frequencies (cm−1) of the Cs form of boric acid, B(OH)3
BO3 BOH BOH BO3 BO HOB HOB BO OH OH
def tors tors oop str def def str str str
def
A′ A″ A″ A″ A′ A′ A′ A′ A′ A′
HF/6-31G* 471 306 508 729 930 1006 1131 1517 4141 4168
488 549 1116 1576 4156
HF/6-31+G* 468 324 514 733 928 979 1099 1495 4136 4166
483 558 1093 1554 4154
HF/6-311+G* 471 313 506 735 923 992 1111 1488 4195 4221
486 560 1107 1543 4211
B3LYP/6-31G* 441 260 476 661 877 947 1058 1430 3808 3830
458 527 1042 1494 3814
B3LYP/6-31+G* 436 295 482 665 873 912 1022 1400 3806 3833
450 539 1012 1463 3816
B3LYP/6-311+G* 438 274 466 668 871 926 1031 1392 3838 3862
452 536 1019 1454 3847
MP2/6-31G* 439 274 491 670 879 963 1074 1446 3825 3853
457 542 1057 1510 3842
MP2/6-31+G* 435 289 487 672 872 928 1036 1410 3797 3830
451 546 1024 1473 3819
MP2/6-311+G* 443 262 467 673 874 942 1047 1411 3875 3904
459 540 1037 1472 3894

3.2 Metaboric Acid, B3O3(OH)3

The structures of α-metaboric acid are analogous in symmetry to orthoboric acid

(Fig. 2). The barriers are quite similar to that of orthoboric acid (Table 7), but the
Cs #2 form is much more stable (only 3–4 kJ/mol above the minimum energy C3h
structure). If metaboric acid is ever observed in the gas phase, it should exist as a
mixture of an appreciable amount of both conformers (assuming it does not
decompose to form the monomer HBO2). It might even be possible to observe the
microwave spectrum of the Cs conformation of the molecule. It is the Cs form that is
observed in the crystal structure (Table 1), so one strategy for gas-phase observa-
tion might be laser ablation.

3.3 Hydrated Orthoboric Acid, H3BO3 ⋅ nH2O

Four Cs forms of hydrated boric acid with a water molecule directly bound to the
boron atom were attempted. Alternatively, this may be viewed as protonated
monoborate anion. In all cases, the water molecule dissociated. In the first two
cases, a solvated boric acid was obtained, neither of which was an energy minimum
(based on Cs #1 and Cs #3 naked boric acid, see Fig. 3). In the final two cases,
An Ab Initio Study of Boric Acid, Borate … 155

C3h C3v Cs #1

Cs #2 C s #3 C1 #1

Fig. 2 Structure of metaboric acid, [B3O3(OH)3]

Table 7 Relative Energies of B3O3(OH)3 (kJ/mol). c/v = excessive core-valence mixing

C3h Cs #2 Cs #1 Cs #3 C1 #1 C3v
HF/6-31G* 0.0 4.0 40.0 38.7 41.4 123.9
HF/6-31+G* 0.0 3.9 39.0 37.7 40.3 120.8
HF/6-311+G* 0.0 4.0 39.7 38.3 41.0 122.9
B3LYP/6-31G* 0.0 3.7 40.9 39.5 42.1 126.7
B3LYP/6-31+G* 0.0 3.4 39.8 38.4 40.8 122.8
B3LYP/6-311+G* 0.0 3.6 40.0 38.6 41.1 122.3
MP2/6-31G* 0.0 4.0 43.4 41.9 44.7 134.4
MP2/6-31+G* 0.0 3.8 42.3 40.8 43.4 c/v
MP2/6-311+G* 0.0 3.9 43.2 41.7 44.4

z-matrix errors occurred as the water molecule rotated. It can be said therefore that
direct protonation of B(OH)4− results in water elimination.
Four forms of hydrated boric acid with a water molecule hydrogen-bonded to the
boric acid were attempted (Fig. 3). To the naked C3h structure, a water molecule
may only hydrogen bond in a donor-acceptor (DA) fashion (C1 #2), whereas to the
naked Cs #2 structure, a water molecule may hydrogen bond in either a double
donor (DD, Cs #1 or C1 #3), donor acceptor (DA, C1 #1), or double acceptor (AA,
Cs #2). All were energy minima, with the exception of Cs #1 at the B3LYP/6-31G*
and all MP2 levels. These desymmetrized to C1 #3, where the water is removed
from the BO3 plane by varying amounts (significantly at the B3LYP and
MP2/6-31G* levels).
For the dihydrate, a Cs and five C1 forms were tried. The C1 #4 and C1 #5 forms
are derived from the naked C3h structure and differ in the orientation of the free
hydrogen of the water molecules, either up,up (uu) or up,down (ud). The other
forms are derived from the naked Cs #2 structure: namely, the Cs/1 #1 (DD,AA), the
C1 #2 (DD,DA), and the C1 #3 (DA,AA). The Cs form was stable at the HF,
156 C. C. Pye

Cs #1 Cs #2 Cs #2a Cs #1/ C1 #3

Cs #2 C1 #1 C1 #2 Cs/1 #1 C1 #1a

C1 #2 C1 #2a C1 #2b C1 #3 C1 #4

C1 #5 C3 #1 C1 #1 C1 #1a C1 #2

Fig. 3 Structure of hydrated boric acid, [B(OH)3] ⋅ nH2O

B3LYP/6-31+G*, B3LYP/6-311+G*, and MP2/6-31+G* levels but reverted to a

C1 #1 form at the other levels. For MP2/6-311+G*, the double-donor water moved
slightly out of the plane, whereas for the other levels it accepted a hydrogen bond
from the other water molecule (C1 #1a). The C1 #2 form is stable at the HF,
B3LYP/6-31+G*, B3LYP/6-311+G*, MP2/6-31+G*, and MP2/6-311+G* levels.
At the B3LYP/6-31G* and MP2/6-31G* levels, the DD water accepts a hydrogen
bond from the free DA water to give a DDwA, DAw C1 #2a structure (w indicates
hydrogen bonding between water molecules) which flatten out at all levels except
B3LYP/6-31G* and MP2/6-31G*(C1 #2b). The C1 #3, 4, and 5 forms were stable
at all levels.
For the trihydrate, a C3 and two C1 forms were tried. The C3 #1 (uuu) and C1 #2
(uud) forms are derived from the naked C3h structure and are stable at all levels,
whereas the C1 #1 is derived from the naked Cs #2 structure. The C1 #1 structure is
stable at the HF, B3LYP/6-31+G*, B3LYP/6-311+G*, MP2/6-31+G*, and
An Ab Initio Study of Boric Acid, Borate … 157

Fig. 4 Variation of B-O distance of B(OH)3 as a function of hydration and structure

MP2/6-311+G* levels, but reverts to C1 #1a by forming a hydrogen bond between

the DD and AA water molecules. This structure also exists at the other levels.
The B-O distances as a function of hydration structure are plotted in Fig. 4. The
average B-O distance slightly increases (<0.005 Å) upon hydration, which
demonstrates how weakly boric acid is hydrated, but the variation in less-symmetric
structures is up to ±0.025 Å.
The vibrational frequencies (MP2/6-311+G*) of hydrated boric acid are plotted
in Figs. 5 and 6. In the OH stretching region (3500–4000 cm−1, Fig. 6), it is clear
that if boric acid, with a OH stretch in the range 3860–3910 cm−1, donates a
hydrogen bond to a DA water, then an OH stretching frequency drops to between
3600–3700 cm−1, whereas if it donates to a AA water, an OH stretching frequency
drops to only 3780–3830 cm−1. If boric acid accepts a hydrogen bond, then the OH
stretching frequency corresponding to the oxygen accepting the hydrogen bond
essentially remains unchanged. In the lower-frequency region (Fig. 5), the BOH
bending frequencies increase from 940–1050 to 1150–1250 cm−1, and the BOH
torsion increases from 260–550 to 800–830 cm−1, upon accepting a hydrogen
bond. The B-O symmetric (∼870 cm−1) and asymmetric stretching (∼1410–
1470 cm−1) frequencies increase slightly upon hydration. The BO3 in plane
deformation increases from ∼425 cm−1 to nearly 500 cm−1 upon hydration,
whereas the out of plane deformation at ∼680 cm−1 is hardly affected.
158 C. C. Pye

Fig. 5 Variation of vibrational frequencies of B(OH)3 (0–1600 cm−1) as a function of hydration

and structure

Our approach to hydration of boric acid is to solvate the boric acid molecule as
completely as possible with a small number of water molecules in order to effi-
ciently model the vibrational spectra. Of course, water molecules might prefer to
hydrogen-bond to other water molecules instead of to boric acid. This work is
therefore somewhat complementary to that of Tachikawa [52], who studied similar
clusters with up to five water molecules and found several in which water molecules
were hydrogen bonded to each other.

3.4 Borate, B(OH)4−

Five forms of monoborate were investigated (Fig. 7). Two of these (D2d #2 or S4 #3,
and S4 #1) were minima, with the S4 #1 structure being lower in energy by 7–10 kJ/mol
at the Hartree-Fock levels (Table 8). The D2d #2 is only a minimum at the Hartree-Fock
levels. We confirm the presence of a second shallow minimum (S4 #3) at the correlated
levels, as first found by Stefani et al. [53] The S4 #2 structure is a transition state that
connects the D2d #2/S4 #3 and S4 #1 structures. It is 2–3 kJ/mol higher in energy than
D2d #2. A D2 structure, derived from D2d #1, ascended in symmetry to give the D2d #2
structure at all levels.
An Ab Initio Study of Boric Acid, Borate … 159

Fig. 6 Variation of vibrational frequencies of B(OH)3 (3500–4000 cm−1) as a function of

hydration and structure

D2d #1 D2d #2/S4 #3 S4 #1 S4 #2

Fig. 7 Structure of borate, B(OH)4−

The tetrahedral borate structure is observed in the crystal structure of sodium

1:1:8 and 1:1:4, and lithium 1:1:16 and 1:1:4 (Table 2). Our calculated B-O bond
lengths range from 1.4702–1.4904 Å, with HF < B3LYP < MP2 (Table 9). These
are slightly longer than the experimentally determined (X-ray) bond lengths of
1.463–1.483 Å and compare favorably with previous calculations.
The vibrational spectra (unscaled) of borate, as well as some experimental
vibrational frequencies from the literature, is given in Table 10. Undistorted borate,
of S4 symmetry, has 21 modes of internal vibration and spans the vibrational
representation
160 C. C. Pye

Table 8 Relative energies of B(OH)4− (kJ/mol)

D2d #1 D2d #2 S4 #1 S4 #2 S4 #3
HF/6-31G* 67.6 7.4 0.0 9.7 n/a
HF/6-31+G* 68.9 8.7 0.0 10.5 n/a
HF/6-311+G* 74.6 9.6 0.0 12.4 n/a
B3LYP/6-31G* 63.3 7.4 0.0 5.6 5.6
B3LYP/6-31+G* 66.0 8.2 0.0 8.0 7.8
B3LYP/6-311+G* 70.9 9.0 0.0 9.0 8.7
MP2/6-31G* 71.3 7.0 0.0 8.1 6.7
MP2/6-31+G* 72.3 8.1 0.0 9.1 8.0
MP2/6-311+G* 77.8 8.8 0.0 11.1 8.7

Table 9 Geometrical parameters of the S4 #1 form of B(OH)4−

Level B-O (Å) O-H (Å) B-O-H angle (deg.)
HF/6-31G* 1.4722 0.9454 105.63
HF/6-31+G* 1.4717 0.9451 107.63
HF/6-311+G* 1.4702 0.9390 107.81
B3LYP/6-31G* 1.4876 0.9674 103.05
B3LYP/6-31+G* 1.4873 0.9673 106.20
B3LYP/6-311+G* 1.4854 0.9626 106.52
MP2/6-31G* 1.4897 0.9689 102.67
MP2/6-31+G* 1.4904 0.9710 105.63
MP2/6-311+G* 1.4829 0.9610 106.00
Literature
HF/STO-3G [69] (D2d) 1.48 114 (fixed)
HF/STO-3G [47] (D2d) 1.492 114 (fixed)
HF/6-31G* [51] (D2d) 1.474 0.946 104
HF/6-31G* [70] 1.472 0.945 105.6
HF/6-31G* [2] 1.472 0.954 105.4
B3LYP/aug-cc-pVDZ [54] 1.486 n/r n/r
B3LYP/6-311++G** 1.489 0.960 n/r
MP2/6-311++G** 1.486 0.960 n/r

Γ vib = 5AðR, pÞ + 6BðIR; R, dpÞ + 5E ðIR; R, dpÞ.

The BOH deformations appear from 1000 to 1300 cm−1, the B-O stretching
motions from 720 to 1100 cm−1, the BO4 deformations from 230 to 570 cm−1, and
the BOH torsions from 180 to 440 cm−1.
Table 10 Theoretical and literature experimental vibrational frequencies (cm−1) of the S4 #1 form of B(OH)4−
BOH BO4 BOH BO4 BOH BO4 BO4 BO BO BO BOH BOH BOH
tors def tors def tors def def str str str def def def
A B E A B E B A E B A B E
Theory
HF/6-31G* 202 273 317 420 431 514 568 775 940 1086 1110 1169 1323
HF/6-31+G* 204 260 307 420 429 511 561 774 924 1056 1086 1141 1285
HF/6-311+G* 212 272 313 429 438 512 568 770 920 1049 1098 1151 1285
B3LYP/6-31G* 190 264 331 345 416 476 521 728 866 1018 1026 1070 1252
B3LYP/6-31+G* 181 235 307 366 407 471 513 724 851 973 1000 1048 1193
B3LYP/6-311+G* 189 245 311 380 408 473 518 721 846 964 1015 1060 1195
MP2/6-31G* 203 272 343 368 423 482 535 732 892 1041 1042 1094 1272
An Ab Initio Study of Boric Acid, Borate …

MP2/6-31+G* 203 249 326 372 414 473 520 725 869 990 1012 1060 1208
MP2/6-311+G* 212 265 323 406 412 477 534 729 871 993 1016 1067 1206
Theory (Literature)
HF/6-31G* [70] 202 273 316 420 431 514 569 775 940 1086 1110 1169 1323
Experiments
NaBO2, 5N [71] 749
NaBO2 (aq) [57] 400– 747 950 1132 1276
500
NaBO2 ⋅ 4H2O(s) 741
H3BO3+NaOH 744–7 949 1150– 1219–
60 20
KB(OH)4(aq) (Td) [72] 379 379 533 533 754 947 947
NaB(OH)4(aq) (Td) [73] 374 374 516 516 744 940 940
sat NaB(OH)4(aq) pH = 5–10 [63] 745
(continued)
161
Table 10 (continued)
162

BOH BO4 BOH BO4 BOH BO4 BO4 BO BO BO BOH BOH BOH
tors def tors def tors def def str str str def def def
A B E A B E B A E B A B E
NaB(OH)4(aq), 1.5M, pH = 6–12 [64] 375 375 522 522 745 940 940
NaB(OH)4(s) [65] 405 405 520 520 738 952 952 1078
Na2B4O7(aq), 0.01–0.8 [65] 744
sat Cs2B4O7 ⋅ 5H2O or CsB5O8 ⋅ 4H2O 480,484 511 742,745
[74]
NaBO2(aq), 4–5 M [54] 740
KBO2 ⋅ nH2O(aq), n = 5, 10, 19 [75] 740
C. C. Pye
An Ab Initio Study of Boric Acid, Borate … 163

C1 C2 C1 C1 S4

Fig. 8 Structure of hydrated borate, B(OH)4− ⋅ nH2O

3.5 Hydrated Borate, B(OH)4− ⋅ nH2O

The monoborate anion may be hydrated. The monohydrate has C1 symmetry; the
dihydrate, either C1 or C2; the trihydrate, C1; and the tetrahydrate, S4 (see Fig. 8).
Upon hydration, the average B-O distance slightly decreases (0.002 Å) but the
deviation from the mean can be as much as 0.025 Å (Fig. 9).
The vibrational frequencies (MP2/6-311+G*) of hydrated monoborate are
plotted in Fig. 10. The OH stretching frequencies (not shown) decrease by about
50 cm−1 upon going from the anhydrate to the tetrahydrate. The restricted trans-
lations of the water molecules appear below 250 cm−1, whereas the restricted
rotations appear at approximately 300–440 cm−1 (rock), 580–730 cm−1 (twist), and

Fig. 9 Variation of B-O distance of B(OH)4− as a function of hydration and structure

164 C. C. Pye

Fig. 10 Variation of vibrational frequencies of B(OH)4− as a function of hydration and structure

770–820 cm−1 (wag). The BOH deformation frequencies are in the range from
1000–1220 cm−1 and tend to increase with hydration as the mode becomes stiffer
upon forming a hydrogen bond to water. The B-O totally symmetric stretch only
increases by a few wavenumbers, whereas the asymmetric stretch increases much
more. The deformation modes tend to increase a bit, but the BOH torsional modes
increase a lot because of the restrictions imposed by hydrogen bonding. There is
significant mixing both between these two types of modes and with the water
rocking modes.

3.6 The Reaction of Borate, B(OH)4− with Hydronium,

H3O+

Attempts to hydrate boric acid by attaching water directly to the boron (see above,
equivalent to direct protonation of monoborate) resulted in the dissociation of the
water molecule. It was thought that the more realistic hydronium ion might interact
with the borate ion by forming an ion pair before losing a water molecule. A scan of
the O…H distance from 2.0 to 0.9 Å in 0.1 Å steps was carried out in an attempt to
carefully protonate the borate at HF/6-31G*; however, a different proton from the
An Ab Initio Study of Boric Acid, Borate … 165

hydronium protonated a different oxygen, resulting in fragmentation of the borate

molecule to form an interacting boric acid and a water dimer. Attempts to optimize
an ion pair structure at the nine levels investigated invariably led to boric acid
interacting with a water dimer. A typical pattern observed is that of protonation,
followed by detachment, and in some cases rotation of the OH group from a Cs #2
boric acid to form C3h boric acid. If such an ion pair exists, more water molecules
would be needed to stabilize it. There can be no enthalpic barrier in the absence of
additional water molecules.

3.7 The Dissociation of Borate, B(OH)4−

The dissociation of one of the B-O bonds of borate to form boric acid and
hydroxide is a simple possibility for their interconversion. Scans were done at each
level, with the zero of energy set to the optimized borate structure and are shown in
Fig. 11. Typically the scans suggest that as the B-O distance increases, two of the
three remaining borate hydroxyls rotate to form stabilizing hydrogen bonds with
the departing hydroxide, which then swings into the plane of a Cs #2 boric acid as

Fig. 11 Scans of the B-O distance of borate, B(OH)4−

166 C. C. Pye

the transition state is passed. However, for the B3LYP/6-31G* and MP2/6-31G*
scans, which proceeded farther than the others, the hydroxide abstracted a proton
from boric acid to give BO(OH)2− + H2O. This suggested the need to explore the
relative energies of hydrated BO(OH)2− and of the boric acid-hydroxide complex.

3.8 Oxodihydroxoborate and Its Hydrates,

BO(OH)2− ⋅ nH2O

The oxodihydroxoborate anion, BO(OH)2−, can potentially exist as one of three

planar conformers, C2v #1 and #2, and Cs. The C2v #2 form is unstable and reverts
to the C2 form (Fig. 12). The order of stability is C2v #1 (most stable) < Cs #1 < C2
(Table 11). Only the Cs from can be derived from the parent C3h boric acid, but all
can be derived from Cs #2 boric acid. To the best of our knowledge, the oxodi-
hydroxoborate structure has not been observed crystallographically, but the related
dioxomonohydroxoborate and trioxoborate have been observed as the sodium salts
(Table 2). The B-O distances (1.28–1.31 Å and 1.42–1.45 Å, Table 12) are much
shorter than those of tetraborate and bracket those of the more negatively charged
deprotonated versions. These results confirm the findings of Stefani et al. [53]. This
ion has been observed in the gas-phase [76].
If the oxodihydroxoborate anion can exist as a transient species in the formation
of tetrahydroxoborate upon basification of boric acid, it might be possible to

C2v #1 C2v #2 Cs C2

Fig. 12 Structure of oxodihydroxoborate, BO(OH)2−

Table 11 Relative energies C2v #1 C2v #2 Cs C2

of BO(OH)2− (kJ/mol)
HF/6-31G* 0.0 49.9 9.8 49.2
HF/6-31+G* 0.0 51.5 9.8 51.4
HF/6-311+G* 0.0 52.4 8.9 52.1
B3LYP/6-31G* 0.0 44.1 8.8 42.6
B3LYP/6-31+G* 0.0 46.6 8.6 46.6
B3LYP/6-311+G* 0.0 47.3 7.7 46.7
MP2/6-31G* 0.0 49.0 9.5 46.8
MP2/6-31+G* 0.0 50.5 9.0 50.2
MP2/6-311+G* 0.0 51.3 8.4 50.1
Table 12 Structure and vibrational frequencies (cm−1) of the C2v #1 form of BO(OH)2−
B-O B-OH BO3 BO3 BOH BOH BO3 oop BOH BOH BOH BOH BO
(Å) (Å) def def tors tors def str def def str str
A1 B2 A2 B1 B1 A1 B2 A1 B2 A1
HF/6-31G* 1.2791 1.4266 454 510 537 545 788 851 1075 1195 1316 1711
HF/6-31+G* 1.2866 1.4217 452 504 550 558 793 853 1072 1172 1295 1648
An Ab Initio Study of Boric Acid, Borate …

HF/6-311+G* 1.2831 1.4199 455 505 542 557 792 847 1077 1183 1287 1639
B3LYP/6-31G* 1.2949 1.4455 412 465 493 517 703 791 972 1106 1235 1605
B3LYP/6-31+G* 1.3041 1.4398 412 459 512 532 708 792 976 1082 1204 1532
B3LYP/6-311+G* 1.2992 1.4379 415 460 504 530 709 787 975 1093 1200 1527
MP2/6-31G* 1.3014 1.4499 415 466 508 531 712 796 989 1117 1250 1617
MP2/6-31+G* 1.3135 1.4450 412 457 511 542 710 794 982 1087 1214 1522
MP2/6-311+G* 1.3062 1.4386 420 463 485 532 708 794 989 1094 1207 1525
167
168 C. C. Pye

C2v/s #1 C1 #1 C2v #2 C2 Cs #3/C1 #5

Cs #2/C1 #4 C1 #2 C1 #3 C2v #1/ C2 #2 Cs #2

C2 #1/ C1 #2 Cs #1 C1 #1 C1 #2a

C1 #3 C1 #4 C1 #5 C1 #6 C2v

C2/1 #1 Cs #1 C1 #2 C1 #3 Cs #2

Fig. 13 Structure of hydrated oxodihydroxoborate, BO(OH)2− ⋅ nH2O

observe it spectroscopically. The vibrational frequencies of the most stable form are
given in Table 12. In the isotropic Raman spectra, one would predict the obser-
vation of the BO3 deformation mode at around 415 cm−1, the BO(H) stretching
mode at around 800 cm−1, the BOH deformation mode at around 1100 cm−1, and
the BO stretching mode at about 1530 cm−1. There is a fair amount of coupling
between the BO stretching and BOH deformation modes.
A monohydrate can be based on any of the three stable anhydrous forms. From
the most stable anhydrous C2v #1 form, both the C2v #1 and the more stable C1 #1
forms can be derived (Fig. 13). The C2v #1 is only stable at the HF levels, and at
B3LYP/6-31G*, converting to the Cs #1 form at the other levels. From the next
most stable anhydrous Cs form, the Cs #2, C1 #2 and C1 #3 forms can be derived.
An Ab Initio Study of Boric Acid, Borate … 169

The C1 #2 form is higher in energy than the nearly isoenergetic Cs #2 and C1 #3

forms. At MP2/6-311+G*, Cs #2 converts to C1 #4. From the high-energy anhy-
drous C2v #2 form, the unstable C2v #2 and Cs #3 forms arise. The lower-energy Cs
#3 form converts to C1 #5. The C2v #2 form could convert to a C2 form (unstable)
or one of two Cs forms. In all of these, the water molecule moves (Cs) or would
move (C2) to a structure close to C1 #5. In all cases, conformers possessing a
hydrogen bond to the oxo group are the most stable within each grouping based on
the naked anion.
For the dihydrate, we can construct several structures based on the three
anhydrous forms. From C2v #1, the C2 #1, Cs #1, and C1 #1 forms may be con-
structed. The C2 #1 form is unstable at HF/6-311+G* and MP2/6-311+G*, giving
C1 #2a and C1 #2 forms, respectively. From Cs #1, the stable C1 #3–6 forms can be
constructed. From C2v #2, the C2v #1 form can be constructed (only stable at HF/
6-31+G*), as well as Cs #2 (stable at MP2/6-311+G*). The C2v #1 form converts to
the stable C2 #2 form (C1 #8 at MP2/6-311+G*), whereas most attempts to obtain
Cs #2 result in migration of the double acceptor water molecule towards the double
donor water molecule.
For the trihydrate, two structures exist for each of the three anhydrous forms.
Both a C2 and a Cs #1 exist for the C2v #1 anhydrous form, although at B3LYP/
6-31G* and the MP2 levels, the C2 structure desymmetrizes to C1 #1. Both C1 #2
and C1 #3 can exist for the Cs #1 anhydrous form. The C2v form could exist for the
C2v #2 anhydrous form, but it has imaginary frequencies. Nearly all attempts to
desymmetrize (Cs #2,3) result in the double acceptor water molecule moving
toward the other double donor water molecules, and at the only level where such a
structure exists (Cs #3, MP2/6-31+G*), an imaginary frequency would desym-
metrize by moving the double acceptor water molecule towards the double donor
water molecules.
In the presence of additional water molecules, the BOH twisting vibrations can
increase to 600–800 cm−1 (if hydrogen bonded), the in-plane BO3 deformations
increase to 480–600 cm−1, the out-of-plane BO3 deformation is hardly affected, the
B-O(H) and B-O symmetric stretching frequency increases slightly, and the coupled
antisymmetric B-OH stretch and BOH deformations all increase in frequency to
about 1100–1300 cm−1 (Fig. 14). For water molecules that are hydrogen-bonded to
the lone oxygen of this strong base, the OH frequency is lowered to around
3000 cm−1.

3.9 Boric Acid-Hydroxide Complex and Its Hydrates,

B(OH)3 ⋅ OH− ⋅ nH2O

The boric acid-boric acid complex was investigated next (Fig. 15). All attempts to
locate the complex between hydroxide and the C3h form of boric acid resulted in
deprotonation to form the oxodihydroxoborate anion-water complex. The Cs
170 C. C. Pye

Fig. 14 Variation of BO(OH)2− vibrational frequencies (0–1700 cm−1) as a function of hydration

and structure

Cs C1

C1 #1 C1 #2 C1 #3 C1 #1 C1 #2

Fig. 15 Structure of boric acid-hydroxide complex, B(OH)3+OH−+nH2O

complex between the Cs #2 form of boric acid acting as a double hydrogen bond
donor to hydroxide was not stable at B3LYP/6-31G* and MP2/6-31G*, reverting
instead to a oxodihydroxoborate-water complex via proton transfer. At the other
levels, an imaginary frequency gave rise to the stable C1 structure, which was 10–
25 kJ/mol less stable than the corresponding oxodihydroxoborate form. Upon
An Ab Initio Study of Boric Acid, Borate … 171

hydration to give one of three possible C1 structures, the B3LYP/6-31G*,

B3LYP/6-31+G*, and MP2/6-31G* converge to the oxodihydroxo form, and with
the exception of C1 #1, this also happens at the HF/6-31G* and B3LYP/6-311+G*
levels. The forms that actually exist are 15–45 kJ/mol less stable than the corre-
sponding oxodihydroxoborate dihydrate. Addition of the second water molecule
only gives stable structures at HF/6-31+G* and HF/6-311+G* levels, reverting to
the oxodihydroxoborate at the other levels. When both can exist, the oxodihy-
droxoborate is 45–50 kJ/mol more stable. Hydrating the boric acid part of the
complex stabilizes the corresponding anion form.

3.10 Transition State Connecting Tetrahydroxoborate

and Oxodihydroxoborate-Water Complex

The transition states connecting hydrated tetrahydroxoborate and the hydrated

oxodihydroxoborate anions are shown in Fig. 16. The transition state essentially
looks like a Cs boric acid interacting with a hydroxyl anion. It is stabilized
somewhat by the interaction between the empty p-orbital of the boron and
the hydroxyl non-bonding electron pairs, and also hydrogen bonding between the
hydroxyl oxygen and the syn-hydrogens of boric acid, which tilt towards the
hydroxyl. The electronic barrier (from tetrahydroxoborate) is between 84 and
114 kJ/mol, depending on level. The transition state could conceivably connect to
either the van der Waals complex between boric acid and hydroxide, or to either the
C1 #1 or #3 forms of the monohydrated oxodihydroxoborate anion.
A water molecule can stabilize the transition state in one of two ways. It can
stabilize the boric acid portion of the transition state (C1 #1, #3, and #5) in one of
the three possible sites (as a hydrogen bond donor-acceptor and two donor-donor
types, respectively). Alternatively, it can stabilize the hydroxyl portion of the
molecule (C1 #2 and #4) where the free boric acid hydrogen is pointing either
toward or away from the water molecule, respectively. The stabilization of the
hydroxide lowers the energy more. When compared with the unhydrated transition
state, addition of the water to the boric acid part of the molecule actually increases
the barrier by between 10 and 25 kJ/mol, whereas stabilization of the hydroxide
part of the molecule lowers the barrier by between 5 and 15 kJ/mol.
Waton and coworkers used the temperature-jump method to study the equilib-
rium between boric acid and borate [77]. The kinetics did not fit a simple equi-
librium, but was analyzed by postulating an intermediate which they called [B
(OH)3,OH−]. Our analysis of their published rate constants (k23) at 4 and 20 °C
suggests an activation barrier of 42 kJ/mol, giving an enthalpy of activation of
39 kJ/mol. We hypothesize that this intermediate is actually [BO(OH)2]− ⋅ H2O.
Our calculated electronic barriers are much too high (101–117 kJ/mol), but these
are lowered upon addition of an extra one (90–103 kJ/mol) or two (69–86 kJ/mol)
water molecules. This leads one to think that additional water molecules might
172 C. C. Pye

C1 C1 #1 C1 #2 C1 #3 C1 #4

C1 #5 C1 #1 C1 #1a C1 #2 C1 #2a

C1 #3a C1 #3b C1 #4 C1 #4a C1 #4b

C1 #5 C1 #6 C1 #7 C1 #8 C1 #8a

C1 #9 C1 #9a C1 #10 C1 #10a

Fig. 16 Structure of boric acid-hydroxide transition structure + nH2O

stabilize the transition state even more to bring the results into even better agree-
ment with experiment, assuming that the postulated intermediate is correct.

3.11 Trioxoborate, BO33−, and Dioxohydroxoborate

BO2(OH)2−

While trioxoborate and dioxohydroxoborate are not expected to exist in aqueous

solution owing to their strong basicity, these ions do exist as molecular entities in
An Ab Initio Study of Boric Acid, Borate … 173

Table 13 Geometrical parameters of BO33− and BO2(OH)2−

Level B-O distances (Å)
BO33− BO2(OH)2− s-BO2(OH)2− a-BO2(OH)2−
HF/6-31G* 1.4186 1.5941 1.3347 1.3188
HF/6-31+G* 1.4135 1.5556 1.3440 1.3251
HF/6-311+G* 1.4127 1.5613 1.3405 1.3209
B3LYP/6-31G* 1.4411 1.6307 1.3551 1.3332
B3LYP/6-31+G* 1.4234 1.5731 1.3637 1.3413
B3LYP/6-311+G* 1.4211 1.5812 1.3587 1.3359
MP2/6-31G* 1.4461 1.6296 1.3616 1.3401
MP2/6-31+G* 1.4363 1.5781 1.3755 1.3513
MP2/6-311+G* 1.4299 1.5741 1.3679 1.3437

Table 14 Vibrational frequencies (cm−1) of the D3h form of BO33−

Level BO3 i.p def E′ BO str. A1′ BO3 o.o.p. def A2″ BO str. E′
HF/6-31G* 581 848 876 1216
HF/6-31+G* 536 829 851 1085
HF/6-311+G* 535 826 844 1062
B3LYP/6-31G* 530 764 773 1107
B3LYP/6-31+G* 271 791 679 897
B3LYP/6-311+G* 227 785 680 867
MP2/6-31G* 533 769 788 1124
MP2/6-31+G* 358 767 659 843
MP2/6-311+G* 344 768 654 824

their sodium salts (Table 1). The calculated B-O bond distance of BO33− (Table 13)
is slightly longer than what is observed experimentally in the sodium salt, and this
is partly due to the neglect of the sodium counterion and crystal packing. For
BO2(OH)2−, the B-O distances for the unprotonated oxygens are in reasonable
agreement with experiment, whereas the B-O(H) distance is quite long. It is
expected that hydrogen bonding present in the crystal structure would shorten the
corresponding B-O(H) distance considerably.
The calculated vibrational frequencies of BO33− are shown in Table 14. There is
a quite surprising dependence of the level of theory and basis set. The stretching
frequencies decrease when going from Hartree-Fock to the correlated levels, as
expected, as well as when going from nondiffuse to diffuse basis sets. It is more
surprising that the deformation modes are affected even more when going from
nondiffuse to diffuse basis sets at the correlated levels. This may be an artifact of
using a correlated level on a system, which, in the gas-phase, is likely unbound with
respect to electron detachment.
The calculated vibrational frequencies of BO2(OH)2− are shown in Table 15.
The dependence of the frequency on the level of theory and basis set is not as
174 C. C. Pye

Table 15 Vibrational frequencies (cm−1) of the Cs #1 form of BO2(OH)2−

Level BO3 i.p BO OHtors BO3 i.p BO3 o.o. OH BO BO
def (H) str def p def def str str
A′ A′ A″ A′ A″ A′ A′ A′
HF/6-31G* 453 491 547 723 812 1052 1134 1661
HF/6-31+G* 471 523 572 719 812 1080 1104 1557
HF/6-311+G* 472 515 560 704 811 1086 1093 1551
B3LYP/6-31G* 391 426 523 656 724 958 1047 1563
B3LYP/6-31+G* 416 476 543 635 712 986 1014 1426
B3LYP/6-311+G* 417 467 528 615 710 990 1007 1424
MP2/6-31G* 397 445 537 668 737 973 1050 1572
MP2/6-31+G* 418 478 547 643 708 991 1001 1406
MP2/6-311+G* 425 480 518 639 703 995 1001 1408

pronounced as for trioxoborate. The B-O and B-OH stretching frequencies decrease
when going from Hartree-Fock to the correlated levels, as expected. When going
from nondiffuse to diffuse basis sets, the B-O stretching frequencies decrease, but
the B-OH frequency increases. The other frequencies do not show any surprising
trends.

4 Conclusions

The calculated bond lengths and vibrational frequencies of boric acid and borate
agree fairly well with that observed experimentally and with previous calculations,
where available, when the comparison is appropriate. The oxodihydroxoborate ion
is much more stable than the boric acid-hydroxide complex, when the latter exists.
The oxodihydroxoborate ion, if it can be observed, should have a strong vibrational
band at approximately 1400–1600 cm−1. A transition state that links the tetrahy-
droxoborate to the hydrated oxodihydroxoborate ion has been found. The addition
of water molecules lowers the barrier significantly, bringing the activation energy to
closer agreement with experiment (assuming an oxodihydroxoborate intermediate).

Acknowledgements The author acknowledges the Atlantic Computational Excellence Network

(ACEnet) for computational support.

References

1. Richens DT (1997) The chemistry of aqua ions. Wiley, Chichester

2. Oi T (2000) Calculations of reducted partition function ratios of monomeric and dimeric boric
acids and borates by the ab initio molecular orbital theory. J Nucl Sci Tech 37(2):166–172
An Ab Initio Study of Boric Acid, Borate … 175

3. Oi T (2000) Ab initio molecular orbital calculations of reduced partition function ratios of

polyboric acids and polyborate anions. Z Naturforsch A 55:623–628
4. Oi T, Yanase S (2001) Calculations of reduced partition function ratios of hydrated
monoborate anion by the ab initio molecular orbital theory. J Nucl Sci Tech 38:429–432
5. Zeebe RE (2005) Stable boron isotope fractionation between dissolved B(OH)3 and B(OH)4−.
Geochim Cosmochim Acta 69:2753–2766
6. Liu Y, Tossell JA (2005) Ab initio molecular orbital calculations for boron isotope
fractionation on boric acids and borates. Geochim Cosmochim Acta 69:3995–4006
7. Tossell JA (2005) Boric acid, “carbonic” acid, and N-containing oxyacids in aqueous
solution: ab initio studies of structure, pKa, NMR shifts, and isotopic fractionation. Geochim
Cosmochim Acta 69:5647–5658
8. Rustad JR, Bylaska EJ (2007) Ab initio calculation of isotopic fractionation in B(OH)3(aq)
and B(OH)4−(aq). J Am Chem Soc 129:2222–2223
9. Rustad JR, Bylaska EJ, Jackson VE, Dixon DA (2010) Calculation of boron-isotope
fractionation between B(OH)3(aq) and B(OH)4-(aq). Geochim Cosmochim Acta 74:2843–
2850
10. Zeebe RE, Sanyal A, Ortiz JD, Wolf-Gladrow DA (2001) A theoretical study of the kinetics
of the boric acid-borate equilibrium in seawater. Marine Chem 73:113–124
11. Kracek FC, Morey GW, Merwin HE (1938) The system, water-boron oxide. Am J Sci A
35:143–171
12. Blasdale WC, Slansky CM (1939) The solubility curves of boric acid and the borates of
sodium. J Am Chem Soc 61:917–920
13. Berger SV (1953) The crystal structure of boron oxide. Acta Chem Scand 7:611–622
14. Strong SL, Kaplow R (1968) The structure of crystalline B2O3. Acta Crystallogr B 24:1032–
1036
15. Gurr GE, Montgomery PW, Knutson CD, Gorres BT (1970) The crystal structure of trigonal
diboron trioxide. Acta Crystallogr B 26:906–915
16. Effenberger H, Lengauer CL, Parthe E (2001) Trigonal B2O3 with higher space-group
symmetry: results of a reevaluation. Monat Chem 132:1515–1517
17. Prewitt CT, Shannon RD (1968) Crystal structure of a high-pressure form of B2O3. Acta
Crystallogr B 24:869–874
18. Zachariasen WH (1934) The crystal lattice of boric acid, BO3H3. Z Kristallogr 88:150–161
19. Cowley JM (1953) Structure analysis of single crystals by electron diffraction. II. Disordered
boric acid structure. Acta Crystallogr 6:522–529
20. Zachariasen WH (1954) The precise structure of orthoboric acid. Acta Crystallogr 7:305–310
21. Shuvalov RR, Burns PC (2003) A new polytype of orthoboric acid, H3BO3-3T1. Acta
Crystallogr C 59:i47–i49
22. Tazaki H (1940) Single crystals of metaboric acid. J Sci Hiroshima Univ A 10:37–54
23. Tazaki H (1940) The structure of orthorhombic metaboric acid, HBO2(a). J Sci Hiroshima
Univ A 10:55–61
24. Peters CR, Milberg ME (1964) The refined structure of orthorhombic metaboric acid. Acta
Crystallogr 17:229–234
25. Zachariasen WH (1952) A new analytical method for solving complex crystal structures. Acta
Crystallogr 5:68–73
26. Zachariasen WH (1963) The crystal structure of monoclinic metaboric acid. Acta Crystallogr
16:385–389
27. Freyhardt CC, Wiebcke M, Felsche J (2000) The monoclinic and cubic phases of metaboric
acid (precise redeterminations). Acta Crystallogr C 56:276–278
28. Zachariasen WH (1963) The crystal structure of cubic metaboric acid. Acta Crystallogr
16:380–384
29. Konig H, Hoppe R (1977) Zur Kenntnis von Na3BO3. Z Anorg Allg Chem 434:225–232
30. Menchetti S, Sabelli C (1982) Structure of hydrated sodium borate Na2[BO2(OH)]. Acta
Crystallogr B 38:1282–1284
176 C. C. Pye

31. Block S, Perloff A (1963) The direct determination of the crystal structure of NaB(OH)42H2O.
Acta Crystallogr 16:1233–1238
32. Csetenyi LJ, Glasser FP, Howie RA (1993) Structure of sodium tetrahydroxyborate. Acta
Crystallogr C 49:1039–1041
33. Touboul M, Betourne E, Nowogrocki G (1995) Crystal structure and dehydration process of
Li(H2O)4B(OH)4.2H2O. J Solid State Chem 115:549–553
34. Zachariasen WH (1964) The crystal structure of lithium metaborate. Acta Crystallogr
17:749–751
35. Hohne E (1964) Die Kristallstruktur des LiB(OH)4. Z Chem 4:431–432
36. Fronczek FR, Aubry DA, Stanley GG (2001) Refinement of lithium tetrahydroxoborate with
low-temperature CCD data. Acta Crystallogr E 57:i62–i63
37. Onak TP, Landesman H, Williams RE, Shapiro I (1959) The B11 nuclear magnetic resonance
chemical shifts and spin coupling values for various compounds. J Phys Chem 63:1533–1535
38. Momii RK, Nachtrieb NH (1967) Nuclear magnetic resonance study of borate-polyborate
equilibria in aqueous solution. Inorg Chem 6:1189–1192
39. How MJ, Kennedy GR, Mooney EF (1969) The pH dependence of the boron-11
chemical-shift of borate-boric acid solutions. J Chem Soc D Chem Commun 267–268
40. Smith HDJ, Wiersema RJ (1972) Boron-11 nuclear magnetic resonance study of polyborate
ions in solution. Inorg Chem 11:1152–1154
41. Covington AK, Newman KE (1973) Base dissociation constant of the borate ion from 11B
chemical shifts. J Inorg Nucl Chem 35:3257–3262
42. Henderson WG, How MJ, Kennedy GR, Mooney EF (1973) The interconversion of aqueous
boron species and the interaction of borate with diols: a 11B N.M.R. study. Carbohydrate Res
28:1–12
43. Janda R, Heller G (1979) 11B–NMR-spektroskopische Untersuchungen an waessrigen
Polyboratloesungen. Z Naturforsch B 34:1078–1083
44. Epperlein BW, Lutz O, Schwenk A (1975) Fourier-Kernresonanzuntersuchungen an 10B und
11B in Waessriger Loesung. Z Naturforsch A 30:955–958
45. Salentine CG (1983) High-field 11B NMR of alkali borates. Aqueous polyborate equilibria.
Inorg Chem 22:3920–3924
46. Frisch MJ et al (2004) Gaussian 03, Revision D.02. Gaussian Inc., Wallingford, CT
47. Gupta A, Tossell JA (1981) A theoretical study of bond distances, X-ray spectra and electron
density distributions in borate polyhedra. Phys Chem Miner 7:159–164
48. Gupta A, Tossell JA (1983) Quantum mechanical studies of distortions and polymerization of
borate polyhedra. Am Miner 68:989–995
49. Zhang ZG, Boisen MBJ, Finger LW, Gibbs GV (1985) Molecular mimicry of the geometry
and charge density distribution of polyanions in borate minerals. Am Miner 70:1238–1247
50. Zaki K, Pouchan C (1995) Vibrational analysis of orthoboric acid H3BO3 from ab initio
second-order perturbation calculations. Chem Phys Lett 236:184–188
51. Tian SX, Xu KZ, Huang M-B, Chen XJ, Yang JL, Jia CC. Theoretical study on infrared
vibrational spectra of boric-acid in gas-phase using density functional methods. J Mol Struct
(Theochem) 459:223–227, 459
52. Tachikawa M (2004) A density functional study on hydrated clusters of orthoboric acid, B
(OH)3(H2O)n (n = 1–5). J Mol Struct (Theochem) 710:139–150
53. Stefani D, Pashalidis I, Nicolaides AV (2008) A computational study of the conformations of
the boric acid (B(OH)3), its conjugate base ((HO)2BO–) and borate anion (B(OH)4–). J Mol
Struct (Theochem) 853:33–38
54. Zhou Y, Fang C, Fang Y, Zhu F (2011) Polyborates in aqueous borate solution: a Raman and
DFT theory investigation. Spectrochim Acta A 83:82–87
55. Ananthakrishnan R (1936) The Raman spectra of some boron compounds (methyl borate,
ethyl borate, boron tri-bromide and boric acid). Proc Indian Acad Sci A 4:74–81
56. Ananthakrishnan R (1937) The Raman spectra of crystal powders. IV. Some organic and
inorganic compounds. Proc Indian Acad Sci A 5:200–221
57. Hibben JH (1938) The constitution of some boric oxide compounds. Am J Sci A 35:113–125
An Ab Initio Study of Boric Acid, Borate … 177

58. Mitra SM (1938) Raman effect in boric acid and in some boron compounds. Ind J Phys
12:9–14
59. Kahovec L (1938) Studien zum Raman-Effekt. Mitteilung LXXXV. Borsauere und Derivate.
Z Phys Chem 40:135–145
60. Miller FA, Wilkins CH (1952) Infrared spectra and characteristic frequencies of inorganic
ions. Anal Chem 24:1253–1294
61. Bethell DE, Sheppard N (1955) The infra-red spectrum and structure of boric acid. Trans
Faraday Soc 51:9–15
62. Servoss RR, Clark HM (1957) Vibrational spectra of normal and isotopically labeled boric
acid. J Chem Phys 26:1175–1178
63. Maya L (1976) Identification of polyborate and fluoropolyborate ions in solution by Raman
spectroscopy. Inorg Chem 15:2179–2184
64. Maeda M, Hirao T, Kotaka M, Kakihana H (1979) Raman spectra of polyborate ions in
aqueous solution. J Inorg Nucl Chem 41:1217–1220
65. Janda R, Heller G (1979) Ramanspektroskopische Untersuchungen an festen und in Wasser
geloesten Polyboraten. Z Naturforsch B 34:585–590
66. Ogden JS, Young NA (1988) The characterisation of molecular boric acid by mass
spectrometry and matrix isolation infrared spectroscopy. J Chem Soc Dalton Trans 1645–
1652
67. Gilson TR (1991) Characterization of ortho- and meta-boric acids in the vapour phase.
J Chem Soc Dalton Trans 2463–2466
68. Andrews L, Burkholder TR (1992) Infrared spectra of molecular B(OH)3 and HOBO in solid
argon. J Chem Phys 97:7203–7210
69. Gupta A, Swanson DK, Tossell JA, Gibbs GV (1981) Calculation of bond distances,
one-electron properties and electron density distributions in first-row tetrahedral hydroxy and
oxyanions. Am Miner 66:601–609
70. Hess AC, McMillan PF, O’Keeffe M (1988) Torsional barriers and force fields in H4TO4
molecules and molecular ions (T = C, B, Al, Si). J Phys Chem 92:1785–1791
71. Nielsen JR, Ward NE (1937) Raman spectrum and structure of the metaborate ion. J Chem
Phys 5:201
72. Edwards JO, Morrison GC, Ross VF, Schultz JW (1955) The structure of the aqueous borate
ion. J Am Chem Soc 77:266–268
73. Oertel RP (1972) Raman study of aqueous monoborate-polyol complexes. Equilibria in the
monoborate-1,2-ethanediol system. Inorg Chem 11:544–549
74. Liu Z, Gao B, Hu M, Li S, Xia S (2003) FT-IR and Raman spectroscopic analysis of hydrated
cesium borates and their saturated aqueous solution. Spectrochim Acta A 59:2741–2745
75. Zhu FY, Fang CH, Fang Y, Zhou YQ, Ge HW, Liu HY (2014) Structure of aqueous
potassium metaborate solution. J Mol Struct 1070:80–85
76. Attina M, Cacace F, Occhiucci G, Ricci A (1992) Gaseous borate and polyborate anions.
Inorg Chem 31:3114–3117
77. Waton G, Mallo P, Candau SJ (1984) Temperature-jump rate study of the chemical relaxation
of aqueous boric acid solutions. J Phys Chem 88:3301–3305
Construction of a Potential Energy Surface
Based on a Diabatic Model for
Proton Transfer in Molecular Pairs

Yuta Hori, Tomonori Ida and Motohiro Mizuno

Abstract We propose a simple construction method of the potential energy surface

based on diabatic model for proton transfer in molecular pairs. Assuming two-state
valence bond electronic wave functions as a diabatic basis, the diagonal and
non-diagonal matrix elements in diabatic potential of water, ammonia, and imida-
zole pairs were obtained. The validity of the construction procedure was confirmed
by comparing two adiabatic potentials: one was transformed from the obtained
diabatic potential and another was calculated by DFT calculation. Diabatic poten-
tials were also obtained using fewer reference points than conventional methods at
various intermolecular distances. Finally, we discuss the resulting diabatic potential
and non-diagonal elements in detail.

Keywords Potential energy surface ⋅ Proton transfer ⋅ Diabatic

1 Introduction

One of the important steps in the theoretical treatment of chemical reactions is

representation of the potential energy surface (PES) [1, 2]. Most approaches to
chemical reactions analyze the PES by quantum chemical calculations derived
under the Born-Oppenheimer approximation, also known as the adiabatic PES.
Once the adiabatic potential is obtained, the scattering cross section, reaction
constant, and reaction path, which are important for understanding chemical reac-
tions, can be obtained from the potential [3]. Though ab initio quantum chemical
calculations are becoming possible for large molecular systems, however, accurate
PES calculations for understanding chemical reaction tend to be unfeasible. In
addition, though analytical function for PES requires to analyze the reaction, the
global function has not been known.

Y. Hori (✉) ⋅ T. Ida ⋅ M. Mizuno

Chemistry Course, Division of Material Chemistry, Graduate School of Natural
Science and Technology, Kanazawa University, Kanazawa 920-1192, Japan
e-mail: [email protected]

© Springer International Publishing AG, part of Springer Nature 2018 179

Y. A. Wang et al. (eds.), Concepts, Methods and Applications of Quantum Systems
in Chemistry and Physics, Progress in Theoretical Chemistry and Physics 31,
https://doi.org/10.1007/978-3-319-74582-4_9
180 Y. Hori et al.

Rather than examine the adiabatic PES, another approach to chemical reactions is
to analyze the diabatic PES. In contrast to the adiabatic potential, the diabatic
potential presents electronic states that change constantly to confine the eigenstates
of the electronic Hamiltonian. The approaches using diabatic picture have been
utilized various area in chemistry and physics where the coupling between nuclei
and electrons such as vibronic coupling [4–6]. There are some approaches to
describing the diabatic potential [7], constructing using some valence bond
(VB) electronic wave functions [8–11]. Especially, empirical valence bond
(EVB) [12] or multistate empirical valence bond (MS-EVB) [13] approach extended
EVB is used the molecular mechanical functions to construct the PES and applied to
the molecular dynamics (MD) simulations for many proton transfer systems [12, 14–
29]. Furthermore, quantum dynamical approach using molecular mechanical func-
tions (double Morse potential) for proton transfer was also performed [30]. How-
ever, to construct PES using diabatic potentials corresponding to reactant and
product states, which is based on VB picture, is important for understanding
chemical reactions in terms on chemical bond character. Although VB structures are
usually not orthogonal, in this study we consider orthonormal VB structures. To
basic idea, consider a two-state VB electronic wave function as the diabatic basis:

jψ⟩ = c1 jϕ1 ⟩ + c2 jϕ2 ⟩, ð1Þ

where jϕ1 ⟩ and jϕ2 ⟩ are VB wave functions that describes the electronic structure of
the reactant and product states, respectively. The lowest adiabatic potential energy
Vad is then given by the lower root of the 2 × 2 secular equation; specifically:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 di 
di 2
V di + V22
di
V11 − V22
V ad = 11 − + V12 di 2 , ð2Þ
2 2

where
di
V11 = ⟨ϕ1 jH jϕ1 ⟩, ð3Þ
di
V22 = ⟨ϕ2 jH jϕ2 ⟩, ð4Þ
di
V12 = ⟨ϕ1 jH jϕ2 ⟩. ð5Þ

di di
V11 and V22 are the potential energies for the two VB structures of the reactant
di di di
and product states, respectively. In this approach V11 , V22 and V12 function forms
including parameters can be obtained to fit in experimental or ab initio data. In these
di di
works, V11 and V22 are related to use of molecular mechanics potential functions,
especially, which are taken as the harmonic normal-mode potential or Morse
potential etc. [14, 15, 17, 18, 25, 26]. On the other hand, although the selection of
di di
V12 is less obvious, Gaussian function as V12 proposed by Chang and Millar [14]
Construction of a Potential Energy Surface … 181

has been widely used. However, they has not been confirmed whether the obtained
diabatic potentials produce the reliable adiabatic potentials or not, although these
functions are obtained to fit in some data. Additionally, to obtain these analytical
functions with parameters uniquely or using fewer data is important to construct the
PES for various chemical reactions. Therefore, a proposal of simple method for
di di
light or more uniquely construction of the diabatic potentials (V11 and V22 ) and
di
non-diagonal matrix element (V12 ) using the analytical functions is important to
analyze the chemical reaction and it can be widely applied to describing the large
molecular systems such as proteins.
In this study, we focus on one-dimensional proton transfer models and suggest a
simple construction method of global PES for intermolecular proton transfer by use
di di di
of Morse potential as V11 and V22 and Gaussian function as V12 in the diabatic
potential matrix and confirm the validity to use these potential functions. In addi-
tion, we investigate whether it is possible to apply to the proton transfer for various
intermolecular distance. Here, we focused on the proton-bonded four bimolecular
models for ammonia (AmH+-Am) and imidazole (ImH+-Im) pairs as symmetrical
homo-molecular proton transfer systems and for imidazole-ammonia (ImH+-Am)
and ammonia-water (AmH+-Wat) pairs as asymmetrical hetero-molecular systems,
the four model structures of which were shown in Fig. 1. We investigate the
portability to use the Morse potential and Gaussian function as the diabatic potential
matrix elements by comparison the transformed adiabatic potential from the dia-
batic one with the calculated by DFT calculation in these systems. Finally, we
di di
discuss the proton transfer characters using obtained diabatic potential (V11 and V22 )
di
and non-diagonal elements (V12 ) for homo- and hetero-molecular pairs.

(a) AmH+-Am (b) ImH+-Im

(c) ImH+-Am (d) AmH+-Wat

Fig. 1 Four model structures of the proton-bonded a AmH+-Am, b ImH+-Im, c AmH+-Wat, and
d ImH+-Am pairs
182 Y. Hori et al.

2 Theoretical and Calculation Methods

2.1 Calculation Models

In the construction of PES for proton transfer, we focused on the four proton transfer
models: (a) AmH+-Am, (b) ImH+-Im as homo-molecular pairs, and (c) ImH+-Am,
(d) AmH+-Wat as hetero-molecular pairs (Fig. 1). Figure 2 shows also these models
and potential energy coordinates. For homo-molecular pairs, coordinate R denotes
the intermolecular distance and x is the translating proton position that defines as a
displacement from the center of the intermolecular distance. For hetero-molecular
pairs, r is used as the displacement between a proton-bonded atom and the proton.

2.2 Diabatic Potential Functions (V di di di

11 , V 22 and V 12 )

In this study, we constructed a PES for proton transfer by using diabatic picture.
Diabatic potentials can be constructed using a variety of valence bond (VB) con-
figurations [7–11]. Here, we considered a two-state VB electronic wave function as
the diabatic basis corresponding to the reactant and product states, i.e., Equa-
tion (1). In particular, the EVB approach can be related to use of molecular
mechanical potential functions describing the molecular vibration, which is rational
for understanding chemical reaction. In most cases, the diagonal matrix elements
di di
(V11 and V22 ) are taken as the harmonic normal-mode or Morse potentials [14, 15,
17, 18, 25, 26]. On the other hand, Chang and Millar [14] suggested the use of a
di
generalized Gaussian function as the non-diagonal matrix element (V12 ).
di di
Therefore, we selected the Morse function as the V11 and V22 , and Gaussian
di
function as the V12 to construct PES for proton transfer. For homo-molecular pairs,
di di di
V11 , V22 and V12 are explicitly defined as

+
Fig. 2 Proton transfer model (a) AmH -Am (b) ImH+-Im
and potential energy
coordinates for a AmH+-Am, H x H H x
b ImH+-Im, c AmH+-Wat, H
N H
N H N N H N N
and d ImH+-Am H
H R H R

(c) ImH+-Am (d) AmH +-Wat

H r H
H r H
N
N H N H N H O
H
R H H H
R
Construction of a Potential Energy Surface … 183

 2
di
V11 ðx; RÞ = D 1 − e − kðx + x0 Þ + c, ð6Þ

 2
di
V22 ðx; RÞ = D 1 − ekðx − x0 Þ + c, ð7Þ
 
di
V12 ðx; RÞ = A exp − bx2 . ð8Þ

di di di
For hetero-molecular pairs, V11 , V22 and V12 are defined as
 2
di
V11 ðr; RÞ = D1 1 − e − k1 ðr − r0 Þ + c, ð9Þ

 0 2
di
V22 ðr; RÞ = D2 1 − e − k2 ðR − r − r0 Þ + c + D3 , ð10Þ
 
di
V12 ðr; RÞ = A exp − bðr − rc Þ2 . ð11Þ

We constructed the PES by optimization of the potential parameters in these

di di di di di
functions (V11 , V22 and V12 ). For V11 and V22 , parameters D, D1 , and D2 are binding
0
energies; x0 , r0 , r0 are the equilibrium bond lengths; k, k1 , and k2 are the decay
di
constants. For V12 , parameter A is the amplitude of the Gaussian function;
parameter b corresponds to the spread of the function; rc is the point of the max-
imum value of the function.

2.3 Optimization of Potential Parameters

Optimization of the potential parameters was conducted according to the following

procedures.
di di
First, we estimated the parameters of V11 and V22 . Parameters D, D1 , and D2
were estimated by dissociation energy of the proton corresponding to one side of
0
the molecular pairs, i.e., ammonium, imidazolium, and oxonium ions. x0 , r0 , r0
were used the equilibrium bond lengths of these molecules. The estimated values of
these parameters are shown in Table 1. For hetero-molecular pairs, parameter D3
 0 
was obtained from the difference between V ad ðr0 ; RÞ and V ad r0 ; R :
 
D3 = V ad r0′ ; R − V ad ðr0 ; RÞ. ð12Þ

di di
The remaining parameters k, k1 , and k2 included in V11 and V22 are considered to
be dependent on the R coordinate. These parameters were estimated by comparison
with the adiabatic potential obtained from DFT calculations according to Eq. (2).
di di
V11 and V22 describes molecular vibration for the reactant and product states,
di di
respectively. Therefore, we assumed that V11 and V22 reproduced the adiabatic
184 Y. Hori et al.

Table 1 Diabatic potential energy parameters for (a) AmH+-Am, (b) ImH+-Im, (c) ImH+-Am,
and (d) AmH+-Wat
(a) AmH+-Am (b) ImH+-Im
−1
D(kJ mol ) r0 (Å) D(kJ mol−1) r0 (Å)
664.758 1.027 668.81 1.0148
(c) ImH+-Am
D1 (kJ mol−1) D2 (kJ mol−1) r0 (Å) 0
r0 (Å)
668.81 664.758 1.015 1.027
(d) AmH+-Wat
D1 (kJ mol−1) D2 (kJ mol−1) r0 (Å) 0
r0 (Å)
725.69 593.731 1.027 0.977

potential, when the proton position was closer to a binding atom than a stable point
di
(r < r0 ). Here, we estimated the parameter k by comparing them with the V22 and
adiabatic potential at x = 2x0 for homo-molecular pairs for fixed R. In the same way,
for hetero-molecular pairs, k1 and k2 were determined by comparing with the
0
adiabatic potential at r = 2r0 − R ̸2 ≡ rk1 and r = 3R ̸2 − 2r0 ≡ rk2 , respectively, for
fixed R. These relations are explicitly defined as
di
V22 ð2x0 Þ = V ad ð2x0 Þ, ð13Þ
di
V11 ðrk1 Þ = V ad ðrk1 Þ, ð14Þ
di
V22 ðrk2 Þ = V ad ðrk2 Þ. ð15Þ

for homo- and hetero-molecular pairs, respectively. Parameters k, k1 , and k2 are

then denoted as
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi!
1 V ad ð2x0 Þ − c
k = ln 1 + , ð16Þ
x0 D
0 sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi1
1 V ad ðrk1 Þ − cA
k1 = ln@1 + , ð17Þ
r 0 − r k1 D1

0sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi1
1 V ad ðrk2 Þ − c − D3 A
k2 = 0 ln@1 + . ð18Þ
r k2 + r 0 − R D2

di
Subsequently, parameters A and b of V12 depending on the coordinate R were
estimated. These parameters were also estimated by comparison with the adiabatic
potential obtained from DFT calculations and using obtained diabatic potentials
di di
(V11 and V22 ). Determination of these parameters was conducted according to the
following steps for each pairs.
Construction of a Potential Energy Surface … 185

First, parameter A was estimated. Because this parameter represents the ampli-
tude of the Gaussian type function, parameter b vanishes at x = 0 for
homo-molecular pairs. Thus, the relationship between the adiabatic and diabatic
potentials at x = 0 and fixed R is given by
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 di di ð0Þ 2

di
V11 ð0Þ + V22
di
ð 0Þ V11 ð0Þ − V22
− + V12 di 2 ð0Þ = V ad ð0Þ. ð19Þ
2 2

di di
Parameter A was then estimated by using obtained V11 and V22 , which was
expressed as
di
A = V11 ð0Þ − V ad ð0Þ, ð20Þ

for homo-molecular pairs. For hetero-molecular pair, parameter A was estimated at

di di
the cross point (r = rc ) of V11 and V22 substituted D3 = 0. Parameter A was then
expressed as
di
A = V11 ðrc Þ − Vad ðrc Þ, ð21Þ

for hetero-molecular pairs.

di
Next, the remaining parameter b of V12 was estimated. This parameter corre-
sponds to the spread of the Gaussian type function. For homo-molecular pair, the
midpoint between the local minimum (x0 ) and local maximum ðx = 0Þ is assumed to
di
define the spread of V12 . Therefore, the relationship between the adiabatic and
diabatic potentials at x = x0 ̸2 and fixed R is also given by

 x0   x0  sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi
 di x0   
di x0 2 x  x 
di
V11 di
+ V22 V11 2 − V22 0 0
2 2
− 2
+ V12 di 2 = V ad . ð22Þ
2 2 2 2

According to this relation, parameter b is expressly defined by

" (     )#
di x0
V11 di x0  2 V di x0  − V di x0 2
2 2 + V22 2 ad x0
b = 2 lnA − ln
2
−V − 11 2 22 2
.
x0 2 2 2
ð23Þ

In the same way, for hetero-molecular pairs, b was determined by comparing

with the adiabatic potential at r = ðrc + r0 Þ ̸ 2 ≡ rb for fixed R, which was defined by
" ( 2  di 2 )#
1 1 di
V11 ðrb Þ + V22
di
ðrb Þ V11 ðrb Þ − V22
di
ðrb Þ
b= ln A − ln
2
− V ðrb Þ −
ad
.
ðrb − rc Þ2 2 2 2
ð24Þ
186 Y. Hori et al.

2.4 Adiabatic Potential Energy Calculation

To obtain the adiabatic PES V ad ðx, RÞ or V ad ðr, RÞ for the proton transfer reaction,
quantum chemical calculations were performed with stepwise movement of the
proton and intermolecular distances by 0.02 and 0.05 Å, respectively. The
geometries of the molecules were then kept in the minimum energy structures,
except for the proton position (x or r) and the intermolecular distance (R). Minimum
energy structures, as shown in Fig. 1, were determined by geometrical optimiza-
tion. Finally, the adiabatic potential energies were fitted using a polynomial series
function of the fourth order with respect to x or r and of the third order with respect
to R [31, 32]. In previous work [33, 34], the proton transfer of ImH+-Im systems
was discussed with the B3LYP approach. Therefore, all DFT calculations were
performed at the B3LYP/aug-cc-pVDZ level using the Gaussian-09 package [35].

3 Results and Discussion

We first estimated the potential parameters k, k1 , k2 , and D3 of V11di di

and V22 using
Eqs. (12) and (16)–(18) and obtained parameters (Table 1) for (a) AmH+-Am,
(b) ImH+-Im, (c) ImH+-Am, and (d) AmH+-Wat. The parameters A and b were then
di
estimated using Eqs. (20), (21), (23), and (24). Figure 3 shows the computed V11 ,
di di
V22 , and V12 values using obtained the potential parameters at some intermolecular
distance for (a) AmH+-Am, (b) ImH+-Im, (c) ImH+-Am and (d) AmH+-Wat,
respectively. To compare with the adiabatic potential by DFT calculation, Figs. 4
and 5 show the transformed adiabatic potential derived from the diabatic potential
di di di
matrix elements (V11 , V22 , and V12 ) using Eq. (2) and one obtained from DFT.
Although the potentials using diabatic model not reproduce the ones using DFT
around the local minimum of the potentials, however, the figures show that the
transformed adiabatic potentials are qualitatively in good agreement with those
calculated by DFT calculations for all proton transfer systems at various inter-
molecular distance R. Furthermore, to compare with the present work and DFT data
Table 2 shows the coefficient of determination (R-squared). Because the values of
R-squared for all models were close to unity, we confirmed the validity of our
approach. Thus, it is indicates that PES for various proton transfer systems can
di di
qualitatively reproduce by using V11 and V22 with Morse potential described the
di
vibrational motion and V12 with the Gaussian function by assumption of two-state
VB wave functions as a diabatic basis.
In our approach, parameters k, A, and b can be uniquely determined not nec-
essary to fit the potential energy. Especially, the construction procedures of the
Construction of a Potential Energy Surface … 187

(a) AmH+-Am
R=2.70 R=3.05
300 300

Energy (kJ/mol)
Energy (kJ/mol)

200 200

100 100

0 0
-0.6 -0.4 -0.2 0 0.2 0.4 0.6 -0.7 -0.5 -0.3 -0.1 0.1 0.3 0.5 0.7
x( ) x( )
(b) ImH+-Im
R=2.66 R=3.01
300 300

Energy (kJ/mol)
Energy (kJ/mol)

200 200

100 100

0 0
-0.6 -0.4 -0.2 0 0.2 0.4 0.6 -0.8 -0.4 0 0.4 0.8
x( ) x( )
(c) ImH+-Am
R=2.77 R=3.12
200 300
Energy (kJ/mol)
Energy (kJ/mol)

250
150
200
100 150
100
50
50
0 0
0.8 1 1.2 1.4 1.6 1.8 2 0.8 1.2 1.6 2 2.4
r( ) r( )
(d) AmH+-Wat
R=2.69 R=3.04
400 400
Energy (kJ/mol)

Energy (kJ/mol)

300 300

200 200

100 100

0
0.8 1 1.2 1.4 1.6 1.8 2 0
0.8 1.2 1.6 2 2.4
r( ) r( )

di di
Fig. 3 Diagonal matrix elements, V11 (blue line) and V22 (red line), and non-diagonal matrix
di
element V12 (yellow line) in diabatic potential matrix computed using obtained potential
parameters at some intermolecular distance R for a AmH+-Am, b ImH+-Im, c ImH+-Am, and
d AmH+-Wat
188 Y. Hori et al.

(a) AmH+-Am R=2.60 R=2.70

180 180

Energy (kJ/mol)
Energy (kJ/mol)

140 140

100 100

60 60

20 20

-20 -20
-0.6 -0.4 -0.2 0 0.2 0.4 0.6 -0.6 -0.4 -0.2 0 0.2 0.4 0.6
x( ) x( )

R=2.85 R=3.05
180 180
Energy (kJ/mol)

Energy (kJ/mol)
140 140
100 100

60 60

20 20

-20 -20
-0.6 -0.4 -0.2 0 0.2 0.4 0.6 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8
x( ) x( )

(b) ImH+-Im R=2.56 R=2.66

260
220
Energy (kJ/mol)

180
Energy (kJ/mol)

180
130
140
80 100
60
30
20
-20 -20
-0.5 -0.3 -0.1 0.1 0.3 0.5 -0.6 -0.4 -0.2 0 0.2 0.4 0.6
x( ) x( )

R=2.81 R=3.01
180 220
Energy (kJ/mol)

Energy (kJ/mol)

140 180

100 140
100
60
60
20
20
-20 -20
-0.6 -0.4 -0.2 0 0.2 0.4 0.6 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8
x( ) x( )

Fig. 4 Transformed adiabatic potential derived from the diabatic potentials using Eq. (2) (red
line) and from DFT (blue dots) at some intermolecular distance R for a AmH+-Am and b ImH+-Im

diabatic potential employed four reference points (x = x0 , 2x0 , 0 and x0 ̸2) at fixed
R for homo-molecular pairs, while five reference points (r = r0 , rk1 , rk2 , rc and rb )
for hetero-molecular pairs. Thus, the whole potentials of proton transfer are
described by using energies of 40 or 50 reference points, while the number of DFT
data points to describe the whole adiabatic potential required approximately 500
points. Therefore, the PES describing the entire proton transfer system for diabatic
picture can be obtained using less than one-tenth of the reference points required for
Construction of a Potential Energy Surface … 189

(a) ImH+-Am R=2.67 R=2.77

200
200

Energy (kJ/mol)
160
Energy (kJ/mol)

160
120 120

80 80

40 40
0 0
-40 -40
0.8 1 1.2 1.4 1.6 1.8 2 0.8 1 1.2 1.4 1.6 1.8 2 2.2
r( ) r( )

R=2.92 R=3.12
200 200
Energy (kJ/mol)

160

Energy (kJ/mol)
160
120 120
80 80
40 40
0 0
-40 -40
0.8 1 1.2 1.4 1.6 1.8 2 2.2 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4

r( ) r( )

(b) AmH +-Wat R=2.59 R=2.69

550
450
Energy (kJ/mol)
Energy (kJ/mol)

450
350
350
250
250
150 150
50 50
-50 -50
0.8 1 1.2 1.4 1.6 1.8 2 0.8 1 1.2 1.4 1.6 1.8 2 2.2
r( ) r( )

R=2.84 R=3.04
450 450
Energy (kJ/mol)

Energy (kJ/mol)

350 350

250 250
150 150
50 50
-50 -50
0.8 1 1.2 1.4 1.6 1.8 2 2.2 0.8 1.2 1.6 2 2.4
r( ) r( )

Fig. 5 Transformed adiabatic potential derived from the diabatic potentials using Eq. (2) (red
line) and from DFT (blue dots) at some intermolecular distance R for c ImH+-Am and d AmH+-
Wat

Table 2 Coefficient of determination (R-squared) between adiabatic potential derived from the
diabatic potentials and from DFT
(a) AmH+-Am (b) ImH+-Im (c) ImH+-Am (d) AmH+-Wat
R-squared 0.966 0.974 0.932 0.965
190 Y. Hori et al.

Fig. 6 Intermolecular 13.2

dependence of potential (a) AmH-Am
di
parameter b of V12 for 11.2
(b) ImH-Im
a AmH -Am, b ImH+-Im,
+
9.2 (c) ImH-Am
c ImH+-Am, and d AmH+-

-2)
Wat 7.2 (d) AmH-Wat

b(
5.2

3.2

1.2
2.5 2.6 2.7 2.8 2.9 3 3.1 3.2
R( )

the adiabatic picture. Furthermore, our method can be applied to proton transfer
system even when the transition state (TS) cannot be calculated, although the
information of TS requires the construction of potential in the previous work.
Therefore, it is concluded that our procedures are useful for PES construction by
using the diabatic picture for proton transfer systems and can be applied to large
molecular systems such as proteins.
Finally, we discuss the obtained potential parameters. Figure 6 shows that
intermolecular dependence of potential parameter b of V12 di
for (a) AmH+-Am,
+ + +
(b) ImH -Im, (c) ImH -Am, and (d) AmH -Wat. The values of parameter b, which
describe the spread of the Gaussian function, decreased as R increased. This result
di
indicates that the non-diagonal matrix element (V12 ) is broadly distributed along the
proton transfer coordinate and the bond mixture between the reactant and the
di
product states occurs over a wide range, not only at the TS. In addition, because V12
is broadly distributed along the intermolecular distance, the proton can be formed
mixture between reactant and product states and transferred at the location formed
di
hydrogen bond. To clarify the effect of non-diagonal matrix element V12 , the ratio
di
of amplitude for the V12 (parameter A) divided by the crossing point energy of the
di di
V11 and V22 was estimated, the results of which were shown in Fig. 7. According to

Fig. 7 Ratio of amplitude for 1

non-diagonal matrix element
divided by the crossing point 0.8
A/V11di(0 or rc)

energy of the diabatic (a) AmH-Am

potential at x = 0 for 0.6 (b) ImH-Im
homo-molecular pairs or (c) ImH-Am
r = rc for hetero-molecular 0.4
(d) AmH-Wat
pairs, i.e.,
11 ðx = 0 or r = rc Þ
A ̸Vdi 0.2

0
2.5 2.6 2.7 2.8 2.9 3 3.1 3.2
R( )
Construction of a Potential Energy Surface … 191

di
Eq. (2), V12 contributes to the stability of the adiabatic potential V ad and the
A ̸V11
di
ðx = 0 or r = rc Þ is assumed to be determined the rate of contribution to the
possibility of proton transfer. For both homo-molecular pairs, the value of A ̸ V11 di

decreased as R increased and was approximately 50 to 80% at various inter-

molecular distance. Thus, the proton can easily transfer at the location formed
hydrogen bond for homo-molecular pairs. On the other hand, for both
hetero-molecular pairs, the value of A ̸V11 di
was constantly about 20% over wide
range intermolecular distance, which was lower than homo-molecular pairs for all
intermolecular distance. It indicates that the proton for homo-molecular pairs can
easily transfer than for hetero-molecular pairs. From the above discussion, we find
that the obtained potentials give the qualitative information about proton transfer
even if a simple two-state diabatic model is used.

4 Conclusion

Potential energy surface (PES) is an important theoretical approach for under-

standing chemical reactions. Diabatic potentials are used to understand proton
transfer reactions. Especially, EVB approach based on the diabatic picture is used to
the molecular mechanical function to construct PES and applied to many appli-
cations including molecular dynamics simulation. In this paper, we constructed the
PES based on diabatic model for proton transfer models: (a) AmH+-Am, (b) ImH+-
Im, (c) ImH+-Am and (d) AmH+-Wat. We confirmed that Morse potentials as the
di di
diagonal matrix element (V11 and V22 ) and Gaussian function as the non-diagonal
di
matrix element (V12 ), which are important to apply widely to understanding the
chemical reaction including the classical or quantum dynamics simulations,
described the proper PES for proton transfer. In addition, we proposed a simple
method to uniquely construct the diabatic potentials using these analytical func-
tions. The diabatic potentials at various intermolecular distance were obtained using
fewer reference points than the adiabatic potentials to describe an entire proton
transfer system. Therefore, our construction method is useful and can be applied to
the large molecular systems such as proteins.
di
From the values of estimated the potential parameters, V12 was broadly dis-
tributed and the proton-bonded mixture between the reactant and product states
occurred over a wide range of reaction coordinates, and not only at the TS. Fur-
thermore according to the relation between diagonal and non-diagonal matrix ele-
ments, it is found that the proton for homo-molecular pairs can transfer easily than
for hetero-molecular pairs.

Acknowledgements This work was supported by a Grant-in-Aid for Scientific Research

(23310063, 26286002) from the Ministry of Education, Science and Technology, Government of
Japan.
192 Y. Hori et al.

References

1. Truhlar DG, Steckler R, Gordon MS (1987) Chem Rev 87:217–236

2. Schatz GC (1989) Rev Mod Phys 61:669–688
3. Levine RD (2009) Molecular reaction dynamics. Cambridge University Press
4. Worth GA, Cederbaum LS (2004) Annu Rev Phys Chem 55:127–158
5. Köppel H, Domcke W, Cederbaum LS (1984) Adv Chem Phys 57:59–246
6. Nicolas G, Gadea FX (1999) J Chem Phys 111:10537
7. Voorhis TV et al (2010) Annu Rev Phys Chem 61:149–170
8. Åqvist J, Warshel A (1993) Chem Rev 93:2523–2544
9. Mo Y, Gao J (2000) J Phys Chem A 104:3012–3020
10. Truhlar DG (2007) J Comp Chem 28:73–86
11. Song L, Gao J (2008) J Phys Chem A 112:12925–12935
12. Warshel A, Weiss RM (1980) J Am Chem Soc 102:6218–6226
13. Lobaugh J, Voth GA (1996) J Chem Phys 104:2056
14. Chang Y-T, Miller WH (1990) J Phys Chem 94:5884–5888
15. Hinsen K, Roux BJ (1997) Comp Chem 18:368–380
16. Minichino C, Voth GA (1997) J Phys Chem B 101:4544–4552
17. Sagnella DE, Tuckerman ME (1998) J Chem Phys 108:2073–2083
18. Vuilleumier R, Borgis D (1998) Chem Phys Lett 284:71–77
19. Wang Y, Gunn JR (1999) Int J Quantum Chem 73:357–367
20. Schmitt UW, Voth GA (1999) J Chem Phys 111:9361–9381
21. Čuma M, Schmitt UW, Voth GA (2000) Chem Phys 258:187–199
22. Čuma M, Schmitt UW, Voth GA (2001) J Phys Chem A 105:2814–2823
23. Brancato G, Tuckerman ME (2005) J Chem Phys 122:224507
24. Hong G, Rosta E, Warshel AJ (2006) Phys Chem B 110:19570–19574
25. Lammers S, Lutz S, Meuwly MJ (2008) Comp Chem 29:1048–1063
26. Chen H, Yan T, Voth GA (2009) J Phys Chem A 113:4507–4517
27. Feng S, Voth GA (2011) J Phys Chem B 115:5903–5912
28. Kamerlin SCL, Warshel A (2011) IREs Comput Mol Sci 1:30–45
29. Li A, Cao Z, Li Y, Yan T, Shen P (2012) J Phys Chem B 116:12793–12800
30. Marinica DC, Gaigeot M-P, Borgis D (2006) Chem Phys Lett 423:390–394
31. Jaroszewski L, Lesyng B, Tanner JJ, McCammon JA (1990) Chem Phys Lett 175:282–288
32. Jaroszewski L et al (1993) J Mol Struct THEOCHEM 283:57–62
33. Tatara W et al (2003) J Phys Chem A 107:7827–7831
34. Mangiatordi GF et al (2011) J Phys Chem A 115:2627–2634
35. Frisch MJ, Trucks GW, Schlegel HB, Scuseria GE, Robb MA, Cheeseman JR, Scalmani G,
Barone V, Mennucci B, Petersson GA, Nakatsuji H, Caricato M, Li X, Hratchian HP,
Izmaylov AF, Bloino J, Zheng G, Sonnenberg JL, Hada M, Ehara M, Toyota K, Fukuda R,
Hasegawa J, Ishida M, Nakajima T, Honda Y, Kitao O, Nakai H, Vreven T, Montgomery Jr.
JA, Peralta JE, Ogliaro F, Bearpark M, Heyd JJ, Brothers E, Kudin KN, Staroverov VN,
Kobayashi R, Normand J, Raghavachari K, Rendell A, Burant JC, Iyengar SS, Tomasi J,
Cossi M, Rega N, Millam JM, Klene M, Knox JE, Cross JB, Bakken V, Adamo C,
Jaramillo J, Gomperts R, Stratmann RE, Yazyev O, Austin AJ, Cammi R, Pomelli C,
Ochterski JW, Martin RL, Morokuma K, Zakrzewski VG, Voth GA, Salvador P,
Dannenberg JJ, Dapprich S, Daniels AD, Farkas Ö, Foresman JB, Ortiz JV, Cioslowski J,
Fox DJ (2009) Gaussian 09, (Revision C.01). Gaussian, Inc., Wallingford CT
Ab Initio Investigations of Stable
Geometries of the Atmospheric Negative
Ion NO3−(HNO3)2 and Its Monohydrate

Atsuko Ueda, Yukiumi Kita, Kanako Sekimoto

and Masanori Tachikawa

Abstract The possible stable geometries of the atmospheric negative core ion
NO3− ðHNO3 Þ2 and its monohydrate were theoretically investigated with the second
order Møller-Plesset perturbation theory (MP2) in consideration of the effect of
electron correlation. For both ionic clusters, we obtained the different stable
geometries from the previous study by Drenck and coworkers (Int J Mass Spectrom
273:126–131, 2008) [1] with the density functional theory of Becke 3-parameters
hybrid functional (B3LYP). The non-planar geometry with two hydrogen-bondings
between one oxygen atom on NO3− and each hydrogen atom of two HNO3 frag-
ments is found as the most stable structure of the core ion at 0 K. For the mono-
hydrate, the most stable geometry at 0 K is found as the H2O-embedded form in
which one water molecule is located at the center of the cluster with
hydrogen-bondings to NO3− and HNO3 fragments. Our results show that the
hydrogen bond network of the core ion can be strongly perturbed by a single water
molecule. We also discussed the relative abundance of conformers of these ionic
clusters under a finite temperature.

1 Introduction

Atmospheric ions are led by multistage reactions, e.g. a charge-transfer reaction, an

ion induced nucleation, etc. [2]. Atmospheric ions affect an atmospheric environ-
ment by generating atmospheric aerosols [3, 4]. Atmospheric ion clusters consisting
of organic compounds, water molecule, etc. have a crucial role in physical and
chemical processes in the atmosphere. Negative ion clusters in the atmosphere lead
to acid rain which has been known to affect the environment, e.g. forest trees,
plants, animals, and buildings [5]. Especially H2SO4 and HNO3 have the dominant
contribution to acid rain, because strong acids release H+ when water vapor exists

A. Ueda ⋅ Y. Kita (✉) ⋅ K. Sekimoto ⋅ M. Tachikawa

Quantum Chemistry Division, Yokohama City University,
Seto 22-2, Kanazawa-ku, Yokohama 236-0027, Japan
e-mail: [email protected]

© Springer International Publishing AG, part of Springer Nature 2018 193

Y. A. Wang et al. (eds.), Concepts, Methods and Applications of Quantum Systems
in Chemistry and Physics, Progress in Theoretical Chemistry and Physics 31,
https://doi.org/10.1007/978-3-319-74582-4_10
194 A. Ueda et al.

[6]. Experimental results observed in situ mass spectrometry show the existence of
atmospheric negative ion NO3− ðHNO3 Þ2 in the upper troposphere (altitude between
9 and 12 km) [7, 8]. This negative ion is also known as a dominant negative ion at
the upper troposphere region [7].
Recently, Sekimoto and Takayama established the atmospheric pressure corona
discharge ionization (APCDI) technique, which enables us to reproducibly generate
negative ions [9]. They reported the existence of stable negative ion water clusters,
NO3− ðHNO3 Þ2 ðH2 OÞn , and the specific stability referred to as a magic number for
n = 8 [10]. Geometric structures of these water clusters are, however, still unclear
even for the core ion and its monohydrate (n = 1).
From theoretical points of view, Drenck and coworkers reported stable structures
of NO3− ðHNO3 Þ2 ðH2 OÞn (n = 0–4) obtained at B3LYP/6-31++G** level of
density functional theory (DFT) calculations [1]. They also reported the stable
structures of NO3− ðHNO3 Þm ðH2 OÞn up to n + m = 6, and assessed the validity of
theoretical calculations by comparing theoretical dissociation energies of the
NO3− ðHNO3 Þm ðH2 OÞn cluster into NO3− , mHNO3, and nH2O fragments to the
experimental results with mass-analyzed ion kinetic energy (MIKE) spectra mea-
surement [11]. Their theoretical results agree with the experiments reasonably, but
assumed only one kind of geometry for each NO3− ðHNO3 Þm ðH2 OÞn clusters despite
that a water cluster should generally have various kinds of conformers. In order to
elucidate the stable geometries of NO3− ðHNO3 Þ2 and its hydrates, a more com-
prehensive geometry searching with first-principles calculations must be
indispensable.
In this study, to elucidate stable geometries of these ionic clusters in details, we
theoretically analyzed stable geometries of the negative core ion NO3− ðHNO3 Þ2 and
its monohydrate in consideration of a lot of possible conformers with the post
Hartree-Fock ab initio method. We also discussed the relative abundance of con-
formers of these ionic clusters under a finite temperature.

2 Computational Details

We employed the second order Møller-Plesset perturbation theory (MP2) with 6-31
++G** Gaussian type basis sets in ab initio calculations of the negative core ion,
and its monohydrate. The basis set superposition error (BSSE) is not corrected
because of the less convergence in BSSE corrected geometrical optimization pro-
cedure. The harmonic approximation was used to evaluate the zero-point vibration
energy (ZPE) and Gibbs free energy. Natural Population Analysis (NPA) [12] was
used to analyze electronic populations on each atom. All calculations were per-
formed with GAUSSIAN 09 program package [13].
In the comformational searching, we picked up the initial geometries to be a
molecular cluster consisting of one NO3− , two HNO3’s, and one H2O fragments,
and optimized all the geometric degrees of freedom of the cluster simultaneously.
Ab Initio Investigations of Stable Geometries … 195

δ O = -0.55 δ O = -0.56 δ O = -0.29

δ H = +0.49

δ N = +0.66 δ N = +0.68
δ H = +0.51
δ O = -0.34 δ O = -0.98

NO3 HNO 3 H2O

Fig. 1 Geometries of molecular fragments, NO3− , HNO3 , and H2O obtained with MP2/6-31+
+G** calculations. The δX means the NPA charge on the element X

Table 1 Intermolecular interaction energies with zero-point vibration correction (EInt, unit in
kcal/mol) between two fragments obtained with MP2/6-31++G** calculation. The EInt value for
the complex X⋯Y is calculated as EInt = EZPE(X) + EZPE(Y) − EZPE(X⋯Y), where EZPE(X) is
the sum of electronic total energy and the zero-point energy (ZPE) of the system X
Complex EInt Exptl. [11]
NO3− ⋯H2 O 14.9 14.6 ± 0.2
NO3− ⋯HNO3 30.6
HNO3 ⋯HNO3 7.9
HNO3 ⋯H2 O 9.2

The stable structures of these molecular fragments are shown in Fig. 1. The ener-
getic stabilities of NO3− ðHNO3 Þ2 and its monohydrate can be mainly determined by
the O⋯H − O type intermolecular hydrogen-bonding (HB) between the fragments,
where the ionic HB between NO3− and HNO3 fragments is the most strong inter-
action among possible two fragments as shown in Table 1. We note here that MP2/
6-31 ++G** level of ab initio calculations reasonably reproduce the corresponding
experimental intermolecular interaction energy between NO3− and H2O with MIKE
spectra [11]. In the geometry optimizations of NO3− ðHNO3 Þ2 , initial geometries
were chosen to have HBs between the fragments as much as possible, where the
total number of initial geometries is 45 including conformers having a different
angle between molecular planes of NO3− and HNO3. In the case of the monohy-
drate, all the possible 218 hydrogen-bonded structures between NO3− , HNO3, and
H2O fragments were chosen as initial geometries.
196 A. Ueda et al.

3 Results and Discussion

3.1 Negative Core Ion NO3− ðHNO3 Þ2

Fifteen different geometries were obtained in the geometry searching of

NO3− ðHNO3 Þ2 . We classified them into two groups, α and β, according to the
number of oxygen atom(s) on NO3− making HBs with two HNO3’s. In the group α
containing three kinds of conformers having a different angle between molecular
planes of NO3− and HNO3, one oxygen atom on NO3− has two HBs with two
HNO3’s. In the group β containing twelve kinds of conformers, two different
oxygen atoms on NO3− have one HBs with different HNO3’s. The stable geometries
of all conformers and their relative energies at 0 K (ΔEZPE) are shown in Fig. 2,
where the most stable structure at 0 K is the structure α1. Drenck and coworkers
reported the stable structure similar to the structure β9 at B3LYP/6-31++G** level
of DFT calculations [1]. At MP2/6-31++G** level of ab initio calculations,
however, the total energy of the structure β9 is 1.6 kcal/mol higher than that of the
most stable structure α1.
The present ab initio calculations suggested that the negative core ion has the
structure of α1 as the most stable structure at 0 K. It is, however, strongly expected
that the relative abundance of these conformers depends on a thermal condition,
because most of the relative energies at 0 K are within a few kcal/mol. The tem-
perature dependence of relative abundance of conformers is shown in Fig. 3, where
the relative abundance of ith conformer (pi) at the temperature T is estimated in the
standard way as

all 
pi ðTÞ = expf − ΔGi ̸kB T g ̸ Z , Z = ∑ exp − ΔGj ̸kB T , ð1Þ
j

where ΔGi is the relative Gibbs energy including the enthalpic (H: the sum of the
electronic energy, ZPE, thermal corrections, etc.) and the entropic terms (−TS). In
Fig. 3, the temperature dependences of pi values for all conformers are shown in the
upper figure (a), and those of the sum of pi values over conformers in each group
are given in the lower figure (b). Figure 3 clearly indicates that the relative abun-
dance of NO3− ðHNO3 Þ2 conformers depends on the temperature, and the inversion
of the total abundance of the groups α and β is found at around 250 K. We note
here that the most stable structure α1 at 0 K is a dominant conformer in the low
temperature region below 90 K, while the structure of β9 has the largest relative
abundance in high temperature region above 350 K due to the large entropic
contribution from the lowest twisting mode of two HNO3 fragments (the harmonic
vibrational frequency ωe = 10 cm−1). Such temperature dependence of the relative
abundances as well as the inversion of the abundance is arising from the entropic
Ab Initio Investigations of Stable Geometries … 197

Fig. 2 Stable geometries of NO3− ðHNO3 Þ2 obtained with MP2/6-31++G** calculations. The
definitions of each group are given in the text. The ΔEZPE (unit in kcal/mol) means the relative
energy with zero-point vibration correction from the most stable structure α1

contributions and the difference in the number of conformers in each group (nu-
merical contribution). Since the conformer α2 has the largest pi value in the tem-
perature range from 90 to 350 K as shown in Fig. 3a, at around the inversion
temperature, the numerical advantage of the group β conformers plays a dominant
role in the inversion of the relative abundance.
198 A. Ueda et al.

(a)

(b)

Fig. 3 The temperature dependence of relative abundances of NO3− ðHNO3 Þ2 conformers: a pi

values defined in Eq. (1) of all conformers (in the common logarithmic scale), b the sum of pi
values in each group (in percent figures). The numbers of conformers in the group α and β are 3
and 12, respectively

3.2 Monohydrated Core Ion NO3− ðHNO3 Þ2 H2 O

Ninety different structures were obtained in the conformational searching for

NO3− ðHNO3 Þ2 H2 O. We classified these conformers into four groups, A−D,
according to the structure of NO3− ðHNO3 Þ2 and/or the hydrogen-bonded structure
between H2O and other fragments. The group A (B) contains 23 (56) kinds of
conformers which have a hydrogen-bonded structure between H2O and the core
ions of the group α (β). In these two groups, H2O is located outside of the cluster
(H2O-attached form). In the group C containing ten conformers, the H2O molecule
is located at the center of a cluster (H2O-embedded form), and has two or three HBs
with NO3− and HNO3 fragments. The group D contains only one conformer in
which the proton transfer is occurred from H2O to NO3− as OH − ðHNO3 Þ3 having
C3 symmetry. The lowest energy geometries in each group and those relative
energies at 0 K are shown in Fig. 4. As seen in Fig. 4, the most stable structure of
NO3− ðHNO3 Þ2 H2 O at 0 K is the H2O-embedded form of the group C. Drenck and
Ab Initio Investigations of Stable Geometries … 199

(a) (b)

(c) (d)

Fig. 4 The most stable structures of NO3− ðHNO3 Þ2 H2 O in the group A–D obtained with MP2/
6-31 ++G** calculations. The definitions of each group are given in the text. The ΔEZPE (unit in
kcal/mol) means the relative energy with zero-point vibration correction from the most stable
structure. The value in the parenthesis is the maximum ΔEZPE value in each group

coworkers reported the structure classified into the group B at B3LYP/6-31++G**

level of DFT calculations [1]. We address here that their structure has a high
relative energy, ΔEZPE = 2.1 kcal/mol, at MP2 level of ab initio calculations.
Our results clearly show that it is not necessary for the most stable geometry in
the monohydrate to maintain the most stable one of the core ion itself. To our
knowledge, the energetic stability of H2O-embedded form has not been reported so
far for NO3− ðHNO3 Þ2 H2 O. Such energetic stability of the group C conformers can
be reasonably explained as resulting from the enhancement of polarization of the
embedded water molecule. In the structures (C) in Fig. 4, for instance, the NPA
charges on two hydrogen (δH ) atoms of the embedded water molecule are +0.52
and +0.54, while those of the outer water molecule are +0.50 and +0.50 in the
structures (A) and (B) in Fig. 4. Such increment of the net charges on two hydrogen
atoms can strengthen the intermolecular HBs between H2O and other fragments.
200 A. Ueda et al.

Since the relative energy of most conformers are within a few kcal/mol (ΔEZPE
values of 76 conformers (84%) are within 3 kcal/mol), the relative abundances of
conformers should depend on a thermal condition as in the case of the core ion.
Figure 5 shows the theoretical temperature dependence of the relative abundances
of conformers: Fig. 5a the temperature dependences of pi values for all conformers,
and Fig. 5b those of the sum of pi values over conformers in each group. As shown
in Fig. 5b, the H2O-embedded conformers of the group C (red solid line) have the
largest relative abundance among all groups in the low-temperature region below
330 K. For the H2O-attached conformers, the conformers of the group A (blue
dashed lines) have a small relative abundance over whole temperature range, but the
group B conformer (green dotted line) has the largest relative abundance above
330 K. The relative abundance of OH − ðHNO3 Þ3 conformer is less than 1% in this

(a)

(b)

Fig. 5 The temperature dependence of relative abundances of NO3− ðHNO3 Þ2 H2 O conformers:

a pi values of all conformers (in the common logarithmic scale), b the sum of pi values in each
group (in percent figures). The inset figure in (b) is the relative abundance of H2O-attached
conformers (the sum of the group A and B), H2O-embedded conformers (the group C), and the
proton transfer structure (the group D). The numbers of conformers in the group A–D are 23, 56,
10 and 1, respectively
Ab Initio Investigations of Stable Geometries … 201

temperature range. As shown in the inset figure in Fig. 5b, the inversion of the
relative abundance of the H2O-embedded and attached conformers occurs at around
270 K. At around this inversion temperature, the numerical advantage of H2O-at-
tached conformers contributes to the inversion dominantly, because the relative
abundances of H2O-embedded conformers are larger than that of H2O-attached
conformers. The large relative abundances of H2O-attached conformers in the high
temperature region are due to the entropic contributions from slow molecular
vibrations relevant to the attached H2O molecule ranging from 20 to 30 cm−1, e.g.
H2O rocking mode.

4 Conclusion

We theoretically analyzed the stable geometries of the atmospheric negative core

ion, NO3− ðHNO3 Þ2 , and its monohydrate, NO3− ðHNO3 Þ2 H2 O, at MP2/6-31++G**
level of ab initio calculations, and discussed the relative energetic stability of
conformers for both ionic clusters. We found a total of 15 and 90 kinds of different
conformers for NO3− ðHNO3 Þ2 and its monohydrate, respectively.
Unlike the previous DFT study by Drenck and coworkers [1], in the most stable
geometry at 0 K, the core ion has the hydrogen-bonded structure in which one
oxygen atom on NO3− has two hydrogen bonds with each HNO3. For the mono-
hydrate, the most stable geometry is the hydrogen-bonded structure in which the
H2O molecule is located at the center of a cluster (H2O-embedded form) rather than
the structure in which H2O is located outside of the cluster (H2O-attached form).
This means that the hydrogen-bonding network of the core ion can be strongly
perturbed by a single water molecule. Analyzing the theoretical temperature
dependence of relative abundances of conformers, we also confirmed that the
conformers having the H2O-attached form become dominant at high temperature
region above 270 K.

Acknowledgements The present study was supported by Grant-in-Aid for Scientific Research
and for Priority Areas by Ministry of Education, Culture, Sports, Science and Technology, Japan
(KAKENHI). A part of the present computations were performed using Research Center for
Computational Science, Okazaki, Japan.

References

1. Drenck K, Hvelplund P, Nielsen SB, Panja S, Støchkel K (2008) Int J Mass Spectrom
273:126–131
2. Yu F, Turco RP (2000) Geophys Res Lett 27:883–886
3. Fend J, Möller D (2004) J Atmos Chem 48:217–233
4. Harrison RG, Carslaw KS (2003) Rev Geophys 41:1012–1027
5. Singh A, Agrawal M (2008) J Environ Biol 29:15–24
202 A. Ueda et al.

6. Jacob DJ (1999) Introduction to atmospheric chemistry, Princeton University

7. Heitmann H, Arnold F (1983) Nature 306:747–751
8. Arnold F (1980) Nature 284:610–611
9. Sekimoto K, Takayama M (2007) Int J Mass Spectrom 261:38–44
10. Sekimoto K, Takayama M (2011) J Mass Spectrom 46:50–60
11. Lee N, Keesee RG, Castleman AW Jr (1980) J Chem Phys 72:1089–1094
12. Reed AE, Weinstock RB, Weinhold F (1985) J Chem Phys 83:735–746
13. Frisch MJ, et al (2010) Gaussian 09, revision C.01, Gaussian, Inc., Wallingford CT
A Theoretical Study of Covalent
Bonding Formation Between
Helium and Hydrogen

Taku Onishi

Abstract In order to investigate chemical bonding between helium and hydrogen

in the He–H model, coupled-cluster calculations were performed. In this study,
three different hydrogen formal charges (positive, neutral and negative) were
considered. In the case of positive hydrogen, it has been concluded that covalent
bonding is formed between helium 1s orbital and hydrogen 1s orbital. Zero-point
vibration energy was smaller than dissociation energy. It has been concluded that
positive hydrogen is kept fixed at optimized structure.

Keywords Helium ⋅
Hydrogen ⋅ Molecular orbital ⋅ Covalent bonding
Chemical bonding rule

1 Introduction

It has been recognized that helium cannot form covalent bonding with other atoms
and molecules. It is because helium has the stable closed shell configuration.
Recently, Helgaker et al. demonstrated that helium clusters are not dispersed under
the strong magnetic field [1]. Several types of helium clusters such as He3, He4 and
He6 were optimized under the circumstance [2]. On the other hand, without mag-
netic field, many quantum chemical calculations were performed, from the interest
of van der Waals interaction [3–5] and potential energy [6–8]. However, molecular
orbital analysis was not performed for helium clusters and helium-including
clusters. In our previous study, the chemical bonding character of helium dimer

T. Onishi (✉)
Center of Ultimate Technology on Nano-electronics, Mie University, Mie, Japan
e-mail: [email protected]; [email protected]
T. Onishi
Centre for Theoretical and Computational Chemistry (CTCC), Department of Chemistry,
University of Oslo, Oslo, Norway
T. Onishi
Department of Applied Physics, Osaka University, Osaka, Japan

© Springer International Publishing AG, part of Springer Nature 2018 203

Y. A. Wang et al. (eds.), Concepts, Methods and Applications of Quantum Systems
in Chemistry and Physics, Progress in Theoretical Chemistry and Physics 31,
https://doi.org/10.1007/978-3-319-74582-4_11
204 T. Onishi

(He–He) [9, 10] was investigated by coupled cluster calculations. At a local min-
imum, bonding and anti-bonding molecular orbitals (MOs) were obtained. By the
use of our chemical bonding rule [11, 12], it was concluded that covalent bonding is
formed at optimized structure, due to orbital overlap between helium 1s orbitals.
Bond order is approximate equation to investigate a stability of chemical
bonding in two-atom system:

Na − Nb
N=
2

where Na and Nb denote the number of electrons in bonding and its anti-bonding
MOs. In the case of helium dimer, bonding and anti-bonding MOs are formed
between helium 1s orbitals. Since two electrons occupy both MOs (Na = Nb = 2),
N becomes zero. It was understood that covalent bonding is formed, even if bond
order is zero [10].
In this study, we have investigated chemical bonding character between helium
and hydrogen. The coupled cluster calculations have been performed for He–H
model. Here, three different hydrogen formal charges have been assumed: (1) pos-
itive (H+); (2) neutral (H); (3) negative (H−).

2 Computations

2.1 Calculation Method and Model

The calculations presented here were performed using the coupled cluster singles
and doubles (CCSD) method. It is because CCSD method accurately reproduces
accurate interatomic distance, electronic state and molecular spectra for simple
molecule [13, 14]. We used aug-cc-pVTZ basis set for helium and hydrogen [15].
All calculations were performed with the Gaussian program [16]. To investigate
chemical bonding character between helium and hydrogen, the simple He–H model
was constructed. Three types of He–H+, He–H and He–H− models were constructed.
Potential energy curves were obtained, changing the interatomic distance. In addi-
tion, zero point vibrational energy has been also obtained at the optimized structure.

2.2 Chemical Bonding Rule

In order to judge chemical bonding character such as covalency and ionicity in the
calculated MOs, chemical bonding rule is very useful and applicable (see Fig. 1).
How to utilize chemical bonding rule
1. In MOs including outer shell electrons, check whether the orbital overlap
between helium and hydrogen exists or not.
A Theoretical Study of Covalent Bonding Formation … 205

Fig. 1 Schematic drawing of

chemical bonding rule

MOs including outer shell electrons:

1s orbitals of helium and hydrogen

Check orbital overlap

2. With orbital overlap, bonding character is covalent. Without orbital overlap,

bonding character is ionic.

3 Results and Discussion

3.1 He–H+ Model

Figure 2 shows potential energy curve of He–H+ model, changing the interatomic
He–H distance. A local minimum is given at 0.776 Å. Two electrons occupy MO1,
which is only one occupied MO. The wave-function of MO1 at a local minimum is

-1790

-1800

-1810
Total Energy [kcal/mol]

-1820

-1830

-1840

-1850

-1860

-1870

-1880
0.5 1.0 1.5 2.0 2.5 3.0
He-H distance [Å]

Fig. 2 Potential energy curve of He–H+ model, changing the interatomic He–H distance
206 T. Onishi

Fig. 3 Schematic drawing of

electrons and orbitals in He– e
H+ model
He H+

e
1s orbital 1s orbital

MO1

ψ MO1 ðHe − H + Þ = 0.13 ϕHð1s′ Þ + 0.12 ϕHð1s′′ Þ + 0.35 ϕHeð1s′ Þ + 0.45 ϕHeð1s′′ Þ + 0.16 ϕHeð1s′′′ Þ

There is orbital overlap between helium 1s orbital and hydrogen 1s orbital. From
chemical bonding rule, it is concluded that helium forms covalent bonding with
hydrogen. Mulliken charge densities of helium and hydrogen are 0.315 and 0.685,
respectively.
Figure 3 depicts the schematic drawing of electrons and orbitals in He–H+
model. Helium 1s electrons are shared by both helium and hydrogen, through
covalent bonding formation between helium and hydrogen. Hence, the interatomic
He–H+ distance becomes smaller.

3.2 He–H Model

Figure 4 shows potential energy curve of He–H model, changing the interatomic
He–H distance. A local minimum is given at 3.577 Å. It is much larger than the
interatomic He–H+ distance (0.776 Å). Alpha and beta electrons are occupied in
different MO1α and MO1β, respectively. The wave-function of MO1α is

ψ MO1α ðHe − H Þ = 0.35 ϕHeð1s′ Þ + 0.48 ϕHeð1s′′ Þ + 0.30 ϕHeð1s′′′ Þ

On the other hand, the wave-function of MO1β is

ψ MO1β ðHe − H Þ = 0.35 ϕHeð1s′ Þ + 0.48 ϕHeð1s′′ Þ + 0.30 ϕHeð1s′′′ Þ

It is found that MO1α and MO1β are paired. Orbital energies of MO1α and
MO1β are slightly different: −0.91792 au for MO1α; −0.91787 au for MO1β. It is
due to broken spin symmetry. MO2α has no paired beta MO. The wave-function of
MO2α is
A Theoretical Study of Covalent Bonding Formation … 207

-2133.770

-2133.775

-2133.780
Total Energy [kcal/mol]

-2133.785

-2133.790

-2133.795

-2133.800

-2133.805

-2133.810

-2133.815
3.0 3.2 3.4 3.6 3.8 4.0 4.2 4.4 4.6 4.8 5.0
He-H distance [Å]

Fig. 4 Potential energy curve of He–H model, changing the interatomic He–H distance

ψ MO2α ðHe − H Þ = 0.24 ϕHð1s′ Þ + 0.51 ϕHð1s′′ Þ + 0.38 ϕHð1s′′′ Þ

There is no orbital overlap between helium and neutral hydrogen. From chemical
bonding rule, no covalent bonding is formed in He–H model. Mulliken spin den-
sities of helium and hydrogen are 0.00 and 1.00, respectively. MO2α is responsible
for the spin density.
Figure 5 depicts the schematic drawing of electrons and orbitals in He–H model.
Electron repulsion between two helium 1s electrons and hydrogen 1s electron is
dominative, though Coulomb interaction exists between helium 1s electrons and
positive hydrogen atomic nucleus. It is considered that the repulsion lets the
interatomic He–H distance elongated.

3.3 He–H− Model

Figure 6 shows potential energy curve of He–H− model, changing the interatomic
He–H distance. A local minimum is given at 6.452 Å. It is much larger than He–H+
and He–H models. It is found that He–H− is weakly bounded. Four electrons
occupy MO1 and MO2. The wave-function of MO1 is

ψ MO1 ðHe − H − Þ = 0.35 ϕHeð1s′ Þ + 0.48 ϕHeð1s′′ Þ + 0.30 ϕHeð1s′′′ Þ

On the other hand, the wave-function of MO2 is

208 T. Onishi

He e H

e
1s orbital 1s orbital

MO2(α)

MO1(β)
MO1(α)

Fig. 5 Schematic drawing of electrons and orbitals in He–H model

-2150.555

-2150.560

-2150.565
Total Energy [kcal/mol]

-2150.570

-2150.575

-2150.580

-2150.585

-2150.590

-2150.595

-2150.600
5.0 5.5 6.0 6.5 7.0 7.5 8.0
He-H distance [Å]

Fig. 6 Potential energy curve of He–H− model, changing the interatomic He–H distance

ψ MO2 ðHe − H − Þ = 0.16 ϕHð1s′ Þ + 0.27 ϕHð1s′′ Þ + 0.41 ϕHð1s′′′ Þ + 0.37 ϕHð2s′ Þ

There is no orbital overlap between helium and hydrogen in He–H− model.

From chemical bonding rule, it is found that no covalent boning is formed. MO1 is
for helium 1s orbital. In MO2, two electrons are delocalized over not only hydrogen
A Theoretical Study of Covalent Bonding Formation … 209

e e

He H-

e e

1s orbital 1s orbital

MO2

MO1

Fig. 7 Schematic drawing of electrons and orbitals in He–H− model

1s orbital but also hydrogen 2s orbital. Mulliken charge densities of helium and
hydrogen are 0.00 and –1.00, respectively.
Figure 7 depicts the schematic drawing of electrons and orbitals in He–H−
model. There is electron repulsion between two helium 1s electrons and two
hydrogen 1s electrons, though Coulomb interaction exists between helium 1s
electrons and positive hydrogen atomic nucleus. It is considered that this effect is
stronger than He–H− model, due to existence of two hydrogen 1s electrons.

3.4 Zero-Point Vibration Energy in the He–H+, He–H

and He–H− Models

The dissociation energies for He–H+, He–H and He–H− models can be estimated
from the total energy difference between local minimum and completely dissociated
point. They were 46.9 kcal/mol, 0.0117 kcal/mol and 0.0191 kcal/mol for He–H+,
He–H and He–H− models, respectively. Except He–H+ model, the values are very
small. It is because the stable covalent bonding between helium and hydrogen is
formed only in He–H+ model. In order to investigate the effect of quantum
vibration on the dissociation, we obtain zero point vibration energy. When zero
point vibration energy is smaller than dissociation energy, hydrogen is kept fixed at
optimized position. On the other hand, when it is larger, the dissociation may be
caused by the external factor.
210 T. Onishi

Zero point vibration energies for He–H+, He–H and He–H− moleds, were 4.58,
0.0467 and 0.0208 kcal/mol, respectively. In He–H+ model, it is much smaller than
dissociation energy. It shows that optimized structure is much stabilized, because
hydrogen is kept fixed at optimized structure. It is concluded that helium can be
strongly bounded with positive hydrogen. On the other hand, zero point vibration
energies are larger than dissociation energies in He–H and He–H− models. In
addition, the dissociation energies are very small and no covalent bonding is
formed.

4 Conclusions

From chemical bonding rule, it was found that helium 1s orbital and hydrogen 1s
orbital forms the stable covalent bonding in He–H+ model. Zero point vibration
energy was estimated to be 4.58 kcal/mol. It is much smaller than dissociation
energy (46.9 kcal/mol). On the other hand, no covalent bonding is formed, and the
small dissociation energy is given in He–H and He–H− models. Helium is weakly
bound with hydrogen.

Acknowledgements This work was supported by the Research Council of Norway

(RCN) through CoE Grant No. 179568/V30 (CTCC) and through NOTUR Grant No. NN4654 K
for HPC resources. The author would like to thank Prof. Trygve Helgaker and Prof. Josef Paldus
for valuable comments.

References

1. Lange KK, Tellgren EI, Hoffmann MR, Helgaker T (2012) Science 337:327
2. Tellgren EI, Reine SS, Helgaker T (2012) Phys Chem Chem Phys 14:9492
3. Tao J, Perdew JP (2005) J Chem Phys 122:114102
4. Zhao Y, Truhlar DG (2006) J Phys Chem A 110:5121
5. Kamiya M, Tsuneda T, Hirao K (2002) J Chem Phys 117:6010
6. Lotrich VF, Bartrett RJ, Grabowski I (2005) Chem Phys Lett 405:43
7. Snook I, Per MC, Russo SP (2008) J Chem Phys 129:164109
8. Allen MJ, Tozer DJ (2002) J Chem Phys 117:11113
9. Onishi T (2016) J Chin Chem Soc 63:83
10. Onishi T (2016) AIP Conf Proc 1790:02002
11. Onishi T (2012) Adv Quant Chem 64:31
12. Onishi T (2015) Adv Quant Chem 70:31
13. Helgaker T, Jorgensen P, Olsen J (2000) Molecular electronic-structure theory. Wiley, p 648
14. Bartlett R, Musial M (2007) Rev Mod Phys 79:291
15. Woon DE, Dunning TH Jr (1994) J Chem Phys 100:2975
16. Gaussian 09, Frisch MJ, Trucks GW, Schlegel HB, Scuseria GE, Robb MA, Cheeseman JR,
Scalmani G, Barone V, Mennucci B, Petersson GA, Nakatsuji H, Caricato M, Li X,
Hratchian HP, Izmaylov AF, Bloino J, Zheng G, Sonnenberg JL, Hada M, Ehara M,
Toyota K, Fukuda R, Hasegawa J, Ishida M, Nakajima T, Honda Y, Kitao O, Nakai H,
Vreven T, Montgomery JA Jr, Peralta JE, Ogliaro F, Bearpark M, Heyd JJ, Brothers E,
A Theoretical Study of Covalent Bonding Formation … 211

Kudin KN, Staroverov VN, Kobayashi R, Normand J, Raghavachari K, Rendell A, Burant JC,
Iyengar SS, Tomasi J, Cossi M, Rega N, Millam JM, Klene M, Knox JE, Cross JB, Bakken V,
Adamo C, Jaramillo J, Gomperts R, Stratmann RE, Yazyev O, Austin AJ, Cammi R,
Pomelli C, Ochterski JW, Martin RL, Morokuma K, Zakrzewski VG, Voth GA, Salvador P,
Dannenberg JJ, Dapprich S, Daniels AD, Farkas Ö, Foresman JB, Ortiz JV, Cioslowski J,
Fox DJ (2009) Gaussian, Inc., Wallingford CT
Small Rhodium Clusters:
A HF and DFT Study–III

M. A. Mora and M. A. Mora-Ramírez

Abstract Small neutral and ionic Rhodium clusters Rhn (n = 6, 8, 13) are investi-
gated by ab initio molecular orbital calculations with full optimization at the
Restricted Open Shell Hartree-Fock (ROHF) level with a LANL2DZ basis set, and
with the methods based on Density Functional Theory, B3LYP/MWB, B3LYP/PBE.
The clusters are found favor close-packed icosahedron structures in contrast to pre-
vious theoretical predictions that rhodium clusters should favor cubic motifs. A range
of spin multiplicities are investigated for each cluster and we present the minimum
energy conformation along with the vertical and adiabatic ionization potentials.

Keywords Rhodium clusters ⋅ ROHF calculations ⋅ Transition metal

Ionization potential

1 Introduction

It is well known that small-sized Rhodium clusters develop a magnetic moment

[1, 2] while larger clusters and Rh-bulk are non-magnetic. Both basic and applied
science researchers have been attracted to this behavior, because of the implications
in applications such as magnetic recording [3]. In fact, while the structural charac-
terization and hence magnetism of Rh clusters is an open problem as its potential for
use as a high-density storage media, Rhodium also has applications in catalysis
[4, 5]. In 2012, 81% of the 30 tons corresponding to the annual world production was
used to produce three-way catalytic converters [6]. Rhodium catalysts are used in

M. A. Mora (✉)
Depto. de Química, Universidad Autónoma Metropolitana, campus Iztapalapa,
Av. Sn. Rafael Atlixco 186, 09340 Mexico, D. F., Mexico
e-mail: [email protected]
M. A. Mora-Ramírez
Depto. Fisicomatemáticas, Facultad de C. Químicas, Benemérita Universidad
Autónoma de Puebla, Sn. Claudio y Sur 22 Col. Sn. Manuel, 72570 Puebla, Mexico
e-mail: [email protected]

© Springer International Publishing AG, part of Springer Nature 2018 213

Y. A. Wang et al. (eds.), Concepts, Methods and Applications of Quantum Systems
in Chemistry and Physics, Progress in Theoretical Chemistry and Physics 31,
https://doi.org/10.1007/978-3-319-74582-4_12
214 M. A. Mora and M. A. Mora-Ramírez

several chemical processes such as manufacturing of certain silicon rubbers [7] and
the reduction of benzene to cyclohexane [8]. Rhodium also finds use in the jewelry
industry and as an agent for hardening and improving corrosion resistance [9].
The main problem to obtain a deep atomic-level understanding of the physical
and chemical properties of clusters relies on an accurate determination of their
equilibrium atomic structure, which is not as simple as it might appear. A direct
identification of the equilibrium atomic structure by experimental techniques is very
difficult and only indirect measurements can provide few clues about the atomic
structure. Thus, the combinations of experimental techniques with first-principles
calculations have been used. For example, vibrational spectroscopies combined
with theoretical calculations have lead to important insights into the atomic struc-
ture of small Rh clusters [10, 11]. Isolated metal clusters have also been investi-
gated by Stern-Gerlach molecular-beam deflection experiments [2, 12–17].
However, there are difficulties in the direct identification of the atomic structure
of clusters by experimental techniques. Thus, most of the structural studies have
been based on theoretical calculations, which can directly determine the atomic
structure of clusters using several well-defined algorithms. Several calculations
based on density functional theory (DFT) have focused on these clusters, for
example, on metal particles containing 13 atoms, Rh13 [18–29]. Furthermore, it is
important to mention that few studies have focused on the search for the
lowest-energy structures with most studies assuming predefined structures.
Sophisticated algorithms have been employed in the search for the lowest-energy
structure, namely, generic algorithms (GA) [30], basin-hopping Monte Carlo
(BHMC) [31–34], Monte Carlo (MC) [35], conformational space annealing [29],
taboo search in descriptor space (TSDE) [23, 36], high-temperature molecular
Dynamics (high-T-MD) [22]. Almost all the studies with these algorithms have
been used in combination with empirical pair potentials. These potentials have
difficulties in providing a correct description of the atomic structure [37–39], and
hence, the ground state structures might not be correct.
In this paper, we present HF and DFT calculations on clusters with 6 and 8
rhodium atoms for comparison with theoretical and experimental results. We also
present results for the 13-atom cluster since it is one of the most studied clusters and
to the best of our knowledge there are no experimental results.

2 Method and Computational Details

It is well known that the method of calculation and the chosen basis set are the two
most important factors in determining the accuracy of results. Ab initio methods
must represent all the electrons in some manner. However, for heavy atoms it is
desirable to reduce the amount of computation burden. One way to do this is by
replacing the core electrons and their basis functions in the wave function by a
potential term in the Hamiltonian. These are called core potentials, effective core
potentials (ECP) or relativistic effective core potentials (RECP). In this work, we
Small Rhodium Clusters: A HF and DFT Study–III 215

use the ECP LANL2DZ [40–42] potential, which is one of the most widely used for
heavy elements.
Another factor to consider in selecting the method of calculation is the spin
contamination. A high spin contamination can affect the geometry and population
analysis and significantly affect the spin density. The error introduced by spin
contamination is unacceptable when systems with transition metals are investigated.
The restricted open-shell Hartree-Fock (ROHF) ab initio method is one of the best
ways to avoid spin contamination and obtain a reliable wave function.
We performed ab initio molecular orbital calculations with full optimization at
the ROHF level with a LANL2DZ basis set. Also in this research we employed the
B3LYP hybrid functional with the small-core quasirelativistic approach of Wood
and Boring [43] MWB ECP, and, the General Gradient Approximation
(GGA) formulated by Perdew, Burcke, and Hernzerhof (PBE) [44, 45] since it has
been shown that these hybrid functional can yield reliable energetics and structural
results for other metal compounds. The geometries were adjusted until a stationary
point on the potential energy surface was found, using the Berny algorithm [46, 47]
for the minimization.

3 Results and Discussion

Figure 1 shows the Rh6 optimized geometry calculated with the B3LYP/MWB
chemical model obtained from different initial geometries; Fig. 1a shows the final
geometry obtained from an initial octahedron, Fig. 1b is a final triangular prism
conformer, Fig. 1c is a deformed pentagonal pyramid which converges to an dis-
torted trigonal prism or capped tetrahedron as in Fig. 1d.
Table 1 shows the relative energy (eV) of the Rh6 cluster obtained with the
B3LYP/MWB chemical model; we searched the spin multiplicity up to 23. The
lowest energy isomer is an octahedron with a multiplicity equal to 9, followed by
the distorted pentagonal pyramid with a multiplicity of 11. Finally the trigonal
prism isomer has M = 9. The minimum energy isomer is in complete agreement
with the experimental results [10, 48], and with previous theoretical results [21, 24,
25, 48–56]. The relative energy of ROHF-isomers with different geometry is 1.039,
3.973 eV for pentagonal pyramid and octahedron respectively. For the
B3LYP-isomers with different energy the relative energy is 0.218 for pentagonal
pyramid, and 0.490 eV for trigonal prism. With the ROHF/LANL2DZ method, the
triangular prism has a minimum energy conformation with multiplicity equal to 15.
The more stable pentagonal pyramid is when M = 17, and the square bi-pyramid
with M = 11. A geometry formed by two perpendicular squares (an incomplete
cube) with M = 15 is the fundamental state for this particular geometry. Among
these four geometries, the minimum energy is the triangular prism. All calculations
in this series were performed with the ROHF/LANL2DZ methods and with mul-
tiplicities from 1 to 23 and full optimization.
216 M. A. Mora and M. A. Mora-Ramírez

Fig. 1 Isomers of Rh6 geometry optimized with the B3LYP/MWB model chemistry, a tetrahe-
dron, b triangular prism, c deformed pentagonal pyramid, d capped tetrahedron

The relative energy (a.u.) calculated with the ROHF/LANL2DZ method for Rh8
clusters are shown in Table 2. We searched the spin multiplicity up to 21. The
relative energy of isomers with different geometry is 23.76, 1.023, and 1.249 eV for
the cubic, bctp, and incomplete-ico respectively. In Fig. 2, we show the optimized
structures of the Rh8 clusters. Figure 2a is the lowest energy isomer, a bicapped
octahedron, bcoh, with a multiplicity equal to 19. Figure 2b corresponds to the
bicapped triangular prism isomer, bctp, with a multiplicity of 19. Figure 2c shows
an isomer with the incomplete icosahedron structure. Finally, a cubic isomer with
M = 13 is presented in Fig. 2d. The octahedron isomer with the minimum energy is
in complete agreement with the experimental results reported by D. J. Harding et al.
[10, 11], and by M. R. Beltran et al. [48], and with theoretical results obtained with
different methods such as local spin density [24, 57], molecular dynamics [26],
effective core potential [57], B3LYP [48, 49], which unlike of the reported here
Small Rhodium Clusters: A HF and DFT Study–III 217

Table 1 Rh6 relative energy (eV) obtained from B3LYP/MWB calculations for isomers of Rh6 in
different states of spin multiplicity, in bold the octahedron isomer
Multiplicity Triangular prism Pentagonal pyramid Square bi-pyramid
1 0.346 2.503 1.309
3 0.354 1.037 0.786
5 0.275 0.786 0.522
7 0.139 0.340 0.375
9 0 0.340 0
11 0.272 0 0.250
13 0.381 0.609 0.675
15 1.167 1.162 1.178
17 3.570 3.763 3.891
19 6.620 nc 7.132
21 nc nc nc
23 14.528 14.749 nc
nc not converged

Table 2 Rh8 relative energy Multiplicity Incomplete-ico bcoh bctp cubic

(eV) obtained from ROHF/
LanL2DZ calculations for 1 17.82 21.52 15.29 3.92
isomers of Rh8 in different 3 16.19 16.14 16.57 0.65
states of spin multiplicity, in 5 11.40 24.81 15.16 1.93
bold the bi-capped octahedron 7 12.00 13.09 10.45 nc
isomer
9 10.58 11.24 nc 2.64
11 7.48 9.03 6.94 1.01
13 4.82 5.93 5.06 0
15 1.09 7.05 nc 5.90
17 1.20 2.37 1.63 5.63
19 0 0 0 1.93
21 nc 2.93 1.63 11.51
nc not converged

were performed by keeping the symmetry, i.e. only optimized bond distances
[19, 56].
For the cluster formed by 13 rhodium atoms, taking as the initial geometry that
presented in Fig. 3a, we performed two series of calculations using density func-
tional theory. The first one using the Becke-Lee-Yang-Parr [58, 59] functional and a
double-Z basis set with effective core potential to represent the electrons close to the
nucleus, and the second using the PBE functional [44, 45]. Both sets of calculations
were carried out by varying the multiplicity from 2 to 26, performed with full
optimization, without symmetry constraints, and following the Berny algorithm for
minimization as it is implemented in the Gaussian 03 computer package.
218 M. A. Mora and M. A. Mora-Ramírez

Fig. 2 Isomers of Rh8 geometry in its minimum energy conformation obtained with the ROHF/
LANL2DZ. a bi-capped octahedron, b bi-capped triangular prism, c incomplete icosahedron,
d cube

The B3LYP/LANL2DZ optimized geometry for Rh13 with M = 2 is shown in

Fig. 3b. We notice that the initial geometry is not retained. The positions of the
atoms 9 and 10 are considerably modified. The plane containing the atoms 3, 4, 7,
8, 11, 12 is slightly modified in the optimization process. The calculations made for
the other multiplicities converge to a geometry similar to that found for M = 2,
which is basically maintained until the cluster with M = 26. The conformation of
minimum energy for this series corresponds to a multiplicity equal to 22, and is
presented in Fig. 3c, where we can see that the plane of the six atoms has been
distorted. Table 3 shows the energy calculated for these clusters as well as the
eigenvalues of the S2 operator. Note that the spin contamination for these calcu-
lations is large. As expected, during the optimization process the initial geometry of
the cluster is not equal to the final geometry, i.e., during the optimization process
the geometry evolves. Geometries known as non-icosahedral or low symmetry have
Small Rhodium Clusters: A HF and DFT Study–III 219

Fig. 3 Rh13 with the functional B3LYP a configuration initial b optimized with M = 2 c the
minimum energy configuration for M = 22

Table 3 Rh13 relative energy Multiplicity Relative energy (eV) <S2>

obtained from B3LYP/MWB
calculations for in different 4 0.095 7.179
states of spin multiplicity and 6 0.248 9.75
the eigenvalue 8 0.093 17.53
of <S2> operator 10 0.003 25.6556
12 0.095 36.364
14 0.035 49.373
16 0.0 64.166
18 0.218 81.228
20 0.299 100.073
22 0.250 121.073
24 1.045 144.030
26 2.041 168.985
28 2.985 195.932
220 M. A. Mora and M. A. Mora-Ramírez

Fig. 4 Final geometry obtained with the Perdew-Burke-Ernzerhof exchange-correlation func-

tional for Rh13

been reported as optimized [27, 55, 57], and are markedly different from those
obtained by us using the same method of calculation. It seems that in those cal-
culations, geometries reported as the minimum energy were not optimized at the
DFT level, probably assuming predefined structures obtained by molecular dynamic
or similar methods.
The series of PBE calculations was also initiated with the geometry presented in
Fig. 3a, without any restrictions during the optimization process. The final geom-
etry, Fig. 4, is similar to that found with the B3LYP functional. The difference is
that the plane formed by the atoms 3, 4, 7, 8, 11, 12 is now more deformed, and the
cube formed by the atoms 1–8 is markedly deformed. Again, the electronic state of
minimum energy corresponds to M = 22.
This same type of calculation was performed earlier [22]. In that report, it is
mentioned that the minimum energy geometry is the so-called non-icosahedral one.
Since our initial geometry is not exactly that considered in those studies, we take as
the initial conformation a double single cube geometry, DSC, previously reported
[19, 28] as the minimum energy conformer. This initial conformation was used to
Small Rhodium Clusters: A HF and DFT Study–III 221

Fig. 5 Double-simple-cubic initial geometry for PBE and B3LYP calculations

Fig. 6 Final geometry obtained for Rh13 with full optimization. The geometry obtained for both
PBE and B3LYP functionals are very similar
222 M. A. Mora and M. A. Mora-Ramírez

perform two other sets of calculations with the same B3LYP and PBE functionals.
Figure 5 shows the initial geometry, the final geometry obtained without any kind
of restriction for both functionals. Figure 6 is very similar. It is a simple double
distorted cube formed by oblique parallelepipeds. Although the geometries are very
similar, that obtained with the functional PBE shows shorter distances than the
corresponding one calculated with the B3LYP functional. The angle formed by the
distorted cubes is 140° and 124° for B3LYP and PBE respectively. The initial
geometry is not preserved during the optimization process. The geometry optimized
by us is 0.13 eV more stable with PBE, and 0.58 eV more stable with B3LYP, than
the capped double cube conformation previously reported. The energies are not the
only differences. Reference [56] reported a magnetic moment of 9 μB for the double
simple cubic, DSC, geometry with the PBE method while we obtain a magnetic
moment of 13 μB with the same method (PBE), and 21 μB using B3LYP, equal to
that obtained by Reddy et al. [18] using the von Barth-Hedin form of the
exchange-correlation contributions in the discrete variational method. It is

Fig. 7 Rh13 cluster in its minimum energy conformation, a centered icosahedron, with
multiplicity 16, obtained with the ROHF method
Small Rhodium Clusters: A HF and DFT Study–III 223

important to emphasize that the DSC geometry reported as the most stable is not
retained once it is subjected to a process of optimization without any kind of
restrictions.
Rh13 is one of the most studied [18–29, 38, 51, 60] clusters because it is
considered as the seed for different cluster growth patterns. A wide dispersion in the
calculated multiplicities is present in the literature.
The ROHF calculations for Rh13 converge to an interesting structure presented
in Fig. 7. It is an interlocking series of pentagons, an irregular icosahedron, with a
rhodium atom in the center. The icosahedron is composed of interlocking pen-
tagonal “caps”. Every vertex of the icosahedron is the top of a pentagonal cap. We
performed calculations for multiplicities in the range from 4 to 30. Table 4 shows
that the energy has an oscillating behavior. Three minimum energy states are
present at multiplicities of 10, 16 and 28. There is a small energy difference of 6.17
and 2.17 eV in relation to the multiplicity equal to 28.
In Fig. 7, the pentagons formed by the Rh1-Rh5, and the Rh6-Rh10 atoms have
different bond distances. These are 2.8 Å and 2.66 Å in average respectively. The
average distance from the vertex, Rh11 to the Rh1-Rh5 pentagon is 2.89 Å and the
distance from the vertex Rh13 to the pentagon Rh6-Rh10 is 3.8 Å. Also shown in
Table 4 is the dipole moment of the cluster in different electronic states.
There are two possibilities for the formation of the anion or the cation. The
cation could be formed when the neutral cluster loses an electron. It may be a
previously paired electron in which case the multiplicity of the cation is increased
by one unit. If the loss is that of a previously unpaired electron, the multiplicity
decreases by one unit in relation to the neutral cluster. On the other hand, the

Table 4 Rh13, energy (a.u.), relative energy (eV), dipolar moment (D), energy of the HOMO,
energy of the LUMO, and the band gap values obtained from ROHF calculations for different
states of spin multiplicity
Mult. Energy R. Energy D HOMO LUMO GAP
4 −1411.243 11.97 2.712 −0.031 −0.010 0.021
6 −1411.314 10.04 2.143 −0.065 −0.003 0.062
8 −1411.317 9.982 1.460 −0.051 −0.014 0.037
10 −1411.457 6.177 1.409 −0.035 −0.010 0.026
12 −1411.416 7.265 2.021 −0.037 0.002 0.039
14 −1411.423 7.075 1.978 −0.032 −0.004 0.028
16 −1411.604 2.176 0.995 −0.044 −0.002 0.042
18 −1411.588 2.585 1.132 −0.045 −0.001 0.041
20 −1411.579 2.830 2.872 −0.024 −0.004 0.020
22 −1411.618 1.769 1.082 −0.031 0.000 0.031
24 −1411.624 1.633 2.618 −0.025 −0.002 0.023
26 −1411.672 0.299 0.968 −0.026 −0.011 0.015
28 −1411.684 0.000 2.254 −0.036 −0.009 0.025
30 −1411.646 1.034 3.048 −0.033 −0.015 0.019
224 M. A. Mora and M. A. Mora-Ramírez

Table 5 Rh13 selected parameters ROHF/LANL2DZ for RH13 neutral with multiplicity 16 and
its ionic species
Specie Charge, Energy (a. E. A. I. P. HOMO LUMO GAP Dip. Mom.
multiplicity u.)
Rh13 +1, 15 −1411.399 0.206 −0.127 −0.115 0.012 0.85
+1, 17 −1411.516 0.087 −0.135 −0.120 0.016 1.05
0, 16 −1411.604 −0.444 −0.003 0.014 0.99
−1, 17 −1411.615 0.012 0.052 0.102 0.049 0.84
−1, 15 −1411.596 0.007 0.599 0.098 0.500 0.62
+1, 27 NC
+1, 29 NC
0, 28 −1411.684 −0.036 −0.009 0.027 2.254
−1, 27 −1411.654 −0.03 −0.046 0.096 0.05 2.656
−1, 29 −1411.641 −0.04 −0.036 −0.009 0.027 2.005

formation of the anion occurs through the gain of an electron. The new electron
may be unpaired in which case the multiplicity increases, or it might pair with a
previously unpaired electron in the neutral cluster thus decreasing the multiplicity.
Table 5 reports the Restricted Open shell Hartree Fock energies of
Rh13 M = 16 and M = 28, and its ionic clusters, the charge and multiplicity, the
electron affinities (EA) and the ionization potential (IP). The electron affinities were
calculated as the energy difference between the neutral and the anionic clusters
while the ionization potential was calculated as the energy difference between the
cation and the neutral molecule. On the other hand the Ionization Potential calcu-
lated from the HOMO following Koopman’s theorem, is 0.116 a.u. = 3.15 eV for
M = 16.

4 Conclusions

We report a study of rhodium clusters using ROHF and DFT methods with full
optimization.
Our results agree with those of other researchers concerning the equilibrium
geometry, but disagree on the magnetic ground state
• The equilibrium structure of the isomers is unambiguously determined with
ROHF methods. For Rh6 and Rh8, the ROHF minimum energy conformation is
in excellent agreement with experiment.
• Different spin states are quite close in energy. All of them have essentially the
same equilibrium structure.
• With different XC functionals, different spin states are obtained for the same
conformation.
Small Rhodium Clusters: A HF and DFT Study–III 225

• The Growth behavior follows the icosahedral structure route.

• In the formation of ions, a larger number of unpaired electrons lead to greater
stability.

References

1. Cox AJ, Louderback JG, Bloomfield LA (1993) Experimental observation of magnetism in

rhodium clusters. Phys Rev Lett 71:923–926
2. Cox AJ, Louderback JG, Apsel SE, Bloomfield LA (1994) Magnetism in 4d-transition metal
clusters. Phys Rev B 4:12295–12298
3. Schmid G (2004) Nanoparticles from theory to applications. Wiley-VCH. ISBN 3527305076
4. Wei J, Iglesia E (2004) Structural requirements and reaction pathways in methane activation
and chemical conversion catalized by rhodium. J Catal 225:116–127
5. Nolte P, Stierle A, Jin-Phillipp NY, Kasper N, Schulli TU, Dosch H (2008) Shape changes of
supported Rh nanoparticles during oxidation and reduction cycles. Science 321:1654–1658
6. Loferty PJ (2013) Commodity report: platinum group metals. United States Geological
Survey
7. Heldings FM, Capka M (2003) Rhodium complexes as catalyst for hydrosilylation
crosslinking of silicone ruber. J Appl Polymer Sci 30(5):1837
8. Halligudi SB et al (1992) Hydrogenation of bencene to cycloexene catalized by Rhodium(I)
complex supported on montmorillonite clay. React Kinet Catal Lett 48(2):547–552
9. Cramer Stepen S Jr, Covino, Bernard (1990) ASM handbook materials park OH: ASM
international report pp 393–396
10. Harding DJ, Gruene P, Haertelt M, Meijer G, Fielicke A, Hamilton SM, Hopkins WS,
Mackenzie RS, Neville SP, Walsh TR (2010) Probing the structures of gas-phase rhodium
cluster cation by far-infrared spectroscopy. J Chem Phys 133:214304–214313
11. Harding DJ, Walsh TR, Hamilton SM, Hopkins WS, Mackenzie SR, Gruene P, Haertelt M,
Maijer G, Fielicke F (2010) Comunications: the structure of Rh+ 8 in the gas phase. J Chem
Phys 132:011101–011104
12. Knickelbin MB (2005) Phys Rev B 71:18444
13. Apsel SE, Emmert JW, Deng J, Bloomfield LA (1996) Surface-enhanced magnetism in nickel
clusters. Phys Rev Lett 76:1441
14. Douglas DC, Bucher JP, Bloomfield LA (1992) Magnetic studies of free ferromagnetic
clusters. Phys Rev B 45:6341–6344
15. Knickelbein MB (2001) Experimental observation of superparamagnetism in manganese
clusters. Phys Rev Lett 86:5255–5257
16. Knickelbein MB (2004) Magnetic ordering in manganese clusters. Phys Rev B 70:014424
17. Xu X, Yin S, Moro R, de Herr WA (2005) Magnetic moments and adiabatic magnetization of
free cobalt clusters. Phys Rev Lett 95:237209
18. Reddy BV, Khanna SN, Dunlap BI (1993) Giant magnetic moments of 4d clusters. Phys Rev
Lett 70:3323–3326
19. Bae YC, Kumar V, Osanai H, Kawazoe Y (2005) Cubic magic clusters of rhodium stabilized
with eight-center bonding: magnetism and growth. Phys Rev B 72:115427–115432
20. Chang CM, Chou MY (2004) Alternative low-symmetry structure for 13-atoms metal
clusters. Phys Rev Lett 93:133401–133404
21. Aguilera-Granja F, Rodríguez-López JL, Michaelian K, Berlanga-Ramírez EO, Vega A
(2002) structure and magnetism of small rhodium clusters. Phys Rev B 66:224410–224419
22. Wang LL, Johnson DD (2007) Density functional study of structural trends for
late-transition-metal 13-atoms clusters. Phys Rev B 75:235405–235409
226 M. A. Mora and M. A. Mora-Ramírez

23. Sun Y, Zhang M, Fournier R (2008) Periodic trends in the geometric structures of 13-atoms
metal clusters. Phys Rev B 77:0754435
24. Jinlong Y, Toigo F, Kelin W (1994) Structural electronic, and magnetic properties of small
rhodium clusters. Phys Rev B 50:7915–7924
25. Reddy BV, Nayak SK, Khanna SN, Rao BK, Jena P (1999) Electronic structure and
magnetism of Rhn (n = 2-13) clusters. Phys Rev B 59:5214–5222
26. Guirado-López R, Villaseñor-González P, Dorantes-Dávila J, Pastor GM (2000) Magnetism
of Rhn clusters. J Appl Phys 87:4906
27. Aguilera-Granja F, Montejano-Carrizales JM, Guirado-López RA (2006) Magnetic properties
of small 3d and 4d transition metal clusters: the role of a noncompact growth. Phys Rev B
73:115422
28. Bae YC, Osansi H, Kumar V, Kawazone Y (2004) Nonicosahedral growth and magnetism
behavior of rhodium clusters. Phys Rev B 70:195413–195419
29. Rogan J, García G, Loyola C, Orellana W, Ramírez R, Kiwi M (2006) Alternative search
strategy for minimal energy nanocluster structures: the case of rhodium, palladium, and silver.
J Chem Phys 125:214708–214712
30. Joswig JO, Springborg M (2003) Generic algorithms search for global minima of aluminium
clusters using Sutton-Chen potential. Phys Rev B 68:085408
31. Zhan L, Chen JZY, Montejano-Carrizalez JM, Guirado-López RA (2006) Phys Rev B
73:115422
32. Aprá E, Ferrando R, Fontunelli A (2006) Density-functional global optimization of gold
clusters. Phys Rev B 73:205414
33. Wales DJ, Doye JPK (1997) Global optimization by basin-hopping and the lowest energy
structures of Lennard-Jones clusters containing up to 110 atoms. J Phys Chem A 101:
5111–5116
34. Kim HG, Choi SK, Lee HM (2008) New algorithm in the basin-hopping Monte Carlo to find
the global minimum structure of unary and binary metallic nano clusters. J Chem Phys
128:144702
35. Erkoc S, Saltaf R (1999) Monte Carlo computer simulations of copper clusters. Phys Rev A
60:3053
36. Cheng J, Fournier R (2004) Structural optimization of atomic clusters by Tabu search in
descriptor spaces. Theor Chem Acc 112:7–15
37. Baletto F, Ferrando R (2005) Structural properties of nanoclusters: energetic, thermodynamic,
and kinetic effects. Rev Mod Phys 77:371
38. Piotrowsky MJ, Piquini P, Da Silva JLF (2010) Density functional theory investigation of 3d,
4d, and 5d 13-atom metal clusters. Phys Rev B 81:155446–155459
39. Piotrowski MJ, Piquini P, Odashima MM, DaSilva JLF (2011) Transition-metal 13-atom
clusterts assessed with solid and surface-biased functionals. J Chem Phys 134:134105–
134110
40. Hay PJ, Wadt WR (1985) Ab initio effective core potentials for molecular calculations.
Potentials for main group elements Na to Bi. J Chem Phys 82:270
41. Wadt WR, Hay PJ (1985) Ab initio effective core potentials for molecular calculations.
Potentials for the transition metal atoms Sc to Hg. J Chem Phys 82:284
42. Hay PJ, Wadt WR (1985) Ab initio effective core potentials for molecular calculations.
Potentials for K to Au including the outermost core orbitals. J Chem Phys 82:299
43. Wood JH, Boring AM (1978) Improved Pauli Hamiltonian for local-potential problems. Phys
Rev B 18:2701
44. Perdew JP, Burke K, Ernzerhof M (1996) General gradient approximation made simple. Phys
Rev Lett 77:3865
45. Perdew JP, Burke K, Ernzerhof M (1997) General gradient approximation made simple [Phys
Rev Lett 77:4884(1996)] Phys Rev Lett 78:1396
46. Peng C, Ayala PY, Schelegel HB, Frish MJ (1996) Using redundant internal coordinates to
optimize equilibrium geometries and transition states. J Comput Chem 17:49
Small Rhodium Clusters: A HF and DFT Study–III 227

47. Peng C, Schelegel HB (1994) Combining synchronous transit and quasi-Newton methods to
find transit states. Israel J Chem 33:449
48. Beltrán MR, Buendía Zamudío F, Chanhan V, Sen P, Wang H, Ko YJ, Bowen K (2013) Ab
initio and anion photoelectron studies of Rhn (n = 1-9) clusters. Eur Phys J D 67:63–70
49. Bertin V, Lopez-Rendón R, del Angel G, Poulain E, Avilés R, Uc-Rosas V (2010)
Comparative theoretical study of small Rhn nanoparticles (2 ≤ n ≤ 8) using DFT methods.
Int J Quantum Chem 110:1152–1164
50. Harding DJ, Davies RDL, Mackenzie SR, Walsh TR (2008) Oxides of small rhodium clusters:
theoretical investigation of experimental reactivities. J Chem Phys 129:124304–124310
51. Mora MA, Mora-Ramírez MA, Rubio-Arroyo Manuel F (2010) Structural and electronic
study of neutral, positive and negative small rhodium clusters [Rhn, Rh+ -
n , Rhn]. Int J Quantum
Chem 110:2541–2547
52. Li ZQ, Yu JZ, Ohno K, Kawazoe Y (1999) Calculations on the magnetic properties of
rhodium clusters. J Phys Condens Matter 7:47–53
53. Hang TD, Hung HM, Thiem LN, Nguyen HMT (2015) Electronic structure and
thermochemical properties of neutral and anionic rhodium clusters Rhn, n = 2-13. Evolution
of structures and stabilities of binary clusters RhmM (M = Fe Co, Ni; m = 1-6). Comput
Theor Chem 1068:30–41
54. Chien CH, Blaisten-Barojas E, Pedersen MR (1998) Magnetic and electronic properties of
rhodium clusters. Phys Rev A 58:2196–2202
55. Harding D, Mackenzie SR, Walsh TR (2006) Structural isomers and reactivity for Rh6, Rh+ 6.
J Phys Chem B 110:18272–18277
56. Da Silva JLF, Piotrowski MJ, Aguilera-Granja F (2012) Phys Rev B 86:125430–125435
57. Sun Y, Fournier R, Zhang M (2009) Structural and electronic properties of 13-atom 4d
transition-metal clusters. Phys Rev B 79:043202–043211
58. Lee C, Yang W, Parr RG (1988) Development Colle-Salvetti correlation-energy formula into
a functional of the electron density. Phys Rev B 37:785
59. Miehlich B, Sabin A, Stoll H, Preus H (1989) Results obtained with the correlation energy
density functional of Becke and Lee, Yang and Parr. Chem Phys Lett 157:200–206
60. Futschek T, Marsman M, Hafner J (2005) Structural and magnetic isomers of small Pd and Rh
clusters: an ab initio functional study. J Phys Condens Matter 17:5927–5963
Spectroscopy of Radiative Decay
Processes in Heavy Rydberg
Alkali Atomic Systems

Valentin B. Ternovsky, Alexander V. Glushkov, Olga Yu. Khetselius,

Marina Yu. Gurskaya and Anna A. Kuznetsova

Abstract We present the results of studying the radiation decay processes and
computing the probabilities and oscillator strengths of radiative transitions in spectra
of heavy Rydberg alkali-metal atoms. All calculations of the radiative decay
(transitions) probabilities have been carried out within the generalized relativistic
energy approach (which is based on the Gell-Mann and Low S-matrix formalism)
and the relativistic many-body perturbation theory with using the optimized
one-quasiparticle representation and an accurate accounting for the critically
important exchange-correlation effects as the perturbation theory second and higher
orders ones. The precise data on spectroscopic parameters (energies, reduced dipole
transition matrix elements, amplitude transitions) of the radiative transitions nS1/2
→ n′P1/2,3/2 (n = 5, 6; n′ = 10–70), nP1/2,3.2 → n′D3/2,5/2 (n = 5, 6; n′ = 10–80) in
the Rydberg Rb, Cs spectra and the transitions 7S1/2-nP1/2,3/2, 7P1/2,3.2-nD3/2,5/2
(n = 20–80) in the Rydberg francium spectrum are presented. The obtained results
are analyzed and discussed from viewpoint of the correct accounting for the rela-
tivistic and exchange-correlation effects. It has been shown that theoretical approach
used provides an effective accounting of the multielectron exchange-correlation
effects, including effect of essentially non-Coulomb grouping of Rydberg levels and
others.

Keywords Radiation decay processes ⋅ Multielectron atoms and multicharged

ions ⋅
Relativistic energy approach

V. B. Ternovsky (✉) ⋅ A. V. Glushkov ⋅ O. Yu. Khetselius ⋅ M. Yu. Gurskaya

A. A. Kuznetsova
Odessa State Environmental University, L’vovskaya Str., 15,
Odessa-9 65016, Ukraine
e-mail: [email protected]

© Springer International Publishing AG, part of Springer Nature 2018 229

Y. A. Wang et al. (eds.), Concepts, Methods and Applications of Quantum Systems
in Chemistry and Physics, Progress in Theoretical Chemistry and Physics 31,
https://doi.org/10.1007/978-3-319-74582-4_13
230 V. B. Ternovsky et al.

1 Introduction

Accurate radiative decay widths and probabilities, oscillator strengths of radiative

transitions in spectra of the Rydberg atomic systems (atoms in the highly excited
states with large values of the principal quantum number n ≫ 1) are of a great
interest for astrophysical analysis, laboratory, thermonuclear plasma diagnostics,
fusion research etc. (see, for example [1–60]). In recent years intensive theoretical
and experimental investigations of spectroscopic properties of the Rydberg atoms
are also stimulated by a great number of their possible important applications in
atomic and molecular optics and spectroscopy, quantum electronics, laser physics
(for example, speech is about new lasing schemes in the short-wave range with
using the Rydberg systems), quantum informatics and computing, astrophysics etc.
It is well known that the Rydberg atoms make the contribution into interstellar
clouds absorption spectrum (Rydberg states with n ∼ 300–700). The unique
properties of the Rydberg atoms are associated with too small ionization potentials,
sufficiently large size, enough long lifetime compared to conventional atomic states,
finally, unprecedented sensitivity to external fields. Really, it is well known that the
Rydberg atomic systems are very sensitive to electromagnetic fields and can be
used for the detection and sensing static and AC electric and magnetic fields.
Strongly interacting Rydberg systems have unique photon emission properties.
These facts stimulate more intensive research of the Rydberg atoms, in particular,
on the basis of new experimental methods of laser spectroscopy, beam-foil spec-
troscopy, using magneto-optical traps, synchrotron radiation sources, cryogenic
devices and so on. It is worth to remind about such unique and interesting physical
objects and phenomena such as the Rydberg matter, Bose-condensate in vapors of
the Rydberg alkali-metal atoms, fountains of cold Rydberg atoms etc.
The well-known quasiclassical and quantum-mechanical approaches such as the
Hartree-Fock (HF) and Dirac-Fock (DF) methods, quantum defect and the Coulomb
approximations, the model potential and pseudopotential methods etc. have been
used to calculate the spectroscopic properties of different light and middle Rydberg
atoms. In a case of the heavy Rydberg atoms in a free state or in an external
electromagnetic field a modern level of description of the Rydberg atoms is not
sufficiently satisfactory. A precise accounting for the relativistic and
exchange-correlation (XC) effects, including an effect of the non-Coulomb
grouping levels in the Rydberg spectra (the effect, which, as a rule, is not con-
sidered within simplified Coulomb and quantum defect models) is of a great interest
and importance.
The purpose of this work is to present the results of studying the radiation decay
processes and computing probabilities and oscillator strengths of the radiative
transitions in the spectra of heavy Rydberg atoms of alkali-metal elements. The
precise data on spectroscopic parameters (energies, reduced dipole transition matrix
elements, amplitude transitions) of the radiative transitions nS1/2 → n′P1/2,3/2 (n = 5,
6; n′ = 10–70), nP1/2,3.2 → n′D3/2,5/2 (n = 5, 6; n′ = 10–80) in the Rydberg Rb, Cs
spectra and the transitions 7S1/2-nP1/2,3/2, 7P1/2,3.2-nD3/2,5/2 (n = 20–80) in the
Spectroscopy of Radiative Decay Processes … 231

Rydberg francium spectrum are listed. The data are discussed from the viewpoint of
the correct accounting for the relativistic and exchange-correlation effects. It has
been shown that the theoretical approach used provides a precise accounting for the
important exchange-correlation effects, including the effect of essentially
non-Coulomb grouping of Rydberg levels, continuum pressure etc.
All calculations of the radiative decay (transitions) probabilities (matrix ele-
ments) in the studied atomic systems have been performed with using the gener-
alized relativistic energy approach and the relativistic many-body perturbation
theory (PT) with using the optimized one-quasiparticle representation and an
accurate accounting of the exchange-correlation effects, including the effect of
essentially non-Coulomb grouping of Rydberg levels [61–63].
Let us remind that the theoretical fundamentals of an energy approach in a case of
the one-electron ions have been considered by Labzovsky et al. [57, 58]. Originally
the energy approach to radiative and autoionization processes in multielectron atoms
and ions has been developed by Ivanova-Ivanov et al. [59–62, 64–67]. More
accurate, advanced version of the relativistic energy approach has been further
developed in Refs. [63, 68–72]. The energy approach is based on the Gell-Mann and
Low S-matrix formalism combined with the relativistic perturbation theory. In
relativistic case the Gell-Mann and Low formula expressed an energy shift ΔE
through the electrodynamical scattering matrix including interaction with as the
photon vacuum field as a laser field. The first case is corresponding to determination
of radiative decay characteristics for atomic systems. Earlier we have applied the
corresponding generalized versions of the energy approach to many problems of
atomic, nuclear and even molecular spectroscopy, including, cooperative
electron-gamma-nuclear “shake-up” processes, electron-muon-beta-gamma-nuclear
spectroscopy, spectroscopy of atoms in a laser field etc. [73–98].

2 Relativistic Energy Approach and Many-Body

Perturbation Theory with the Dirac-Kohn-Sham Zeroth
Approximation

Let us describe in brief the key moments of our theoretical approach (look for more
details in Refs. [63, 65–69, 73–76]). As usually, the wave functions zeroth basis is
found from the Dirac equation solution with self-consistent total potential.
The bare Hamiltonian is as follows:



 1 − αi αj
H = ∑ fαcp − βmc2 + Uðri jZÞg + + ∑ exp iωij rij ⋅ , ð1Þ
i i>j rij

where αi , αj —the Dirac matrices, ωij —the transition frequency, c—the light
velocity, Z is a charge of the atomic nucleus. Within relativistic perturbation theory
[3, 4] we introduce the zeroth–order Hamiltonian as:
232 V. B. Ternovsky et al.

H0 = ∑ fαcpi − βmc2 + ½ − Z ̸ri + UMF ðri jbÞ + VXC ðri Þg, ð2Þ
i

where VXC ðri Þ—one-particle exchange-correlation potential, UMF ðri jbÞ—a

self-consistent Coulomb-like mean-field potential (b is the potential parameter,
which is further determined within ab initio procedure), that potential interaction
“quasiparticles-core” in the case of atomic system consisting of closed electron
shells and external quasiparticles.
The relativistic wave functions are calculated by solution of the Dirac equation
with the potential, which includes the Coulomb potential of the closed electron
shells core of an alkali atomsplus the exchange Kohn-Sham potential and corre-
lation Lundqvist-Gunnarsson potential (see details in Refs. [63, 70, 73–76]).
In order to provide the construction of the optimized one-quasiparticle repre-
sentation and improve an effectiveness of the numerical code we have used special
ab initio procedure within relativistic energy approach [68] (see also [69, 70]). It
reduces to accurate treating the lowest order multielectron effects, in particular, the
gauge dependent radiative contribution into imaginary part of the electron system
energy Im δEninv for the certain class of the photon propagator calibrations and
minimization of the corresponding density functional Im δEninv . Some known
alternative approaches to construction of an optimized one-quasiparticle represen-
tation for multielectron atom can be found in Refs. [11–22].
Within the relativistic energy approach [61, 62, 64, 65] an imaginary part of the
electron energy shift of an atom is directly connected with the radiation decay
possibility (transition amplitude). An approach, using the Gell-Mann and Low
formula with the QED scattering matrix, is used in treating the relativistic atom. The
total energy shift of the state is usually presented in the form:

ΔE = ReΔE + iΓ ̸ 2 ð3Þ

where Γ is interpreted as the level width, and the decay possibility P = Γ.

The imaginary part of an electron energy of the atomic system can be determined
in the lowest second order of perturbation theory as:

e2 jωαn j
ImΔEðBÞ = − ∑  Vαnαn , ð4Þ
4π 
α>n>f
α>n≤f

where (α > n > f) for electron and (α < n < f) for vacancy. The matrix element is
determined as follows:
ZZ
jωj sinjωjr12
Vijkl = dr1 dr2 Ψ *i ðr1 ÞΨ *j ðr2 Þ ð1 − α1 α2 ÞΨ *k ðr2 ÞΨ *l ðr1 Þ ð5Þ
r12
Spectroscopy of Radiative Decay Processes … 233

When calculating the matrix elements (5), one should use the angle symmetry of
the task and write the corresponding expansion for sinjωjr12 ̸ r12 on spherical
harmonics as follows [62]:

sin jωjr12 π ∞
= pffiffiffiffiffiffiffiffi ∑ ðλÞJλ + 1 ̸2 ðjωjr1 ÞJλ + 1 ̸2 ðjωjr2 ÞPλ ðcos rd
1 r2 Þ ð6Þ
r12 2 r1 r2 λ = 0

where J is the Bessel function of first kind and ðλÞ = 2λ + 1.

This expansion is corresponding to usual multipole one for probability of
radiative decay (an amplitude approach of quantum mechanics). Substitution of the
expansion (5) to matrix element allow to get the following expression:
 
ω 1 ̸2 μ j1 j3 λ
V1234 = ½ðj1 Þðj2 Þðj3 Þðj4 Þ ∑ ð − 1Þ × ImQλ ð1234Þ
λμ m1 − m3 μ , ð7Þ
Qul Br
Qλ = Qλ + Qλ

where ji is the total single electron momentums, mi—the projections; QQul is the
Coulomb part of interaction, QBr—the Breit part.
The total radiation width of the one-quasiparticle state can be presented in the
following form:


ΓðγÞ = − 2 ImM 1 ðγÞ = − 2 ∑ ð2j + 1ÞImQλ nγ lγ jγ nlj
λnlj

Qλ = QCul Br ð8Þ
λ + Qλ .
QBr Br Br
λ = Qλ, λ − 1 + Qλ, λ + QBr
λ, λ + 1

The individual terms of the Σnlj sum correspond to the partial contribution of the
nλ lλ jλ → nlj transitions; Σλ is a sum of the contributions of the different multiplicity
transitions. The detailed expressions for the Coulomb and Breit parts can be found
in Refs. [62–66].
The imaginary parts of the Coulomb part QCul λ and the Breit part contain the
radial Rλ and angular Sλ integrals as follows (in the Coulomb units) [65]:
n

−1 e e e e
Im QCul
λ ð 12; 43 Þ = Z Im Rλ ð 12; 43 ÞS λ ð 12; 43 Þ + Rλ 12; 4 3 S λ 12; 4 3 +




o ð9Þ
+ Rλ 1e 2; e43 Sλ 1e 43 + Rλ e
2; e 1e2; e
4e3 Sλ e 1e
2; e
4e3 .

1 n e e
l e e

ImQBr
λ, l = Im Rλ 12; 4 3 Sλ 12; 4 3 + Rλ e1e2; 43 Slλ e1e2; 43 +
Z



o ð10Þ
+ Rλ e
12; e
43 Sl e
12; e
λ 43 + Rλ e1e
2; e
4e3 Sl e1e2; e
4e3 .
λ

Here fλ l1 l3 g means that λ, l1 and l3 must satisfy the triangle rule and the sum
λ + l1 + l3 must be an even number. The rest terms in (9), (10) include the small
components of the Dirac functions. The tilde designates that the large radial
234 V. B. Ternovsky et al.

component f must be replaced by the small one g, and instead of li , lĩ = li − 1 should
be taken for ji < li and lĩ = li + 1 for ji > li . The detailed definitions for the radial Rλ
and angular Sλ integrals can be found in Refs. [59, 64–67].
The total probability of a λ—pole transition is usually represented as a sum of
the electric PEλ and magnetic PM λ parts. The electric (or magnetic) λ—pole transition
γ → δ connects two states with parities which by λ (or λ +1) units. In our desig-
nations (the radiative γ → δ transition) one could write:

PEλ ðγ → δÞ = 2ð2j + 1ÞQEλ ðγδ; γδÞ QEλ = QCul Br Br

λ + Qλ, λ − 1 + Qλ, λ + 1
. ð11Þ
PMλ ðγ → δÞ = 2ð2j + 1ÞQλ ðγδ; γδÞ
M M Br
Qλ = Qλ, λ

The adequate, precise computation of radiative parameters of the heavy Rydberg

alkali-metal atoms within relativistic perturbation theory requires an accurate
accounting for the multi-electron exchange-correlation effects (including polariza-
tion and screening effects, a continuum pressure etc.). These effects within our
approach are treated as the effects of the perturbation theory second and higher
orders. Using the standard Feynman diagram technique one should consider
two kinds of diagrams (the polarization and ladder ones), which describe the
polarization and screening exchange-correlation effects. The detailed description of
the polarization diagrams and the corresponding analytical expressions for matrix
elements of the polarization interelectron interaction (through the polarizable core
of an alkali atom) potential is presented in Refs. [63, 73–76].
An effective approach to accounting of the polarization diagrams contributions is
in adding the effective two-quasiparticle polarizable operator into the perturbation
theory first order matrix elements. In Ref. [65] the corresponding non-relativistic
polarization functional has been derived. More correct relativistic expression has
been presented in the Refs. [34, 35] and used in our theory. The corresponding
two-quasiparticle polarization potential looks as follows:
8
1 ̸3
<Z dr ′ ρð0Þ
> ′
c ðr Þ θðr ′ Þ
d
Vpol ðr1 r2 Þ = X
>
: jr1 − r ′ j ⋅ jr ′ − r2 j


1 ̸ 3


1 ̸3
 9
Z dr ′ ρð0Þ
c r′
ð0Þ
θ r ′ Z dr ′′ ρc r ′′ θ r ′′
1 ̸3
>
=
− ̸⟨ ρð0Þ ⟩
j r1 − r′ j jr ′′ − r2 j c
>
;

ð12aÞ

1 ̸ 3 Z
1 ̸ 3
⟨ ρð0Þ
c ⟩= dr ρð0Þ
c ðrÞ θðrÞ, ð12bÞ

h  i2 1 ̸2
̸3
θðrÞ = 1 + 3π 2 ⋅ ρð0Þ
c ðrÞ ̸c 2
ð12cÞ
Spectroscopy of Radiative Decay Processes … 235

where ρ0c is the core electron density (without account for the quasiparticle), X is
numerical coefficient, c is the light velocity. The contribution of the ladder diagrams
(these diagrams describe the immediate interparticle interaction) is summarized by a
modification of the perturbation theory zeroth approximation mean-field central
potential (look [35, 65]), which include the screening (anti-screening) of the core
potential of each particle by the two others. The details of calculating this contri-
bution can be found in Refs. [63, 65–76]. All computing was performed with using
the modified PC code “Superatom-ISAN” (version 93).

3 Results and Conclusions

In Tables 1 and 2 we present the experimental and theoretical values (in atomic
units: a.u.) of the reduced dipole transition matrix elements for the Fr and Cs atoms:
experimental data—Exp; theoretical data: perturbation theory (PT)-DFSD—PT with
the Dirac-Hartree-Fock zeroth approximation (single-double SD approximation in
which single and double excitations of Dirac-Hartree-Fock wave functions are

Table 1 The reduced dipole transition matrix elements for Fr (see text)
Transition/ PT-DFSD PT-DFSD EMP PT-RHF PT-RHF DF RPT-EA Exp.
(corr)
method (corr)
7p1/2-7s 4.256 – – 4.279 4.304 4.179 4.272, 4.277
4.274
8p1/2-7s 0.327 0.306 0.304 0.291 0.301 – 0.339
9p1/2-7s 0.110 0.098 0.096 – – – 0.092
10p1/2-7s – – – – – – 0.063
7p3/2-7s 5.851 – – 5.894 5.927 5.791 5.891 5.898
8p3/2-7s 0.934 0.909 0.908 0.924 – – 0.918 –
9p3/2-7s 0.436 0.422 0.420 – – – 0.426 –
10p3/2-7s – – – – – – 0.284 –
7p1/2-8s 4.184 4.237 4.230 4.165 4.219 4.196 4.228 –
8p1/2-8s 10.02 10.10 10.06 10.16 10.00 10.12 –
9p1/2-8s 0.985 – 0.977 – – – 0.972 –
10p1/2-8s – – – – – – 0.395 –
7p3/2-8s 7.418 7.461 7.449 7.384 7.470 7.472 7.453 –
8p3/2-8s 13.23 13.37 13.32 13.45 13.26 13.35 –
9p3/2-8s 2.245 – 2.236 – – – 2.232 –
10p3/2-8s – – – – – – 1.058 –
7p1/2-9s 1.016 – 1.010 – – – 1.062 –
8p1/2-9s 9.280 – 9.342 – – – 9.318 –
9p1/2-9s 17.39 – 17.40 – – – 17.42 –
(continued)
236 V. B. Ternovsky et al.

Table 1 (continued)
Transition/ PT-DFSD PT-DFSD EMP PT-RHF PT-RHF DF RPT-EA Exp.
(corr)
method (corr)
10p1/2-9s – – – – – 1.836 –
7p3/2-9s 1.393 – 1.380 – – – 1.41 –
8p3/2-9s 15.88 – 15.92 – – – 15.96 –
9p3/2-9s 22.59 – 22.73 – – – 22.68 –
10p3/2-9s – – – – – – 3.884 –

Table 2 The reduced dipole transition matrix elements for Cs (see text)
Transition PT-DFSD PT-DFSD DF PT-RHF QDA EF-RMP Exp.
(corr)
6p1/2-6s 4.482 4.535 4.510 – 4.282 4.489 4.4890(7)
6p3/2-6s 6.304 6.382 6.347 – 5.936 6.323 6.3238(7)
7p1/2-6s 0.297 0.279 0.280 0.2825 0.272 0.283 0.284(2)
7p3/2-6s 0.601 0.576 0.576 0.582 0.557 0.583 0.583(9)
8p1/2-6s 0.091 0.081 0.078 – 0.077 0.088 –
8p1/2-6s 0.232 0.218 0.214 – 0.212 0.228 –
6p1/2-7s 4.196 4.243 4.236 4.237 4.062 4.234 4.233(22)
6p3/2-7s 6.425 6.479 6.470 6.472 6.219 6.480 6.479(31)
7p1/2-7s 10.254 10.310 10.289 10.285 9.906 10.309 10.309
(15)
7p3/2-7s 14.238 14.323 14.293 14.286 13.675 14.323 14.325
(20)

included to all PT orders); PT-DFSD—PT with the Dirac-Hartree-Fock zeroth

approximation (plus compilation), EMP—empirical model potential method;
PT-RHF—PT with the relativistic Hartree-Fock (RHF) zeroth approximation and
PT-RHF(corr)—corrected version with using empirical data; DF-PT with the
Dirac-Fock zeroth approximation; RPT-EA—our method (relativistic PT with the
Dirac-Kohn-Sham zeroth approximation combined with an energy approach (EA)).
All data have been taken from Refs. [1–6, 10, 30, 31]).
In Fig. 1 we present a dependence of the calculated reduced dipole matrix
elements upon a principal quantum number for different states of the Rydberg atom
Rb: 5P3/2-nD5/2 (n ∼ 70): available experimental data—the circles; Theory: con-
tinuous line—our data; the dotted line- data by Piotrowicz et al., obtained within the
quasiclassical Dyachkov-Pankratov model [4, 10, 12, 30, 31]. In Figs. 2 and 3 we
present the same dependences for the Rydberg states of the Cs atom: 6P3/2 → nD5/2
and the Fr atom: 7P3/2 → nD5/2, n = 10–70.
The detailed analysis of the computation data shows very important role of the
relativistic and interelectron exchange-correlation effects (for example the contri-
bution due to the perturbation theory second and higher orders, including the
Spectroscopy of Radiative Decay Processes … 237

Fig. 1 A dependence of the calculated reduced dipole matrix elements upon principal quantum
number for Rydberg atom Rb: 5P3/2 → nD5/2 (n ∼ 70). The available experimental data are listed
as a circle; Theory: continuous line—our data, dotted line- data by Piotrowicz et al. within the
quasi-classical Dyachkov-Pankratov model (see text)

Fig. 2 A dependence of the calculated reduced dipole matrix elements upon principal quantum
number for Rydberg atom Cs: 6P3/2 → nD5/2 (n ∼ 70). The available experimental data are listed
as a circle; Theory: continuous line—our data, dotted line- data by Piotrowicz et al. within the
quasi-classical Dyachkov-Pankratov model (see text)

interelectron polarization interaction and mutual screening ones reaches ∼40%), as

well as the effect of the non-Coulomb grouping levels in the Rydberg spectra.
To conclude, we have presented the results of studying the radiation decay
processes and computing reduced dipole matrix elements (radiative amplitudes) of
transitions in spectra of heavy Rydberg atoms of alkali elements (Rb, Cs, Fr; for the
states with the principal quantum number n = 10–80) on the basis of the gener-
alized relativistic energy approach and the relativistic many-body perturbation
theory with the optimized one-quasiparticle representation. A critically important
238 V. B. Ternovsky et al.

Fig. 3 A dependence of the calculated reduced dipole matrix elements upon principal quantum
number for Rydberg atom Fr: 7P3/2 → nD5/2, n = 10–80 (our data)

role of the many-body exchange-polarization effects and effect of the non-Coulomb

grouping levels in the Rydberg spectra has been found. The detailed numerical data
on the dipole matrix elements and transition probabilities are listed in Refs.
[99, 100].

Acknowledgements The authors are very much thankful to Prof. J. Maruani and Dr. Y. A. Wang
for invitation to make contributions on the QSCP-XVI workshop (Vancouver, Canada). The useful
comments of the anonymous referees are very much acknowledged too.

References

1. Martin W (2004) NIST spectra database, version 2.0. NIST, Washington. http://physics.nist.
gov.asd
2. Moore C (1987) NBS spectra database. NBS, Washington
3. Weiss A (1977) J Quant Spectrosc Radiat Transf 18:481; Phys Scripta T65:188 (1993)
4. Nadeem A, Haq SU (2010) Phys Rev A 81:063432; (2011) Phys Rev A 83:063404
5. Simsarian JE, Orozco LA, Sprouse GD, Zhao WZ (1998) Phys Rev A 57:2448
6. Curtis L (1995) Phys Rev A 51:4574
7. Li Y, Pretzler G, Fill EE (1995) Phys Rev A 52:R3433–R3435
8. Feng Z-G, Zhang L-J, Zhao J-M, Liand C-Y, Jia S-T (2009) J Phys B At Mol Opt Phys
42:145303
9. Marinescu M, Vrinceanu D, Sadeghpour HR (1998) Phys Rev A 58:R4259
10. Safronova UI, Johnson W, Derevianko A (1999) Phys Rev A 60:4476; Dzuba V,
Flambaum V, Safranova MS (2006) Phys Rev A 73:02211
11. Grant IP (2007) Relativistic quantum theory of atoms and molecules, theory and
computation. In: Springer series on atomic, optical, and plasma physics, vol 40. Springer,
Berlin, pp 587–626
12. Glushkov AV (2008) Relativistic quantum theory. Quantum mechanics of atomic systems.
Astroprint, Odessa, p 700
Spectroscopy of Radiative Decay Processes … 239

13. Wilson S (2007) Recent advances in theoretical physics and chemistry systems. In:
Maruani J, Lahmar S, Wilson S, Delgado-Barrio G (eds) Series: Progress in theoretical
chemistry and physics, vol 16. Springer, Berlin, pp 11–80
14. Feller D, Davidson ER (1981) J Chem Phys 74:3977
15. Froelich P, Davidson ER, Brändas E (1983) Phys Rev A 28:2641
16. Rittby M, Elander N, Brändas E (1983) Int J Quantum Chem 23:865
17. Wang YA, Yam CY, Chen YK, Chen GH (2011) J Chem Phys 134:241103
18. Maruani J (2016) J Chin Chem Soc 63:33
19. Pavlov R, Mihailov L, Velchev C, Dimitrova-Ivanovich M, Stoyanov Z, Chamel N, Maru-
ani J (2010) J Phys Conf Ser 253:012075
20. Dietz K, Heβ BA (1989) Phys Scripta 39:682
21. Kohn JW, Sham LJ (1964) Phys Rev A 140:1133
22. Gross EG, Kohn W (2005) Exchange-correlation functionals in density functional theory.
Plenum, New York
23. Froese Fischer C, Tachiev G, Irimia A (2006) GAtom Data Nucl Data Tables 92:607
24. Cheng K, Kim Y, Desclaux J (1979) At Data Nucl Data Tables 24:11
25. Safranova UI, Safranova MS, Johnson W (2005) Phys Rev A 71:052506
26. Ternovsky EV, Antoshkina OA, Florko TA, Tkach TB (2017) Photoelectronics 26:139
27. Indelicato P, Desclaux JP (1993) Phys Scripta T 46:110
28. Dzuba VA, Flambaum VV, Sushkov OP (1995) Phys Rev A 51:3454
29. Sapirstein J (1998) Rev Modern Phys 70:55
30. Piotrowicz M, MacCormick C, Kowalczyk A et al (2011) arXiv:1103.0109v2 [quant-ph];
Beterov I, Mansell CW, Yakshina EA et al (2012) arXiv:1207.3626v1 [physics.atom-ph]
31. Dyachkov LG, Pankratov PM (1994) J Phys B 27(3):461; Piotrowicz M, MacCormick C,
Kowalczyk A, Bergamini S, Yakshina EA (2011) New Journ Phys 13:093012
32. Quinet P, Argante C, Fivet V, Terranova1 C, Yushchenko AV, Biémont É (2007) Astrophys
Astron 474:307
33. Biémont É, Fivet V, Quinet P (2004) J Phys B At Mol Opt Phys 37:4193
34. Glushkov AV (1991) Opt Spectrosc 70:555
35. Glushkov AV (1990) Soviet Phys J 33(1):1
36. Khetselius OY (2009) Int J Quantum Chem 109:3330
37. Khetselius OY (2009) Phys Scripta T135:014023
38. Malinovskaya SV, Glushkov AV, Khetselius OY, Svinarenko AA, Mischenko EV,
Florko TA (2009) Int J Quant Chem 109(4):3325
39. Glushkov AV, Loboda AV, Gurnitskaya EP, Svinarenko AA (2009) Phys Scripta
T135:014022
40. Glushkov AV, Khetselius OY, Svinarenko AA (2013) Phys Scripta T153:014029
41. Svinarenko AA (2014) J Phys Conf Ser 548:012039
42. Glushkov AV, Mansarliysky VF, Khetselius OY, Ignatenko AV, Smirnov A, Prepelitsa GP
(2017) J Phys Conf Ser 810:012034
43. Glushkov AV, Loboda AV (2007) J Appl Spectrosc 74:305. (Springer)
44. Malinovskaya SV, Glushkov AV, Khetselius OY, Loboda AV, Lopatkin YuM,
Svinarenko AA, Nikola LV, Perelygina TB (2011) Int J Quantum Chem 111:288
45. Glushkov AV, Khetselius OY, Loboda AV, Ignatenko AV, Svinarenko AA,
Korchevsky DA, Lovett L (2008) Spectral line shapes. AIP Conf Proc 1058:175
46. Khetselius OY (2012) J Phys Conf Ser 397:012012
47. Khetselius OY, Florko TA, Svinarenko AA, Tkach TB (2013) Phys Scripta T153:014037
48. Khetselius OY (2008) Spectral line shapes. AIP Conf Proc 1058:363
49. Khetselius OY, Glushkov AV, Gurnitskaya EP, Loboda AV, Mischenko EV, Florko T,
Sukharev D (2008) Spectral line shapes. AIP Conf Proc 1058:231
50. Khetselius OY (2007) Photoelectronics 16:129
51. Khetselius OY, Gurnitskaya EP (2006) Sens Electron Microsyst Technol N 3:35
52. Khetselius OY, Gurnitskaya EP (2006) Sens Electron Microsyst Technol N 2:25
53. Johnson WR, Lin CD, Cheng KT (1980) Phys Scr 21:409
240 V. B. Ternovsky et al.

54. Johnson WR, Sapistein J, Blundell S (1988) Phys Rev A 37:307

55. Khetselius OY (2011) Quantum structure of electroweak interaction in heavy finite
Fermi-systems. Astroprint, Odessa
56. Hibbert A, Hansen JE (1994) J Phys B At Mol Opt Phys 27:3325
57. Braun MA, Dmitriev YuYu, Labzovsky LN (1969) JETP 57:2189
58. Tolmachev VV (1969) Adv Quantum Chem 4:331
59. Ivanov LN, Ivanova EP (1979) At Data Nucl Data Tables 24:95
60. Driker MN, Ivanova EP, Ivanov LN, Shestakov AF (1982) J Quant Spectrosc Radiat Transf
28:531
61. Ivanov LN, Letokhov VS (1985) Com Mod Phys D At Mol Phys 4:169
62. Vidolova-Angelova E, Ivanov LN, Ivanova EP, Angelov DA (1986) J Phys B At Mol Opt
Phys 19:2053
63. Glushkov AV (2006) Relativistic and correlation effects spectra of atomic systems.
Astroprint, Odessa
64. Ivanov LN, Ivanova EP, Aglitsky EV (1988) Phys Rep 166:315
65. Ivanova EP, Ivanov LN, Glushkov AV, Kramida AE (1985) Phys Scripta 32:513
66. Ivanova EP, Glushkov AV (1986) J Quant Spectrosc Radiat Transf 36:127
67. Ivanov LN, Ivanova EP, Knight L (1993) Phys Rev A 48:4365
68. Glushkov AV, Ivanov LN (1992) Phys Lett A 170:33
69. Glushkov AV, Ivanov LN, Ivanova EP (1986) Autoionization phenomena in atoms.
Moscow University Press, Moscow, pp 58–160
70. Glushkov AV (2012) Quantum systems in chemistry and physics: progress in methods and
applications. In: Nishikawa K, Maruani J, Brändas E, Delgado-Barrio G, Piecuch P
(eds) Series: Progress in theoretical chemistry and physics, vol 26. Springer, pp 231–252
71. Glushkov AV, Ivanov LN (1993) J Phys B At Mol Opt Phys 26:L379
72. Glushkov AV (1992) JETP Lett 55:97
73. Khestelius OYu (2008) Hyperfine structure of atomic spectra. Astroprint, Odessa, 210p
74. Svinarenko AA, Glushkov AV, Khetselius OY, Ternovsky VB, Dubrovskaya YV,
Kuznetsova AA, Buyadzhi VV (2017) In: Orjuela JEA (ed) Rare earth element. InTech,
pp 83–104
75. Glushkov AV, Khetselius OY, Svinarenko AA, Buyadzhi VV, Ternovsky VB, Kuznetsova,
Bashkarev PG (2017) In: Uzunov DI (ed) Recent studies in perturbation theory. InTech,
pp 131–150
76. Glushkov AV, Ambrosov SV, Loboda AV, Chernyakova Yu G, Svinarenko AA,
Khetselius OY (2004) Nucl Phys A Nucl Hadr Phys 734:21
77. Glushkov AV, Malinovskaya SV, Sukharev DE, Khetselius OY, Loboda AV, Lovett L
(2009) Int J Quantum Chem 109:1717
78. Glushkov AV (2013) Advances in quantum methods and applications in chemistry, physics,
and biology. In: Hotokka M, Maruani J, Brändas E, Delgado-Barrio G (ed) Series: Progress
in theoretical chemistry and physics, vol 27B. Springer, pp 161–178
79. Glushkov AV, Khetselius OY, Svinarenko AA, Prepelitsa GP (2010) In: Duarte FJ
(ed) Coherence and ultrashort pulsed emission. InTech, Rijeka, pp 159–186
80. Buyadzhi VV, Glushkov AV, Lovett L (2014) Photoelectronics 23:38
81. Buyadzhi VV, Glushkov AV, Mansarliysky VF, Ignatenko AV, Svinarenko AA (2015) Sens
Electron Microsyst Technol 12(4):27
82. Glushkov AV, Khetselius OY, Malinovskaya SV (2008) Frontiers in quantum systems in
chemistry and physics. In: Wilson S, Grout PJ, Maruani J, Delgado-Barrio G, Piecuch P
(eds) Series: Progress in theoretical chemistry and physics, vol 18. Springer, pp 525–541
83. Glushkov AV, Khetselius OY, Malinovskaya SV (2008) Eur Phys J Spec Top 160:195
84. Glushkov AV, Khetselius OY, Malinovskaya SV (2008) Mol Phys 106:1257
85. Glushkov AV, Khetselius OY, Svinarenko AA (2012) Advances in the theory of quantum
systems in chemistry and physics. In: Hoggan P, Maruani J, Brandas E, Delgado-Barrio G,
Piecuch P (eds) Series: Progress in theoretical chemistry and physics, vol 22. Springer,
pp 51–68
Spectroscopy of Radiative Decay Processes … 241

86. Glushkov AV, Khetselius OY, Lovett L (2010) Advances in the theory of atomic and
molecular systems dynamics, spectroscopy, clusters, and nanostructures. In: Piecuch P,
Maruani J, Delgado-Barrio G, Wilson S (eds) Series: Progress in theoretical chemistry and
physics, vol 20. Springer, pp 125–152
87. Glushkov AV, Khetselius OY, Loboda AV, Svinarenko AA (2008) Frontiers in quantum
systems in chemistry and physics. In: Wilson S, Grout PJ, Maruani J, Delgado-Barrio G,
Piecuch P (eds) Series: Progress in theoretical chemistry and physics, vol 18. Springer,
pp 543–560
88. Glushkov AV, Khetselius O, Gurnitskaya E, Loboda A, Florko T, Sukharev D, Lovett L
(2008) Frontiers in quantum systems in chemistry and physics. In: Wilson S, Grout P,
Maruani J, Delgado-Barrio G, Piecuch P (eds) Series: Progress in theoretical chemistry and
physics, vol 18. Springer, pp 507–524
89. Glushkov AV, Ambrosov SV, Loboda AV, Gurnitskaya EP, Khetselius OY (2006) Recent
advances in theoretical physics and chemistry systems. In: Julien JP, Maruani J, Mayou D,
Wilson S, Delgado-Barrio G (eds) Series: Progress in theoretical chemistry and physics, vol
15. Springer, pp 285–299
90. Malinovskaya SV, Glushkov AV, Dubrovskaya YV, Vitavetskaya LA (2006) Recent
advances in theoretical physics and chemistry systems. In: Julien J-P, Maruani J, Mayou D,
Wilson S, Delgado-Barrio G (eds) Series: Progress in theoretical chemistry and physics, vol
15. Springer, pp 301–307
91. Khetselius OY (2012) Quantum systems in chemistry and physics: progress in methods and
applications. In: Nishikawa K, Maruani J, Brandas E, Delgado-Barrio G, Piecuch P
(eds) Series: Progress in theoretical chemistry and physics, vol 26. Springer, pp 217–229
92. Khetselius OY (2015) Frontiers in quantum methods and applications in chemistry and
physics. In: Nascimento M, Maruani J, Brändas E, Delgado-Barrio G (eds) Series: Progress
in theoretical chemistry and physics, vol 29. Springer, pp 55–76
93. Glushkov AV, Svinarenko AA, Khetselius OY, Buyadzhi VV, Florko TA, Shakhman AN
(2015) Frontiers in quantum methods and applications in chemistry and physics. In:
Nascimento M, Maruani J, Brändas E, Delgado-Barrio G (eds) Series: Progress in theoretical
chemistry and physics, vol 29. Springer, pp 197–217
94. Khetselius OY, Zaichko PA, Smirnov AV, Buyadzhi VV, Ternovsky VB, Florko TA,
Mansarliysky VF (2017) Quantum systems in physics, chemistry, and biology. In: Tadjer A,
Pavlov R, Maruani J, Brändas E, Delgado-Barrio G (eds) Series: Progress in theoretical
chemistry and physics, vol 30. Springer, pp 271–281
95. Glushkov AV, Rusov VD, Ambrosov SV, Loboda AV (2003) New projects and new lines of
research in nuclear physics. In: Fazio G, Hanappe F (eds). World Scientific, Singapore,
pp 126–132
96. Glushkov AV, Malinovskaya SV, Loboda AV, Shpinareva IM, Prepelitsa GP (2006) J Phys
Conf Ser 35:420
97. Glushkov AV, Malinovskaya SV, Chernyakova YG, Svinarenko AA (2004) Int J Quantum
Chem 99:889
98. Glushkov AV, Ambrosov SV, Ignatenko AV, Korchevsky DA (2004) Int J Quantum Chem
99:936
99. Glushkov AV, Ignatenko AV, Khetselius OY, Ternovsky VB (2017) Spectroscopy of
Rydberg atoms and relativistic quantum chaos, OSENU, Odessa, p 152
100. Ternovsky VB, Buyadzhi VV, Gurskaya MY, Kuznetsova AA (2015) Preprint OSENU,
NAM-3 (Odessa, 2015), p 32; Preprint OSENU, OSENU, NAM-4 (Odessa, 2015), p 36
Enhancement Factors for Positron
Annihilation on Valence and Core
Orbitals of Noble-Gas Atoms

D. G. Green and G. F. Gribakin

Abstract Annihilation momentum densities and vertex enhancement factors for

positron annihilation on valence and core electrons of noble-gas atoms are calcu-
lated using many-body theory for s, p and d-wave positrons of momenta up to the
positronium-formation threshold. The enhancement factors parametrize the effects
of short-range electron-positron correlations which increase the annihilation proba-
bility beyond the independent-particle approximation. For all positron partial waves
and electron subshells, the enhancement factors are found to be relatively insensitive
to the positron momentum. The enhancement factors for the core electron orbitals
are also almost independent of the positron angular momentum. The largest enhance-
ment factor (∼10) is found for the 5p orbital in Xe, while the values for the core
orbitals are typically ∼1.5.

Keywords Positron annihilation ⋅ Annihilation momentum density

Many-body theory ⋅ Enhancement factors ⋅ Noble-gas atoms

1 Introduction
Low-energy positrons annihilate in atoms and molecules forming two 𝛾 rays whose
Doppler-broadened spectrum is characteristic of the electron velocity distribution in
the states involved, and thus of the electron environment. This makes positrons a
unique probe in materials science. For example, vacancies and defects in semicon-
ductors and other industrially important materials can be studied [1–6]. Positron-
induced Auger-electron spectroscopy (PAES) [7–11] and time-resolved PAES
[11, 12] enable studies of surfaces with extremely high sensitivity, including dynam-
ics of catalysis, corrosion, and surface alloying [13]. The 𝛾 spectra are also sensitive
to the positron momentum at the instant of annihilation. This is exploited in Age-

D. G. Green (✉) ⋅ G. F. Gribakin

Centre for Theoretical Atomic, Molecular and Optical Physics,
Queen’s University Belfast, Belfast, Northern Ireland BT71NN, UK
e-mail: [email protected]
G. F. Gribakin
e-mail: [email protected]

© Springer International Publishing AG, part of Springer Nature 2018 243

Y. A. Wang et al. (eds.), Concepts, Methods and Applications of Quantum Systems
in Chemistry and Physics, Progress in Theoretical Chemistry and Physics 31,
https://doi.org/10.1007/978-3-319-74582-4_14
244 D. G. Green and G. F. Gribakin

MOmentum Correlation (AMOC) experiments (see, e.g., [14–16]), in which the 𝛾

spectra are measured as a function of the positron “age” (i.e., time after emission
from source). AMOC enables study of positron and positronium cooling (and, more
generally, transitions between positron states, e.g., for different trapping states, or
via chemical reactions) [14–16].
Interpretation of the experiments relies heavily on theoretical input, e.g., in PAES
one requires accurate relative annihilation probabilities for core electrons of various
atoms [17]. Such quantities, however, are not easy to calculate, as the annihilation
process is strongly affected by short-range electron-positron and long-range positron-
atom correlations. These effects significantly enhance the annihilation rates [18, 19]
and alter the shape and magnitude of the annihilation 𝛾 spectra [20–24], compared
to independent-particle approximation (IPA) calculations.
A powerful method that allows for systematic inclusion of the correlations in
atomic systems is many-body theory (MBT). MBT enables one to calculate the so-
called enhancement factors (EF), which quantify the increase of the electron den-
sity at the positron due to the effect of correlations. The EF can be used to correct
the IPA annihilation probabilities and 𝛾-spectra [2, 17]. They are particularly large
(∼10) for the valence electrons, but are also significant for the core electrons [25].1
EF were introduced in early MBT works involving positron annihilation in metals
that were based on considering positrons in a homogeneous electron gas [28, 29].
Subsequently, density functional theories (DFT) were developed to describe positron
states and annihilation in a wider class of condensed-matter systems [30, 31]. These
methods usually rely on some input in the form of the correlation energy and EF for
the positron in electron gas from MBT [32]. When applied to real, inhomogeneous
systems, position-dependent EF can lead to spurious effects in the spectra [5], and
show deficiencies when benchmarked against more accurate calculations [33].
A recent study of a model system of eight electrons and one positron confined
in a harmonic potential [34] highlighted significant discrepancies between the anni-
hilation momentum densities (AMD) and EF obtained using exact diagonalization
and those found using common DFT approaches. The best agreement for the shapes
of AMD was observed for position-dependent EF [35, 36] calculated in the Kahana
formalism [36, 37], though there was a factor-of-two difference for the total annihi-
lation rate. It was also suggested in [34] (see also [38]) that the EF could be defined
rigorously using natural geminals. These quantities are electron-positron pair wave-
functions which diagonalise the two-body reduced density matrix, and which can be
extracted from the accurate many-particle wavefunction. It would be very useful if
the natural geminals could be used without the knowledge of the total wavefunction,
and it is possible that this can be done using MBT.
In the context of the positron-atom problem, the MBT calculations provided an
accurate and essentially complete picture of low-energy positron interaction with

1
Positron annihilation with core electrons is also affected by exchange-assisted tunnelling [26, 27].
This is a manifestation of electron exchange, which increases the wavefunctions of inner electrons in
the range of distances of the valence electrons. For this effect to be properly included in a calculation,
one needs to use true nonlocal exchange potentials, e.g., at the Hartree-Fock level, as is the case in
the present calculations.
Enhancement Factors for Positron Annihilation on Valence . . . 245

noble-gas atoms [19], with excellent agreement between the theoretical results and
experimental scattering cross sections and annihilation rates. The MBT work was
extended recently [24] to the 𝛾-spectra for (thermal) positron annihilation on noble-
gas atoms. It provided an accurate description of the measured spectra for Ar, Kr and
Xe [3] and firmly established the relative contributions of various atomic orbitals to
the spectra. The calculations also yielded “exact” ab initio EF 𝛾̄nl for individual elec-
tron orbitals nl, and found that they follow a simple scaling with the orbital ionization
energy [24].
In this work we provide a more detailed analysis and report EF for annihilation of
s-, p- and d-wave positrons with momenta up to the positronium formation threshold.
We demonstrate that the EF for a given electron orbital and positron partial wave are
insensitive to the positron momentum (in spite of the strong momentum dependence
of the annihilation probability [19]). Moreover, we show that whilst the EF for the
core orbitals are almost independent of the positron angular momenta, those for the
valence subshells vary between the positron s, p and d waves. In addition to their
use in correcting IPA calculations of positron annihilation with core electrons in
condensed matter, the positron-momentum dependent EF calculated here can be used
to determine accurate pick-off annihilation rates for positronium in noble gases [39].

2 Theory of Positron Annihilation in Many-Electron Atoms

2.1 Basics
Consider annihilation of a low-energy (𝜀 ∼ 1 eV) positron with momentum 𝐤 in a
many-electron system, e.g., an atom. In the dominant process, the positron anni-
hilates with an electron in state n to form two 𝛾-ray photons of total momentum
𝐏 [40]. In the centre-of-mass frame, where the total momentum 𝐏 is zero, the
two photons are emitted in opposite directions and have equal energies E𝛾 = p𝛾 c =
mc2 + 12 (Ei − Ef ) ≃ mc2 ≃ 511 keV, where Ei and Ef denote the energy of the initial
and final states of the system (excluding rest mass). When 𝐏 is non-zero, however,
the two photons no longer propagate in exactly opposite directions and their energy
is Doppler shifted. For example, for the first photon E𝛾1 = E𝛾 + mcV cos 𝜃, where 𝜃
is the angle between the momentum of the photon and the centre-of-mass velocity of
the electron-positron pair 𝐕 = 𝐏∕2m (assuming that V ≪ c, and p𝛾1 = E𝛾1 ∕c ≈ mc).
The Doppler shift of the photon energy from the centre of the line then is

Pc
𝜖 = E𝛾1 − E𝛾 = mc V cos 𝜃 = cos 𝜃. (1)
2
The typical momenta of electrons bound with√energy 𝜀n determine the characteristic
width of the annihilation spectrum 𝜖 ∼ Pc ∼ |𝜀n |mc2 ≫ |𝜀n |. Hence the shift 𝜀n ∕2
of the line centre E𝛾 from mc2 = 511 keV can usually be neglected, even for the core
electrons. The 𝛾 spectrum averaged over the direction of emitted photons (or that of
the positron momentum 𝐤) takes the form (see, e.g., [20])
246 D. G. Green and G. F. Gribakin

∞
1 PdPd𝛺𝐏
wn (𝜖) = |An𝐤 (𝐏)|2 , (2)
c ∫ ∫2|𝜖|∕c (2𝜋)3

where An𝐤 (𝐏) is the annihilation amplitude, whose calculation using MBT is
described below. The quantity |An𝐤 (𝐏)|2 is the annihilation momentum density.2
The annihilation rate 𝜆 for a positron in a gas of atoms or molecules with number
density nm is usually parametrized by

𝜆 = 𝜋r02 cnm Zef f , (3)

where r0 = e2 ∕mc2 is the classical radius of the electron (in CGS units) and Zef f
is the effective number of electrons per target atom or molecule that contribute to
∑
annihilation [42, 43]. It is found as a sum over electron states Zef f = n Zef f ,n , where

d3 𝐏
Zef f ,n = wn (𝜖) d𝜖 = |An𝐤 (𝐏)|2 (4)
∫ ∫ (2𝜋)3

is the partial contribution due to positron annihilation with electron in state n, and
where it is assumed that the incident positron wavefunction used in the calculation
of An𝐤 (𝐏) is normalized to a plane wave. In general, the parameter Zef f is different
from the number of electrons in the target atom Z. In particular, positron-atom and
electron-positron correlations can make Zef f ≫ Z [19, 24, 44–46].

2.2 Many-Body Theory for the Annihilation Amplitude

The incident positron wavefunction is taken in the form of a partial-wave expansion3

√
4𝜋 𝜋 ∑ 𝓁 i𝛿𝓁 ∗ ̂
𝜓𝐤 (𝐫) = i e Y𝓁m (𝐤)Y𝓁m (̂𝐫 )P𝜀𝓁 (r), (5)
r k 𝓁m

where 𝛿𝓁 is the scattering phaseshift [47], Y𝓁m is the spherical harmonic, and
where the radial function with orbital angular momentum 𝓁 is normalized by
its asymptotic behaviour P𝜀𝓁 (r) ≃ (𝜋k)−1∕2 sin(kr − 𝜋𝓁∕2 + 𝛿𝓁 ). In the simplest
approximation the radial wavefunctions are calculated in the static field of the
ground-state (Hartree-Fock, HF) atom. This approximation is very inaccurate for the
positron-atom problem. It fails to describe the scattering cross sections and grossly

2 Alternatively to the Doppler-shift spectrum, experiments measure the one-dimensional angular

correlation of annihilation radiation (1D-ACAR), i.e., the small angle 𝛩 between the direction of
one photon and the plane containing the other. The corresponding distribution can be obtained from
w(𝜀) using 𝛩 = 2𝜖∕mc2 . Not also that if the positron wavefunction is constant, then the annihilation
momentum density is proportional to the electron momentum density, and the 𝛾 spectrum becomes
similar to the Compton profile [22, 23, 41].
3 In this and subsequent sections we make wide use of atomic units (a.u.).
Enhancement Factors for Positron Annihilation on Valence . . . 247

underestimates the annihilation rates. Much more accurate positron wavefunctions

(Dyson orbitals) are obtained by solving the Dyson equation which includes the non-
local, energy-dependent positron-atom correlation potential [19, 48] (Sect. 2.2.2).
In the lowest-order approximation the annihilation amplitude is given by

An𝐤 (𝐏) = e−i𝐏⋅𝐫 𝜓𝐤 (𝐫)𝜑n (𝐫)d3 𝐫, (6)

∫

where 𝜑n (𝐫) ≡ 𝜑nlm (𝐫) = 1r Pnl (r)Ylm (̂𝐫 ) is the wavefunction of electron in subshell
nl. Equation (6) is equivalent to IPA. After integration over the directions of 𝐏 in the
spectrum (2), all positron partial waves contribute to the AMD incoherently. This
means that the annihilation amplitude can be calculated independently for each 𝓁,
replacing 𝜓𝐤 (𝐫) in Eq. (6) by the corresponding positron partial wave orbital 𝜓𝜀 (𝐫).
Omitting the index 𝓁, we denote such amplitude An𝜀 (𝐏).
As described below, the main corrections to the zeroth-order amplitude originate
from the electron-positron Coulomb interaction which increases the probability of
finding the electron and positron at the same point in space.

2.2.1 The Annihilation Vertex

Figure 1 shows the amplitude An𝜀 (𝐏) in diagrammatic form [20, 21, 24, 49].4 The
total amplitude is depicted on left-hand side of the diagrammatic equation, with
the double line (𝜀) corresponding to incoming positron that annihilates an elec-
tron in orbital n, producing two 𝛾-rays (double-dashed line), and the circle with a
cross denoting the full annihilation vertex. The main contributions to the amplitude
are shown on the right-hand side of the equation. Diagram (a) is the zeroth-order
amplitude [IPA, Eq. (6)], diagram (b) is the first-order correction and diagram (c)
is the nonperturbative ‘virtual-positronium’ correction. This correction contains the
shaded ‘𝛤 -block’ which represents the sum of an infinite series of electron-positron
ladder diagrams shown in the lower part of Fig. 1.
The ladder diagrams represented by the 𝛤 -block are important because the
electron-positron Coulomb attraction supports bound states of the positronium (Ps)
atom. To form Ps, the energy of the incident positron needs to be greater than the
Ps-formation threshold EPs = I − 6.8 eV, where I is the ionization potential of the
atom. However, even at lower energies where the Ps can only be formed virtually, this
process gives a noticeable contribution. In practice, the 𝛤 -block is found from the
linear equation 𝛤 = V + V𝜒𝛤 , shown diagrammatically in the lower part of Fig. 1,
where V is the electron-positron Coulomb interaction and 𝜒 is the propagator of the
intermediate electron-positron state. Discretizing the electron and positron continua
by confining the system in a spherical cavity reduces this to a linear matrix equation,
which is easily solved numerically (see [19, 21, 48, 50] for further details).

4 It
is also possible to develop a diagrammatic expansion for Zef f [19, 20, 44, 45, 48] that enables
one to calculate the annihilation rate directly, rather than from Eq. (4).
248 D. G. Green and G. F. Gribakin

(a) (b) (c)

P P P P
ε ε ε ε ν1 ν2
+ ν
= + +
Γ
n n μμ μ1 μ2
n
n

ν1 ν2 ν1 ν2 ν1 ν2 ν1 ν2 ν1 ν2 ν1 ν ν2
Γ ≡ + + + ... = + Γ
μ1 μ2 μ1 μ2 μ1 μ2 μ1 μ2 μ1 μ2 μ1 μ μ2

Fig. 1 Amplitude of positron annihilation with an electron in state n: a zeroth-order, b first-order,

and c with virtual-positronium corrections. Double lines labelled 𝜀 represent the incident positron;
single lines labelled 𝜈 (𝜇) represent positron (excited electron) states; lines labelled n represent
holes in the atomic ground state; wavy lines represent the electron-positron Coulomb interaction,
and double-dashed lines represent the two 𝛾-ray photons. The 𝛤 -block is the sum of the electron-
positron ladder diagram series. Summation over all intermediate positron, electron, and hole states
is assumed.

The total amplitude takes the form

{ }
An𝜀 (𝐏) = e−i𝐏⋅𝐫 𝜓𝜀 (𝐫)𝜑n (𝐫) + 𝛥̃𝜀 (𝐫; 𝐫1 , 𝐫2 )𝜓𝜀 (𝐫1 )𝜑n (𝐫2 )d3 𝐫1 d3 𝐫2 d3 𝐫. (7)
∫

Here, the first term, corresponding to the diagram Fig. 1a, is simply the Fourier trans-
form of the product of electron and positron wavefunctions, taken at the same point.
The second term, involving the non-local annihilation kernel 𝛥̃𝜀 (of non-trivial form),
describes the vertex corrections. Note that An𝜀 (𝐏) is the Fourier transform of the cor-
related pair wavefunction (the term in the braces5 ). References [49, 50] present the
partial-wave analysis and corresponding working analytic expressions for the matrix
elements involving the vertex corrections.

2.2.2 Dyson Equation for the Positron Wavefunction

As mentioned above, accurate annihilation rates and 𝛾-spectra can be obtained only
by taking into account the positron-atom correlation potential. This potential is
described by another class of diagrams that “dress” the positron wavefunction. The
corresponding positron quasiparticle wavefunction (or Dyson orbital, double line in
Fig. 1) is calculated from the Dyson equation (see, e.g., [51–53])

5 The term in braces can also be compared with the expression for the natural geminal corresponding
√
to the positron state 𝜀 and electron orbital n, 𝛼𝜀n (𝐫, 𝐫) = 𝛾𝜀n (𝐫)𝜓𝜀 (𝐫)𝜑n (𝐫) (cf. Eq. (9) in Ref. [34]),
which can be used to determine the position dependent EF 𝛾𝜀n (𝐫), see Sect. 4.
Enhancement Factors for Positron Annihilation on Valence . . . 249

(a) (b) (c) (d)

ε ν ε ε ν1 ν2 ε ε ν ε ε ν ε
Γ + μ1
μ + μ1 μ2 μ1 μ2
μ2
n1
n n n1 n2 n2

(e) (f) (g)

εε ν εε ε ν ε ε ν1 ν2 ε
μ1
+ n1 + μ2 + μ
μ2 n1
μ1
n2 n1 n2
n2

Fig. 2 Main contributions to the positron self-energy matrix ⟨𝜀′ |𝛴̂ E |𝜀⟩. The lowest, second-order
diagram a describes the effect of polarization; diagram b accounts for virtual Ps formation repre-
sented by the 𝛤 -block. Diagrams c–g represent leading third-order corrections not included in (b).
Top lines in the diagrams describe the positron. Other lines with the arrows to the right are excited
electron states, and to the left, holes, i.e., electron states occupied in the target ground state. Wavy
lines represent Coulomb interactions. Summation over all intermediate states is assumed

( )
Ĥ 0 + 𝛴̂ 𝜀 𝜓𝜀 (𝐫) = 𝜀𝜓𝜀 (𝐫). (8)

Here Ĥ 0 is the Hamiltonian of the positron in the static field of the N-electron atom
in the ground state (described at the HF level). The nonlocal, energy-dependent cor-
relation potential 𝛴̂ 𝜀 is equal to the self-energy of the positron Green’s function [54],
and acts as an integral operator 𝛴̂ 𝜀 𝜓𝜀 (𝐫) = ∫ 𝛴𝜀 (𝐫, 𝐫 ′ )𝜓𝜀 (𝐫 ′ )d3 𝐫 ′ .
The main contributions to 𝛴̂ 𝜀 are shown in Fig. 2. At large positron-atom dis-
tances the correlation potential reduces to the local polarization potential 𝛴𝜀 (𝐫, 𝐫 ′ ) ≃
−𝛼d 𝛿(𝐫 − 𝐫 ′ )∕2r4 , where 𝛼d is the dipole polarizability of the atom. If only diagram
Fig. 2a is included, the polarizability is given by the HF approximation

2 ∑ |⟨𝜇|𝐫|n⟩|2
𝛼d = . (9)
3 n,𝜇 𝜀𝜇 − 𝜀n

Diagrams Fig. 2c–f are third-order corrections to the polarization diagram (a) of the
type described by the random-phase approximation with exchange [55]. Including
these gives asymptotic behaviour of 𝛴𝜀 (𝐫, 𝐫 ′ ) with a more accurate value of 𝛼d . The
diagram Fig. 2b describes the virtual Ps-formation contribution. Adding it to diagram
Fig. 2a nearly doubles the strength of the correlation potential in heavier noble-gas
atoms (Ar, Kr and Xe). The diagram Fig. 2g describes the positron-hole repulsion.
Including the diagrams of Fig. 2 in the positron-atom correlation potential provides
accurate scattering phaseshifts and cross sections for all noble-gas atoms [19].
250 D. G. Green and G. F. Gribakin

The positron self-energy diagrams and the annihilation amplitude contain sums
over the intermediate excited electron and positron states. In practice we calculate
them numerically using sets of electron and positron basis states constructed using
40 B-splines of order 6, in a spherical box of radius 30 a.u. We use an expoten-
tial knot sequence for the B-splines, which provides for an efficient spanning of the
electron and positron continua in the sums over intermediate states [48]. The max-
imum angular momentum of the intermediate states is lmax =15, and we extrapolate
to lmax → ∞ as in [19, 21, 48, 50].

3 Annihilation Momentum Densities for Valence and Core

Electron Orbitals in Noble Gases

Figures 3 and 4 show the AMD |An𝜀 (𝐏)|2 [spherically averaged, as in Eq. (2)] for
thermal (k = 0.04 a.u.) s-wave positrons annihilating on individual core and valence
subshells of the noble gas atoms, calculated using different approximations for the
annihilation amplitude and positron wavefunction. The range of two-𝛾 momenta P =
0–6 a.u. corresponds to the maximum Doppler energy shift 𝜖 ≈ 11 keV.
The simplest approximation shown uses the zeroth-order (IPA) annihilation ampli-
tude (6) with positron wavefunctions in the static field of the HF atom. Better approx-
imations involve using the full annihilation vertex of Fig. 1 [Eq. (7)], or the best
(Dyson) positron wavefunction, or both. In general, including correlations of either
types increases the AMD and the annihilation probability.
General trends are observed throughout the noble-gas sequence. The AMD are
broader for the core orbitals for which the typical electron momenta are greater,
leading to greater Doppler shifts. The core AMD (and the core annihilation proba-
bilities [24, 50]) also have noticeably smaller magnitudes compared with those of the
valence electrons. For all electron orbitals whose radial wavefunctions have nodes
(e.g., 2s in Ne, 2s, 3s and 3p in Ar, etc.) the AMD display deep minima related to the
nodes of the annihilation amplitude An𝜀 (𝐏). Their number and positions are related
to the number and positions of the nodes in the orbital’s radial wavefunction (i.e., the
radial nodes that occur closer to the nucleus result in the nodes of An𝜀 (𝐏) at higher
momenta). This behaviour is easy to understand from the zeroth-order amplitude (6),
which is the Fourier transform of the product of the electron and positron wavefunc-
tions. For low positron energies, its wavefunction inside the atom decreases mono-
tonically towards the nucleus (suppressed by the repulsive electrostatic potential at
smaller distances), and has no nodes. Hence, the nodal structure of the annihilation
amplitude is determined by the behaviour of the electron wavefunction. Inclusion of
the correlation corrections to the annihilation vertex, as described by Eq. (7), leads
only to a small shift in the positions of the nodes.
For a given approximation for the annihilation vertex, the AMD calculated using
the positron Dyson orbitals (red curves) are larger than those calculated using the
HF positron wavefunction (blue curves). The corresponding increase is nearly the
Enhancement Factors for Positron Annihilation on Valence . . . 251

2 2
10 10
1
1 10
10
He 1s 10
0 Ar 3p
0
10 -1
10

2
-1 -2
2

|A(P)|
10 10
|A(P)|

-3
-2 10
10
-4
-3 10
10 -5
10
-4
10 -6
10
-5 -7
10 10
1 2
10 10
1
0 10
10 Ne 2p 0 Ar 3s
10
-1 -1
10 10
2

2
-2
|A(P)|

|A(P)|
-2 10
10 -3
10
-3 -4
10 10
-5
-4 10
10 -6
10
-5 -7
10 10
1 2
10 10
1
0 10
10 Ne 2s 0 Ar 2p
10
-1 -1
10 10
2

-2
|A(P)|

|A(P)|

-2 10
10 -3
10
-3 -4
10 10
-5
-4 10
10 -6
10
-5 -7
10 10
1 2
10 10
1
0 10
10 Ne 1s 0 Ar 2s
10
-1 -1
10 10
2

-2
|A(P)|

|A(P)|

-2 10
10 -3
10
-3 -4
10 10
-5
-4 10
10 -6
10
-5 -7
10 10
0 1 2 3 4 5 6 0 1 2 3 4 5 6
P (a.u.) P (a.u.)

Fig. 3 Annihilation momentum density as a function of the total 2𝛾 momentum P, for s-wave
positron annihilation in He, Ne and Ar, calculated in different approximations for the annihilation
vertex: zeroth-order vertex (dashed lines); full vertex (0 + 1 + 𝛤 ) (solid lines), for HF (thin blue
lines) and Dyson (thick red lines) positron of momentum k = 0.04 a.u. (For a given approximation
for the vertex, the lines for the calculation with Dyson positron lie above those for HF positron)
252 D. G. Green and G. F. Gribakin

3 4
10 10
3
2 10
10 2
1 Kr 4p 10 Xe 5p
10 1
0 10
10 0
10

2
2

-1

|A(P)|
|A(P)|
10 -1
-2 10
10 -2
-3
10
10 -3
10
-4 -4
10 10
-5 -5
10 10
-6 -6
10 10
3 4
10 10
3
2 10
10 2
1 Kr 4s 10 Xe 5s
10 1
0 10
10 0
10

2
2

-1

|A(P)|
|A(P)|

10 -1
-2 10
10 -2
-3
10
10 -3
10
-4 -4
10 10
-5 -5
10 10
-6 -6
10 10
3 4
10 10
3
2 10
10 2
1 Kr 3d 10 Xe 4d
10 1
0 10
10 0
10
2
2

-1
|A(P)|
|A(P)|

10 -1
-2 10
10 -2
-3
10
10 -3
10
-4 -4
10 10
-5 -5
10 10
-6 -6
10 10
3 4
10 10
3
2 10
10 2
1 Kr 3p 10 Xe 4p
10 1
0 10
10 0
10
2
2

-1
|A(P)|
|A(P)|

10 -1
-2 10
10 -2
-3
10
10 -3
10
-4 -4
10 10
-5 -5
10 10
-6 -6
10 10
3 4
10 10
3
10
2 10
2
1 Kr 3s 10 Xe 4s
10 1
0 10
10 0
10
2
2

-1
|A(P)|
|A(P)|

10 -1
-2 10
10 10
-2
-3 -3
10 10
-4 -4
10 10
-5 -5
10 10
-6 -6
10 0 1 2 3 4 5 6
10 0 1 2 3 4 5 6
P (a.u.) P (a.u.)

Fig. 4 Annihilation momentum densities as a function of the total 2𝛾 momentum P, for s-wave
positron annihilation in Kr and Xe. Various lines as described in Fig. 3
Enhancement Factors for Positron Annihilation on Valence . . . 253

same for the valence and core orbitals in a given atom. It ranges from a factor of
∼2 in Helium to ∼100 in Xe, and is only weakly dependent on P. (Note that the
AMD are plotted on the logarithmic scale.) This increase is due to the action of the
attractive correlation potential 𝛴̂ 𝜀 on the positron (Dyson) wavefunction. It leads to
a build-up of the positron density in the vicinity of the atom and greater overlap with
the atomic electron density. This effect is stronger for the heavier, more polarizable
atoms, leading to much greater Zef f values. In Ar, Kr and Xe the attractive positron-
atom potential supports low-lying virtual s levels, leading to a characteristic resonant
growth of the annihilation rates at low positron energies [19, 44, 45, 56].
When the vertex corrections are included in the annihilation amplitude (solid
curves), the AMD is enhanced above the zeroth-order (IPA) result (dashed curves).
This is due to the Coulomb attraction within the annihilating pair, which increases
the probability of finding the electron and positron at the same point in space. The
size of the enhancement is similar for the HF and Dyson positron wavefunctions. At
the same time, the enhancement is much greater for the valence electrons than for the
core electrons, as the former are more easily perturbed by the positron’s Coulomb
field. For the core electrons, the vertex correction is dominated by the first-order
diagram Fig. 1b (similar to the case of hydrogen-like ions [21]). For the valence
electrons, the nonperturbative 𝛤 -block contribution Fig. 1c is also very important.
From Figs. 3 and 4 one can also see that the vertex enhancement is significantly
stronger at low momenta P of the electron-positron pair (which leads to narrowing
of the 𝛾-ray spectra in comparison with those obtained with the zeroth-order ampli-
tude [20, 24]). This can be seen most clearly in the AMD of the valence electrons.
Their high-P content is due to positron annihilation with the electron when the latter
is closer to the nucleus, and where its local velocity is higher, making it less suscep-
tible to the positron’s attraction. (Note that for large P the calculated 𝛤 -block vertex
corrections contain numerical errors which manifest themselves as extra oscillations
visible in AMD for valence electrons. This, however, has a negligible effect on the
annihilation spectra, since the AMD for the valence electrons at such momenta are
very small.) The vertex enhancement is considered in more detail below.
We conclude this section by noting that calculations of the corresponding 𝛾 spec-
tra were reported in [24, 50]. They showed excellent agreement with the measured
spectra for Ar, Kr and Xe [3], and firmly established the fraction of core annihilation
for these atoms.

4 Vertex Enhancement Factors

The enhancement of the AMD and annihilation rates due to the correlation cor-
rections to the vertex with respect to those obtained using the zeroth-order (IPA)
approximation [cf. Eqs. (7) and (6)] can be parameterized through so-called vertex
enhancement factors. The MBT enables a direct ab initio calculation of ‘exact’ EF:
one simply compares the results obtained using the annihilation amplitude calculated
in different approximations, as described here.
254 D. G. Green and G. F. Gribakin

The EF were introduced originally to correct the IPA annihilation rates for
positrons in condensed matter (see, e.g., [31] and references therein),

𝜆 = 𝜋r02 c n− (𝐫)n+ (𝐫)𝛾(𝐫)d3 𝐫, (10)

∫

where n− (𝐫) and n+ (𝐫) are the electron and positron densities, respectively, and
𝛾(𝐫) is the EF. The latter is typically computed for a uniform electron gas (e.g.,
using MBT [32, 57]) and parameterized in terms of the electron density, e.g., as
𝛾 = 1 + 1.23rs − 0.0742rs2 + 16 rs3 , where rs = (3∕4𝜋n− )1∕3 [58] (see also [30, 59]).
An approximation commonly used to account for the vertex enhancement of the IPA
annihilation amplitude (6) is [5, 60]
√
An𝜀 (𝐏) = e−i𝐏⋅𝐫 𝜓𝜀 (𝐫)𝜓n (𝐫) 𝛾(𝐫)d3 𝐫. (11)
∫

However, this method is known to give spurious effects in the high-momentum

regions of the 𝛾 spectra [5].
The general MBT expression for the annihilation amplitude in a finite rather than
infinite and homogeneous system, Eq. (7), shows that the correlation contribution to
the vertex is nonlocal, i.e., it involves the positron and electron wavefunctions 𝜓𝜀 (𝐫1 )
and 𝜓n (𝐫2 ) at different points in space. The corresponding enhancement is described
by the three-point function 𝛥̃𝜀 (𝐫; 𝐫1 , 𝐫2 ). This allows one to formally define the EF
for the electron in orbital n and positron of energy 𝜀 by

√ ∬ 𝛥̃𝜀 (𝐫; 𝐫1 𝐫2 )𝜓𝜀 (𝐫1 )𝜑n (𝐫2 )d3 𝐫1 d3 𝐫2

𝛾n𝜀 (𝐫) ≡ 1 + . (12)
𝜓𝜀 (𝐫)𝜑n (𝐫)

However, the presence of nodes in the wavefunctions in the denominator renders this
quantity of limited use and we must opt for a more pragmatic approach.
It is clear from Figs. 3 and 4 that the vertex enhancement of the AMD |An𝜀 (𝐏)|2 ,
i.e., full-vertex results compared with zeroth-order, has a weak dependence on the
momentum P (except near the nodes of the amplitude). This momentum dependence
of the vertex enhancement has little effect on the annihilation 𝛾 spectra for the noble-
gas atoms, especially for the core orbitals [50]. It is thus instructive to define a two-𝛾
momentum-averaged vertex EF as the ratio of the full-vertex partial annihilation rate
to that calculated using the zeroth-order (IPA) vertex:

Zef(0+1+𝛤
f ,nl
)
(k)
𝛾̄nl (k) = , (13)
Zef(0)f ,nl (k)

where the superscript denotes the vertex order (see Fig. 1) and nl labels the subshell
of the electron that the positron of momentum k annihilates with. Analogous EF are
commonly used to analyse and predict the annihilation rates and 𝛾 spectra in solids
Enhancement Factors for Positron Annihilation on Valence . . . 255

(see, e.g., [5, 61–66]). The true spectrum for annihilation on a given subshell for a
given positron momentum can then be approximated by

wnl (𝜖) ≈ 𝛾̄nl w(0)

nl
(𝜖) (14)

where w(0) nl
is the 𝛾 spectrum calculated using the zeroth-order vertex. Accurate
reconstruction of the true spectra for s-wave thermal positrons using Eq. (14) has
been demonstrated for noble-gas atoms in [50].
In a recent paper [24] we showed that for thermal s-wave positrons (k = 0.04 a.u.)
𝛾̄nl follows a near-universal scaling with the orbital ionization energy Inl ,
√
𝛾̄nl = 1 + A∕Inl + (B∕Inl )𝛽 , (15)

where A, B and 𝛽 are constants.6 The second term on the RHS of Eq. (15) describes
the effect of the first-order correction, Fig. 1b. Its scaling with Inl was motivated
by the 1∕Z scaling for positron annihilation in hydrogen-like ions [21]. The third
term is phenomenological and describes the effect of the 𝛤 -block correction that is
particularly important for the valence subshells.
Here we extend the calculations of the enhancement factors to s-, p- and d-wave
positrons with momenta up to the positronium-formation threshold. At small, e.g.,
room-temperature, thermal positron momenta k ∼ 0.04 a.u., the contributions of the
positron p and d waves to the annihilation rates are very small, owing to Zef f (k) ∝ k2𝓁
low-energy behaviour. (This is a manifestation of the suppression of the positron
wavefunction in the vicinity of the atom by the centrifugal potential 𝓁(𝓁 + 1)∕2r2 .)
However, for higher momenta close to the Ps-formation threshold, the s-, p- and d-
wave contributions to the annihilation rates become of comparable magnitude (see
Fig. 16 and Tables III–VII in Ref. [19]).
Figures 5–9 show the enhancement factors for positron annihilation with elec-
trons in the valence (np and ns) and core [(n − 1)s, (n − 1)p, (n − 1)d, as applicable]
orbitals of He, Ne, Ar, Kr, and Xe, as functions of the positron momentum.
The EF for positron annihilation with 1s electrons in He (Fig. 5) are 2.6–3.0 for
the s-wave, 3.8–4.1 for the p-wave, and 5.2–5.9 for the d-wave. They show only
a weak dependence on the positron momentum, which is a typical feature of all
the data. There is also little difference between the EF obtained with the static-
field (HF) positron wavefunctions (dashed lines) and those found using the positron
Dyson orbitals (solid lines). This is in spite of the fact that the use of the correlated
Dyson positron wavefunctions increases the AMD (and the annihilation rates [19])
by almost an order of magnitude for s-wave positrons (Fig. 3).
The weak dependence of the EF on the positron energy and the type of positron
wavefunction used is related to the nature of the vertex enhancement. The inter-
mediate electron and positron states in diagrams Fig. 1b and c that describe the

6
For HF positron wavefunctions the values of the parameters are A = 1.54 a.u. = 42.0 eV, B =
0.92 a.u. = 24.9 eV, and 𝛽 = 2.54. For Dyson positron wavefunctions the values are A = 1.31 a.u. =
35.7 eV, B = 0.83 a.u. = 22.7 eV, and 𝛽 = 2.15 [24].
256 D. G. Green and G. F. Gribakin

6
s-wave p-wave
5

4 d-wave
γ

0
0 0.2 0.4 0.6 0.8 1 1.2 0 0.2 0.4 0.6 0.8 1 1.2 0 0.2 0.4 0.6 0.8 1 1.2
k (a.u.) k (a.u.) k (a.u.)

Fig. 5 Enhancement factors for s-, p- and d-wave positrons annihilating on the 1s electrons in He,
obtained with HF (dashed lines) and Dyson (solid lines) positron wavefunctions

short-range vertex enhancement (𝜈, 𝜇, 𝜈1 , 𝜇1 , etc.) are highly virtual, i.e., have rel-
atively large energies. For example, the energy denominator of diagram Fig. 1b is
𝜀 − 𝜀𝜈 − 𝜀𝜇 + 𝜀n (see Refs. [20, 50]). Estimating the typical electron and positron
energies as 𝜀𝜈,𝜇 ∼ |𝜀n | (the ionization energy of electron orbital n), we see that for
few-electronvolt positrons, the positron energy 𝜀 can be neglected. For the same rea-
son, the vertex correction function 𝛥̃𝜀 (𝐫; 𝐫1 , 𝐫2 ) is only weakly nonlocal, i.e., it is large
only for |𝐫1 − 𝐫2 | ≪ |𝐫1,2 | ∼ |𝐫| (see the “annihilation maps” in Figs. 4.14–4.16 of
Ref. [67]). The situation becomes different at large momenta close to the Ps forma-
tion threshold. Here the p- and d-wave EF show an upturn related to the virtual Ps
formation becoming “more real” [68]. This is also seen in 𝛾̄np for heavier atoms.
The increase of the EF with the positron orbital angular momentum 𝓁 seen in
Fig. 5 can be related to the behaviour of the low-energy positron wavefunctions near
the atom. Due to the action of the centrifugal potential, the p- and d-wave radial
wavefunctions are suppressed as (kr)𝓁 with 𝓁 = 1 and 2, compared with the s wave.
The nonlocal correlation corrections Fig. 1b and c “help” the positron to pull the
atomic electron towards larger distances, which has a greater advantage for the higher
partial waves.
It is interesting to compare the values of 𝛾̄1s for He with the EF for positron anni-
hilation with atomic hydrogen: 6–7, 10–12, and 15–17, for the s-, p-, and d-wave
positrons, respectively, with k ≤ 0.4 a.u. (see Fig. 13 in Ref. [48]). The greater val-
ues of the EF for hydrogen are related to the smaller binding energy of the 1s electron
in hydrogen (13.6 eV) compared with that in He (24.6 eV). The vertex corrections
are generally greater for the more weakly bound electrons that have more diffuse
orbitals and are more easily perturbed by the positron’s Coulomb interaction. The
same trend will be seen throughout the noble-gas-atom sequence, with more strongly
bound electron orbitals, in particular those in the core, displaying smaller EF [cf.
Eq. (15)].
Enhancement Factors for Positron Annihilation on Valence . . . 257

5.00
s-wave p-wave
4.00 2p
d-wave
2p
3.00
2p
2s
γ

2s
2.00 2s

1s 1s 1s
1.00

0.00
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
k (a.u.) k (a.u.) k (a.u.)

Fig. 6 Enhancement factors for s-, p- and d-wave positrons annihilating on the 1s, 2s and 2p sub-
shells in Ne, obtained with HF (dashed lines) and Dyson (solid lines) positron wavefunctions.

Turning to Ne (Fig. 6), we observe that the EF for the outer valence 2p subshell
are slightly smaller than those for 1s in He, in spite of the binding energy of the 2p
electrons (21.6 eV) being lower than that of He 1s. Ne also has the broadest 𝛾 ray
spectrum of all the noble gases (see AMD in Fig. 3, and the data for the calculated
and measured spectra [50, 69]). The latter indicates that the 2p electrons in Ne have
large typical momenta, which makes the correlation correction to the annihilation
vertex relatively small. The EF for the inner valence 2s subshell is around 2, while
for the deeply bound 1s electrons, 𝛾̄1s ≈ 1.2. We also note that for the core orbitals,
the values of the EF for the positron s, p and d waves are quite close. This is in fact
a general trend observed for all atoms that the relative difference between the values
of 𝛾̄nl − 1 for the positron s, p and d waves is becoming small with the increase in the
binding energy. The smaller effect of the orbital angular momentum of the positron
on the EF for core orbitals is due to the vertex correction becoming “more local”,
and hence, less sensitive to the variation of the positron radial wavefunction.
The EF in Ar, Kr and Xe (Figs. 7, 8 and 9) become progressively larger, for
both the valence and core electrons. For example, the vertex EF for s-wave positron
annihilation with the outer valence np electrons increases from 𝛾̄3p = 5.2 (Ar), to
𝛾̄4p = 6.6 (Kr), to 𝛾̄5p = 9.2 (Xe) (for the HF positron wavefunction at low momenta
k ≲ 0.1 a.u.). The EF for the (n − 1)l core orbitals also increase to 𝛾̄(n−1)l ∼ 1.5–2,
with the values for the 3d and 4d orbitals being noticeably larger than those of the
3s/3p and 4s/4p orbitals, for Kr and Xe, respectively.
Another feature of the data is the growing difference between the EF for the np
electrons obtained with the Dyson positron wavefunction (solid lines) and those
found using the static-field (HF) positron wavefunction (dashed lines). This effect
is related to the increase in the strength of the positron-atom correlation potential
𝛴̂ 𝜀 for the heavier noble-gas atoms [19, 44, 45]. For s-wave positrons it results in
the creation of positron-atom virtual levels [47] whose energies 𝜀 = 𝜅 2 ∕2 become
lower for heavier atoms, with values of 𝜅 = −0.23, −0.10 and −0.012 a.u. for Ar,
258 D. G. Green and G. F. Gribakin

10.00
9.00 s-wave p-wave d-wave
8.00 3p
7.00
3p
6.00
5.00 3p
γ

4.00
3s 3s
3.00 3s
2.00 2s, 2p 2s, 2p 2s,2p
1.00
0.00
0 0.2 0.4 0.6 0.8 0 0.2 0.4 0.6 0.8 0 0.2 0.4 0.6 0.8
k (a.u.) k (a.u.) k (a.u.)

Fig. 7 Enhancement factors for s-, p- and d-wave positrons annihilating on the 2s, 2p, 3s and 3p
subshells in Ar, obtained with HF (dashed lines) and Dyson (solid lines) positron wavefunctions

12.00

10.00 s-wave p-wave d-wave

4p

8.00

4p
6.00
γ

4p

4.00 4s
4s 4s

2.00 3d 3d 3d

3s, 3p 3s, 3p 3s, 3p

0.00
0 0.2 0.4 0.6 0 0.2 0.4 0.6 0 0.2 0.4 0.6
k (a.u.) k (a.u.) k (a.u.)

Fig. 8 Enhancement factors for s-, p- and d-wave positrons annihilating on the 3s, 3p, 3d, 4s and
4p subshells in Kr, obtained with HF (dashed lines) and Dyson (solid lines) positron wavefunctions

Kr and Xe, respectively [19]. This is accompanied by a rapid growth of the positron
wavefunction near the atom, with the Dyson orbitals being enhanced by a factor
∼1∕|𝜅| compared to the static-field positron wavefunctions at low energies. Hence,
the inclusion of the correlation potential makes the radial dependence of the positron
wavefunction more vigorous. This is evidenced by some broadening of the 𝛾 spectra
obtained with the Dyson rather than the HF positron wavefunction [50]. This also
Enhancement Factors for Positron Annihilation on Valence . . . 259

18

16
s-wave p-wave d-wave
14

12 5p

10
γ

5p
8
5p
6

5s
4 5s 5s

4d 4d 4d
2
4s, 4p 4s, 4p 4s, 4p
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0 0.1 0.2 0.3 0.4 0.5 0 0.1 0.2 0.3 0.4 0.5
k (a.u.) k (a.u.) k (a.u.)

Fig. 9 Enhancement factors for s-, p- and d-wave positrons annihilating on the 4s, 4p, 4d, 5s and
5p subshells in Xe, obtained with HF (dashed lines) and Dyson (solid lines) positron wavefunctions

results in a reduction of the EF, which is most noticeable for the valence electrons,
and is largest in Xe, which has the strongest correlation potential for the positron.
Besides the rise in the valence EF for high positron momenta (related to the prox-
imity of the Ps-formation threshold), one other exception from the weak momentum-
dependence of the EF is seen at low momenta for d-wave positron annihilating on the
3d and 4d orbitals in Kr and Xe. The enhancement factors in this case are approxi-
mately constant from the Ps-formation threshold down to ∼0.3 a.u., but then deviate
at lower momenta, especially in Xe. Both the zeroth-order and full-vertex Zef f values
for these orbitals are calculated to be smooth functions of k. However, they obey the
∼k4 behaviour and become very small at low k (e.g., for Xe, using the Dyson wave-
function we find Zef f,4d ∼ 10−3 at k ∼ 0.3 a.u., decreasing to ∼10−7 for k ∼ 0.03 a.u.
It appears that for such small k numerical inaccuracies arise in the calculation of the
𝛤 -block contribution, leading to errors when evaluating the ratio in Eq. (13).

5 Conclusions

We used many-body theory methods to calculate the annihilation momentum den-

sities and vertex enhancement factors for s-, p- and d-wave positrons annihilat-
ing on valence and core electrons in noble-gas atoms. The general trends of the
EF is their weak dependence on the positron momentum and decrease with the
260 D. G. Green and G. F. Gribakin

increasing electron binding energy. We find that the type of the positron wavefunc-
tion used, i.e., Dyson orbital which accounts for the positron-atom correlation attrac-
tion vs repulsive static-field (HF) wavefunction, has relatively little effect on the EF,
except for the valence orbitals in most polarisable targets. We also find a relatively
weak dependence of the EF for the core and inner-valence electrons on the positron’s
orbital angular momentum.
The weak momentum-dependence of the EF obtained in positron-atom calcula-
tions suggests that they can be used to improve the calculations of positron anni-
hilation in more complex environments. One such system is positronium colliding
with noble-gas atoms, where calculations of Ps-atom pick-off annihilation rates that
neglect the short-range vertex enhancement strongly underestimate the measured
rates [70]. Another context where similar EF can be used is positron annihilation in
molecules. Here there is a sharp contrast between the large amount of experimen-
tal information, including 𝛾-spectra, for a wide range of molecule [69] and paucity
of credible theoretical data [23, 71]. The positron-molecule problem is particularly
interesting because the Zeff values for most polyatomic molecules show orders-of-
magnitude increases due to resonant positron annihilation [71]. In such molecules
the positron annihilates from a temporarily formed weakly-bound state. Attempts to
calculate such states using standard quantum-chemistry methods have been numer-
ous but not very successful [71] (i.e., there is only a small number of systems where
theory and experiment can be compared, and the agreement is mostly qualitative
[72]).
The calculations presented in this paper could be extended to other atoms, includ-
ing those with open shells. Partial filling of electron shells can be taken into account
in the many-body-theory sums using fractional occupation numbers (cf. Ref. [73]),
and the positron wavefunction is insensitive to such details of the electronic
structure at the static (HF) level. (At the level of Dyson orbitals, one will need
to calculate the positron self-energy using fractional electron-shell occupancies.)
Calculations are particularly straightforward for annihilation with core electrons, as
their enhancement factor is described well by the first-order correction to the anni-
hilation vertex.

Acknowledgements DGG is supported by a United Kingdom Engineering and Physical Sciences

Research Council Fellowship, grant number EP/N007948/1.

References

1. Asoka-Kumar P, Alatalo M, Ghosh V, Kruseman A, Nielsen B, Lynn K (1996) Phys Rev Lett
77:2097. https://doi.org/10.1103/PhysRevLett.77.2097
2. Lynn KG, MacDonald JR, Boie RA, Feldman LC, Gabbe JD, Robbins MF, Bonderup E,
Golovchenko J (1977) Phys Rev Lett 38:241. https://doi.org/10.1103/PhysRevLett.38.241
3. Iwata K, Gribakin GF, Greaves RG, Surko CM (1997) Phys Rev Lett 79:39. https://doi.org/10.
1103/PhysRevLett.79.39
Enhancement Factors for Positron Annihilation on Valence . . . 261

4. Lynn K, Dickman J, Brown W, Robbins M, Bonderup E (1979) Phys Rev B 20:3566. https://
doi.org/10.1103/PhysRevB.20.3566
5. Alatalo M, Barbiellini B, Hakala M, Kauppinen H, Korhonen T, Puska M, Saarinen K, Hauto-
järvi P, Nieminen R (1996) Phys Rev B 54:2397. https://doi.org/10.1103/PhysRevB.54.2397
6. Tuomisto F, Makkonen I (2013) Rev Mod Phys 85:1583. https://doi.org/10.1103/RevModPhys.
85.1583
7. Weiss A, Mayer R, Jibaly M, Lei C, Mehl D, Lynn KG (1988) Phys Rev Lett 61:2245. https://
doi.org/10.1103/PhysRevLett.61.2245
8. Ohdaira T, Suzuki R, Mikado T, Ohgaki H, Chiwaki M, Yamazaki T (1997) Appl Surf Sci
116:177. https://doi.org/10.1016/S0169-4332(96)01049-5
9. Weiss AH, Fazleev NG, Nadesalingam MP, Mukherjee S, Xie S, Zhu J, Davis BR (2007) Radiat
Phys Chem 76:285. https://doi.org/10.1016/j.radphyschem.2006.03.053
10. Mayer J, Hugenschmidt C, Schreckenbach K (2010) Surf Sci 604:1772. https://doi.org/10.
1016/j.susc.2010.07.003
11. Hugenschmidt C (2016) Surf Sci Rep 71:547. https://doi.org/10.1016/j.surfrep.2016.09.002
12. Hugenschmidt C, Lwe B, Mayer J, Piochacz C, Pikart P, Repper R, Stadlbauer M, Schreck-
enbach K (2008) Nucl Instrum Methods A 593:616. https://doi.org/10.1016/j.nima.2008.05.
038
13. Mayer J, Hugenschmidt C, Schreckenbach K (2010) Phys Rev Lett 105:207401. https://doi.
org/10.1103/PhysRevLett.105.207401
14. Stoll H, Koch KMM, Major J (1991) Nucl Instrum Methods B 582:56
15. Coleman P (ed) (2000) Positron beams and their applications. World Scientific
16. Sano Y, Kino Y, Oka T, Sekine T (2015) J Phys Conf Ser 618:012010. http://stacks.iop.org/
1742-6596/618/i=1/a=012010
17. Jensen KO, Weiss A (1990) Phys Rev B 41:3928. https://doi.org/10.1103/PhysRevB.41.3928
18. Iwata K, Greaves RG, Murphy TJ, Tinkle MD, Surko CM (1995) Phys Rev A 51:473. https://
doi.org/10.1103/PhysRevA.51.473
19. Green DG, Ludlow JA, Gribakin GF (2014) Phys Rev A 90:032712. https://doi.org/10.1103/
PhysRevA.90.032712
20. Dunlop LJM, Gribakin GF (2006) J Phys B 39:1647. https://doi.org/10.1088/0953-4075/39/7/
008
21. Green DG, Gribakin GF (2013) Phys Rev A 88:032708. https://doi.org/10.1103/PhysRevA.88.
032708
22. Green DG, Saha S, Wang F, Gribakin GF, Surko CM (2010) Mater Sci Forum 666:21. https://
doi.org/10.4028/www.scientific.net/MSF.666.21
23. Green DG, Saha S, Wang F, Gribakin GF, Surko CM (2012) New J Phys 14:035021. http://
stacks.iop.org/1367-2630/14/i=3/a=035021
24. Green DG, Gribakin GF (2015) Phys Rev Lett 114:093201. https://doi.org/10.1103/
PhysRevLett.114.093201
25. Bonderup E, Andersen JU, Lowy DN (1979) Phys Rev B 20:883. https://doi.org/10.1103/
PhysRevB.20.883
26. Flambaum VV (2009) Phys Rev A 79:042505. https://doi.org/10.1103/PhysRevA.79.042505
27. Kozlov MG, Flambaum VV (2013) Phys Rev A 87:042511. https://doi.org/10.1103/PhysRevA.
87.042511
28. Kahana S (1963) Phys Rev 129:1622. https://doi.org/10.1103/PhysRev.129.1622
29. Carbotte JP (1967) Phys Rev 155:197. https://doi.org/10.1103/PhysRev.155.197
30. Boroński E, Nieminen R (1986) Phys Rev B 34:3820. https://doi.org/10.1103/PhysRevB.34.
3820
31. Puska MJ, Nieminen RM (1994) Rev Mod Phys 66:841. https://doi.org/10.1103/RevModPhys.
66.841
32. Arponen J, Pajanne E (1979) Ann Phys 121:343. https://doi.org/10.1016/0003-
4916(79)90101-5
33. Mitroy J, Barbiellini B (2002) Phys Rev B 65:235103. https://doi.org/10.1103/PhysRevB.65.
235103
262 D. G. Green and G. F. Gribakin

34. Makkonen I, Ervasti MM, Siro T, Harju A (2014) Phys Rev B 89:041105. https://doi.org/10.
1103/PhysRevB.89.041105
35. Daniuk S, Kontrym-Sznajd G, Rubaszek A, Stachowiak H, Mayers J, Walters PA, West RN
(1987) J Phys F: Metal Phys 17:1365. https://doi.org/10.1088/0305-4608/17/6/011
36. Jarlborg T, Singh AK (1987) Phys Rev B 36:4660. https://doi.org/10.1103/PhysRevB.36.4660
37. Rubaszek A , Stachowiak H (1984) Physica Status Solidi (B) 124:159. https://doi.org/10.1002/
pssb.2221240117
38. Zubiaga A, Ervasti MM, Makkonen I, Harju A, Tuomisto F, Puska MJ (2016) J Phys B: At
Mol Opt Phys 49:064005
39. Swann AR, Green DG, Gribakin GF arXiv: 1709.00394
40. Berestetskii VB, Lifshitz EM, Pitaevskii LP (1982) Quantum electrodynamics, 2nd edn. Perg-
amon, Oxford
41. Kaijser P, Smith Jr VH (1977) Adv Quantum Chem 10:37. https://doi.org/10.1016/S0065-
3276(08)60578-X
42. Fraser PA (1968) Adv At Mol Phys 4:63
43. Pomeranchuk I, Eksp Zh (1949) Teor Fiz 19:183
44. Dzuba VA, Flambaum VV, King WA, Miller BN, Sushkov OP (1993) Phys Scripta T46:248.
https://doi.org/10.1088/0031-8949/1993/T46/039
45. Dzuba VA, Flambaum VV, Gribakin GF, King WA (1996) J Phys B 29:3151. https://doi.org/
10.1088/0953-4075/29/14/024
46. Surko CM, Gribakin GF, Buckman SJ (2005) J Phys B 38:R57. https://doi.org/10.1088/0953-
4075/38/6/R01
47. Landau LD, Lifshitz EM (1977). Quantum mechanics (Non-relativistic theory). In: Course of
theoretical physics, 3rd edn, vol 3. Pergamon, Oxford
48. Gribakin GF, Ludlow J (2004) Phys Rev A 70:032720. https://doi.org/10.1103/PhysRevA.70.
032720
49. Green DG (2011) PhD thesis, Queen’s University Belfast
50. Green DG, Gribakin GF (2015) arXiv:1502.08045
51. Abrikosov AA, Gorkov LP, Dzyalonshinkski IE (1975) Methods of quantum field theory in
statistical physics. Dover, New York
52. Fetter AL, Walecka JD (2003) Quantum theory of many-particle systems. Dover, New York
53. Dickhoff WH, Neck DV (2008) Many body theory exposed!—Propagator description of quan-
tum mechanics in many-body systems, 2nd edn. World Scientific, Singapore
54. Bell JS, Squires EJ (1959) Phys Rev Lett 3:96. https://doi.org/10.1103/PhysRevLett.3.96
55. Amusia MY, Cherepkov NA (1975) Case studies in atomic physics 5:47
56. Goldanski VI, Sayasov YS (1968) Phys Lett 13:300
57. Arponen J, Pajanne E (1979) J Phys F 9:2359. https://doi.org/10.1088/0305-4608/9/12/009
58. Barbiellini B, Puska MJ, Torsti T, Nieminen RM (1995) Phys Rev B 51:7341. https://doi.org/
10.1103/PhysRevB.51.7341
59. Stachowiak H, Lach J (1993) Phys Rev B 48:9828. https://doi.org/10.1103/PhysRevB.48.9828
60. Daniuk S, Kontrym-Sznajd G, Rubaszek A, Stachowiak H, Mayers J, Walters PA, West RN
(1987) J Phys F 17:1365. https://doi.org/10.1088/0305-4608/17/6/011
61. Jensen KO (1989) J Phys Condens Matter 1:10595. https://doi.org/10.1088/0953-8984/1/51/
027
62. Alatalo M, Kauppinen H, Saarinen K, Puska MJ, Mäkinen J, Hautojärvi P, Nieminen RM
(1995) Phys Rev B 51:4176. https://doi.org/10.1103/PhysRevB.51.4176
63. Barbiellini B, Puska MJ, Alatalo M, Hakala M, Harju A, Korhonen T, Siljamäki S, Torsti T,
Nieminen RM (1997) Appl Surf Sci 116:283. https://doi.org/10.1016/S0169-4332(96)01070-
7
64. Barbiellini B, Hakala M, Puska MJ, Nieminen RM, Manuel AA (1997) Phys Rev B 56:7136.
https://doi.org/10.1103/PhysRevB.56.7136
65. Makkonen I, Hakala M, Puska MJ (2006) Phys Rev B 73:035103. https://doi.org/10.1103/
PhysRevB.73.035103
Enhancement Factors for Positron Annihilation on Valence . . . 263

66. Kuriplach J, Morales AL, Dauwe C, Segers D, Šob M (1998) Phys Rev B 58:10475. https://
doi.org/10.1103/PhysRevB.58.10475
67. Ludlow J (2003) PhD thesis, Queen’s University Belfast
68. Gribakin GF, Ludlow J (2002) Phys Rev Lett 88:163202. https://doi.org/10.1103/PhysRevLett.
88.163202
69. Iwata K, Greaves RG, Surko CM (1997) Phys Rev A 55:3586. https://doi.org/10.1103/
PhysRevA.55.3586
70. Mitroy J, Ivanov IA (2001) Phys Rev A 65:012509. https://doi.org/10.1103/PhysRevA.65.
012509
71. Gribakin GF, Young JA, Surko CM (2010) Rev Mod Phys 82:2557. https://doi.org/10.1103/
RevModPhys.82.2557
72. Tachikawa M (2014) J Phys Conf Ser 488:012053. https://doi.org/10.1088/1742-6596/488/1/
012053
73. Dzuba VA, Kozlov A, Flambaum VV (2014) Phys Rev A 89:042507. https://doi.org/10.1103/
PhysRevA.89.042507
Geometric Phase and Interference Effects
in Ultracold Chemical Reactions

N. Balakrishnan and B. K. Kendrick

Abstract Electronically non-adiabatic effects play an important role in many chemi-

cal reactions and light induced processes. Non-adiabatic effects are important, when
there is an electronic degeneracy for certain nuclear geometries leading to a con-
ical intersection between two adiabatic Born-Oppenheimer electronic states. The
geometric phase effect arises from the sign change of the adiabatic electronic wave
function as it encircles the conical intersection between two electronic states (e.g.,
a ground state and an excited electronic state). This sign change requires a corre-
sponding sign change on the nuclear motion wave function to keep the overall wave
function single-valued. Its effect on bimolecular chemical reaction dynamics remains
a topic of active experimental and theoretical interrogations. However, most prior
studies have focused on high collision energies where many angular momentum par-
tial waves contribute and the effect vanishes under partial wave summation. Here,
we examine the geometric phase effect in cold and ultracold collisions where a sin-
gle partial wave, usually the s-wave, dominates. It is shown that unique properties of
ultracold collisions, including isotropic scattering and an effective quantization of the
scattering phase shift, lead to large geometric phase effects in state-to-state reaction
rate coefficients. Illustrative results are presented for the hydrogen exchange reaction
in the fundamental H+H2 system and its isotopic counterparts.

Keywords Geometric phase ⋅ Ultracold molecules ⋅ Ultracold chemistry

Ultracold collisions

N. Balakrishnan (✉)
Department of Chemistry, University of Nevada, Las Vegas, NV 89154, USA
e-mail: [email protected]
B. K. Kendrick
Los Alamos National Laboratory, Theoretical Division (T-1, MS B221),
Los Alamos, NM 87545, USA

© Springer International Publishing AG, part of Springer Nature 2018 265

Y. A. Wang et al. (eds.), Concepts, Methods and Applications of Quantum Systems
in Chemistry and Physics, Progress in Theoretical Chemistry and Physics 31,
https://doi.org/10.1007/978-3-319-74582-4_15
266 N. Balakrishnan and B. K. Kendrick

1 Introduction

The Born-Oppenheimer approximation is the basis of much of the development

in electronic structure theory, quantum dynamics of nuclear motion and molecular
spectroscopy. The approach exploits the vast difference in timescale for electronic
and nuclear motion due to the small electron/nuclei mass ratio. In this approach, the
Schrödinger equation for electronic motion is solved for various fixed nuclear config-
urations and the resulting electronic energy as a function of the nuclear degrees of
freedom is called the electronic potential energy surface (PES). The nuclei evolve
under the influence of this electronic PES (or PESs) and subsequent solution of
the nuclear Schrödinger equation yields energy levels for rotational, vibrational and
translational motion of the nuclei. This two-step procedure for quantum chemical
dynamics has been wildly successful for many elementary chemical reactions. How-
ever, when there is an electronic degeneracy for certain nuclear configurations, i.e.,
a conical intersection between two electronic PESs, this adiabatic solution of the
Schrödingier equation for nuclear motion breaks down and a fully non-adiabatic
treatment is desirable. Such non-adiabatic treatments which include the coupling
between the ground and excited electronic PESs are computationally challenging
and not practical for the vast majority of chemical reactions.
An important consequence of the electronic degeneracy is that the real-valued
ground state electronic wave function changes sign when the nuclear motion encir-
cles the conical intersection (CI) between two electronic states. The sign change
requires a corresponding sign change on the nuclear motion wave function to keep the
overall wave function single-valued. In other words, the nuclear motion Schrödinger
equation acquires a vector potential as originally pointed out by Mead and
Truhlar [1]. The effect of the vector potential is equivalent to a magnetic solenoid
centered at the conical intersection [2–4]. Flux of this magnetic field through the
surface enclosed by the CI yields a phase shift. This phase shift, due to its geomet-
ric origin, is referred to as the geometric phase (GP) or the Berry phase [5]. The
geometric phase resulting from the vector potential is analogous to the Aharonov-
Bohm effect [6] and Mead initially referred to this as the molecular Aharonov-Bohm
effect [7]. There have been numerous attempts in the literature to include the geo-
metric phase in both bound state [8–10] and scattering calculations of triatomic sys-
tems [11–23]. While bound state studies of alkali metal trimers such as Li3 [24],
Na3 [25] and transition metal systems like Cu3 [26] showed much better agreement
with experimental results when the GP effect is included, experimental verification
of the GP effects in a bimolecular chemical reaction has not been successful yet
[27–33].
Almost all of the experimental studies of GP effects in bimolecular chemical reac-
tions have so far been limited to H or D atom exchange reactions in H+HD/D+HD
systems at energies close to the conical intersection [27–33]. At these high colli-
sion energies, many angular momentum partial waves contribute and any small GP
effect present in a partial wave resolved cross section washes out when a summation
over all partial waves is carried out to evaluate the total differential or integral cross
Geometric Phase and Interference Effects in Ultracold Chemical Reactions 267

sections [14–23]. Thus, it appears that an energy regime where only a single par-
tial wave contributes is the most relevant regime to explore GP effects in a chemical
reaction. This regime, referred to as the cold and ultracold regime, has gained much
interest in recent years, thanks to the dramatic progress in cooling and trapping of
molecules in the mK and 𝜇K regimes. Here, we will focus on our recent studies
of the GP effect in chemical reactions in the ultracold regime taking the hydrogen
exchange reaction as an illustrative example.
The ultracold regime [34–38] provides a fascinating domain to explore quan-
tum effects in chemical reactions. Because s-wave scattering dominates at ultracold
temperatures (for bosons and distinguishable particles), only the l = 0 partial wave
contributes and the GP effect is not smeared out by partial wave summation. Further-
more, isotropic scattering in the s-wave regime allows for maximum constructive or
destructive interference between wave functions along alternative paths around the
CI (direct and looping/exchange paths). These properties combined with an effec-
tive quantization of the scattering phase-shift in the ultracold regime (Levinson’s
theorem [39] 𝛿(0) = n𝜋 where 𝛿 is the phase shift and n is the number of bound
states supported by the potential well) entail maximum constructive or destruc-
tive interference between the direct and exchange/looping scattering amplitudes.
This leads to a large enhancement or suppression of reactivity, as recently demon-
strated for O+OH(v = 0, 1) → H+O2 (v′ , j′ ) [40, 41] and the hydrogen exchange
processes in H+H2 (v = 4, j = 0), H+HD(v = 4, j = 0) and D+HD(v = 4, j = 0) reac-
tions [42–45]. The H+H2 reaction has an energy barrier for vibrational levels v < 3
but becomes barrierless for v > 3 [43, 46–48]. Indeed, vibrationally adiabatic poten-
tials for the H+H2 reaction for v = 4 and higher vibrational levels depict an effective
potential well. The bound state structure of this potential well has a dramatic effect
on the scattering process at ultracold temperatures as discussed below. Also, barrier-
less reactions occur with much larger rate coefficients at ultracold temperatures and
are more amenable to experiments than barrier reactions that proceed via tunneling.
The chapter is organized as follows. In Sect. 2 we briefly discuss the mecha-
nism of the GP effect in ultracold reactions. Section 3 outlines the coupled channel
method employed in the scattering calculations. Illustrative results of GP effects in
H+H2 /H+HD/D+HD reactions are presented in Sect. 4 followed by conclusions in
Sect. 5.

2 Mechanism of the GP Effect in the Ultracold Regime

In our previous work [40, 42] we showed that to observe the GP effect in reactive
and inelastic collisions two criteria should be satisfied: (i) the adiabatic PES must
exhibit a conical intersection; (ii) the scattering amplitudes along the two scattering
pathways (direct and exchange/looping) must have comparable magnitude and scat-
ter into the same angular region. Isotropic scattering in the ultracold regime and the
effective quantization of the scattering phase shift as required by Levinson’s theorem
provide the criterion for maximum constructive and destructive interference between
268 N. Balakrishnan and B. K. Kendrick

10-11 10-11
(a) (b)
v = 0, j = 8

`
v = 1, j = 4

`
10-12 10-12

10-13
Rate Coefficient (cm3/s)

10-13

Rate Coefficient (cm3/s)

NGP J = 0-4
GP
NGP J = 0-4
10-14
GP
10-14
10-11
(c) (d)
v = 2, j = 4
`
`

10-11 v = 3, j = 4

`
10-12
10-12

10-13
10-13 J = 0-4
NGP J = 0-4 NGP
GP GP

10-14 -6 10-14 -6
10 10-4 10-2 100 102 10 10-4 10-2 100 102
Energy (K) Energy (K)

Fig. 1 Rotationally resolved reaction rate coefficients for the H + H2 (v = 4, j = 0) → H + H2 (v′ , j′ )

(para-para) as functions of the incident collision energy: a v′ = 0, j′ = 8, b v′ = 1, j′ = 4, c v′ = 2,
j′ = 4, and d v′ = 3, j′ = 4. In all panels the red curves include the geometric phase (GP) and the
black curves do not (NGP). Contributions from all values of total angular momentum J = 0–4 are
included in the rates. Reproduced with permission from [45]

the two scattering amplitudes [40–42]. The details of the interference mechanism
have been discussed in prior works [40–42] and only a brief description is given
here.
In H+HD collisions, due to the presence of two identical H atoms, a correct treat-
ment of the H+HD channel should include both purely non-reactive collisions such
as Ha +Hb D(v, j)→Ha +Hb (v′ , j′ )D (where a and b labels the two hydrogen atoms for
illustrative purpose) and exchange collisions such as Ha +Hb D(v, j)→Hb +Ha D(v′ , j′ )
where the two identical H atoms exchange with one another. The same applies to the
D+HD system where the identical D atoms are exchanged in the exchange scatter-
ing amplitude. Both reactions also include purely reactive channels leading to D+H2
and H+D2 products. These reactive pathways may involve a “direct” path (traverses
over one transition state) or a “looping path” (traverses over two transition states).
For a schematic illustration of these reaction pathways, see Fig. 1a, b of Kendrick et
al. [42]. Our recent studies have shown large GP effects in the H+HD and D+HD
channels due to strong interference between the inelastic and exchange components
of the scattering amplitudes. The GP effect was found to be not significant for the
purely reactive channels due to small values of the scattering amplitudes for the loop-
ing pathway resulting in negligible interference with the dominant “direct” pathway.
Geometric Phase and Interference Effects in Ultracold Chemical Reactions 269

Here, we focus on the exchange channel that shows the largest GP effect. Our the-
oretical description is based on the exchange pathways depicted in Fig. 1a of Ref.
[42] and earlier works of Althorpe and collaborators [20–23].
For H+HD and D+HD systems (or in general, A+AB collisions), the GP and
NGP (no geometric phase) scattering amplitudes can be written in terms of the
“inelastic” and “exchange” scattering amplitudes, finel and fex , respectively,

1
fNGP∕GP = √ (finel ± fex ). (1)
2

The square modulus of the scattering amplitudes for the NGP and GP calculations
may be written as

1
|fNGP∕GP |2 = (|f |2 + |fex |2 ± 2|finel | |fex | cos Δ), (2)
2 inel

where the complex scattering amplitudes finel = |finel |ei𝛿inel and fex = |fex |ei𝛿ex and
Δ = 𝛿ex − 𝛿inel is the phase difference between the exchange and inelastic pathways.
For comparable values of the two scattering amplitudes, i.e., |fex | = |finel | = |f |,
Eq. (2) becomes |fNGP∕GP |2 = |f |2 (1 ± cos Δ). Furthermore, if cos Δ = +1 then max-
imum (constructive) interference occurs for the NGP case and |fNGP |2 ∼ 2|f |2 and
|fGP |2 ∼ 0. On the other hand, if cos Δ = −1 then maximum (constructive) inter-
ference occurs for the GP case and |fGP |2 ∼ 2|f |2 and |fNGP |2 ∼ 0. Recalling that
Δ = n𝜋 can occur in the ultracold regime (Levinson’s theorem) where n is an inte-
ger, the reaction can be turned on or off depending simply on the sign of the inter-
ference term (since | cos Δ| ∼ 1). In contrast, if one of the scattering amplitudes is
much greater than the other, |fex |2 ≫ |finel |2 or |finel |2 ≫ |fex |2 , then Eq. (2) becomes
|fNGP∕GP |2 ∼ |fex |2 ∕2 or |fNGP∕GP |2 ∼ |finel |2 ∕2. The GP effect vanishes in this case
and the interference term containing | cos Δ| plays no role. In the high partial wave
limit (high collision energies), the interference term averages out to zero (cos Δ ∼ 0)
and there is no GP effect. This description is also valid for the pure reactive case,
e.g., H+HD→D+H2 except the two scattering amplitudes for the different paths are
replaced by |fex | = |floop | and |finel | = |fdirect |. In our previous work, we have shown
that the phase quantization of Δ = n𝜋 can be understood by considering scattering in
a simple spherical well potential for the different pathways (i.e., Levinson’s theorem
𝛿ex = nex 𝜋 and 𝛿inel = ninel 𝜋 but with a different number of bound states nex and ninel
for the spherical well potentials traversed by the two pathways) [40].

3 Quantum Scattering Method

The reactive scattering calculations were carried out using hyperspherical coor-
dinates. Two sets of hyperspherical coordinates are employed: the adiabatically-
adjusting principle axis hyperspherical (APH) coordinates of Pack and Parker
270 N. Balakrishnan and B. K. Kendrick

[49, 50] in the inner hyper-radial region where the three-body interaction is strong
and the Delves hyperspherical coordinates in the outer region where the three-body
forces vanish and different atom-diatom configurations emerge. The APH coordi-
nates are independent of the different atom-diatom arrangement channels and allow
an evenhanded description of all three arrangement channels in an A+BC system
compared to the Delves hyperspherical coordinates. The method accurately treats the
body-frame Eckart singularities [50] associated with non-zero total angular momen-
tum quantum number J and includes the geometric phase using the general vector
potential approach [15, 16, 18]. The geometric phase is included only in the APH
coordinates as it is relevant only in the region of three-body interaction where the
CI is located. Regardless of the choice of the hyperspherical coordinates, the basic
numerical approach involves a sector-adiabatic formalism. The hyper radius (𝜌) is
divided into a large number of sectors and at the center of each sector, the total wave
function is expanded in terms of five-dimensional hyperspherical surface functions.
The surface functions are in turn expanded in primitive angular functions. Conver-
gence is sought with respect to the number of primitive functions included in the
expansion. A sequential truncation/diagonalization procedure is used to reduce the
size the surface function matrix. The expansion coefficients depend on the hyper
radius but within a sector they are assumed to be independent of 𝜌. Coupled channel
equations resulting from the Schrödinger equation with this expansion of the total
wave function in terms of hyperspherical surface functions are solved from sector-
to-sector. Asymptotic boundary conditions are applied in Jacobi coordinates at the
last sector in 𝜌 to evaluate the reactance and scattering matrices from which cross
sections and rate coefficients are computed using standard expressions [49].

4 Results

In a series of papers [42–45], we have carried out a detailed analysis of geometric

phase effects in the H+H2 , H+HD and D+HD reactions with the H2 /HD molecule
excited to the v = 4 vibrational level. As discussed previously, the vibrationally adia-
batic potential curves display a barrierless path for v > 3 with a small potential well
compared to v = 0 which proceeds through an energy barrier. For the symmetric
H+H2 reaction, the geometric phase can be accounted for by properly symmetrizing
the scattering amplitude and including a phase factor. For para-para transition (even
j to even j′ transitions) the properly symmetrized differential cross section is given
by [18, 51]

d𝜎 | k̄ v′ j′ N
| = |f ′ ′ ′ − (−1) fvjm→v′ j′ m′ |2 ,
igp R
(3)
dΩ |vjm→v′ j′ m′ k̄ vj vjm→v j m
Geometric Phase and Interference Effects in Ultracold Chemical Reactions 271

Fig. 2 Total reaction rate 10-10

coefficients for the H + (a)
H2 (v = 4, j = 0) → H + H2
(para-para) reaction as a
10-11
function of the collision
energy: a summed over all

Rate Coefficient (cm3/s)

values of total angular J = 0-4
momentum J = 0–4, and b 10-12 NGP
individual contributions GP
from each J. Solid curves
J = 0, dashed curves J = 1, 10-10 (b)
dot dashed J = 2, dotted J=0
J = 3, and double-dot 10-12
dashed J = 4. The red curves
J=1 J=3
include the geometric phase 10-14
(GP) and the black curves do J=2
not (NGP). Reproduced with 10-16
permission from [45] J=4
10-18 -6
10 10-5 10-4 10-3 10-2 10-1 100 101 102
Energy (K)

where f N and f R are the scattering amplitudes for the non-reactive and reactive chan-
nels, k̄ vj are the appropriately normalized wave vector magnitudes, and igp = 1 or 0
for calculations which include or do not include the GP, respectively (see Ref. [18] for
details). As demonstrated in previous [18] studies using the vector potential approach
[1] the GP effect is captured almost entirely by the sign change (i.e., igp = 1) given
in Eq. (3). Thus, the GP can be accurately included for H + H2 by performing cal-
culations without the vector potential [18, 51]. The computed f N and f R are then
properly combined using Eq. (3) to include (igp = 1) or not include (igp = 0) the GP.
We used this approach for the H + H2 calculations reported in this work. For H+HD
and D+HD reactions, the GP effect was included using the vector potential approach.
The H3 PES of Boothroyd et al. [52] is used in the calculations reported here. We
have verified that the H3 PES of Mielke et al. [53] yields comparable results [42].
Figure 1 shows the rotationally resolved reaction rate coefficients for the
H+H2 (v = 4, j = 0) → H+H2 (v′ , j′ ) reaction for v′ = 0, j′ = 8, v′ = 1, j′ = 4, v′ =
2, j′ = 4 and v′ = 3, j′ = 4 [45]. It is seen that the GP rates dominate the NGP rates
for all four rotational levels shown in Fig. 1. Indeed, a similar trend is found for all
the state-to-state rotational transitions leading to even rotational levels of H2 for all v′
levels (see Table I of Kendrick et al. [45]). As a result, a similar GP effect is found for
vibrationally resolved rate coefficients for para-para transitions when summed over
all rotational levels in a given vibrational state. The trend also prevails in the total
rate coefficients as illustrated in Fig. 2. The upper panel of Fig. 2 shows the total rate
coefficient and the lower panel displays contributions from different partial waves
(orbital angular momentum l = J for initial rotational level j = 0).
Figure 3 shows a comparison between GP and NGP rate coefficients for the
H+HD(v = 4, j = 0) → HD(v′ , j′ )+ H reaction for v′ = 0, j′ = 3, v′ = 1, j′ = 2, and
272 N. Balakrishnan and B. K. Kendrick

10-13 Evn Ex-Sym (a) 10-13 Odd Ex-Sym (d)

10-14 10-14
v = 0, j = 3

`
10-15 v = 0, j = 3

`
10-15
10-16
Rate Coefficient (cm3/s)

10-16

Rate Coefficient (cm3/s)

10-12
10-12
Evn Ex-Sym (b) Odd Ex-Sym (e)
10-14 10-14
v = 1, j = 2 v = 1, j = 2

`
`
`

10-16 10-16
10-12
Evn Ex-Sym (c) 10-12 Odd Ex-Sym (f)
10-13
10-13
10-14 10-14 v = 3, j = 2

`
v = 3, j = 2
`

10-15 10-15

10-16 10-1610-6
10-6 10-4 10-2 100 102 10-4 10-2 100 102
Energy (K) Energy (K)

Fig. 3 Rotationally resolved rate coefficients (cross section times the relative velocity) for the
H+HD(v = 4, j = 0) → H+HD(v′ , j′ ) reaction for v′ = 0, j′ = 3, v′ = 1, j′ = 2, and v′ = 3, j′ = 2 as
functions of the collision energy. Even and odd exchange symmetry results are given in the left
and right panels, respectively. In each panel, the red curves show the GP results and the black
curves denote the NGP results. The results include all values of total angular momentum J = 0–4.
Reproduced with permission from [43]

v′ = 3, j′ = 2 as a function of the incident collision energy. Total angular momentum

quantum numbers J = 0–4 have been included the calculations to yield converged
results up to 20 K though results are presented for energies up to 100 K. Results for
even exchange symmetry are shown in the left panel and those for odd exchange
symmetry are given in the right panel. It is seen that the GP and NGP results differ
dramatically in the ultracold regime but they merge and show little difference for
energies above 20 K. The GP/NGP effect is sensitive to the final rovibrational lev-
els of the HD molecule and the exchange parity symmetry. Even and odd exchange
symmetry results exhibit opposite GP/NGP effects and the overall GP effect becomes
smaller when the total rate is computed by combining even/odd exchange symmetry
contributions with appropriate nuclear spin statistics factors. This is illustrated in
Fig. 4 for the H+HD(v = 4, j = 0) → HD+ H reaction. A similar trend is observed
for the corresponding D+HD reaction.
The dominance of the GP/NGP effect for a given state-to-state rate coefficient can
be explained based on the sign and magnitude of cos Δ and the scattering amplitudes
for the direct and exchange pathways. This can be extracted from the GP and NGP
scattering amplitudes (see Eqs. (1) and (2)). Figure 5 shows the ratio of the squares
Geometric Phase and Interference Effects in Ultracold Chemical Reactions 273

10-10
Evn+Odd Symm
NGP
GP
Rate Coefficient (cm3/s) 10-12

10-14 J = 0-4
J=0
J=1
J=2
J=3
10-16 J=4

10-18 -7
10 10-6 10-5 10-4 10-3 10-2 10-1 100 101 102
Energy (K)

Fig. 4 Total angular momentum resolved (partial wave resolved since l = J for j = 0) rate coef-
ficients for the H+HD(v = 4, j = 0) → H+HD reaction summed over all energetically accessible
rotational and vibrational levels of the product HD molecule as well as even and odd exchange
symmetry contributions weighted by appropriate nuclear spin statistics factors. Reproduced with
permission from [43]

of the scattering amplitudes for the exchange and direct paths (upper panel) and the
⟨cos Δ⟩ values (lower panel) as a function of the collision energy for the v′ = 0, j′ =
3 transition shown in Fig. 3 for the H+HD reaction. It is seen that the NGP rates
dominate when ⟨cos Δ⟩ = +1 and the GP rates dominate when ⟨cos Δ⟩ = −1. This
is most clearly seen at energies below 1 mK where s-wave scattering dominates.
At higher collision energies where non-zero partial waves contribute, the ⟨cos Δ⟩
values oscillate around zero making the contribution from the interference term in
Eq. (2) less significant leading to a negligible GP effect. This trend is observed in all
state-to-state rotationally resolved rate coefficients for H+HD and D+HD reactions
[43, 44].
The large GP effect in state-to-state cross sections when isotropic scattering dom-
inates in the s-wave regime is illustrated in Fig. 6 where differential cross sections
(DCSs) for the v′ = 3, j′ = 0 final state in the D+HD(v = 4, j = 0) reaction are plot-
ted as a function of the collision energy and the scattering angle. It is seen that the
NGP results are enhanced for even exchange symmetry while the GP results are
enhanced for the odd exchange symmetry through constructive interference between
the scattering amplitudes for the direct and exchange pathways. The interference is
destructive when the DCSs for these cases are suppressed. The interference pattern
changes when resonances are present as seen at energies near 1 K where a l = 2
shape resonance occurs for the D+HD reaction. A detailed discussion of the reso-
nances and parameters characterizing them (position, width and lifetimes) are given
in Hazra et al. [43] and Kendrick et al. [44] for the H+HD and D+HD reactions.
274 N. Balakrishnan and B. K. Kendrick

1.4
Evn Ex-Sym (a) Odd Ex-Sym (c)
1.3 1.35

v = 0, j = 3

`
1.2 1.3 J=0
J=1
v = 0, j = 3

`
J=0 J = 0-4
1.1 J=1
J = 0-4 1.25

1.5 1.5
Evn Ex-Sym (b) Odd Ex-Sym (d)
1 1
v = 0, j = 3

`
0.5 0.5
v = 0, j = 3
`

`
COS Δ

COS Δ
0 0
J=0 J=0
J=1 J=1
-0.5 J = 0-4 -0.5 J = 0-4

-1 -1

-1.5 -1.5
10-6 10-4 10-2 100 102 10-6 10-4 10-2 100 102
Energy (K) Energy (K)

Fig. 5 Ratio of the square of the scattering amplitudes for the exchange (reactive) and inelastic
(non-reactive) pathways (upper panels) and average value of cos Δ (lower panels) for both even (left
panels) and odd (right panels) exchange symmetries as functions of the collision energy. Results are
presented for the H+HD(v′ = 0, j′ = 3) product channel. Reproduced with permission from [43]

5 Summary and Conclusions

We have discussed the importance of the geometric phase effect in ultracold hydro-
gen exchange reactions in collisions of H and D atoms with vibrationally excited
H2 and HD molecules. For vibrational levels v > 3 these reactions occur through
a barrierless path. Results presented for the v = 4 vibrational levels of the H2 and
HD molecules illustrate strong interference between the direct and exchange com-
ponents of the scattering amplitudes leading to enhancement or suppression of the
reactivity. Isotropic scattering in the ultracold s-wave regime allows maximum con-
structive/destructive interference leading to large GP effects in state-to-state reaction
rate coefficients. The effect persists but to a lesser extent in the total reaction rate, due
in part, to the cancellation of the GP effects when even and odd exchange symmetry
results are added to yield the total rates for H+HD and D+HD reactions. The results
presented here illustrate that the GP effect may be experimentally observable by the
selection of a particular nuclear spin-state of the HD molecule in H+HD/D+HD
collisions.
Geometric Phase and Interference Effects in Ultracold Chemical Reactions 275

(a) Even Ex-Sym

NGP v’ = 3 , j’ = 0
log(DCS)
log(DCS) cm /Sr
2

180

(deg)
140
GP
100

angle
60

ering
scatt
20

log(Ec) K

(b) Odd Ex-Sym

GP
v’ = 3 , j’ = 0
log(DCS)
log(DCS) cm /Sr
2

180
(deg)

140
NGP
100
ngle
ring a

60
e
scatt

20

log(Ec) K

Fig. 6 The DCS is plotted as a function of collision energy and scattering angle for the D +
HD(v = 4, j = 0) → D + HD(v′ = 3, j′ = 0) reaction. Panel a is even exchange symmetry and b
is odd exchange symmetry. The DCS plotted with the red mesh includes the geometric phase (GP)
while the one with the black mesh does not (NGP). The results include all values of total angular
momentum J = 0–4. Reproduced with permission from [44]
276 N. Balakrishnan and B. K. Kendrick

Acknowledgements N. B. acknowledges support from the Army Research Office, MURI grant
No. W911NF-12-1-0476 and the National Science Foundation, grant No. PHY-1505557. B. K. K.
acknowledges that part of this work was done under the auspices of the US Department of Energy
under Project No. 20140309ER of the Laboratory Directed Research and Development Program
at Los Alamos National Laboratory. Los Alamos National Laboratory is operated by Los Alamos
National Security, LLC, for the National Security Administration of the US Department of Energy
under contract DE-AC52-06NA25396.

References

1. Mead CA, Truhlar DG (1979) J Chem Phys 70:2284; 78:6344E (1983)

2. Simon B (1983) Phys Rev Lett 51:2167
3. Bohm A, Boya LJ, Kendrick B (1991) Phys Rev A 43:1206
4. Mead CA (1992) Rev Mod Phys 64:51
5. Berry MV (1984) Proc R Soc Lon Ser A 392:45
6. Aharonov Y, Bohm D (1959) Phys Rev 115:485
7. Mead CA (1980) Chem Phys 49:23
8. Kendrick BK (1997) Phys Rev Lett 79:2431
9. Kendrick BK (1997) Int J Quantum Chem 64:581
10. Babikov D, Kendrick BK, Zhang P, Morokuma K (2005) J Chem Phys 122:044315
11. Kuppermann A, Wu YSM (1993) Chem Phys Lett 205:577; 213:636E (1993)
12. Wu YSM, Kuppermann A (1995) Chem Phys Lett 235:105
13. Kuppermann A, Wu YSM (1995) Chem Phys Lett 241:229
14. Kendrick BK (2001) J Chem Phys 114:8796
15. Kendrick B, Pack RT (1996) J Chem Phys 104:7475
16. Kendrick B, Pack RT (1996) J Chem Phys 104:7502
17. Kendrick BK, (2000) J Chem Phys 112:5679; 114:4335E (2001)
18. Kendrick BK (2003) J Phys Chem 107:6739
19. Kendrick BK (2003) J Chem Phys 118:10502
20. Juanes-Marcos C, Althorpe SC (2005) J Chem Phys 122:204324
21. Juanes-Marcos JC, Althorpe SC, Wrede E (2005) Science 309:1227
22. Althorpe SC (2006) J Chem Phys 124:084105
23. Althorpe SC, Stecher T, Bouakline F (2008) J Chem Phys 129:214117
24. von Busch H, Eckel V, Dev H, Kasahara S, Wang J, Demtröder W, Sebald P, Meyer W (1998)
Phys Rev Lett 81:4584
25. Keil M, Krämer H-G, Kudell A, Baig MA, Zhu J, Demtröder W, Meyer W (2000) J Chem Phys
113:7414
26. Rohlfing EA, Valentini JJ (1986) Chem Phys Lett 126:113
27. Kliner DAV, Adleman DE, Zare RN (1991) J Chem Phys 95:1648
28. Adelman DE, Shafer NE, Kliner DAV, Zare RN (1992) J Chem Phys 97:7323
29. Kitsopoulos TN, Buntine MA, Baldwind DP, Zare RN, Chandler DW (1993) Science 260:1605
30. Jankunas J, Sneha M, Zare RN, Bouakline F, Althorpe SC (2013) J Chem Phys 139:144316
31. Jankunas J, Sneha M, Zare RN, Bouakline F, Althorpe SC, Phys Z (2013) Chemistry 227:1281
32. Jankunas J, Sneha M, Zare RN, Bouakline F, Althorpe SC, Herráez-Aguilar D, Aoiz FJ (2014)
PNAS 111:15
33. Sneha M, Gao H, Zare RN, Jambrina PG, Menéndez M, Aoiz FJ (2016) J Chem Phys
145:024308
34. Krems RV, Stwalley WC, B. Friedrich B (eds) (2009) Cold molecules: theory, experiment,
applications. CRC Press, Taylor & Francis Group
35. Carr LD, DeMille D, Krems RV, Ye J (2009) New J Phys 11:055049
Geometric Phase and Interference Effects in Ultracold Chemical Reactions 277

36. Ospelkaus S, Ni K-K, Wang D, de Miranda MHG, Neyenhuis B, Quéméner G, Julienne PS,
Bohn JL, Jin DS, Ye J (2010) Science 327:853
37. Knoop S, Ferlaino F, Berninger M, Mark M, Nägerl H-C, Grimm R, D’Incao JP, Esry BD
(2010) Phys Rev Lett 104:053201
38. Balakrishnan N (2016) J Chem Phys 145:150901
39. Levinson N (1949) Kgl. Danske Videnskab Selskab Mat Fys Medd 25:9
40. Kendrick BK, Hazra J, Balakrishnan N (2015) Nat Commun 6:7918
41. Hazra J, Kendrick BK, Balakrishnan N (2015) J Phys Chem A 119:12291
42. Kendrick BK, Hazra J, Balakrishnan N (2015) Phys Rev Lett 115:153201
43. Hazra J, Kendrick BK, Balakrishnan N (2016) J Phys B At Mol Opt Phys 49:194004
44. Kendrick BK, Hazra J, Balakrishnan N (2016) New J Phys 18:123020
45. Kendrick BK, Hazra J, Balakrishnan N (2016) J Chem Phys 145:164303
46. Simbotin I, Ghosal S, Côté R (2011) Phys Chem Chem Phys 13:19148
47. Simbotin I, Ghosal S, Côté R (2014) Phys Rev A 89:040701
48. Simbotin I, Côté R (2015) N J Phys 17:065003
49. Pack RT, Parker GA (1987) J Chem Phys 87:3888
50. Kendrick BK, Pack RT, Walker RB, Hayes EF (1999) J Chem Phys 110:6673
51. Mead CA (1980) J Chem Phys 72:3839
52. Boothroyd AI, Keogh WJ, Martin PG, Peterson MR (1996) J Chem Phys 104:7139
53. Mielke SL, Garrett BC, Peterson KA (2002) J Chem Phys 116:4142
 Part III
Biochemistry and Biophysics
Adducts of Arzanol with Explicit Water
Molecules: An Ab Initio and DFT Study

Liliana Mammino

Abstract Arzanol (C22H26O7) is a naturally occurring acylphloroglucinol present

in Helichrysum italicum. It is the major responsible of its medicinal properties,
which include anti-oxidant properties. In the arzanol molecule, the R of the COR
group characterising acylphloroglucinols is a methyl group, and the two sub-
stituents in meta to COR are an α-pyrone ring, bonded to the benzene ring through a
methylene bridge, and a prenyl chain. The high number of hydrogen bond donor
and acceptor sites in the molecule entails an investigation taking into account
solute-solvent hydrogen bonds in an explicit manner. The current work considers
adducts of arzanol with explicit water molecules for a representative selection of its
conformers. Adducts with one water molecule attached in turn to each of the
H-bond donors or acceptors were calculated to estimate the strength with which
each site can bind a water molecule. Adducts with varying numbers of water
molecules were calculated to identify preferred arrangements of the water mole-
cules around the various sites and around the molecule as a whole. These adducts
also suggest possible geometries for the first solvation layer. All the adducts were
calculated at the HF/6-31G(d, p) and the DFT/B3LYP/6-31+G(d, p) levels, with
fully relaxed geometry.

Keywords Acylphloroglucinols ⋅Adducts with explicit water molecules

Arzanol ⋅
Hydrogen bonding ⋅ Solute-solvent interactions

Electronic supplementary material The online version of this chapter (https://doi.org/10.1007/

978-3-319-74582-4_16) contains supplementary material, which is available to authorized users.

L. Mammino (✉)
Department of Chemistry, University of Venda, P/Bag X5050, Thohoyandou 0950,
South Africa
e-mail: [email protected]

© Springer International Publishing AG, part of Springer Nature 2018 281

Y. A. Wang et al. (eds.), Concepts, Methods and Applications of Quantum Systems
in Chemistry and Physics, Progress in Theoretical Chemistry and Physics 31,
https://doi.org/10.1007/978-3-319-74582-4_16
282 L. Mammino

Fig. 1 Structure of the O14

13
arzanol molecule and atom H17 7
numbering utilized in this
work [5]. The C atoms are O12 1 O8 H15
denoted by their numbers. 6 2
The figure shows the carbon PHL O23
skeleton of the molecule, the 5 3
O atoms, and the H atoms 9
11 4 18
pertaining to OH groups. The 17
other H atoms are hidden, to 31 29 O19
O10 PYR
better highlight the molecular 30 22
structure H16 20 28
O26 21 24
32
H27 25

Fig. 2 General structure of R O

acylphloroglucinols. The
molecules are characterised H
by the presence of three
equally spaced OH and a O
COR group O
H

R" R'

O
H

1 Introduction

Arzanol (C22H26O7, Fig. 1) is a naturally-occurring acylphloroglucinol, and is the

major responsible of the anti-inflammatory, anti-oxidant, antibiotic and antiviral
activities of Helichrysum italicum [1–3]. Acylphloroglucinols (ACPL, Fig. 2, [4])
are derivatives of phloroglucinol (1,3,5-trihydroxybenzene) characterised by the
presence of a COR group. In the arzanol molecule, R is a methyl group, R′ is an
α-pyrone ring attached to the phloroglucinol moiety through a methylene bridge
and R″ is a prenyl chain (Fig. 1).
Figure 1 shows the atom numbering utilized in this work, which is consistent
with the numbering utilised in a thorough conformational study of the molecule [5];
for the phloroglucinol moiety, it is also consistent with the numbering utilized in
previous works on ACPLs [6–10], to facilitate cross-references and comparisons.
For the sake of conciseness, the two moieties and the molecule are denoted by the
following acronyms in the rest of the text: PHL for the acylphloroglucinol moiety
Adducts of Arzanol with Explicit Water Molecules … 283

(comprising the prenyl chain attached to C5), PYR for the α-pyrone moiety, and
ARZ for arzanol.
The computational study of the ARZ molecule [5] showed that its conforma-
tional preferences are influenced by the patterns of intramolecular hydrogen bonds
(IHB), which are the dominant stabilising factor, by the mutual orientation of
the PHL and PYR moieties (which also determines part of the IHB patterns), by the
orientation of the phenol OHs (as is true for ACPLs in general [6–10]) and by the
orientation of the prenyl chain (which is generally true for prenylated ACPLs [8]).
The IHBs comprise the IHB formed by O14 and either H15 or H16 (here termed
“first IHB” [6–10]), the IHBs between the two moieties (which will be categorised
as IMHB, for ‘intermoiety H-bonds’, when it is relevant to underline this role, [5])
and the O10-H16⋯π or O12-H17⋯π interactions, when either O10-H16 or
O12-H17 and the prenyl chain have favourable orientations. The distribution of
donor and acceptor sites in the ARZ molecule enables the formation of two
simultaneous IMHBs, one on either side of the methylene bridge, and all the lowest
energy conformers are characterised by the presence of the first IHB and two
IMHBs [5]; when the first IHB engages H15, it is cooperative with the IMHB
engaging O8.
The computational study of ARZ [5] included calculations in three solvents
(chloroform, acetonitrile and water) utilising the Polarizable Continuum Model
(PCM, [11–13]). In general, “continuum solvation models are the ideal conceptual
framework to describe solvent effects within the QM approach” [13]. However,
PCM does not take into explicit account directional solute-solvent interactions such
as hydrogen bonding [14] (except implicitly for some effects [15]). On the other
hand, solute-solvent H-bonding is important for solute molecules containing
H-bond donors or acceptors and solvent molecules capable of forming H-bonds.
The most important of these solvents is water, which constitutes the highest pro-
portion of the mass of living organisms. The consideration of adducts with explicit
water molecules is the most informative option on solute-solvent H-bonding util-
ising QM approaches. It can provide information about preferential arrangements of
water molecules in the vicinity of the various donor or acceptor sites. It can also
contribute information on the outcome of the competition between intramolecular
H-bonding and intermolecular solute-water H-bonding through energetics com-
parisons (by comparing an adduct maintaining a certain IHB and an adduct in
which its donor or acceptor is engaged in a solute-solvent H-bond), and also
through the optimisation itself, which may ‘open’ (break) specific IHBs, as verified,
e.g., in the study of adducts of caespitate [16] or other ACPLs [17] with explicit
water molecules.
This work considers adducts of various conformers of ARZ with explicit water
molecules, trying to identify patterns for the energy of the solute-solvent interaction
at different binding sites and for preferred arrangements of water molecules around
different sites of the ARZ molecule. The study appears to be particularly interesting
because of the high number of H-bond donors and acceptors in the ARZ molecule
and because of the presence of IHBs (including cooperative ones), which influences
the way in which water molecules approach the corresponding regions. The work
284 L. Mammino

pertains to an ongoing study of antioxidant ACPLs which, for ARZ, has so far
produced the results reported in [5, 18].

2 Calculation Details

2.1 Selection of Adducts

ARZ can have a high number of conformers, with different mutual orientations of
the PHL and PYR moieties and different IHB patterns. More than 90 conformers
were considered in [5] (which included all the lowest energy ones and disregarded
only some very high energy ones without the first IHB). Four mutual orientations of
the two moieties were identified and denoted with the numbers 1, 2, 3 and 4 [5].
Corresponding conformers of the #1 and #2 series differ only by the fact that the
orientation of the PYR ring is symmetrical with respect to the plane identified by
the PHL ring and have very close energies; the same is true for corresponding
conformers of the #3 and #4 series. Therefore, it is sufficient to select one series
from each pair to ensure the consideration of the different IHMB patterns. The #1
and #3 series were selected, to include the minimum energy conformers of each pair
(Fig. 3). It can be expected that adducts of conformers of the #2 series with the
same number and input-arrangement of water molecules as for corresponding
conformers of the #1 series will be very similar, as the PYR-related portion is
symmetrical in the two cases and the IHB and steric patterns are the same.
Equivalently, adducts of corresponding conformers of the #3 and #4 series are
expected to be very similar. A few adducts of conformers of the #2 and #4 series
were considered to verify this expectation.

1-d-r-ξ-αδ 3-s-w-η-γτ

Fig. 3 The two lowest energy conformers with different mutual orientations of the two ring
systems, as identified in [5], and the. Acronyms denoting them
Adducts of Arzanol with Explicit Water Molecules … 285

The variety of possible adducts with explicit water molecules is extremely high
for a solute molecule like ARZ, which contains a high number of H-bond donor or
acceptor sites and can form a variety of conformers in which one or another site
may be more or less available to form H-bonds with water molecules. It was opted
to consider all the lower energy conformers of the #1 and #3 series and repre-
sentative higher energy ones, to ensure that the most interesting arrangements of
water molecules around the various sites of ARZ are captured.
Two sets of adducts were considered. One set involves most of the conformers
of the #1 and #3 series, and entails adducts in which one water molecules is
attached in turn to each donor or acceptor site of each conformer. These adducts
enable a comparison of the energy with which each site can bind a water molecule,
and also offer indications about how close a water molecule can approach the given
site. In the real situation within the solvent, this is determined also by the inter-
actions between water molecules and, therefore, a water molecule attached to a
given site might remain at a greater distance than in the models with only one water
molecule. On the other hand, it may happen that one (or, sometimes, more) water
molecules remain attached to the solute molecule when it enters the active site of
the biological target and may contribute to the binding between the molecule and
the target; in such cases, the water molecule will likely remain as close as possible
to the ARZ site to which it binds. The knowledge of the strength with which a water
molecule binds to a certain donor or acceptor site of ARZ may thus be useful also
for a better understanding of its permanence (when it occurs) when ARZ binds to its
target, or its role in such binding.
The second set comprises adducts with several explicit water molecules,
attempting to approximate a first solvation layer or portions of it. Like in previous
studies on adducts of ACPLs with explicit water molecules [16, 17, 19], the ‘first
solvation layer’ concept is expanded to include not only the water molecules
directly H-bonded to suitable sites of the solute molecule, but also water molecules
that might bridge them (the presence of a third water molecule bridging two
molecules directly H-bonded to the solute often has a stabilizing effect [17]). The
distribution and spacing of the several H-bond donors or acceptors in ARZ enable
the possibility of considering adducts in which the water molecules attached to
ARZ, and those bridging them, approximate a continuous layer in the vicinity of
extensive portions of ARZ.
The inputs were prepared placing water molecules in the vicinity of H-bond
donor or acceptor sites of the selected conformers of ARZ. Different numbers and
arrangements of water molecules were considered for each conformer, also taking
into account the resulting arrangements of already optimised outputs. For instance,
when one or more water molecules ‘moved’ into a second solvation layer on
optimization (out of contact with the ARZ molecule and with no bridging role
between water molecules attached to it), those molecules were removed and the
resulting input was optimised as a new adduct.
The selection of the number/s of water molecule in the adducts likely to better
contribute the desired information is a rather delicate issue. Too small a number
would not enable the incorporation of the effects of water-water interactions
286 L. Mammino

relevant for the first solvation layer, such as the stabilising effect of a water
molecule bridging two water molecules H-bonded to ARZ. On the other hand, the
tendency of water molecules to cluster together limits the number of water mole-
cules in an adduct, if one wishes them to ‘remain’ in the first solvation layer on
optimisation. For the case of ARZ, it was found that when more than 9–10 water
molecules are present, their tendency to cluster becomes dominant and several of
them may move away from their initial binding sites of ARZ, yielding arrangements
in which they ‘crowd’ in the vicinity of only a portion of ARZ, and one or more of
them may move beyond the first solvation layer.

2.2 Computational Approaches

All the adducts were calculated in vacuo, performing optimisation with fully
relaxed geometry at the same levels utilised in all the previous calculations on
ACPLs [6–10, 16, 17, 19–25], i.e., Hartree Fock (HF) with the 6-31G(d,p) basis set
and Density Functional Theory (DFT) with the B3LYP functional [26–28] and the
6-31+G(d,p) basis set. The reasons for the selection of the two levels of theory and
basis sets are explained in the previous works [6–10, 16–24]; it is considered
important to maintain them in this and further studies involving ACPLs, to enable
informative and straightforward comparisons.
In the previous studies [5–10], DFT calculations have mostly been performed as
post-HF calculations; random testing had shown that the same inputs optimise to
the same conformers with the two methods; thus, treating DFT as post-HF calcu-
lations was expedient to decrease computational costs. In the case of the adducts of
ARZ considered here, HF and DFT calculations were performed independently for
each input because the presence of several H-bond donor and acceptor sites and the
non-covalent nature of the solute-solvent H-bonds suggests the possibility that the
two methods may lead to different arrangements of water molecules. In most cases,
the same input optimised to similar arrangements, but, in a number of cases, the
optimised adducts differed substantially. Such outputs were then utilised as inputs
for the other method, what enabled the consideration of additional geometries that
had not been envisaged on the initial input-preparation.
The interaction energy (ΔEarz-n ⋅ aq) between the ARZ molecule and the water
molecules bonded to it was calculated for each adduct. The general equation is [29]


ΔEarz−n ⋅ aq = Eadduct − Earz + n Eaq − ΔEaq−aq ð1Þ

where Eadduct is the energy of the adduct, n is the number of water molecules in the
adduct, Earz is the energy of the isolated ARZ conformer, Eaq is the energy of an
isolated water molecule and ΔEaq-aq is the overall interaction energy between water
molecules, resulting mainly from water-water H-bonds.
Adducts of Arzanol with Explicit Water Molecules … 287

Fig. 4 Example of an adduct of arzanol with 9 explicit water molecules and the system of the sole
water molecules in the same arrangement as in the adduct, used to calculate ΔEaq-aq

For adducts with only one water molecule, the equation becomes simply

ΔEarz−aq = Eadduct − Earz − Eaq ð2Þ

For adducts with n non-interacting water molecules, the equation would be

ΔEarz−n ⋅ aq = Eadduct − Earz + n Eaq ð3Þ

For adducts with interacting water molecules (which is the most common case
when there are several water molecules), ΔEaq-aq is evaluated through a single point
(SP) calculation on a group of water molecules arranged exactly as in the adduct,
but without the ARZ molecule ([29], Fig. 4). If Eaq-set is the energy of this set of
water molecules, then

ΔEaq−aq = Eaq−set − n Eaq

and Eq. (1) becomes

ΔEarz−n ⋅ aq = Eadduct − Earz − Eaq−set ð4Þ

Both Eadduct and Eaq-set have been corrected for basis set superposition error
(BSSE), using the counterpoise method [30], when the adduct contains more than
one water molecule. The BSSE correction was not applied to adducts containing
only one water molecule because it would be small for these adducts and because
these calculations are meant for comparisons. The values for the calculated adducts
with four or more water molecules show that the BSSE correction increases sub-
stantially as the number of water molecules in the adduct increases; therefore, it will
be smallest for adducts with one water molecule (also in view of the absence of
288 L. Mammino

water-water interactions). The main objective of calculating these adducts was that
of enabling comparison of the binding strengths of individual sites. Neglecting
BSSE corrections does not appear to greatly affect comparisons among analogous
molecular systems when the correction is sufficiently small [31] (all the adducts
with one water molecule consist of one ARZ molecule and one water molecule, so,
they are analogous molecular systems). It was thus assumed that the comparisons
among these adducts remain reliable also without BSSE correction. Neglecting the
correction proved expedient also in view of the high number of calculated adducts
of this type.
All the calculations were performed with GAUSSIAN 03, Revision D 01 [32].
All the energy values reported are in kcal/mol and all the distances are in Å. For
conciseness sake, the calculation methods will be denoted simply as HF for HF/
6-31G(d,p) and DFT for DFT/B3LYP/6-31+G(d,p) in the text.
Detailed information, including tables with all energy values and H-bond
parameters for the calculated adducts, and figures showing the geometries of the
calculated adducts, is provided in the Supplementary Information.

3 Results

3.1 Naming of Conformers and Adducts

Following a practice introduced since the initial studies of ACPLs [6–10], the
conformers are denoted by acronyms which provide information about their
characteristics, to enable easy tracking of geometric characteristics and energy-
influencing features. The acronyms for the ARZ conformers start with the number
denoting the mutual orientation of the moieties. The other geometry features are
denoted by letters, whose meanings are listed in Table 1 [5]. The two lowest energy
conformers and their acronyms are shown in Fig. 3.
In order to identify them in a straightforward way, the adducts are denoted by the
acronym of the ARZ conformer at their centre, followed by information about the
water molecules. For the adducts with one water molecule, the name of the con-
former is followed by ‘1aq’ and by the position to which the water molecule is
attached. Thus, 1-d-r-ξ-αδ-1aq-O14 denotes the adduct of the 1-d-r-ξ-αδ conformer
in which one water molecule is attached to O14; 1-d-r-αδ-1aq-H17-π1 denotes the
adduct of 1-d-r-αδ in which a water molecule is attached to H17 and is also
interacting with the π bond of the prenyl chain; and so on. The adducts containing
more than one water molecule are denoted with the name of the conformer followed
by the number of molecules in the adduct. For instance, 1-d-r-ξ-αδ-5aq denotes an
adduct of conformer 1-d-r-ξ-αδ with 5 water molecules. Since different arrange-
ments of the water molecules are possible for adducts of the same conformer and
with the same number of water molecules, letters are added at the end of the
acronym to distinguish them from one another. However, the information on the
site to which each water molecule is attached would be too bulky to summarise it in
an acronym, and 3D models showing the arrangements of the water molecules are
Adducts of Arzanol with Explicit Water Molecules … 289

Table 1 Symbols utilised to specify geometrical characteristics in the acronyms denoting the
conformers. The symbols d, s, r, w, u, η and ξ have the same meanings as in other studies on
ACPLs [6–10] and the symbols α, β, γ, δ, ε and τ had been introduced in [5]
Symbol Meaning
d The H15⋯O14 first IHB is present
s The H17⋯O14 first IHB is present
r H16 is oriented towards the α-pyrone ring
w H16 is oriented towards the prenyl chain
u H15 or H17, not engaged in the first IHB, is oriented toward the COR group
(‘upwards’)
η Presence of O-H⋯π interaction between H16 and the C29=C30 double bond
ξ Presence of O-H⋯π interaction between H17 and the C29=C30 double bond
a No O-H⋯π interaction is present, and the prenyl chain is oriented ‘upwards’
b No O-H⋯π interaction is present, and the prenyl chain is oriented ‘downwards’
α The H27⋯O8 intermonomer hydrogen bond is present
β The H15⋯O26 intermonomer hydrogen bond is present
γ The H15⋯O23 intermonomer hydrogen bond is present
δ The H16⋯O23 intermonomer hydrogen bond is present
ε The H16⋯O26 intermonomer hydrogen bond is present
τ The H27⋯O10 intermonomer hydrogen bond is present
π1 The C29=C30 double bond, when acting as binding site for a water molecule

the only option to convey a clear image of each adduct (the 3D models are included
in the Supplementary Information).

3.2 Results for the Adducts of Arzanol with One Water

Molecule

Given the high number of conformers of the ARZ molecule, and the high number of
sites to which a water molecule can bind (including the possibility of simultaneous
binding to two geometrically suitable sites), considering all the possible adducts of
this type would be unaffordable. It was opted to calculate a selection of sufficiently
representative adducts for each binding site. While some sites (e.g., O14 or O23)
are accessible in all conformers, other sites are not available in some conformers,
and the combinations for simultaneous binding to two sites depend on the type of
conformer; therefore, only a limited number of adducts may be obtainable for
certain combinations. Changes during optimisation reduce the number of adducts
for some sites while increasing it for others (which informs that the former sites or
site-combinations are less favourable than others).
A total of more than 200 adducts were calculated. Figure 5 shows the main
geometries obtained, according to the type of conformer of the ARZ molecule and
to the site to which the water molecule binds. All the geometries are shown in the
290 L. Mammino

1-d-r-ξ-αδ-1aq-O14 1-d-r-ξ-αδ-1aq-O14-O26 1-d-r-ξ-αδ-1aq-O26

1-d-r-ξ-αδ-1aq-O23 1-d-r-ξ-αδ-1aq-O10 1-d-r-ξ-αδ-1aq-O12

3-d-r-ξ-ε-1aq-H27 1-d-r-αδ-1aq-H17-π1 1-d-w-α-1aq-H17 3-s-w-η-γτ-1-aq-O14

3-s-w-η-γτ-1-aq-O12 1-s-r-u-δ-1aq-O12-π1 1-s-r-δ-1aq-H15-O26

3-s-w-a-γτ-1aq-O8 1-s-r-u-δ-1aq-H15 1-s-r-βδ-1aq-H27-O8 3-s-w-η-γ-1aq-O10-H27

Fig. 5 Representative adducts of arzanol with one water molecule attached to different donor or
acceptor sites. The first two rows show adducts of the lowest energy conformer of arzanol
(1-d-r-ξ-αδ), the other rows show other relevant types of arrangement of the water molecule,
for different types of conformers

Supplementary Information. Nearly all the geometries are obtained with both HF
and DFT, with similar shapes.
In many cases, the optimisation does not change the site to which the water
molecule binds; it is the case of inputs in which it binds to O14 or to O23, but also
to other favourable sites. There are, however, also several cases in which the
Adducts of Arzanol with Explicit Water Molecules … 291

Table 2 Examples of changes occurring during optimization, highlighting the binding prefer-
ences of the water molecule
Reference Input geometry Output geometry
acronym
ch-1 1-d-r-ξ-αδ-1aq-O19-O23 1-d-r-ξ-αδ-1aq-O23
ch-2 1-d-r-ξ-αδ-1aq-O8 1-d-r-ξ-αδ-1aq-O14
ch-3 3-s-r-b-γε-1aq-O12 3-s-r-b-γε-1aq-O14
ch-4 3-s-w-η-γτ-1aq-O8 3-s-w-η-γτ-1aq-O23
ch-5 1-d-r-b-αδ-1aq-O8 1-d-r-b-αδ-1aq-O26
ch-6 3-d-w-η-τ-1aq-O8 3-d-w-η-τ-1aq-O8-O23
ch-7 3-d-w-ξ-1aq-O23 3-d-w-ξ-1aq-O8-O23
ch-8 3-d-r-ξ-ε-1aq-O26 3-d-r-ξ-ε-1aq-H27
ch-9 3-d-w-η-τ-1aq-O12 3-d-w-η-τ-1aq-H17
ch-10 3-d-w-ξ-τ-1aq-O10 3-d-w-ξ-τ-1aq-H16
ch-11 1-s-r-βδ-1aq-O8 1-s-r-βδ-1aq-O8-H27
ch-12 1-s-r-βδ-1aq-O26 1-s-r-βδ-1aq-O8-H27
ch-13 3-d-w-η-τ-1aq-O26 3-d-w-η-τ-1aq-O26-H16
ch-14 3-d-r-u-b-ε-1aq-O26 3-d-r-u-b-ε-1aq-O10-H27
ch-15 1-s-r-u-δ-1aq-O8-O26 1-s-r-u-βδ-1aq-O8-H27
ch-16 3-s-r-γε-1aq-O10-π1 3-s-r-γε-1aq-O10-H27
ch-17 3-s-w-η-γ-1aq-O10-O26 3-s-w-η-γ-1aq-O10-H27
ch-18 1-d-w-ξ-α-1aq-O10-O23 1-d-ξ-α-1aq-H16-O23
ch-19 1-d-w-η-α-1aq-O10-O23 1-d-w-η-α-1aq-H16-O23
ch-20 1-s-w-u-η-α-1aq-H15 1-s-w-u-η-α-1aq-H15-O26
ch-21 3-s-w-u-η-τ-1aq-H15 3-s-w-u-η-τ-1aq-H15-O23
ch-22 1-d-w-α-1aq-H16 1-d-w-α-1aq-H16-O23
ch-23 1-s-r-βδ-1aq-H27 1-s-r-βδ-1aq-O8-H27
ch-24 1-d-w-ξ-1aq-H27 1-d-w-ξ-1aq-O14-H27
ch-25 3-s-w-η-γ-1aq-H27 3-s-w-η-γ-1aq-O10-H27
ch-26 1-d-w-ξ-α-1aq-O12 1-d-w-α-1aq-O12
ch-27 1-d-r-b-αδ-1aq-O10-π1 1-d-r-ξ-αδ-1aq-O10’
ch-28 1-d-w-η-α-1aq-O10-O23 1-d-w-ξ-α-1aq-O12
ch-29 1-d-r-b-αδ-1aq-H17 1-d-r-αδ-1aq-H17-π1
ch-30 3-s-w-a-γτ-1aq-H16 3-s-w-γτ-1aq-H16-π1
ch-31 3-d-w-τ-1aq-H16-π1 3-d-w-ξ-τ-1aq-H16
ch-32 1-d-w-1aq-H17-π1 3-d-w-η-τ-1aq-H17’
ch-33 3-s-w-a-γτ-1aq-O12-π1 3-s-w-η-γτ-1aq-O12
ch-34 3-d-r-b-ε-1aq-O10-π1 3-d-r-ξ-ε-1aq-O10-H27

optimisation changes the site to which the water molecule binds, with respect to the
input. Since these changes highlight binding preferences, they are given detailed
attention in the current analysis. Examples are reported in Table 2, considering
types of changes which appeared more than once. The changes are denoted with
292 L. Mammino

acronyms (‘ch’, which stands for ‘change’, followed by a number) to enable easy
referencing to them within this text.
The water molecule appears to prefer to bind to an sp2 O rather than to an sp3 O.
Thus, inputs in which the water molecule is attached to O19, or to both O19 and
O23, optimize to adducts with the water molecule attached to O23 (ch-1). Inputs in
which the water molecule is attached to the donor of the first IHB (O8 or O12, for d
and s conformers respectively), in conformers in which O14 is the only sp2 O in the
vicinity, often optimize to adducts in which water binds to O14 (ch-2, ch-3). If O23
is also available in the vicinity (e.g., in inputs involving 3-d conformers), the water
molecule often shifts from O8 to O23 (ch-4) or, sometimes, to O26 (ch-5). These
changes may be related to the hydrophobic character of IHB regions for hydrox-
ybenzenes in general [33], and of the first IHB of ACPLs in particular [17].
A clear preference appears for the water molecule to bind to two sites simul-
taneously, when two sites are geometrically suitable; this may lead to adducts in
which it binds to an sp2 O and an sp3 O simultaneously (e.g., O8 and O23; ch-6,
ch-7). The water molecule also appears to prefer to be acceptor to an OH group
(consistently also with the known tendency of phenol OHs to be donors in inter-
molecular H-bonds [16, 31]). This may involve rotation of H15, H16, H17 or H27,
to enable the formation of the H-bond with the water molecule (it is interesting to
recall that the phenol OHs in ACPLs do not usually rotate to form IHBs [16],
whereas the adducts calculated here show that the OHs in ARZ may rotate to form
an intermolecular H-bond with the solvent). The water molecule may change
binding site completely in order to be acceptor to the H of an OH (ch-8, ch-9,
ch-10). On the other hand, its tendency to bind to two geometrically suitable sites
simultaneously may lead to adducts in which it binds simultaneously to the H of an
OH and to a suitably close O atom (ch-11 to ch-19). The tendency to bind to two
sites appears also for inputs in which the water molecule is initially placed as
acceptor to an OH (ch-20 to ch-25).
The water molecule does not break O-H⋯O IHBs. It appears, however, to be
able to act on the O-H⋯π interaction, by breaking it (ch-26) or prompting it
(ch-27), or changing its pattern (ch-28, in which it changes from η to ξ). The water
molecule itself may interact with the C29=C30 π bond; the interaction usually
appears during optimisation (ch-29, ch-30), whereas inputs having the interaction
do not often optimise to adducts in which it is maintained (ch-31 to ch-34).
The changes just outlined appear both with HF and with DFT optimization. In
most cases, the same change occurs with the same input, i.e., HF and DFT opti-
misations lead to the same changed output. In some cases, the changes are different
with the two methods, leading to different outputs. Figure 6 shows a case in which
HF and DFT lead to outputs that are different from the input and different from each
other. In such cases, it is not easy to estimate a priori which of the two outcomes
might be closer to reality; for the specific case shown in Fig. 6, the HF result seems
more probable, as it shows preference for an sp2 O, which can form a stronger
H-bond than an sp3 O like O8 or O26.
Table 3 reports the relative energy and the ARZ-water interaction energy for
representative adducts, selected in such a way as to comprise nearly all the
Adducts of Arzanol with Explicit Water Molecules … 293

HF DFT

Fig. 6 An example in which the HF and DFT optimisations of the same input with one water
molecule lead to different results. In the input, the water molecule was attached to O8. The HF
optimisation moves it to O14 and the DFT optimisation moves it to O26. The conformer is the
lowest energy conformer of arzanol (1-d-r-ξ-αδ)

identified binding options of the water molecule. The relative energy depends to a
considerable extent on the type of conformer and its relative energy in vacuo,
whereas the interaction energy depends largely on the binding site (although the
geometry of the conformer influences the approach of the water molecule to a given
binding site). Table 4 reports the ranges of the interaction energy for the different
binding sites. The interaction energy values confirm the binding preferences of the
water molecule highlighted by the changes occurring during optimisation, such as
the preference for simultaneous binding to two atoms of ARZ (and, among these,
for one of the sites being a donor OH), and the preference for H atoms of OH
groups, followed by sp2 O atoms, when it binds only to one site.
Table 5 reports the ranges of the length of the ARZ-water H-bonds. The binding
sites are listed in the same sequence as in Table 4 to facilitate comparison in terms
of length of the H-bond and strength of the interaction energy (a few sites present in
Table 4 are not reported in Table 5 for space reasons). The correspondence is
meaningful because, for adducts with only one water molecule, the molecule-water
interaction energy can be viewed as the energy of the molecule-water H-bond, and
an H-bond length is an indication of its strength. The H-bond lengths are shorter
when the water molecule is acceptor to an OH group of ARZ. The H-bond lengths
for sp2 O are shorter than those of sp3 O, consistently with the ability of sp2 O to
form stronger H-bonds. Although the trends (comparison of lengths across adducts)
are largely similar in the HF and DFT results, the HF values are longer than the
DFT values for the same adducts. This is consistent with the known tendency of HF
to underestimate the strength of H-bonds and of DFT to overestimate it; therefore, it
appears reasonable to assume that the actual H-bond distance for a given adduct
will be somehow intermediate between the HF and the DFT values.
Table 3 Relative energy and arzanol-water interaction energy of representative adducts of arzanol with one explicit water molecule, in the HF/6-31G(d,p) and
294

DFT/B3LYP/6-31+G(d,p) results in vacuo

HF/6-31G(d,p) results DFT/B3LYP/6-31+G(d,p) results
Adduct Relative energy Interaction energy Adduct Relative energy Interaction energy
(kcal/mol) (kcal/mol) (kcal/mol) (kcal/mol)
1-d-r-ξ-αδ-1aq-O23 0.000 −6.523 1-d-r-ξ-αδ-1aq-O23 0.000 −6.521
1-d-r-ξ-αδ-1aq-O14 0.111 −5.502 1-d-r-ξ-αδ-1aq-O14 0.814 −5.707
1-d-r-αδ-1aq-H17-π1 1.210 −11.926 1-d-r-ξ-αδ-1aq-O10 1.300 −5.221
1-d-r-b-αδ-1aq-O12 1.357 −5.259 1-d-r-αδ-1aq-H17-π1 1.511 −7.511
3-s-w-η-γτ-1aq-O23 1.453 −6.412 3-s-w-a-γτ-1aq-H16-π1 2.129 −9.689
1-d-r-ξ-αδ-1aq-O10 1.575 −5.029 1-d-r-ξ-αδ-1aq-O26 2.329 −4.192
3-s-w-γτ-1aq-H16-π1 1.788 −8.695 3-s-w-η-γτ-1aq-O12 3.066 −5.430
1-d-r-ξ-αδ-1aq-O14-O26 2.915 −3.689 1-d-w-η-α-1aq-O23-H16 4.911 −6.850
3-s-w-η-γτ-1aq-O26 3.075 −4.790 3-s-w-u-η-τ-1aq-O23-H15 5.449 −16.818
1-d-w-ξ-α-1aq-O23-H16 4.256 −15.101 1-d-r-u-b-αδ-1aq-H17 5.819 −6.722
3-s-r-γε-1aq-O10-H27 4.526 −12.600 3-s-r-γε-1aq-O10-H27 6.463 −12.089
3-s-w-u-η-τ-1aq-O23-H15 4.755 −17.118 3-s-w-a-γτ-1aq-O8 7.358 −4.461
3-s-r-γε-1aq-O10-π1 5.510 −11.616 3-s-r-u-ε-1aq-O10-H27 7.484 −11.111
3-s-r-b-γε-1aq-O10-O26 5.510 −11.601 3-d-w-η-1aq-O10-O26 9.467 −6.220
3-s-w-a-γτ-1aq-O8 6.232 −4.250 1-s-r-βδ-1aq-O8-H27 9.477 −9.879
3-s-w-η-γ-1aq-O10-H27 7.285 −9.8685 3-s-w-η-γ-1aq-H27 10.687 2.191
1-s-r-βδ-1aq-O8-H27 8.020 −10.144 1-r-ξ-βδ-1aq-O14-H27 12.480 −19.134
1-d-w-α-1aq-H17 11.500 −7.857 3-s-r-b-γε-1aq-O12-π1 13.431 −5.165
3-s-r-γε-1aq-O12-π1 12.038 −5.088 1-s-w-u-η-α-1aq-O26-H15 14.059 −8.849
1-s-w-u-η-α-1aq-O26-H15 12.622 −10.546 3-d-w-ξ-τ-1aq-H16 15.863 −9.941
(continued)
L. Mammino
Table 3 (continued)
HF/6-31G(d,p) results DFT/B3LYP/6-31+G(d,p) results
Adduct Relative energy Interaction energy Adduct Relative energy Interaction energy
(kcal/mol) (kcal/mol) (kcal/mol) (kcal/mol)
3-w-ξ-γτ-1aq-H16 15.167 −8.338 1-s-r-u-αδ-1aq-H15 17.161 −8.936
3-d-w-η-τ-1aq-O8-O23 15.354 −7.793 3-d-w-u-η-τ-1aq-O8-O23 17.971 −7.504
3-d-w-η-τ-1aq-O26-H16 16.297 −6.850 1-s-r-u-δ-1aq-O8-O26 18.197 −6.378
1-d-w-η-1aq-O8-H27 17.040 −12.144 3-d-w-ξ-τ-1aq-O8-O23 18.246 −7.558
1-d-w-η-1aq-O14-H27 20.485 −8.699 3-w-ξ-γτ-1aq-O12-O14 19.408 −7.010
1-r-ξ-βδ-1aq-O12-O14 20.620 −7.202 1-d-w-η-1aq-O23-H16 19.972 −9.698
Adducts of Arzanol with Explicit Water Molecules …
295
296 L. Mammino

Table 4 Ranges of the magnitude (absolute values) of the arzanol-water interaction energy for the
adducts of arzanol with one explicit water molecule, in the HF/6-31G(d,p) and DFT/B3LYP/6-31
+G(d,p) results in vacuo. When only two values are available, they are reported individually,
separated by a comma. When only one value is available, it is reported individually
Binding HF range DFT range Binding HF range DFT range
site (kcal/mol) (kcal/mol) site (kcal/mol) (kcal/mol)
O23-H15 15.4, 17.1 15.2, 16.8 H16-π1 7.2–8.7 7.3–9.9
O10-H27 7.6–15.6 9.3–12.1 O12 3.2–8.1 3.0–8.5
O23-H16 10.2–15.3 9.2–16.3 O8-O23 6.5–7.9 7.2–7.8
O8-H27 10.1–13.5 9.9 O23 3.8–7.8 4.0–11.1
O8-O26 12.3 5.4, 5.4 O14-O23 7.8 7.0
H17-π1 7.3–11.9 7.0–8.8 O12-O14 7.2, 7.5 7.5–7.8
O10-π1 6.1, 11.6 O14 3.9–7.1 3.7–7.4
O10-O26 11.6 6.1, 6.2 O8-O14 7.1 7.0
O26-H15 9.5–10.5 6.1–8.9 O10 3.9–7.0 5.2–8.2
H15 8.4–10.2 7.2 O26-H16 6.9
O10-O23 10.1 7.0–9.9 O12-π1 5.1–5.3 5.1–7.5
H27 5.7–10.0 6.6–11.2 O26 2.4–5.2 3.5–8.0
H16 8.3–9.5 7.6–9.9 O12-H17 4.6
O14-H27 8.7, 9.2 19.1 O8 4.3–4.4 4.5–5.1
H17 6.7–9.1 6.7–7.5 O14-O26 3.7

3.3 Results for the Adducts of Arzanol with More Than One
Water Molecule

A total of 84 adducts of ARZ with 4–12 water molecules were calculated, with
greater number of adducts with 7, 8 or 9 water molecules. What the calculation of
such adducts may contribute in a study of this type is not an exhaustive inclusion of
all possible options, but a sufficiently representative selection highlighting the
variety of possibilities and some recurrent patterns. The geometries present a high
variety of possible arrangements of the water molecules around the ARZ molecule.
Figure 7 shows some geometries of adducts with different numbers of water
molecules.
Figure 8 illustrates some relevant aspects in the arrangement of water molecules,
including some recurrent patterns. The water molecules tend to keep away from the
region of an IHB, as it is a hydrophobic region [17, 33]. Shapes already encoun-
tered for adducts of other ACPLs [17] appear with a certain frequency, such as a
pentagon of O atoms formed by the atoms engaged in an IHB and the O atoms of
the water molecules keeping away from it. The whole region of two cooperative
IHBs appears to be hydrophobic, and the water molecules may form a larger ring
(e.g., seven O atoms) while keeping away from it.
Differently from the adducts with one water molecule, a water molecule may
break an O-H⋯O IMHB and insert itself between the donor and the acceptor; the
Adducts of Arzanol with Explicit Water Molecules … 297

Table 5 Ranges of the length of the hydrogen bonds between the arzanol molecule and the water
molecule for selected binding sites in the calculated adducts of arzanol with one water molecule.
When only two values are available, they are reported individually, separated by a comma. When
only one value is available, it is reported individually
Binding site/s Length considered Range of values (Å)
HF results DFT results
O23, H15 Oaq⋯H15 1.878, 1.892 1.713, 1.754
Haq⋯O23 1.918, 1.953 1.747, 1.819
O10, H27 Oaq⋯H27 1.913–2.229 1.784–2.174
Haq⋯O10 1.983–2.298 1.911–2.282
O23, H16 Oaq⋯H16 1.851–2.190 1.690–1.751
Haq⋯O23 1.904–2.164 1.731–1.984
O8, H27 Haq⋯O8 1.966–2.125 –
Oaq⋯H27 1.877–1.932 –
O8, O26 Haq⋯O8 – 2.177–2.401
Haq⋯O26 – 2.033–2.177
H17, π1 H17⋯ Oaq 1.885–1.901 1.792–1.848
O26, H15 Oaq⋯H15 1.943–1.991 1.871–1.930
Haq⋯O26 2.365–2.429 2.035–2.199
H15 Oaq⋯H15 1.909–2.030 1.888–1.895
H27 Oaq⋯H27 1.890–2.009 1.822–1.933
H16 Oaq⋯H16 1.926, 1.935 1.905, 1.914
O14, H27 Haq⋯O14 2.253, 2.332 –
Oaq⋯H27 2.011, 2.025 –
H17 Oaq⋯H17 1.939–2.005 1.873–1.941
H16, π1 Haq⋯O10 1.879–1.901 1.752–1.870
O12 Haq⋯O12 2.132–2.240 1.931–2.060
O8, O23 H′aq⋯O23 2.120–2.158 1.965–2.037
Haq⋯O8 2.311–2.378 2.130–2.351
O23 Haq⋯O23 2.038–2.178 1.911–2.024
O14, O23 H′aq⋯O23 2.129, 2.182 2.008
Haq⋯O14 2.573, 2.604 2.662
O12, O14 H′aq⋯O14 2.217, 2.289 2.438–2.570
Haq⋯O12 2.285, 2.309 2.000–2.056
O14 Haq⋯O14 2.040–2.089 1.870–1.929
O8, O14 H′aq⋯O14 2.209 2.097
Haq⋯O8 2.366 2.220
O10 Haq⋯O10 2.106–2.158 1.952–2.023
O12, π1 Haq⋯O12 2.366–2.478 1.837–2.145
O26 Haq⋯O26 2.152–2.257 1.987–2.052
O8 Haq⋯O8 2.087–2.112 1.926–1.957
298 L. Mammino

1-d-r-ξ-αδ-4aq 1-d-r-ξ-αδ-5aq 3-d-r-b-ε-5aq-x

1-d-α-5aq-x 2-d-r-αδ-6aq 1-d-r-ξ-αδ-6aq-x

2-d-w-η-α-6aq 3-d-r-ξ-ε-6aq-x 2-d-r-αδ-7aq

1-d-r-ξ-αδ-7aq 1-d-r-ξ-αδ-7aq-e-x 1-s-r-u-a-αδ-7aq

1-s-r-u-a-αδ-8aq 2-d-r-δ-8aq 3-s-w-b-γτ-9aq-e

Fig. 7 Illustration of the variety of possible geometries for the arrangement of water molecules in
the adducts
Adducts of Arzanol with Explicit Water Molecules … 299

2-s-r-βδ-9-aq-x 2-d-w-η-α-6aq 4-s-r-a-γε-8aq-x

3-s-w-a-γτ-9aq 3-s-w-b-γτ-9aq-e 3-s-w-b-γτ-11aq

Fig. 8 Some relevant aspects in the arrangement of water molecules around the arzanol molecule.
The water molecules tend to keep away from the IHB regions; a frequent result is a pentagon of O
atoms (including those of the IHB), as in 2-s-r-βδ-9-aq-x, 4-s-r-a-γε-8aq-x and 3-s-w-a-γτ-9aq; in
some cases, the ring may contain more O atoms, as the ring around two consecutive IHBs in
2-d-w-η-α-6aq. A water molecule may break an inter-monomer H-bond, as in 3-s-w-a-γτ-9aq,
where it breaks the H27⋯O10 IHB. In a number of cases, the tendency of the water molecules to
cluster together fosters arrangements in which some donor or acceptor sites of the arzanol
molecule are not binding a water molecule, as in 3-s-w-b-γτ-9aq-e (where no water molecule
attaches to O14, although it is a strong acceptor) and in 3-s-w-b-γτ-11aq. Although in adducts with
one or few water molecules no water molecule binds to O19, it may happen that a water molecule
binds to it if there are enough bridging water molecules to facilitate the arrangement, as in
3-s-w-b-γτ-11aq

cases observed concern IMHBs where H27 is the donor (H27⋯O8 or H27⋯O10).
The length of the H-bond between H27 and the O of the water molecule attached to
it is the shortest ARZ-water H-bond length in these adducts, suggesting that this
H-bond might be among the strongest if not the strongest. No breaking has been
observed for the first IHB (which is considerably stronger than H27⋯O8 or
H27⋯O10), consistently with the results for the adducts of other ACPLs [16, 17].
For similar reasons (the acceptor being an sp2 O), no breaking is observed for
IMHBs in which O23 is the acceptor. The O-H⋯π interaction is broken in a number
of cases, with a water molecule inserting itself between the H and the C29=C30
π-bond. The tendency of the water molecules to cluster together may prompt major
changes during optimization, leading to chains of water molecules alternatingly
binding to a site of ARZ and playing bridging roles.
The relative energies of adducts with the same number of water molecules
depend on their geometrical arrangement and binding sites more than on the relative
energy of the isolated conformer. The BSSE correction increases as the number of
300 L. Mammino

Table 6 Ranges of the BSSE correction to the energy of the adduct and to the energy of the
system of the sole water molecules, and percentage of the latter correction with respect to the
former, for the calculate adducts of arzanol with explicit water molecules. The ranges are
considered according to the number of water molecules in the adduct
Number of BSSE correction BSSE correction range for Range of percentage
water range for the adduct the sole water molecules of the contribution
molecules (kcal/mol) (kcal/mol) of water
HF DFT HF DFT HF DFT
5 8.50–11.05 5.15–6.46 0.95–3.99 0.99–1.84 9.63–47.25 16.56–55.52
6 10.32–12.84 6.14–8.11 2.25–6.64 1.61–3.51 17.84–53.74 25.17–50.87
7 12.52–15.45 7.34–9.71 3.00–7.70 2.65–5.83 21.00–52.65 31.93–62.99
8 12.45–18.28 8.43–11.17 4.85–8.91 4.17–6.01 30.57–57.59 45.51–71.32
9 14.62–21.96 8.94–13.06 5.53–11.12 4.48–7.99 36.58–57.60 50.11–64.28
10 19.87–20.42 11.16–14.52 11.86–12.29 7.25–9.39 50.84–61.86 60.18–64.97
11 20.00–25.85 14.68–15.70 9.23–13.97 9.37–10.52 45.41–55.07 53.68–67.01
12 24.71–25.77 15.81–16.51 11.46–12.62 8.42–9.75 46.31–49.18 53.30–60.65

water molecules in the adduct increases, although it can vary broadly for adducts
with the same number of water molecules. The ranges of values are somewhat
narrower in the DFT results and somewhat broader in the HF results (Table 6). The
correction is often greater for adducts with lower relative energy.
The ARZ-water interaction energy depends on the conformer of ARZ, on the
sites to which the water molecules bind and on the number of water molecules. The
dominant factor appears to be the arrangement of the water molecules, which
includes the binding sites and also the types of water-water interactions. Table 7
reports the ranges of the ARZ-water interaction energy as identified from the cal-
culated adducts; the values show a tendency to higher upper-limit of the range as
the number of water molecules increases, but without a straightforward relationship
between the two. The dominance of qualitative aspects (characteristics of the ARZ
conformer, types of binding sites, geometry of the water molecules arrangements)
may hamper the possibility of identifying more definite types of relationship. The
magnitude of the DFT values tend to be greater than that of the HF values for
corresponding conformers, which may be ascribed to the tendency of HF and DFT
to respectively underestimate and overestimate the strength of H-bonds.
The length of the ARZ-water H-bonds depends on the binding site and appears to
be fairly consistent with the patterns highlighted by the adducts with one water
molecule, although the other water molecules binding to a water molecule attached to
ARZ may influence the length of its H-bond with the ARZ site. This is the case of
H-bonds between H27 and the O of a water molecule, which are the shortest in
adducts with several water molecules, but not in the adducts with one water molecule.
Adducts of Arzanol with Explicit Water Molecules … 301

Table 7 Ranges of the magnitude of the arzanol-water interaction energy (corrected for BSSE)
according to the number of water molecules in the adduct
Number Magnitude of the Number Magnitude of the
of water interaction energy of water interaction energy
molecules (kcal/mol) molecules (kcal/mol)
HF DFT HF DFT
5 13.64–30.52 15.82–36.70 9 16.14–38.63 21.10–45.31
6 13.40–35.77 17.93–39.47 10 16.79–24.24 18.92–33.62
7 6.32–35.95 8.64–48.69 11 22.16–37.55 24.23–47.61
8 11.97–41.16 15.16–41.93 12 26.13–39.39 31.12–57.73

3.4 Results in Water Solution for the Adducts of Arzanol

with More Than One Water Molecule

When solute–solvent intermolecular H-bonds are possible, the combination of

explicit consideration of those solvent molecules that are expected to be more
closely linked to the solute molecule (as those H-bonded to it or bridging them), and
a bulk effect for the rest of the solvent, can provide a better picture of the situation
in solution than the sole PCM calculations on the central molecule, while remaining
within QM calculations affordability because of the limited number of discrete
solvent molecules. Calculations in water solution were performed on selected
adducts considering the entire adduct as a solute and utilising the PCM model. They
were performed as SP calculations, because of affordability reasons in view of the
size of the supermolecular structures of the adducts.
The results show a decrease in the relative energy for most adducts, with greater
decrease for adducts having higher relative energy in vacuo. This is consistent with
common behaviours for isolated molecules. Few exceptions may appear for lower
energy adducts, more frequently in the DFT results. The identification of the lowest
energy adduct may differ from that in vacuo, but remains among the lower energy
adducts in vacuo. No significant patterns for the energy decrease can be identified in
terms of types of the conformers of ARZ.
The solvent effect (free energy of solvation, ΔGsolv) is mostly negative in the
DFT results; in the HF results, it is mostly negative for adducts with smaller
numbers of water molecules and positive for several adducts with higher numbers
of water molecules. Given the expectation that a molecular unit incorporating water
molecules in its outer region would have some solubility in water, the DFT results
are likely more realistic. The electrostatic component of ΔGsolv (Gel) is always
negative, with values always smaller than −20 kcal/mol; its magnitude varies rather
randomly with adducts of different conformers, but shows a tendency to an average
increase as the number of water molecules increases.
302 L. Mammino

4 Discussion and Conclusions

A study of the type considered in this work involves a variety of challenges because
of the nature of the adducts and of the character of the information that can be
obtained.
The high number of conformers of the ARZ molecule and the high number of its
H-bond donor or acceptor site implies an enormous number of possible adducts
with a given number of water molecules, and this number increases rapidly as the
number of water molecules in the adducts increases. Calculations show high sen-
sitivity of the optimisation procedure to small differences in the inputs, which
would recommend the consideration of several adducts with similar or very similar
(but not identical) input geometries for each relevant geometry, thus further mul-
tiplying the number of potentially interesting adducts. On the other hand, the ten-
dency of water molecules to cluster on optimisation may yield similar adducts from
different inputs, which decreases the informative role of the result. It may also lead
to adducts where some water molecules cluster beyond the boundaries of the
adopted criterion for the definition of ‘first solvation layer’, resulting in high
water-water interaction energy and poor solute-water interaction energy; since the
latter phenomenon is more extensive as the number of water molecules in the
adduct increases, it prevents the possibility of considering a higher number of water
molecules than the one for which their clustering becomes extensive or dominant.
The huge number of adducts that would be needed to provide a comprehensive
panoramic taking into account all the relevant energy-influencing features (all the
geometrical features of all the conformers of ARZ, and all the possible arrange-
ments of water molecules around each conformer) would imply enormous com-
putational costs. Therefore, it was opted to select representative adducts for each
relevant characteristic. This corresponds to a sampling approach more than to an
unaffordable exhaustive approach. All the same, a sampling approach can be
informative for a variety of aspects.
Within the reality of a water solution, the supermolecular structures of the
adducts are not ‘fixed’ in time, because the water molecules H-bonded to a solute
molecule do not remain the same in time (there is continuous fast interchange with
the surrounding water molecules). Therefore, the calculated adducts represent
time-averaged probable possibilities rather than permanent structures.
Despite all these challenges, the calculation of adducts with explicit water
molecules provides information on the relative strength with which a water mole-
cule can bind to each donor or acceptor site of the solute molecule, on the distance
to which a water molecule preferably approaches each site, and on the preferred
arrangements of water molecules around each donor or acceptor sites or around the
region of two or more spatially close donors or acceptors. Since it results from
optimisation and H-bonds are directional, this information can be considered as
responding to the more common situations in the vicinity of the donors or acceptors
of the solute molecule. The adducts also provide indications about whether a certain
IHB tends to remain or to break in water solution—a type of information which
Adducts of Arzanol with Explicit Water Molecules … 303

may be relevant for other investigations, including the investigation of possible

mechanisms for the biological activity of the molecule.
For the specific case of ARZ, the adducts confirm that the binding of a water
molecule to the central molecule is stronger when it binds to two sites simultane-
ously, or when it is acceptor to an OH, or donor to an sp2 O. They also confirm that
the regions around IHBs are largely hydrophobic. They show the possibility that the
clustering of water molecules (when these remain within the first solvation layer)
may lead to a continuous frame of alternating water molecules H-bonded to ARZ
and water molecules bridging them—an arrangement with enhanced stability.
Finally, the results also contribute to the general ensemble of information on the
interactions of ACPLs with explicit water molecules and with water as a solvent.

References

1. Appendino G, Ottino M, Marquez N, Bianchi F, Giana A, Ballero M, Sterner O, Fiebich BL,

Munoz E (2007) J Nat Prod 70:608–612
2. Bauer J, Koeberle A, Dehm F, Pollastro F, Appendino G, Northoff H, Rossi A, Sautebin L,
Werz O (2011) Biochem Pharmacol 81:259–268
3. Rosa A, Pollastro F, Atzeri A, Appendino G, Melis MP, Deiana M, Incani A, Loru D, Dessì
MA (2011) Chem Phys Lipids 164:24–32
4. Singh IP, Bharate SB (2006) Nat Prod Rep 23:558–591
5. Mammino L, (2017) Molecules.22, 1294. https://doi.org/10.3390/molecules22081294
6. Mammino L, Kabanda MM (2009) J Mol Struct (Theochem) 901:210–219
7. Mammino L, Kabanda MM (2009) J Phys Chem A 113(52):15064–15077
8. Mammino L, Kabanda MM (2012) Int J Quant Chem 112:2650–2658
9. Kabanda MM, Mammino L (2012) Int J Quant Chem 112:3691–3702
10. Mammino L, Kabanda MM (2013) Molec Simul 39(1):1–13
11. Tomasi J, Persico M (1994) Chem Rev 94:2027–2094
12. Tomasi J, Mennucci B, Cammi R (2005) Chem Rev 105:2999–3093
13. Mennucci B (2010) J Phys Chem Lett 1:1666–1674
14. Alagona G, Ghio C (2002) Int J Quant Chem 90:641–656
15. Mammino L (2009) Chem Phys Letters 473:354–357
16. Mammino L, Kabanda MM (2007) J Mol Struct (Theochem) 805:39–52
17. Mammino L, Kabanda MM (2010) Int J Quant Chem 110(13):2378–2390
18. Mammino L (2017) J Mol Model. DOI. https://doi.org/10.1007/s00894-017-3443-4
19. Mammino L, Kabanda MM (2009) WSEAS Transact Biol Biomed 6(4):79–88
20. Mammino L, Kabanda MM (2008) Int J Quant Chem 108:1772–1791
21. Mammino L, Kabanda MM (2012) Int J Biol Biomed Engin 1(6):114–133
22. Mammino L (2013) Int J Biol Biomed Engin 2(7):15–25
23. Mammino L (2013) J Molec Model 19:2127–2142
24. Mammino L (2014) Curr Bioact Compd 10(3):163–180
25. Mammino L (2015) Current Phys Chem 5:274–293
26. Becke AD (1992) J Chem Phys 96:9489
27. Becke AD (1993) J Chem Phys 98:5648–5652
28. Lee C, Yang W, Parr RG (1998) Phys Rev B 37:785–789
29. Alagona G, Ghio C (2006) J Phys Chem A 110:647–659
30. Boys SF, Bernardi F (1970) Mol Phys 19:553
31. Alagona G, Ghio C (2009) J Phys Chem A 113:15206–15216
304 L. Mammino

32. Frisch MJ, Trucks GW, Schlegel HB, Scuseria GE, Robb MA, Cheeseman JR, Mont-
gomery JA, Vreven T, Kudin KN, Burant JC, Millam JM, Iyengar SS, Tomasi J, Barone V,
Mennucci B, Cossi M, Scalmani G, Rega N, Petersson GA, Nakatsuji H, Hada M, Ehara M,
Toyota K, Fukuda R, Hasegawa J, Ishida M, Nakajima T, Honda Y, Kitao O, Nakai H,
Klene M, Li X, Knox JE, Hratchian HP, Cross JB, Adamo C, Jaramillo J, Gomperts R,
Stratmann RE, Yazyev O, Austin AJ, Cammi R, Pomelli C, Ochterski JW, Ayala PY,
Morokuma K, Voth GA, Salvador P, Dannenberg JJ, Zakrzewski VG, Dapprich S,
Daniels AD, Strain MC, Farkas O, Malick DK, Rabuck AD, Raghavachari K, Foresman JB,
Ortiz JV, Cui Q, Baboul AG, Clifford S, Cioslowski J, Stefanov BB, Liu G, Liashenko A,
Piskorz P, Komaromi I, Martin RL, Fox DJ, Keith T, Al-Laham MA, Peng CY,
Nanayakkara A, Challacombe M, Gill PMW, Johnson B, Chen W, Wong MW, Gonzalez C,
Pople JA (2003) Gaussian 03. Gaussian Inc, Pittsburgh
33. Mammino L, Kabanda MM (2011) Int J Quant Chem 111:3701–3716
Computational Study of Jozimine A2,
a Naphthylisoquinoline Alkaloid
with Antimalarial Activity

Mireille K. Bilonda and Liliana Mammino

Abstract Jozimine A2 is a dioncophyllaceae-type naphthylisoquinoline alkaloid

isolated from the root bark of an Ancistrocladus species from the Democratic
Republic of Congo and exhibiting high antimalarial activity. It is the first naturally
occurring dimeric naphthylisoquinoline of this type to be discovered. Its molecule
consists of two identical 4′-O-demethyldioncophylline A units, with each unit
containing an isoquinoline moiety and a naphthalene moiety. A thorough confor-
mational study of this molecule was performed in vacuo and in three solvents with
different polarities and different H-bonding abilities (chloroform, acetonitrile and
water), using two levels of theory, HF/6-31G(d,p) and DFT/B3LYP/6-31+G(d,p).
Intramolecular hydrogen bond (IHB) patterns were investigated considering all the
possible options. Preferences for the mutual orientations of the moieties were
identified through the potential energy profiles for the rotation of the single bonds
between moieties. Harmonic vibrational frequencies were calculated to confirm the
true-minima nature of stationary points, to obtain the zero point energies and to get
indications about IHB strengths from red shifts. Intramolecular hydrogen bonds
(O−H⋯O IHBs and O−H⋯π interaction) are the most stabilizing factors. The
mutual orientations of the four moieties also have considerable influence and they
prefer to be perpendicular to each other.

Keywords Alkaloids ⋅
Antimalarials ⋅
Intramolecular hydrogen bond
Jozimine A2 ⋅ Naphthyl-isoquinoline alkaloids ⋅
Mutual orientation
Solute-solvent interactions

Electronic supplementary material The online version of this chapter (https://doi.org/10.1007/

978-3-319-74582-4_17) contains supplementary material, which is available to authorized users.

M. K. Bilonda ⋅ L. Mammino (✉)

Department of Chemistry, University of Venda, P/bag X5050, Thohoyandou, South Africa
e-mail: [email protected]
M. K. Bilonda
e-mail: [email protected]

© Springer International Publishing AG, part of Springer Nature 2018 305

Y. A. Wang et al. (eds.), Concepts, Methods and Applications of Quantum Systems
in Chemistry and Physics, Progress in Theoretical Chemistry and Physics 31,
https://doi.org/10.1007/978-3-319-74582-4_17
306 M. K. Bilonda and L. Mammino

1 Introduction

Malaria is an infectious disease caused by plasmodia, among which Plasmodium

falciparum is the most dangerous and responsible for most deaths. According to
WHO 2014 reports, 97 countries and territories are affected by malaria transmission
[1]. 98 million cases and 584 000 deaths were recorded in 2013 [1] and 212 million
new cases and 429 000 deaths in 2015 [2]. 90% of the cases occurred in African
countries and most of the deaths concern children under 5 years [1, 2].
The major challenge for malaria treatment is the fast development of resistance
to new drugs, just within few years after their introduction into clinical use. Some
drugs have already lost their efficacy and resistance has started appearing for
artemisinin, which is the current drug of choice. The development of new drugs
should at least keep the pace with the rate at which Plasmodium develops resistance
to existing drugs. This implies urgent need for new compounds having original
modes of action. Natural products constitute a potentially immense source of new
compounds, with diverse molecular structures and pharmacophores.
Jozimine A2 (Fig. 1, denoted by the acronym JZM in the rest of the text) is a
dioncophyllaceae-type naphthylisoquinoline alkaloid isolated from a plant
belonging to the ancistrodaceae family. It has confirmed antimalarial activity, with
the lowest IC50 (0.0014 μm) among antimalarial naphthylisoquinoline alkaloids [3].
(The IC50 indicates how much of a particular drug is needed to inhibit 50% of a
given biological process or component of a process such as an enzyme, cell, cell
receptor or microorganism). The molecule has a dimeric structure: it consists of the
two units, each consisting of a naphthalene moiety and an isoquinoline moiety. The
two units are identical (have the same substituents in corresponding positions)
making the molecule a C2 symmetric dimer [3]. These characteristics make the
computational study of this molecule particularly interesting. The analysis of the
results will give specific attention to symmetry aspects.
The electronic structure of a molecule and the properties related to it—such as
dipole moment, molecular electrostatic potential, molecular orbital energies—pro-
vide important information for a better understanding of molecular interactions when
biological recognition processes are involves [4–8]. Some studies have also already
shown relationships between the electronic structure and the antimalarial activity of
alkoxylated and hydroxylated chalcones [9], tetrahydropyridines [10] and the cin-
chona alkaloids [8]. Some Quantitative Structure-Activity Relationship (QSRA)
studies have shown correlation between electronic structure, antimalarial activity
and phototoxicity of selected quinolinemethanol derivatives and their analogs [11].
The current study is the first study of the electronic structure of JZM, and it is part
of an ongoing computational investigation of naphthylisoquinoline alkaloids [12,
13]. The conformational preferences of JZM were studied in vacuo and in three
solvents with different polarities and different H-bonding abilities (chloroform,
acetonitrile and water). Two levels of theory, Hartree-Fock (HF/6-31G(d,p)) and
Density Functional Theory with the B3LYP functional (DFT/B3LYP/6-31+G(d,p)),
Computational Study of Jozimine A2, a Naphthylisoquinoline … 307

57
CH3

58 35
H 36
N 34
A'
37 39
H359C
40 33
B'
60
H 38 32
46
O 31
56
24 23 CH3
25 29 22 52 51
H CH3
C' D' 43 42
O O
26 30 21
27 28 14 15
13 19 16
O45 O44 D C
H355C 54
H 12 20 17
H353C 11 18
41
5 O
6 4 H50
B
7 9 49
CH3
10 3
A
2
8 N
1 H48
47
CH3

Fig. 1 Two representations of the jozimine A2 molecule and atom numbering utilized in this
work. In the structure on the left, the C atoms in the rings are represented only by the numbers
denoting their positions. Only the H atoms attached to O or N are numbered separately, while the
H atoms attached to C atoms are given the same number as the C atom and are not shown in the
structure. The two units are identical. Identical rings are denoted by the same letters, primed for
one of the units (A, B, C, D and C′, D′, A′ and B′)

were used to perform the calculations. Calculations in solution were performed using
the PCM model.
The results highlight the dominant stabilizing effect of the O−H⋯O
intramolecular hydrogen bonds (IHBs) and of other H-bond-type intramolecular
interactions (for instance, O−H⋯π interaction) and the preference for all the
moieties to be mutually perpendicular. Other computable molecular properties
(dipole moments, HOMO-LUMO energy gaps, solvent effect, etc.) also show some
dependence on IHB patterns.
Comparisons are also made with the results of a study of michellamine A [12]—
a dimeric naphthylisoquinoline alkaloid with anti-HIV activity [13] and with the
same ring systems as JZM. The two molecules differ by the substituents in the
moieties and by the fact that the two naphthalene moieties in michellamine A are
not identical. The stabilising effects show interesting similarities. The main dif-
ference is that no IHB between the two naphthalene moieties is possible in JZM,
whereas it is possible in michellamine A.
308 M. K. Bilonda and L. Mammino

2 Computational Details

Calculations in vacuo were performed with fully relaxed geometry using two levels
of theory: Hartree-Fock (HF) with the 6-31G(d,p) basis set, and Density Functional
Theory (DFT) with the B3LYP functional [14, 15] and the 6-31+G(d,p) basis set.
HF is a moderately cheap quantum mechanical method which can yield accurate
information regarding conformational analysis. Previous studies on other molecules
[12, 16–18] showed that HF can successfully handle intramolecular H-bonding and
yields HOMO-LUMO energy gaps approaching those of experiments. DFT—an
alternate method to wavefunction approaches—is often used in conformational
search because it takes into account part of the correlation effects at a relatively low
cost. Among the numerous functionals available for the DFT framework, B3LYP
[14, 15, 19] is the most widely utilized; it can provide better quality results in
combination with basis sets containing diffuse functions, above all for molecular
systems containing IHBs [12, 16–18].
Harmonic vibrational frequencies were calculated in vacuo at the HF/6-31G(d,p)
level to verify that the stationary points from optimization results corresponded to
true minima and to obtain the zero-point energy (ZPE) corrections. The frequency
values were scaled by 0.9024 [20].
A preliminary identification of conformers of interest was carried out by con-
sidering the potential energy profiles for the rotation of the C5−C11 bond (showing
the minima for the mutual orientation of the two moieties within one unit, Fig. 2)
and for the rotation of the C13−C21 bond (showing the minima for the mutual
orientation of the two units). The rotation of the two bonds was carried out
simultaneously, yielding a 3D potential energy profile (Fig. 2). It was carried out

100 360 -10.00

3.750
Rela

17.50
315
80 31.25
tive e

45.00
270 58.75
n

60
erg y

72.50
Torsion angle D1

225 86.25
40 100.0
(kcal/

180
20
m
ol)

0 360 135
0 315
45 270 90
D2

90 225
gle

135 180 45
To
an

rs 180 135
io n
io n

225
an 90 0
rs

g le 270 0 45 90 135 180 225 270 315 360

To

45
D1 315
360 0 Torsion angle D2

Fig. 2 3D potential energy profile for the scan of the rotation of the C5–C11 and C13–C21 bonds
(energy versus D1 and D2, with D1 being the C4–C5–C11–C12 torsion angle and D2 being the
C14–C13–C21–C28 torsion angle) and 2D potential energy profile for the same rotation scan. The
complete 360° rotations were performed at 15° pace
Computational Study of Jozimine A2, a Naphthylisoquinoline … 309

for a complete rotation (360°) of the bond with a 15° pace. A conformer without
O−H⋯π interaction between the moieties (a conformer with the O43−H52⋯O42
and O44−H54⋯O45 IHBs) was chosen as input, to avoid the effect of the removal
of the interaction during the rotation. The scan highlighted five minima (Fig. 2,
considering that angles differing by 360° are the same). Each of the corresponding
geometries was optimized to find the actual geometry corresponding to a given
minimum (as the optimization involves fully relaxed geometry). Since interactions
between S and S′ were not present in the input, this provides information only on
the preferences for the mutual orientations of the moieties and identifies five dif-
ferent combinations of orientations. More conformers were then identified by
changing the IHB patterns in each of the five conformers obtained from the
potential energy profile.
Calculations in solution utilized the Polarizable Continuum Model (PCM, [21–
26]). In this model, the solvent is considered immeasurable and is modelled by a
continuous isotropic dielectric into which the solute is inserted. Thus, the solute
molecule is embedded in a cavity surrounded by the continuum solvent. The
geometry of the cavity follows the geometry of the solute molecule, considering its
solvent accessible surface. The calculations utilised the default settings of Gaus-
sian03 [27] for PCM, namely, Integral Equation Formalism model (IEF, [23–26])
and Gepol model for building the cavity around the solute molecule [28–30], with
simple United Atom Topological Model (UAO) for the atomic radii and 0.200 Å2
for the average area of the tesserae into which the cavity surface is subdivided.
The SCFVAC option was selected to obtain more thermodynamic data.
Calculations in solution were performed as single point (SP) calculations on the
in-vacuo optimised geometries, with the same levels of theory utilised in vacuo. It
was opted to use SP calculations because the size of the molecule makes
re-optimisation in solution computationally expensive. Although SP calculations
cannot provide information on the geometry changes caused by the solvent, they
can provide reasonable information on the energetics, such as the conformers’
relative energies in solution and the energy aspects of the solution process (the free
energy of solvation, ΔGsolv, and its components).
The three solvents considered (chloroform, acetonitrile and water) cover the
ranges of polarity and of hydrogen bonding abilities interesting for biologically
active molecule. Chloroform is an apolar aprotic solvent with low relative per-
mittivity (εr = 4.90) and low dipole moment (µ = 1.04 D [31]). Acetonitrile is a
dipolar aprotic solvent with large relative permittivity (εr = 36.64) and high dipole
moment (µ = 3.92 D [31]). Water is a protic solvent with high relative permittivity
(εr = 78.39) and a sizeable dipole moment (µ = 1.83 D [31]).
Calculations were performed using GAUSSIAN 03, Revision D 01 [27].
All the energy values reported are in kcal/mol and all the distances are in Å.
Acronyms are utilized for the calculation methods and for the media, for con-
ciseness sake on reporting values: HF for HF/6-31G(d,p), DFT for DFT/B3LYP/
6-31+G(d,p), ‘vac’ for vacuum, ‘chlrf’ for chloroform, ‘actn’ for acetonitrile and
‘aq’ for water. Tables with all the numerical values of the properties of the cal-
culated conformers (relative energy, dipole moments, free energy of solvation in the
310 M. K. Bilonda and L. Mammino

solvents considered, HOMO-LUMO energy gaps, etc.), in the results of both cal-
culation methods, and figures showing all the conformers, are included in the
Supplementary Information.

3 Results

3.1 Naming of Conformers

A system of symbols is introduced to be able to clearly and easily identify the

geometric characteristics of each conformer when analysing the results; thus, each
conformer is denoted by an acronym summarising its characterising features. Since
it is useful to be able to mention each ring individually and concisely, the rings are
denoted by uppercase letters [12]. Since the two units are identical, identical rings
are denoted with the same letter, primed for one of the units. Thus, the rings are
denoted as A, B, C, D, A′, B′, C′ and D′ as shown in Fig. 1. In order to be able to
mention each of the two units of the dimeric molecular structure individually, the
whole unit comprising the A, B, C and D rings is called S and the whole unit
comprising the A′, B′, C′ and D′ rings is called S′.
The O−H⋯O IHBs and other IHB-type interactions such as O−H⋯π and C
−H⋯O are the most important factors stabilizing the conformers. Different con-
formers have different combinations of these intramolecular interactions and are
characterized by them. A specific lowercase letter is used to denote each IHB or
IHB-type interaction in the acronyms denoting the conformers. The letters (a–h)
and their meanings are listed in Table 1. Their use is illustrated by the names of the
representative conformers of JZM shown in Fig. 3.

Table 1 Letters utilized in the acronyms denoting the conformers of Jozimine A2 in this work,
and their meanings
Letter Meaning Letter Meaning
a Presence of the O43 h Presence of C−H59⋯O46
−H52⋯O42 IHB
b Presence of the O44 t C14−C13−C21−C28 torsion angle
−H54⋯O45 IHB close to +90°
c Presence of the O43−H52⋯π v C14−C13−C21−C28 torsion angle
interaction close to −90°
d Presence of the O44−H54⋯π p C4−C5−C1−C12 torsion angle close
interaction to +90°
e Presence of the O41−H50⋯π q C4−C5−C1−C12 torsion angle close
interaction to −90°
f Presence of the O46−H60⋯π x C22−C23−C31−C38 torsion angle
interaction close to +90°
g Presence of C−H49⋯O41 y C22−C25−C31−C38 torsion angle
close to −90°
Computational Study of Jozimine A2, a Naphthylisoquinoline … 311

a-b-e-f-g-h-p-v-x a-b-e-f-g-h-q-v-x a-b-e-f-g-h-q-t-x a-b-e-f-g-h-q-v-y a-b-e-f-g-h-p-t-x

a-b-f-h-q-v-x a-d-e-f-g-h-q-v-x b-c-e-f-g-h-q-t-x a-b-e-g-q-t-x b-c-e-f-g-h-p-v-x

a-d-e-f-g-h-p-v-x a-b-f-h-p-v-x a-b-e-g-p-v-x a-b-e-g-v-y a-b-f-h-v-y

a-b-e-g-p-t-x a-b-f-h-p-t-x c-d-e-f-g-h-p-v-x a-d-e-g-p-t-x a-b-q-v-y

a-b-p-t-x c-d-f-h-p-v-x c-d-f-h-q-v-x b-c-q-v-y

a-d-q-v-y c-d-q-v-y c-d-p-t-x

Fig. 3 Optimized geometries of representative conformers of jozimine A2 and letter-combinations

identifying their characteristics in the acronyms denoting them. Geometries from HF/6-31G(d,p)
results. The images show the main combinations of interaction types and moieties’ orientations
which may be present in the conformers of jozimine A2. The initial part of the acronyms is not
reported under the images, because of space reasons; it is the same for all the conformers (JZM)
312 M. K. Bilonda and L. Mammino

a-b-e-f-g-h-p-v-x a-b-e-f-g-h-q-v-x a-b-e-f- g-h -q-t-x a-b-e-f-g-h-q-v-y

a-b-e-f-g-h-p-t-x b-c-e-f-g-h-p-t-x a-d-e-f-g-h-p-t-x c-d-e-f-g-h-p-t-x

Fig. 4 Optimized geometries of selected conformers of jozimine A2 highlighting the perpendic-

ularity of the moieties

The mutual orientations of the four moieties also have significant influence on the
energy of the conformers and, therefore, they are also denoted by specific letters.
Lowercase letters are utilised to provide information about the torsion angles between
moieties, i. e, information about their mutual orientation. The letters ‘p’ and ‘q’ refer to
the torsion angle within the S unit (angle between the A, B and the D, C rings systems,
i.e., C4−C5−C11−C12); the letters ‘x’ and ‘y’ refer to the torsion angle within S′ unit
(angle between the C′, D′ and the A′, B′ rings systems, i.e., C22−C25−C31−C38);
the letters ‘t’ and ‘v’ refer to the torsion angle between the two units S and S′ (i.e., C14
−C13−C21−C28); the meaning of these letters is explained in Table 1. Figure 4
highlights the perpendicularity of the moieties through selected examples. Differently
from michellamine A [12], there is no specific correspondence between the orientation
of the two moieties (within each unit) and the IHB patterns.
When the conformers are mentioned in the text, their acronyms start with JZM,
followed by the letters denoting their characteristics. In tables and figures, ‘JZM’ is not
reported for space-saving reasons. It may be expedient to illustrate how the acronyms
describe the characteristics of individual conformers through some examples.
JZM-a-b-e-f-g-h-p-v-x is a conformer characterized by the presence of the O43
−H52⋯O42 and O44−H54⋯O45 IHBs, the O41−H50⋯π and O46−H60⋯π
interactions and the C−H49⋯O41 and C−H59⋯O46 interactions, with the C4−C5
−C11−C12 torsion angle close to 90°, the C14−C13−C21−C28 torsion angle close
to −90° and C22−C25−C31−C38 torsion angle close to 90°; JZM-c-d-e-f-g-h-q-v-y
is a conformer with the O43−H52⋯π, O44−H54⋯π, O41−H50⋯π and O46
−H60⋯π interactions and the C−H49⋯O41 and C−H59⋯O46 interactions, with the
C4−C5−C11−C12 torsion angle close to −90°, the C14−C13−C21−C28 torsion
angle close to −90° and the C22−C25−C31−C38 torsion angle close to −90°;
JZM-b-c-q-t-x is a conformer characterized by the presence of the O44−H54⋯O45
IHB and the O43−H52⋯π interaction, and with the C4−C5−C11−C12 torsion angle
close to −90°, the C14−C13−C21−C28 torsion angle close to +90° and the C22
−C25−C31−C38 torsion angle close to +90°.
Computational Study of Jozimine A2, a Naphthylisoquinoline … 313

3.2 Results in Vacuo

80 conformers were identified, and the frequency calculations confirm that they
correspond to true minima. The relative energies of selected calculated conformers
of JZM in vacuo and in the three solvents considered are reported in Table 2.

Table 2 Relative energies of the selected conformers of jozimine A2 shown in Fig. 3. HF/6-31G
(d,p) and DFT/B3LYP/6-31+G(d,p) results in vacuo and in the solvents considered, respectively
denoted as HF and DFT in the columns’ headings. The results in vacuo are from full optimization
calculations; the results in solution are from single point PCM calculations on the
in-vacuo-optimized geometries. The conformers are listed in order of increasing relative energies
in the HF results in vacuo
Conformer Relative energy (kcal/mol)
HF DFT
vac chlrf actn aq vac chlrf actn aq
a-b-e-f-g-h-p-v-x 0.000 0.000 0.000 0.112 0.000 0.000 0.000 0.095
a-b-e-f-g-h-q-v-x 0.743 0.467 0.278 0.020 0.608 0.404 0.320 0.000
a-b-e-f-g-h-q-t-x 0.793 0.591 0.303 0.243 0.606 0.411 0.290 0.058
a-b-e-f-g-h-q-v-y 1.463 0.637 0.618 0.000 1.202 0.777 0.567 0.005
a-b-e-f-g-h-p-t-x 1.466 0.920 0.707 0.075 1.180 0.825 0.578 0.187
a-b-f-h-q-v-x 5.783 5.411 3.856 1.772 5.634 4.645 3.970 1.860
a-d-e-f-g-h-q-v-x 5.857 5.469 5.429 3.921 5.944 5.799 −a −a
b-c-e-f-g-h-q-t-x 5.934 5.759 5.547 3.902 5.990 5.840 5.679 4.132
a-b-e-g-q-t-x 5.962 4.497 4.003 1.987 5.770 4.757 4.121 1.900
b-c-e-f-g-h-p-v-x 6.209 5.756 5.725 4.095 6.267 6.053 5.902 4.341
a-d-e-f-g-h-p-v-x 6.209 5.756 5.725 4.095 6.267 6.053 5.897 4.375
a-b-e-g-p-v-x 6.269 5.549 4.132 2.037 6.028 4.978 4.266 2.160
a-b-f-h-p-v-x 6.269 4.914 4.192 2.096 6.028 5.004 −a 2.134
a-b-e-g-v-y 6.476 5.168 4.123 1.625 6.202 −a −a 1.819
a-b-f-h-v-y 6.476 5.157 4.037 1.706 6.202 4.917 4.289 1.813
a-b-e-g-p-t-x 6.552 5.316 4.308 1.969 6.204 5.008 4.295 1.999
a-b-f-h-p-t-x 6.552 5.194 4.294 1.945 6.204 5.021 4.278 1.959
c-d-e-f-g-h-p-v-x 11.627 11.590 11.467 8.363 11.508 11.728 11.623 8.798
a-d-e-g-p-t-x 11.682 10.404 9.397 5.723 11.568 10.460 9.692 6.632
a-b-q-v-y 11.899 9.483 7.954 3.038 11.602 −a −a 3.878
a-b-p-t-x 12.096 10.142 8.111 3.817 11.684 9.638 8.295 4.437
c-d-f-h-p-v-x 15.532 14.858 14.216 10.362 15.508 15.181 14.688 10.984
c-d-f-h-q-v-x 15.687 14.974 14.302 3.432 15.648 15.128 14.575 −a
b-c-q-v-y 16.262 13.764 12.531 6.784 16.266 14.080 12.760 7.637
a-d-q-v-y 16.262 13.982 12.425 6.961 16.266 14.099 12.753 7.714
c-d-q-v-y 20.097 18.547 17.013 11.069 20.169 18.675 17.585 11.609
c-d-p-t-x 20.432 18.889 17.427 11.707 20.216 18.933 18.027 12.417
a
The calculation for this conformer did not converge in the given solvent
314 M. K. Bilonda and L. Mammino

The reported conformers comprise the first fourteen lower-energy conformers and
representative conformers of other types (other combinations of IHBs) not
appearing among the first fourteen. The IHB patterns (numbers and types of IHBs
present in a conformer) are the dominant stabilizing factors. Others factors
influencing conformational preferences are the mutual orientation of the two
naphthalene moieties (C, D and C′, D′ ring systems), which determines the mutual
orientation of the two units (S and S′) and the orientation of the two moieties
(naphthalene and isoquinoline) within each unit. The isoquinoline moiety prefers to
be perpendicular to the naphthalene moiety and the two naphthalene moieties
(linked by the inter-units biaryl axis) also prefer to be perpendicular to each another;
this results in the two units (S and S′) preferring to be mutually perpendicular,
which excludes the possibility of an IHB between them.
As mentioned previously, three types of IHBs interactions may be present:
O−H⋯O IHBs (O43−H52⋯O42 and O44−H54⋯O45), O−H⋯π interactions
between the O−H in an isoquinoline moiety and the closest π system in the
naphthalene moiety of the same unit (O41−H50⋯π, O46−H60⋯π) or between an
O−H in the naphthalene moiety of one unit and the closest π system in the
naphthalene moiety of the other unit (O43−H52⋯π, O44−H54⋯π); and C−H⋯O
interactions within an isoquinoline moiety (C−H49⋯O41 and C−H59⋯O46). The
O−H⋯O IHBs have the strongest stabilizing effect. The two O−H⋯O IHBs (O43
−H52⋯O42 and O44−H54⋯O45) have practically the same geometric parameters
in a conformer having both of them simultaneously.
The lowest energy conformers are the conformers having all the IHB-types
interactions simultaneously. The first five lowest energy conformers have relative
energy below 1.5 kcal/mol in both the HF and the DFT results. They have the same
types of IHBs interactions and differ only by the orientations of the moieties.
Altogether, they account for 99.95% of the population in vacuo (Table S12).
The reported X-ray structure of JZM [3] corresponds to the a-b-f-h-p-v-x con-
former, which has 6.269 kcal/mol relative energy and 0.0014 population (Table 2).
This conformer has the two O−H⋯O IHBs (O43−H52⋯O42 and O44
−H54⋯O45), one O−H⋯π interaction and one C−H⋯O interaction. Compared to
the five lowest energy ones (which have all the interactions), this conformer has one
O−H⋯π and C−H⋯O interaction less, which leaves one OH group free. This
phenomenon is observed frequently (e.g., in the case of caespitate [32]): the
crystalline structure differs from the geometry in vacuo by not having one or more
weaker IHBs or other IHB-type interactions that are present in the gas phase.
A possible reason is the need, for molecules in the crystal structure, to have some
donors or acceptors not engaged in intramolecular interactions, so that they are
available for intermolecular interactions with the surrounding molecules.
The symmetry of the molecule results in a number of pairs of conformers with
the same characteristics (IHBs) in the S moiety of one conformer and in the S′
moiety of the other conformer. Pairs of this type will be termed S/S′ symmetric
pairs in the rest of the text.
Computational Study of Jozimine A2, a Naphthylisoquinoline … 315

The removal of the O43−H52⋯O42 IHB brings about the O43−H52⋯π

interaction and the removal of the O44−H54⋯O45 IHB brings about the O44
−H54⋯π interaction. Each of these removals causes an increase in the energy of the
conformer. Comparison of the energies of conformers JZM-a-b-e-f-g-h-p-v-x and
JZM-b-c-e-f-g-h-p-v-x (removal of O43−H52⋯O42 and formation of O43
−H52⋯π) shows an energy increase of ≈6 kcal/mol. Since the two units are
identical, the presence or the absence of a given IHB in only one of the units has the
same effects whether it occurs in unit S (O43−H52⋯O42) or in unit S′ (O44
−H54⋯O45). Therefore, the conformers JZM-a-d-e-f-g-h-p-v-x and JZM-b-c-e-f-
g-h-p-v-x (or the conformers JZM-a-d-e-f-g-h-q-v-y and JZM-b-c-e-f-g-h-q-v-y,
or JZM-a-d-e-f-g-h-p-t-x and JZM-b-c-e-f-g-h-p-t-x, or JZM-a-d-q-t-x and
JZM-b-c-q-t-x, or JZM-b-c-q-v-y and JZM-a-d-q-v-y, or JZM-b-c-p-t-x and
JZM-a-d-p-t-x, or JZM-b-c-p-v-x and JZM-a-d-p-v-x, or JZM-b-c-q-t-x and
JZM-a-d-q-v-x (Table S13) have the same relative energy and similar molecular
properties such as dipole moment and HOMO-LUMO energy gap, and also rather
close values of ΔGsolv.
Since the removal of an O−H⋯O IHB brings about an O−H⋯π IHB and vice
versa, it is not possible to evaluate the energy of an IHB by comparison with a
conformer in which it is removed by 180° rotation of the donor [33–42]. It is only
possible to compare the strength of the O−H⋯O and O−H⋯π IHBs utilizing
conformers in which only one or the other is present and the other geometry
characteristics are identical, as in the previously considered case of
JZM-a-b-e-f-g-h-p-v-x and JZM-b-c-e-f-g-h-p-v-x. Other analogous cases can be
considered. For instance, JZM-a-b-e-f-g-h-p-v-x and JZM-c-d-e-f-g-h-p-v-x differ
only because the former has two O−H⋯O IHBs and two O−H⋯π IHBs and the
latter has four O−H⋯π IHBs. The relative energy of the latter is 11.627/HF and
11.508/DFT kcal/mol higher than that of the former, suggesting that in this case the
stabilisation brought by an O−H⋯π IHB is ∼5.8 kcal/mol smaller than that of an
O−H⋯O IHB. Conformers JZM-a-b-e-f-g-h-q-v-x and JZM-c-d-e-f-g-h-q-v-x,
JZM-a-b-e-f-g-h-q-t-x and JZM-c-d-e-f-g-h-q-t-x, JZM-a-b-e-f-g-h-q-v-y and
JZM-c-d-e-f-g-h-q-v-y, and JZM-a-b-e-f-g-h-p-t-x and JZM-c-d-e-f-g-h-p-t-x are
other pairs of this type with the relative energy of the latter in each pair ∼10 kcal/
mol higher than that of the corresponding former, with both HF and DFT.
Table 3 reports the ranges of the lengths of all the IHB types present in the
conformers. Complete data of the parameters of all the conformers are reported in
the Supplementary Information. The parameters of the O43−H52⋯O42 and O44
−H54⋯O45 IHBs in the same conformer are very close. For S/S′ symmetric pairs,
the parameters are identical whether these IHBs are in S or S′. The bond length (Å)
is close to 1.73/DFT and 1.77/HF, the O⋯O distance (Å) is close to 2.58 with both
HF and DFT and the OĤO bond angle is ∼143°/HF and ∼145°/DFT for con-
formers having also other types of IHBs. On the basis of their bond lengths, O⋯O
distances and bond angles, the O−H⋯O IHBs in JZM can be classified in the lower
region of moderate H-bonds [43].
316 M. K. Bilonda and L. Mammino

Table 3 Ranges of the IHB H⋯O (Å) or H⋯C (Å)

length of the IHBs in the
HF DFT
conformers of Jozimine A2.
For the O−H⋯π interaction, H52⋯O42 1.764−1.770a 1.722−1.731a
the length is taken as distance 1.771−1.773b 1.731−1.735b
between the H atom and the H54⋯O45 1.764−1.770a 1.722−1.732a
closest C atom in the π
1.768−1.773b 1.731−1.735b
system, which is indicated in
parentheses after the π symbol H49⋯O41 2.367−2.372 2.866−3.008
H59⋯O46 2.369−2.372 3.000−3.008
H50⋯π (C11) 2.323−2.343 2.284−2.312
H52⋯π (C21) 2.222−2.233 2.189−2.203
H54⋯π (C13) 2.221−2.325 2.186−2.288
H60⋯π (C23) 2.324−2.404 2.288−2.315
a
In conformers having other IHBs
b
In conformers having only O−H⋯O IHBs

For the O−H⋯π interactions, a bond length is not defined, as the acceptor is a
whole π system and not an individual atom. It may be convenient to consider the
distance between the H atom of the donor OH and the C atom in the aromatic
system closest to it. The ranges of the distances thus defined are reported in Table 3.
It can be inferred from these ranges that the H⋯C distance is slightly shorter when
the interaction involves an O−H in the naphthalene moiety of one unit and the
closest π system in the naphthalene moiety of the other unit (O43−H52⋯π and O44
−H54⋯π) than when it involves an O−H in the isoquinoline moiety and the closest
π system in the naphthalene moiety of the same unit (O41−H50⋯π and O46
−H60⋯π).
The ranges of the H⋯O distance for the C−H⋯O interactions within the iso-
quinoline moiety (C−H49⋯O41 and C−H59⋯O46) are also reported in Table 3.
This distance is considerably longer than the H⋯O distance for O−H⋯O IHBs,
consistently with the fact that the C−H⋯O interaction is considerably weaker. The
donor–acceptor C⋯O distance for C−H49⋯O41and C−H59⋯O46 is ≈3.0 Å in
both the HF and DFT results. The CĤO bond angle is ≈115.0°/HF and ≈116.0°/
DFT for both C−H49⋯O41 and C−H59⋯O46. Their parameters show that the C
−H⋯O IHBs are weak H-bonds.
Table 4 lists the ranges of the calculated harmonic vibrational frequencies of the
O–H bonds. Only HF frequencies are available in this study because frequency
calculations with the DFT method did not complete (which is probably due to the
high number of atoms in this molecule). When IHBs are present, it is interesting to
consider also the red-shift (lowering of the vibrational frequency of the donor OH)
caused by the IHBs. The red shift is evaluated with respect to the frequency of a
free OH of the same type. Since the OHs in JZM are never free, two model
structures with free OHs and with the other features similar to those of the moieties
in JZM (including a CH3 to mimic the presence of other moieties attached to the
one considered) were used to calculate a reference frequency. These structures are
Computational Study of Jozimine A2, a Naphthylisoquinoline … 317

Table 4 Ranges of the harmonic vibrational frequencies of the OH groups in jozimine A2 and of
the red shifts caused by IHBs
OH Frequency (cm−1) Red shift when engaged in O Red shift when engaged in O−H⋯π
−H⋯O IHB interactions
O41−H50 3735.02−3804.55 − 39.08−64.31
O43−H52 3707.53−3758.99 27.37−60.59 9.13−46.53
O44−H54 3707.73−3759.84 30.03−60.39 8.23−51.28
O46−H60 3736.13−3838.26 − 40.34−65.17

Fig. 5 Model structures used (a) CH3 (b)

to calculated a reference
frequency for the vibration of 41
5 O
H
the free O41−H50 and O46 6 4 H50 42
−H50 (a) or O43−H52 and B H O
O44−H54 (b) 7 9 49
CH3 14 15 CH3
10 3 13 19 16
A D C
2
8 N 12 20 17
1 H48 11 18 CH3
47
CH3 CH3

shown in Fig. 5. The ranges of the red shifts of O43−H52 and O44−H54 when
engaged in the different types of IHBs are reported in Table 4. The values of the red
shifts for O−H⋯O IHBs and the O−H⋯π interactions are comparable, which is
consistent with the stabilizing effect of both types of IHBs (although the strength of
the O−H⋯O IHB is expected to be greater than that of the O−H⋯π IHB).
Table 5 reports the values of the ZPE and of the relative energies corrected for
ZPE for the conformers listed in Table 2. The ZPE corrections are very close for all
the conformers. Their values (kcal/mol) are in the 571.572−572.602 kcal/mol
range, with the greater values corresponding to lower energy conformers. The
relative energies corrected for ZPE have the same trends as the uncorrected ones.
Table 6 reports the values of the dipole moment for the conformers listed in
Table 2. The ranges (debye) of the values of the dipole moments in vacuo are 0.50
−77.53/HF and 0.59−77.33/DFT. Both the magnitude and the direction of the
dipole moment vector change according to the conformer and are largely influenced
by the orientation of the OH groups (although the orientation of the aromatic rings
also plays a role).
Conformers with O−H⋯O IHBs have higher dipole moment than conformers
with O−H⋯π interactions, with few exceptions. Conformers with only O−H⋯π
interactions (no O−H⋯O IHBs) are among the conformers with smaller dipole
318 M. K. Bilonda and L. Mammino

Table 5 Relative energy (ΔEcorrect, kcal/mol) corrected for ZPE and ZPE corrections (kcal/mol)
for conformers of jozimine A2 selected among those reported in Table 2. Results from HF/6-31G
(d,p) frequency calculations
Conformer ΔEcorrect ZPE Conformer ΔEcorrect ZPE
correction correction
a-b-e-f-g-h-p-v-x 0.000 572.602 a-b-e-g-p-t-x 6.100 572.150
a-b-e-f-g-h-q-v-x 0.644 572.503 c-d-e-f-g-h-p-v-x 11.036 572.011
a-b-e-f-g-h-q-t-x 0.688 572.497 a-d-e-g-p-t-x 11.053 571.973
a-b-e-f-g-h-q-v-y 1.268 572.407 a-b-q-v-y 11.089 571.791
a-b-e-f-g-h-p-t-x 1.294 572.430 a-b-p-t-x 11.313 571.819
a-d-e-f-g-h-q-v-x 5.545 572.289 c-d-f-h-p-v-x 14.776 571.846
b-c-e-f-g-h-q-t-x 5.616 572.284 c-d-f-h-q-v-x 14.866 571.781
a-b-e-g-q-t-x 5.575 572.215 b-c-q-v-y 15.280 571.621
b-c-e-f-g-h-p-v-x 5.871 572.264 c-d-q-v-y 19.007 571.572
a-d-e-f-g-h-p-v-x 5.872 572.264 c-d-p-t-x 19.433 571.603

moments. The same phenomenon was observed in the previous study of naph-
thylisoquinoline alkaloids with antimalarial activity [16] and in the study of
michellamine A [12]. The mutual orientation of the moieties also has considerable
influence on the dipole moment. Different orientations of the S and S′ units may
cause 1−4 D difference in the dipole moments, and different orientations of the
isoquinoline and naphthalene moieties within each unit may cause 1−2 D difference
in the dipole moments. This trend is reversed with respect to what was observed for
the conformers of michellamine A [12], where the orientation of the units caused a
1−2 D difference and the different orientation of the isoquinoline and naphthalene
moieties within each unit caused a 1−4 D difference. When two conformers of JZM
differ both by the orientation of the S and S′ units and by the orientations of the
isoquinoline and naphthalene moieties, the dipole moment difference is ≈2 D.
Conformers of S/S′ symmetric pairs have the same dipole moment.
Table 7 reports the HOMO-LUMO energy gap for the conformers listed in
Table 2. The gap is influenced by the IHB patterns. Conformers with only one
O−H⋯O IHB and other types of IHBs interactions (O−H⋯π and C−H⋯O), such
as conformers JZM-b-c-e-g-p-v-x, JZM-c-f-h-q-t-x, JZM-a-d-e-g-p-t-x,
JZM-a-d-e-g-q-v-y, have the smallest HOMO-LUMO energy gap. The gap is
slightly greater for conformers with the two O−H⋯O IHBs and other IHB-type
interactions and highest for conformers with only O−H⋯π interactions (which are
accompanied by C−H⋯O interactions if they involve the two moieties within the
same unit). The presence of the C−H⋯O interactions in a conformer slightly
decreases the HOMO-LUMO energy gap with respect to a corresponding con-
former where it is absent; for instance, the gap in JZM-c-d-f-h-p-v-x is ≈1 kcal/mol
less than in JZM-c-d-p-v-x. Conformers of S/S′ have the same HOMO-LUMO
energy.
The estimation of the HOMO-LUMO energy gap shows marked difference
between HF and DFT values. This is a known phenomenon, as DFT substantially
Computational Study of Jozimine A2, a Naphthylisoquinoline … 319

Table 6 Dipole moment of the conformers of jozimine A2 listed in Table 2, in vacuo and in the
three solvents considered. HF/6-31G(d,p) and (DFT/B3LYP/6-31+G(d,p) results, respectively
denoted as HF and DFT in the column headings. The results in vacuo are from full optimization
calculations, the results in solution are from single point PCM calculations on the
in-vacuo-optimized geometries
Conformer Dipole moment (Debye)
HF DFT
vac chlrf actn aq vac chlrf actn aq
a-b-e-f-g-h-p-v-x 2.65 2.67 2.55 2.44 2.81 2.87 2.74 2.58
a-b-e-f-g-h-q-v-x 5.46 5.96 6.06 5.97 5.32 5.93 6.09 5.99
a-b-e-f-g-h-q-t-x 5.03 5.41 5.43 5.34 5.01 5.50 5.57 5.45
a-b-e-f-g-h-q-v-y 6.83 7.46 7.55 7.44 6.59 7.36 7.52 7.39
a-b-e-f-g-h-p-t-x 7.53 8.29 8.50 8.43 7.33 8.28 8.57 8.49
a-b-f-h-q-v-x 4.25 4.95 4.51 4.31 4.04 4.36 4.41 4.22
a-d-e-f-g-h-q-v-x 4.56 4.85 4.87 4.73 4.70 5.11 −a −a
b-c-e-f-g-h-q-t-x 5.03 6.12 6.29 6.28 5.52 6.22 6.46 6.41
a-b-e-g-q-t-x 4.20 4.45 4.47 4.27 4.10 4.39 4.42 4.21
b-c-e-f-g-h-p-v-x 4.11 4.53 4.63 4.69 4.15 4.65 4.81 4.83
a-d-e-f-g-h-p-v-x 4.11 4.53 4.64 4.69 4.15 4.66 4.80 4.83
a-b-e-g-p-v-x 3.93 4.24 4.29 4.37 4.38 4.79 4.85 4.91
a-b-f-h-p-v-x 3.93 4.23 4.28 4.36 4.38 4.78 −a 4.90
a-b-e-g-v-y 4.84 5.08 5.04 4.89 4.36 −a −a 4.27
a-b-f-h-v-y 4.84 5.07 5.04 4.89 4.36 4.59 4.55 4.27
a-b-e-g-p-t-x 4.96 5.21 5.18 4.88 4.78 5.12 5.10 4.78
a-b-f-g-p-t-x 4.95 5.22 5.17 4.88 4.77 5.11 5.09 4.78
c-d-e-f-g-h-p-v-x 0.50 0.20 0.02 0.17 0.70 0.40 0.20 0.21
a-d-e-g-p-t-x 4.62 5.10 5.24 5.20 4.03 4.54 4.70 4.62
a-b-q-v-y 1.38 1.13 −a 0.65 0.59 −a −a 0.41
a-b-p-t-x 1.42 1.19 0.96 0.50 1.08 0.85 0.60 0.15
c-d-f-h-p-v-x 3.83 4.28 4.49 4.65 4.04 4.60 4.82 4.98
c-d-f-h-q-v-x 3.79 4.22 4.39 4.39 3.71 4.56 4.29 −a
b-c-q-v-y 3.91 4.40 4.56 4.67 3.97 4.69 5.00 5.25
a-d-q-v-y 3.91 4.40 4.56 4.67 3.97 4.70 5.00 5.24
c-d-q-v-y 1.28 1.81 2.03 2.40 2.07 2.83 3.19 3.70
c-d-p-t-x 1.33 1.86 2.09 2.64 1.84 2.52 2.87 3.48
a
The calculation for this conformer did not converge in the given solvent

underestimates the values of the gaps [44, 45]. However, the two methods show
similar trends.
Figure 6 shows the shapes of the HOMO and LUMO orbitals for the five lowest
energy conformers, some representative higher energy ones and the conformers of
some S/S′ symmetric pairs. The shapes indicate greater electron density and similar
distribution in the two naphthalene moieties than in the isoquinoline moieties for
Table 7 HOMO-LUMO energy gap of the conformers of jozimine A2 reported in Table 2, in vacuo and in the three solvents considered. HF/6-31G(d,p) and
320

(DFT/B3LYP/6-31+G(d,p) results, respectively denoted as HF and DFT in the column headings. The results in vacuo are from full optimization calculations,
the results in solution are from single point PCM calculations on the in-vacuo-optimized geometries
Conformer HOMO-LUMO energy gap (kcal/mol)
HF DFT
vac chlrf actn aq vac chlrf actn aq
a-b-e-f-g-h-p-v-x 232.762 232.850 232.881 232.919 98.343 98.506 98.563 98.613
a-b-e-f-g-h-q-v-x 253.376 233.929 233.960 234.048 98.908 99.065 99.115 99.203
a-b-e-f-g-h-q-t-x 233.810 233.923 233.967 234.029 98.757 98.946 99.027 99.102
a-b-e-f-g-h-q-v-y 234.921 235.046 235.096 235.159 99.391 99.567 99.642 99.730
a-b-e-f-g-h-p-t-x 234.745 234.883 234.933 234.977 99.579 99.774 99.862 99.931
a-b-f-h-q-v-x 232.341 232.800 232.517 233.007 97.007 97.094 97.389 98.155
a-d-e-f-g-h-q-v-x 224.843 228.733 230.126 231.927 89.031 92.922 −a −a
b-c-e-f-g-h-q-t-x 224.661 228.608 230.158 231.896 88.868 92.752 94.327 96.266
a-b-e-g-q-t-x 232.191 232.272 232.505 233.051 96.963 96.982 97.226 97.923
b-c-e-f-g-h-p-v-x 224.912 228.840 230.359 232.166 89.050 92.991 94.515 96.511
a-d-e-f-g-h-p-v-x 225.157 228.859 230.365 232.134 89.044 92.984 94.515 96.505
a-b-e-g-p-v-x 231.846 231.978 232.235 232.749 96.655 96.718 97.025 97.596
a-b-f-h-p-v-x 231.839 231.971 232.241 232.718 96.655 96.718 −a 97.615
a-b-e-g-v-y 232.938 232.856 232.969 233.709 96.837 −a −a 98.036
a-b-f-h-v-y 232.938 232.850 232.969 233.697 96.843 96.944 97.214 98.036
a-b-e-g-p-t-x 232.498 232.618 232.925 233.640 96.756 96.850 97.163 97.954
a-b-f-h-p-t-x 232.498 232.611 232.900 233.659 96.756 96.850 97.151 97.973
c-d-e-f-g-h-p-v-x 236.558 236.508 236.483 236.433 100.552 100.502 100.489 100.502
a-d-e-g-p-t-x 221.166 225.226 227.064 229.618 85.435 89.257 91.020 93.536
a-b-q-v-y 234.017 234.306 234.506 234.707 98.776 −a −a 99.460
(continued)
M. K. Bilonda and L. Mammino
Table 7 (continued)
Conformer HOMO-LUMO energy gap (kcal/mol)
HF DFT
vac chlrf actn aq vac chlrf actn aq
a-b-p-t-x 234.237 234.462 234.569 234.776 98.889 99.184 99.335 99.567
c-d-f-h-p-v-x 233.051 233.032 233.239 233.747 97.251 97.088 97.295 97.904
c-d-f-h-q-v-x 234.205 233.873 233.967 234.557 97.760 97.835 97.835 0.000
b-c-q-v-y 225.000 229.028 230.591 232.442 88.999 93.016 94.547 96.442
a-d-q-v-y 225.000 229.041 230.559 232.398 89.006 92.978 94.509 96.448
c-d-q-v-y 236.320 236.307 236.289 236.245 100.4077 100.364 100.326 100.282
c-d-p-t-x 236.452 236.401 236.370 236.301 100.3073 100.276 100.257 100.213
a
The calculation for this conformer did not converge in the given solvent
Computational Study of Jozimine A2, a Naphthylisoquinoline …
321
322 M. K. Bilonda and L. Mammino

HOMO

a-b-e-f-g-h-p-v-x a-b-e-f-g-h-q-v-x a-b-e-f- g-h-q-t-x a-b-e-f-g-h-q-v-y a-b-e-f-g-h-p-t-x

a-b-f-h-q-v-x a-b-e-g-q-t-x b-c-e-f-g-h- q-t-x b-c-q-t-x a-d-q-t-x

LUMO

a-b-e-f-g-h-p-v-x a-b-e-f-g-h-q-v-x a-b-e-f-g-h-q-t-x a-b-e-f-g-h-q-v-y a-b-e-f-g-h-p-t-x

a-b-f-h-q-v-x b-c-e-f-g-h- q-t-x a-b-e-g-q-t-x b-c-q-t-x a-d-q-t-x

Fig. 6 Shapes of the HOMO and LUMO frontier orbitals of the lowest energy conformers of
jozimine A2. HF/6-31G(d,p) results. The initial part of the acronyms is not reported under the
images, because of space reasons; it is the same for all the conformers (JZM)

the five lowest energy conformers—a phenomenon that was also observed for
monomeric naphthylisoquinoline alkaloids with antimalarial activity [16] and for
michellamine A [12]. For higher energy conformers, the electron distribution
becomes considerably different for the two naphthalene moieties. For conformers of
S/S′ symmetric pairs, electrons concentrate on one or the other of the naphthalene
moieties.

3.3 Results in Solution

Table 2 shows the relative energy of the conformers in the three solvents consid-
ered (chloroform, acetonitrile and water). The relative energy of the conformer
decreases with increasing solvent polarity—which is consistent with common
behaviours. (Acetonitrile has a greater dipole moment than water, but its effect on
solutes is often intermediate between that of chloroform and that of water; therefore,
“increasing solvent polarities” refers to the chloroform-acetonitrile-water sequence
Computational Study of Jozimine A2, a Naphthylisoquinoline … 323

in this discussion). The first five lowest energy conformers have relative energy
below 1 kcal/mol in chloroform and acetonitrile, and below 0.24 kcal/mol in water.
The identification of the lowest energy conformer does not change in chloroform
and acetonitrile with respect to in vacuo, but may be different in water; however, the
relative energies of the five lower-energy conformers are so small in water that a
different identification of the lowest-energy one among them does not affect the
interpretation of the results. Major changes in water occur for some conformers
having relative energy in vacuo greater than 5 kcal/mol. These cases concern
conformers in which O41−H50 or O46−H60 are not engaged in the O−H⋯π
interaction and, therefore, are available to form H-bonds with water molecules.
Although PCM does not take into explicit account solute-solvent H-bonds, it
appears sometimes to take it into account implicitly through the energetics (possibly
as the effect of the point-charges distribution in the areas of the cavity surface
corresponding to an H-bond donor or acceptor) [46]. A confirmation about the
interactions of these conformers with water molecules can be obtained through the
consideration of adducts with explicit water molecules, which might be the object
of a separate study.
The five lowest energy conformers account for 99.89% of the population in
chloroform, 99.66% in acetonitrile and 90.71% in water. The population distribution
of the other conformers (besides the five lowest energy ones) is also different. In
vacuo, no other conformer has a population greater than 0.003%; in chloroform,
there are two conformers with population of 0.02 and 0.01% respectively, and all
the other populations are <0.007%; in acetonitrile, ten other conformers have a
population between 0.02 and 0.05%; in water, three other conformers have popu-
lation between 1.0 and 1.3% and seven conformers between 0.1 and 0.9%.
Conformers of S/S′ symmetric may have slightly different relative energy in
solution, suggesting that the effect of their symmetric situation may somewhat
decrease in solution.
If 3.5 kcal/mol is taken as a cautious threshold value for conformers which
might be responsible for the biological activity, the results in water solution suggest
that most conformers of JZM (all those with relative energy ≤ 3.5 kcal/mol in water
solution) might be considered as potential responsibles for the antimalarial activity
of JZM.
Table 8 reports the solvent effect (free energy of solvation, ΔGsolv) for the
conformers listed in Table 2. ΔGsolv is positive in chloroform and acetonitrile and
negative in water for all the conformers. The values in acetonitrile are considerably
greater than those in chloroform. The electrostatic component of ΔGsolv (Gel) has
negative values in all the three solvents, but considerably more negative in water.
A quick estimation of the octanol/water partition coefficient of JZM (6.3743, [47])
suggests that JZM could be more soluble in non-polar solvents than in water. This
could be due to presence of many aromatic rings and to the high molecular mass.
On the other hand, the negative values of ΔGsolv in water suggest the possibility of
some (although limited) solubility in water or, at least, the possibility that the
molecule may be present also in the water phases of living organisms cannot be
324 M. K. Bilonda and L. Mammino

Table 8 Solvent effect (free energy of solvation, ΔGsolv) of selected conformers of jozimine A2 in
the three solvents considered Results from HF/6-31G(d,p) and DFT/6-31+G(d,p) single point
PCM calculations on the in-vacuo-optimized geometries
Conformer ΔGsolv (kcal/mol)
HF DFT
chlrf actn aq chlrf actn aq
a-b-e-f-g-h-p-v-x 7.19 17.60 −4.12 7.37 18.30 −4.12
a-b-e-f-g-h-q-v-x 6.95 17.66 −4.39 7.29 18.16 −4.64
a-b-e-f-g-h-q-t-x 7.15 17.77 −4.04 7.32 18.14 −4.57
a-b-e-f-g-h-q-v-y 6.81 17.67 −4.70 7.19 17.93 −5.09
a-b-e-f-g-h-p-t-x 7.17 17.89 −4.50 7.22 17.96 −4.87
a-b-f-h-q-v-x 5.76 16.21 −7.62 6.65 16.98 −7.54
a-d-e-f-g-h-q-v-x 6.97 17.86 −5.37 7.61 −a −a
b-c-e-f-g-h-q-t-x 7.19 17.91 −5.46 7.61 18.44 −5.50
a-b-e-g-q-t-x 5.78 16.21 −7.54 6.57 16.93 −7.72
b-c-e-f-g-h-p-v-x 6.84 18.05 −5.26 7.50 18.36 −5.60
a-d-e-f-g-h-p-v-x 7.17 17.95 −5.55 7.50 18.36 −5.57
a-b-e-g-p-v-x 6.38 15.89 −7.96 6.44 16.72 −7.83
a-b-f-h-p-v-x 5.74 15.95 −7.90 6.47 −a −7.86
a-b-e-g-p-t-x 6.01 15.96 −8.09 6.61 16.94 −7.73
a-b-f-h-p-t-x 5.88 15.94 −8.13 6.62 16.93 −7.76
c-d-e-f-g-h-p-v-x 7.57 18.43 −6.35 7.95 18.87 −6.33
a-d-e-g-p-t-x 6.13 16.10 −9.25 6.68 16.93 −8.49
a-b-q-v-y 4.89 −a −12.31 −a −a −11.36
a-b-p-t-x 5.65 14.61 −11.38 5.61 15.32 −10.92
c-d-f-h-p-v-x 6.74 17.09 −8.42 7.30 17.83 −8.25
c-d-f-h-q-v-x 6.80 17.12 −8.53 7.48 18.00 −a
b-c-q-v-y 5.36 15.08 −12.30 5.85 15.53 −11.91
a-d-q-v-y 5.76 15.07 −12.03 5.88 15.54 −11.81
c-d-q-v-y 6.10 15.53 −12.02 6.59 16.50 −11.77
c-d-p-t-x 5.91 15.47 −11.86 6.56 16.70 −11.23
a
The calculation for this conformer did not converge in the given solvent

excluded. It is probable that the ΔGsolv takes into account the presence of OHs,
which favours the solubility in water.
The presence and number of O−H⋯O IHBs, and the O−H⋯π and C−H⋯O
interaction seem to influence ΔGsolv and this effect is different for different solvents.
The first five lowest energy conformers (with a-b-e-f-g-h in common) have the
smallest absolute values of ΔGsolv in water. Conformers in which either O41−H50
or O46−H60 is not engaged in the O−H⋯π IHB have the greatest absolute values
of ΔGsolv in water. Conformers in which only O41−H50 or O46−H60 is engaged
Computational Study of Jozimine A2, a Naphthylisoquinoline … 325

in O−H⋯π interaction have greater values of ΔGsolv than the conformers in which
O41−H50 and O46−H60 are simultaneously engaged in O−H⋯π interaction.
In acetonitrile, the conformers with all the OHs engaged in IHBs have the
greatest values of ΔGsolv. The conformers in which neither O41−H50 nor O46
−H60 are engaged in O−H⋯π IHBs have the smallest absolute values of ΔGsolv
and conformers in which only either O41−H50 or O46−H60 is engaged in the O
−H⋯π IHB have values of ΔGsolv greater than the conformers in which they are not
engaged in them. The behaviour in acetonitrile is different with respect to the
behaviour in water. The importance of solute-solvent H-bonds for acetonitrile may
not be as relevant as in water, both because acetonitrile can only be H-bond
acceptor and because N is a weaker acceptor than O. The trend in chloroform is
almost similar to the trend in acetonitrile, but the conformers having all the possible
O−H⋯π interactions have slightly greater ΔGsolv than the other conformers.
Conformers of S/S′ symmetric pairs may have slightly different ΔGsolv values.
The dipole moments (Table 6) increases in chloroform with respect to in vacuo
and in acetonitrile with respect to chloroform, but decreases slightly in water with
respect to acetonitrile. This might be related to the fact that acetonitrile has greater
dipole moment than water (although several of its effects on solute molecules
appear to be intermediate between those of chloroform and those of water). Con-
formers of S/S′ symmetric pairs have the same value of dipole moment in the
solvents considered.
The HOMO-LUMO energy gap increases with the solvent polarity. Conformers
with all the O−H⋯π and C−H⋯O interactions (c-d-e-f-g-h) have the highest values
of the HOMO-LUMO energy gap in all the media, and conformers with only one O
−H⋯O IHB (O43−H52⋯O42 or O44−H54⋯O45) have the lowest energy
gap. The range of the HOMO-LUMO energy gap in vacuo is mostly 221.17
−236.56/HF and 85.43−100.79/DFT. The gap mostly increases in solution, but it
decreases in some cases; the change with respect to in vacuo is −0.41−4.34/HF and
−0.25−4.15/DFT in chloroform; −0.43−6.04/HF and −3.39−1.84/DFT in ace-
tonitrile and −0.15−8.56/HF and −0.13−8.34/DFT in water. Conformers of S/S′
symmetric pairs may have slightly different HOMO-LUMO gaps.

4 Discussion and Conclusions

A computational study of jozimine A2—a naphthylisoquinoline alkaloid with an-

timalarial activity—has been carried out at the HF/6-31G(d,p) and the DFT/B3LP/
6-31+G(d,p) levels in vacuo and in chloroform, acetonitrile and water. The results
highlight the main properties of this molecule. Conformational preferences and
other properties are influenced by the presence of O−H⋯O IHBs (in the naph-
thalene moieties) and other IHB-type interactions, such as O−H⋯π (between some
OH and a π ring in the same or in another moiety) and C−H⋯O (between a CH and
an OH of the same isoquinoline moiety), and by the mutual orientation of the
326 M. K. Bilonda and L. Mammino

moieties. The O−H⋯O IHBs are the dominant factor determining conformational
preferences and energetics.
The low relative energy of many conformers in water solution suggests that a
comparatively high number of conformers may be potential responsibles for the
molecule’s biological activity. The values of the free energy of solvation suggest
greater solubility in water than in the other two solvents considered in this study,
which would be consistent with the high polarity of the JZM molecule, but not with
the octanol/water partition coefficient of JZM or with its high molecular mass.
It is interesting to note that (differently from a number of other molecules), JZM
appears to respond to the three solvents considered according to their increasing
dipole moment, so that the responses to water are slightly smaller than those to
acetonitrile.
A typical feature of this molecule is the presence of axial chirality (stereoiso-
merism resulting from the non-planar arrangement of four groups in pairs about a
chirality axis). The presence of axial chirality suggests that rotation about the C13
−C21 bond is not free because of steric hindrance. A space-filling visualisation with
Chem3D [47] appears to confirm the hindrance, at least for lower energy con-
formers. On the other hand, the rotation scans about this bond (and about the other
two biaryl bonds) were not disrupted (i.e., the steric hindrance did not appear in the
bond-rotation model). A possible reason is the fact that the rotation scan input was a
conformer with 12.811 kcal/mol relative energy (2.32 × 10−8 population);
although expedient to identify minima in the orientation of the moieties, this
conformer is poorly populated and, therefore, it does not contribute to the mole-
cule’s actual (or experimentally determined) behaviour.
An analysis of the differences with the behaviours and properties of michel-
lamine A may be useful for a better understanding of the difference in the biological
activities of these largely similar molecules.

Acknowledgements M. K. Bilonda expresses her gratitude to the National Research Foundation

(NRF) of South Africa for a bursary to support her Ph.D. studies.

References

1. World Health Organization (2014) World malaria report 2014 summary. http://www.who.int/
malaria/publications/world_malaria_report_2014/wmr-2014-no-profiles.pdf
2. World Health Organization (2016) World malaria report 2016. http://www.who.int/malaria/
media/world-malaria-report-2016/en/
3. Bringmann G, Zhang G, Büttner T, Bauckmann G, Kupfer T, Braunschweig H, Brun R,
Mudogo V (2013) Chem Eur J 19:916–923
4. Bierer DE, Dener JM, Dubenko LG, Gerber RE, Litvak J, Peterli S, Peterli-Roth P,
Truong TV, Mao G, Bauer BE (1995) J Med Chem 38:2628
5. Ganellin CR, Mitchell RC, Young RC (1988) In: Melchiorre C, Giannella M (eds) Recent
advances in receptor chemistry. Elsevier Science Publishers B.V., Amsterdam, pp 289–306
6. Guha S, Majumdar D, Bhattacharjee AK (1992) J Mol Struct (Theochem) 256:61
7. Sjoberg P, Murray JS, Brinck T, Evans P, Politzer P (1990) J Mol Graph 8:81
Computational Study of Jozimine A2, a Naphthylisoquinoline … 327

8. Karle JM, Bhattacharjee AK (1999) Bioorg Med Chem 7:1769–1774

9. Batagin-Neto A, Lavarda FC (2014) Med Chem Res 23:580–586
10. Naranjo-Montoya OA, Martins LM, da Silva-Filho LC, Batagin-Neto A, Lavarda FC (2015) J
Braz Chem Soc 26:832–836
11. Purcell WP, Sundaram K (1969) J Med Chem 12:18–21. http://dx.doi.org/10.5935/0103-
5053.20140263
12. Mammino L, Bilonda MK (2017) In: Tadjern A, Pavlov R, Maruani J, Brändas EJ,
Delgado-Barrio G (eds) Quantum systems in physics, chemistry, and biology—advances in
concepts and applications. Book series Progress in theoretical chemistry and physics.
Springer, pp 303–316
13. Boyd MR (1994) J Med Chem 37:1740–1745
14. Becke AD (1993) J Chem Phys 98:5648–5662
15. Lee C, Yang W, Parr RG (1988) Phys Rev B 37:785–789
16. Mammino L, Bilonda MK (2016) Theor Chem Acc 135:101. https://doi.org/10.1007/s00214-
016-1843-7
17. Mammino L, Kabanda MM (2009) J Mol Struct (Theochem) 901:210–219
18. Mammino L, Kabanda MM (2012) Int J Quantum Chem 112:2650–2658
19. Harvey JN (2004) In: Nkaltsoyanis N, McGrady JE (eds) Principles and applications of
density functional theory in inorganic chemistry II. Springer, p 170
20. Irikura K, Johnson III RD, Kacker RN (2005) J Phys Chem A 109:8430–8437
21. Barone V, Cossi M (1997) J Chem Phys 107:3210–3221
22. Tomasi J, Mennucci B, Cammi R (2005) Chem Rev 105:2999–3093
23. Barone V, Cossi M, Tomasi J (1998) J Comput Chem 19:404–417
24. Cossi M, Scalmani G, Rega N, Barone V (2002) J Chem Phys 117:43–54
25. Cancès E, Mennucci B, Tomasi J (1997) J Chem Phys 107:3032–3041
26. Tomasi J, Mennucci B, Cancès E (1999) (Theochem) 464:211–226
27. Frisch MJ, Trucks GW, Schlegel HB, Scuseria GE, Robb MA, Cheeseman JR, Mont-
gomery JA, Vreven T, Kudin KN, Burant JC, Millam JM, Iyengar SS, Tomasi J, Barone V,
Mennucci B, Cossi M, Scalmani G, Rega N, Petersson GA, Nakatsuji H, Hada M, Ehara M,
Toyota K, Fukuda R, Hasegawa J, Ishida M, Nakajima T, Honda Y, Kitao O, Nakai H,
Klene M, Li X, Knox JE, Hratchian HP, Cross JB, Adamo C, Jaramillo J, Gomperts R,
Stratmann RE, Yazyev O, Austin AJ, Cammi R, Pomelli C, Ochterski JW, Ayala PY,
Morokuma K, Voth GA, Salvador P, Dannenberg JJ, Zakrzewski VG, Dapprich S,
Daniels AD, Strain MC, Farkas O, Malick DK, Rabuck AD, Raghavachari K, Foresman JB,
Ortiz JV, Cui Q, Baboul AG, Clifford S, Cioslowski J, Stefanov BB, Liu G, Liashenko A,
Piskorz P, Komaromi I, Martin RL, Fox DJ, Keith T, Al-Laham MA, Peng CY,
Nanayakkara A, Challacombe M, Gill PMW, Johnson B, Chen W, Wong MW, Gonzalez C,
Pople JA (2003) Gaussian 03. Gaussian Inc, Pittsburgh
28. Pascual-Ahuir JL, Silla E (1990) J Comput Chem 11:1047–1047
29. Silla E, Villar F, Nilsson O, Pascual-Ahuir JL, Tapia O (1990) J Mol Graph 8:168–172
30. Silla E, Tunon I, Pascual-Ahuir JL (1991) J Comput Chem 12:1077–1088
31. Reichardt C (2003) Solvents effects in organic chemistry, 3rd edn. Wiley-VCH Verlag
GmbH, Boschstr. 12, D-69469 Weinheim, Germany, p 82
32. Mammino L, Kabanda MM (2007) J Mol Struct (Theochem) 805:39–52
33. Buemi G, Zuccarello F (2002) J Mol Struct (Theochem) 581:71–85
34. Simperler A, Lampert H, Mikenda W (1998) J Mol Struct 448:191–199
35. Gilli G, Bellucci F, Ferretti V, Bertolasi V (1989) J Am Chem Soc 111:1023–1028
36. Bertolasi V, Gilli P, Ferretti V, Gilli G (1991) J Am Chem Soc 113:4017–4925
37. Gilli P, Bertolasi V, Ferretti V, Gilli G (1994) J Am Chem Soc 116:909–915
38. Nolasco MM, Ribeiro-Claro PJA (2005) Chem Phys Chem 6:496–502
39. Buemi G (2002) Chem Phys 282:181–195
40. Posokhov Y, Gorski A, Spanget-Larsen J, Duus F, Hansen PE, Waluk (2004) J Chem Phys
Chem 5:495–502
41. Sobczyk L, Grabowski SJ, Krygowski TM (2005) Chem Rev 105:3513–3560
328 M. K. Bilonda and L. Mammino

42. Jablonski M, Kaczmarek A, Sadlej AJ (2006) J Phys Chem A 110:10890–10898

43. Schalley CA, Springer A (2009) Mass spectrometry and gas-phase chemistry of non-covalent
complexes. Wiley, Hoboken, NJ, p 17
44. https://www.wiki.ed.ac.uk/display/EaStCHEMresearchwiki/How+to+analyse+the+orbitals
+from+a+Gaussian+calculation=25, cited in Synthetic Metals (2010) 160(7):643. https://doi.
org/10.1016/j.synthmet.2009.12.026
45. Zhang G, Musgrave CB (2007) J Phys Chem A 111:1554–1561
46. Mammino L (2009) Chem Phys Lett 473:354–357
47. Chem3D Ultra Version 8.0.3., ChemOffice, Cambridge Software, 2003
 Part IV
Fundamental Theory
Functional Derivatives and Differentiability
in Density-Functional Theory

Ping Xiang and Yan Alexander Wang

Abstract Based on Lindgren and Salomonson’s analysis on Fréchet differentiability

[Phys Rev A 67:056501 (2003)], we showed a specific variational path along which
the Fréchet derivative of the Levy-Lieb functional does not exist in the unnormal-
ized density domain. This conclusion still holds even when the density is restricted
within a normalized space. Furthermore, we extended our analysis to the Lieb func-
tional and demonstrated that the Lieb functional is not Fréchet differentiable. Along
our proposed variational path, the Gâteaux derivative of the Levy-Lieb functional or
the Lieb functional takes a different form from the corresponding one along other
more conventional variational paths. This fact prompted us to define a new class of
unconventional density variations and inspired us to present a modified density vari-
ation domain to eliminate the problems associated with such unconventional density
variations.

Keywords Density functional ⋅ Density variation ⋅ Functional differentiability

Functional derivative

1 Introduction

Functional differentiability plays crucial roles in density-functional theory (DFT)

[1]. Within DFT, the total electronic energy is expressed as a functional of the elec-
tron density 𝜌(𝐫) [1, 2]. To find the ground-state (GS) density and its corresponding
energy, we need to perform variational calculations based on the functional deriva-
tive [1–4]. Naturally, researchers have been interested in studying the differentiability
of density functionals since the beginning of DFT.
More than twenty years ago, based on Lieb’s early work [5], Englisch and Englisch
proved the Gâteaux differentiability of a large class of density functionals and settled

P. Xiang ⋅ Y. A. Wang (✉)

Department of Chemistry, University of British Columbia, 2036 Main Mall,
Vancouver, BC V6T 1Z1, Canada
e-mail: [email protected]

© Springer International Publishing AG, part of Springer Nature 2018 331

Y. A. Wang et al. (eds.), Concepts, Methods and Applications of Quantum Systems
in Chemistry and Physics, Progress in Theoretical Chemistry and Physics 31,
https://doi.org/10.1007/978-3-319-74582-4_18
332 P. Xiang and Y. A. Wang

the issue regarding the Gâteaux differentiability of density functionals [6, 7]. How-
ever, the Fréchet differentiability of density functionals remained unresolved until
Lindgren and Salomonson claimed its plausibility recently [8–10]. In this paper, we
reexamine Lindgren and Salomonson’s analysis to gain a better understanding about
the Fréchet differentiability of density functionals in DFT.
Mathematically speaking, a functional is a mapping from a function to a number.
G[f ], a functional of function f (x), can be expressed as f (x) ↦ G[f ]. The differential
of a functional, dG[f , 𝛿f ], is the part of the difference,

𝛿G[f , 𝛿f ] = G[f + 𝛿f ] − G[f ] , (1)

that depends on 𝛿f linearly:

⟨ ⟩
𝛿G
dG[f , 𝛿f ] = 𝛿f (x) , (2)
𝛿f (x)

where 𝛿G∕𝛿f (x) is the functional derivative of G[f ] with respect to f at point x. For
the sake of brevity, ⟨⋅⟩ is adopted as a shorthand notation for integration throughout
the text.
In DFT, there are two kinds of functional derivative: the Gâteaux derivative and
the Fréchet derivative [6–10]. Following Lindgren and Salomonson [8–10], all the
density functionals are defined on a convex space of densities,
√
 = {𝜌 | 𝜌 ≥ 0, 𝜌 ∈  1 (3 )} , (3)

where  1 (3 ) is a Sobolev space [5]:

 1 (3 ) = {q | q ∈ 2 (3 ), 𝛁q ∈ 2 (3 )} , (4)

and 2 and 3 denote the spaces of square-integrable functions and three-

dimensional real coordinates, respectively. Rather than the general definitions given
in Appendix 1, we have slightly different definitions for functional differentiability
and functional derivatives [11].
Let G be a functional from  to the real numbers. If the limit

G(𝜌0 + 𝛽𝛿𝜌) − G(𝜌0 )

dG(𝜌0 ; 𝛿𝜌) = lim (5)
𝛽→0+ 𝛽

exists, it is called the Gâteaux differential of G at 𝜌0 in the direction 𝛿𝜌. If the limit
exists for any 𝛿𝜌 such that 𝜌0 + 𝛽𝛿𝜌 ∈ , we say G is Gâteaux differentiable at 𝜌0 .
Stated in another way, provided the functional G is Gâteaux differentiable at 𝜌0 ,
the functional difference upon a density variation, 𝜌0 (𝐫) → 𝜌0 (𝐫) + 𝛽𝛿𝜌(𝐫), has two
terms,

𝛿G[𝜌0 , 𝛽𝛿𝜌] = G[𝜌0 + 𝛽𝛿𝜌] − G[𝜌0 ] = 𝛽dG[𝜌0 , 𝛿𝜌] + R[𝜌0 , 𝛽𝛿𝜌] , (6)
Functional Derivatives and Differentiability in Density-Functional Theory 333

where the last term satisfies the following limiting condition:

R[𝜌0 , 𝛽𝛿𝜌]
lim =0, (7)
𝛽→0+ 𝛽

for a scaling parameter 0 ≤ 𝛽 ≤ 1. Usually we choose another density 𝜌′ ∈  to

be 𝛿𝜌 and require 𝜌 = 𝜌0 + 𝛽𝜌′ to be always in  during the variational process;
it follows immediately from the definition of a convex subspace (see Appendix 1)
that the range of 𝛽 must be [0, 1]. In a rigorous sense, the differential dG needs to
be neither linear nor continuous in 𝛿𝜌. However, most applied literature adopt the
convention that dG is linear and continuous in 𝛿𝜌. Consequently, dG can be written
as ⟨ ⟩
𝛿G
dG[𝜌0 , 𝛿𝜌] = 𝛿𝜌(𝐫) , (8)
𝛿𝜌0 (𝐫)

in which 𝛿G∕𝛿𝜌0 (𝐫) is called the Gâteaux derivative of G at 𝜌0 (𝐫). Similar to a direc-
tional derivative, the Gâteaux derivative is a functional of 𝜌0 (𝐫) only, although along
various directions it might take different expressions (e.g., different functions of the
variable 𝐫).
If the last term of the functional difference,
⟨ ⟩
𝛿G
𝛿G[𝜌0 , 𝛿𝜌] = G[𝜌0 + 𝛿𝜌] − G[𝜌0 ] = 𝛿𝜌(𝐫) + R[𝜌0 , 𝛿𝜌] , (9)
𝛿𝜌0 (𝐫)

instead satisfies
R[𝜌0 , 𝛿𝜌]
lim =0, (10)
||𝛿𝜌||→0 ||𝛿𝜌||

for the norm of 𝛿𝜌(𝐫), ||𝛿𝜌|| = ⟨|𝛿𝜌(𝐫)|⟩, 𝛿G∕𝛿𝜌0 (𝐫) is called the Fréchet derivative.
The Fréchet derivative is a global derivative: all directions approaching 𝜌0 (𝐫) yield
the same derivative (with the same expression). By default, Fréchet differentiability
is stronger than Gâteaux differentiability.

2 Controversy over Fréchet Differentiability

Consider an N-electron quantum system under the influence of a local electron-

nuclear potential v(𝐫). In the adiabatic connection formulation [12–17], the total
Hamiltonian operator Ĥ is a sum of three terms, the kinetic-energy operator T,
̂ the
̂ ̂
potential-energy operator V, and the inter-electron coulombic repulsion operator W
with an adiabatic connection parameter 𝜔:

∑
N
∑
N
∑N
̂ + V̂ = − 1
Ĥ = T̂ + 𝜔W ∇2i + 𝜔
1
+ v(𝐫i ) , (11)
2 i=1 i<j
|𝐫i − 𝐫j | i=1
334 P. Xiang and Y. A. Wang

where i and j are dummy electron indices. For convenience, the first two operators in
Eq. (11) can be grouped into a single Hohenberg-Kohn (HK) universal operator [2],

F̂ 𝜔 = T̂ + 𝜔W
̂ . (12)

In Eqs. (11) and (12), when 𝜔 = 0, we have the noninteracting system; when 𝜔 = 1,
we have the fully interacting system instead. It is then straightforward to show that
the Schrödinger equation governing this system,

̂ = Ev Ψ ,
HΨ (13)

can be reduced to a set of single-electron Kohn-Sham equations [3, 4] with a local

effective external potential veff (𝐫),
[ ]
1
− ∇2 + veff (𝐫) 𝜙i (𝐫) = 𝜀i 𝜙i (𝐫) , (14)
2
where electron i is described by spin orbital 𝜙i with orbital energy 𝜀i . It is understood
that any wavefunction, Ψ, considered here is constructed from spin orbitals. How-
ever, for simplicity, we do not indicate any spin dependence explicitly throughout
the entire text unless otherwise noted.
Consider first the nondegenerate case. The corresponding total energy can be
expressed as a density functional,

Ev [𝜌] = F 𝜔 [𝜌] + V[𝜌] , (15)

with the electron density defined as

𝜌(𝐫) = N⟨Ψ|Ψ⟩N−1 , (16)

where the subscript “N − 1” indicates the integration to be carried out over all spatial
and spin coordinates except for one spatial coordinate of a single electron.
In Eq. (16), the normalization constraint of the electron density can be relaxed to
allow ⟨Ψ|Ψ⟩ ≠ 1. As a result, the potential-energy density functional V[𝜌] and the
HK universal density functional F 𝜔 [𝜌] are defined in the domain of unnormalized
densities:
̂
V[𝜌] = ⟨Ψ|V|Ψ⟩ = ⟨v(𝐫) 𝜌(𝐫)⟩ , (17)

and
F 𝜔 [𝜌] = FLL
𝜔
[𝜌] = inf ⟨Ψ|T̂ + 𝜔W|Ψ⟩
̂ , (18)
Ψ→𝜌

where “inf ” is the infimum or the greatest lower bound and the subscript “LL”
denotes the Levy-Lieb functional [5, 18]. It should be noted that the density of con-
cern, 𝜌, belongs to , not the convex set of N-representable densities, N :
Functional Derivatives and Differentiability in Density-Functional Theory 335
√
N ≡ {𝜌 | 𝜌 ≥ 0, ⟨𝜌⟩ = N, 𝜌 ∈  1 (3 )} . (19)

If 𝜔 = 0, Eq. (18) reduces to the definition of the noninteracting kinetic-energy den-

sity functional,
̂
Ts [𝜌] = inf ⟨Ψ|T|Ψ⟩. (20)
Ψ→𝜌

Built upon the following paradoxical statement [4]:

𝛿Ts [𝜌]
= 𝜀i − veff (𝐫), (21)
𝛿𝜌(𝐫)

Nesbet asserted that the functional derivative of the noninteracting kinetic-energy

functional in unnormalized density domain is a Gâteaux derivative rather than a
Fréchet derivative [19, 20]. Shortly after, Lindgren and Salomonson [8–10] refuted
Nesbet by pointing out that the noninteracting kinetic-energy functional used by Nes-
bet was not a proper density functional and further reasoned the Fréchet differentia-
bility of the noninteracting kinetic-energy functional.
Based on Eqs. (16) and (17), Lindgren and Salomonson [8–10] obtained the
expression of the HK universal density functional for the GS density 𝜌0 (𝐫) and wave-
function Ψ0 ,
⟨[ ] ⟩
E0
F 𝜔 [𝜌0 ] = ⟨Ψ0 |F̂ 𝜔 |Ψ0 ⟩ = ⟨Ψ0 |Ĥ − V|Ψ
̂ 0⟩ = − v(𝐫) 𝜌0 (𝐫) , (22)
N

where E0 is the GS energy. Because of the identity for arbitrary Ψ = Ψ0 + 𝛿Ψ,

̂
⟨Ψ|F̂ 𝜔 |Ψ⟩ = ⟨Ψ|H|Ψ⟩ − ⟨v(𝐫) 𝜌(𝐫)⟩
= E0 ⟨Ψ|Ψ⟩ + ⟨𝛿Ψ|Ĥ − E0 |𝛿Ψ⟩ − ⟨v(𝐫) 𝜌(𝐫)⟩
⟨[ ] ⟩
E0
= − v(𝐫) 𝜌(𝐫) + ⟨𝛿Ψ|Ĥ − E0 |𝛿Ψ⟩ , (23)
N

the corresponding Levy-Lieb density functional takes the following form:

𝜔
FLL [𝜌0 + 𝛿𝜌] = inf ⟨Ψ|F̂ 𝜔 |Ψ⟩
Ψ→𝜌0 +𝛿𝜌
⟨[ ] ⟩
E0 [ ]
= − v(𝐫) 𝜌0 (𝐫) + 𝛿𝜌(𝐫) + inf ⟨𝛿Ψ|Ĥ − E0 |𝛿Ψ⟩ . (24)
N Ψ0 +𝛿Ψ→𝜌0 +𝛿𝜌

Here, the GS density, 𝜌0 , is a pure-state v-representable (PS-v-representable) density,

which corresponds to a single GS wavefunction of a Hamiltonian with a physically
reasonable potential v in the space of ∞ + 3∕2 [5, 9, 21]. Subtracting Eq. (22)
from Eq. (24) leads to
336 P. Xiang and Y. A. Wang

𝜔 𝜔 𝜔
𝛿FLL = FLL [𝜌0 + 𝛿𝜌] − FLL [𝜌0 ]
⟨[ ] ⟩
E0
= − v(𝐫) 𝛿𝜌(𝐫) + inf ⟨𝛿Ψ|Ĥ − E0 |𝛿Ψ⟩ . (25)
N Ψ0 +𝛿Ψ→𝜌0 +𝛿𝜌

Then, the condition for Fréchet differentiability requires the last term in Eq. (25) to
satisfy
⟨𝛿Ψ|Ĥ − E0 |𝛿Ψ⟩
inf → 0 , as ||𝛿𝜌|| → 0 , (26)
Ψ0 +𝛿Ψ→𝜌0 +𝛿𝜌 ||𝛿𝜌||

for all density variations in the neighborhood of 𝜌0 (𝐫). Lindgren and Salomonson
argued that Eq. (26) was plausible because the numerator is quadratic in 𝛿Ψ whereas
𝛿𝜌 is only linear in 𝛿Ψ [8], and they analyzed this issue further based on their proof
of the Gâteaux differentiability of the Levy-Lieb functional [9].
When the ground state is degenerate, it can no longer be represented by a PS-
v-representable density. Instead, we should use an ensemble v-representable (E-v-
representable) density [6, 21],
∑ ∑
𝜌0 = sk 𝜌k0 , sk ≥ 0 , sk = 1, (27)
k k

where 𝜌k0 is a PS-v-representable density of the Hamiltonian in Eq. (11). For concise-
ness, we use the same notation 𝜌0 to represent either a PS-v-representable density or
an E-v-representable density [6, 21], and whenever needed, we will specify the type
of the density. For this degenerate case, by using a similar method as above, Lindgren
and Salomonson [9] has also shown the plausibleness of the Fréchet differentiability
of the Lieb functional [5],
∑ ⟨ ⟩
| ̂ || Ψk ,
FL𝜔 [𝜌] = inf sk Ψk |T̂ + 𝜔W (28)
sk ,Ψ →𝜌
k | |
k

where {Ψk } is any set of orthonormal eigenfunctions of the Hamiltonian in Eq. (11).

3 Analysis of Lindgren-Salomonson’s Assessment

Unfortunately, Lindgren and Salomonson’s assessment is incomplete. Before expos-

ing the limitation of their assessment, we would like to introduce the concept of
strongly orthogonal function (SOF) [22]. Two functions f (x1 , x2 , … , xm , xm+1 , … , xN )
and g(x1 , x2 , … , xm , xm+1 , … , xN ) are mutually order-m strongly orthogonal, if they
satisfy the following strongly orthogonal condition for any arbitrary (N − m) vari-
ables:
⟨f |g⟩N−m = 0 , (29)
Functional Derivatives and Differentiability in Density-Functional Theory 337

where N is the total number of variables and m is the number of variables excluded
in the integration. If f is chosen to be the seed function, we can define the order-m
SOF space that consists all functions of order-m strongly orthogonal to the same seed
function f .
Because spin operator commutes with the Hamiltonian, they share a common
complete set of eigenfunctions. In the forthcoming discussion, we only consider
wavefunctions in Hilbert space spanned by this complete set. Let the seed function
be the N-electron GS wavefunction Ψ0 and the integration in Eq. (29) be over all spa-
tial and spin coordinates except m spatial coordinates, and label  as its associated
order-1 SOF space and  as its order-0 SOF space with  excluded. For a general
Hamiltonian (0 ≤ 𝜔 ≤ 1), there is no doubt that  exists. Since we integrate over
all spin coordinates, if any non-GS eigenfunction Ψ has a different spin multiplicity
from that of Ψ0 , we have
⟨Ψ0 |Ψ⟩N−1 = 0 , (30)

then Ψ is in . Therefore,  does exist in general. For the noninteracting system

(𝜔 = 0) in particular, Ψ0 is a single Slater determinant,  includes all singly excited
determinants, and  has all doubly and higher excited determinants. Thus, the com-
plete set of eigenfunctions {Ψi } of the Hamiltonian can be partitioned into three
parts: Ψ0 , , and .
Let a normalized wavefunction Ψ be a linear combination of eigenfunctions in
. Hereafter, we are going to show that along one particular variational path defined
by
Ψp = Ψ0 + 𝛽Ψ , (31)

the Fréchet derivative does not exist. For the wavefunction Ψp , its corresponding
electron density is
⟨ ⟩
|
𝜌p (𝐫) = N Ψp |Ψp
|
⟨ ⟩ N−1 ⟨ ⟩ (⟨ ⟩ )
= N Ψ0 |Ψ0 N−1 + 𝛽 2 N Ψ ||Ψ N−1 + 2𝛽NRe Ψ ||Ψ0 N−1
|
= 𝜌0 (𝐫) + 𝛽 2 𝜌 (𝐫) , (32)
( )
where Re ⟨Ψ |Ψ0 ⟩N−1 , the real part of ⟨Ψ |Ψ0 ⟩N−1 , is zero because of the nature
of Ψ . When 𝛽 approaches 0, 𝜌p (𝐫) also approaches 𝜌0 (𝐫). Clearly, 𝜌p (𝐫) lies in the
neighborhood of 𝜌0 (𝐫) within . Equation (32) specifies the density variation path
for the forthcoming discussion. For later convenience, we label  as the set of all
legitimate 𝜌p (𝐫) for a given Ψ0 or 𝜌0 (𝐫). Throughout the text, Re(⋅) will be used to
denote the real part of the quantity involved.
̃ be a trial wavefunction yielding the
Along this particular variational path, let Ψ
same density 𝜌p (𝐫):
Ψ̃ = Ψ0 + 𝜆Ψt ⟼ 𝜌p (𝐫) , (33)
338 P. Xiang and Y. A. Wang

where Ψt is normalized to 1 and 𝜆 is a real scaling parameter (potentially different

from 𝛽). We can expand Ψt in terms of the complete set of normalized eigenfunctions
̂
{Ψi } of H,
∞
∑
Ψt = ci Ψi , (34)
i=0

where ci is the expansion coefficient of Ψi . Without loss of generality, we can choose

all expansion coefficients {ci } to be real because any complex phases in {ci } can be
attributed to the corresponding {Ψi } instead. The electron density for Ψ̃ then takes
the following form:

̃ Ψ⟩
𝜌̃(𝐫) = N⟨Ψ| ̃ N−1
( )
= N⟨Ψ0 |Ψ0 ⟩N−1 + 𝜆2 N⟨Ψt |Ψt ⟩N−1 + 2𝜆NRe ⟨Ψ0 |Ψt ⟩N−1
( )
= 𝜌0 (𝐫) + 𝜆2 𝜌t (𝐫) + 2𝜆NRe ⟨Ψ0 |Ψt ⟩N−1 . (35)

Clearly, as 𝜆 → 0, 𝜌̃(𝐫) also approaches 𝜌0 (𝐫).

Because the density variation is along the path with the same density (but with
different wavefunctions),
𝜌̃(𝐫) = 𝜌p (𝐫) → 𝜌0 (𝐫) , (36)

the normalization of the two densities must be identical,

⟨ ⟩
𝜌(𝐫)⟩ = 𝜌p (𝐫) .
⟨̃ (37)

Substituting Eqs. (32) and (35) into Eq. (37), one has
(⟨ ⟩)
𝜆2 ⟨𝜌t (𝐫)⟩ + 2𝜆NRe Ψ0 ||Ψt = 𝛽 2 ⟨𝜌 (𝐫)⟩ . (38)

Since ⟨𝜌 (𝐫)⟩ and ⟨𝜌t (𝐫)⟩ are all equal to N, Eq. (38) can be readily simplified to
∞
(⟨ ⟩) ∑ ( ⟨ ⟩)
2 2 | 2
𝛽 = 𝜆 + 2𝜆Re Ψ0 |Ψt = 𝜆 + 2𝜆 Re ci Ψ0 ||Ψi = 𝜆2 + 2𝜆c0 . (39)
i=0

At any specific point along the variational path, the value of 𝛽 is fixed, and we can
solve 𝜆 in terms of 𝛽 based on Eq. (39):
√
𝜆 = −c0 ± c20 + 𝛽 2 . (40)

If c0 ≠ 0 when 𝛽 approaches 0, 𝜆 takes a Taylor-series expansion,

[ ]
1 2 1
𝜆 = −c0 ± c0 + 𝛽 − 3 𝛽4 + ⋯ . (41)
2c0 8c0
Functional Derivatives and Differentiability in Density-Functional Theory 339

Because both 𝜆 and 𝛽 approach 0 concurrently near the end of the variational path,
only the positive sign (see Appendix 3) is allowed in Eq. (41). Thus, as 𝛽 → 0, we
get
1 2 1
𝜆= 𝛽 − 3 𝛽4 + ⋯ . (42)
2c0 8c0

Immediately, we can conclude that towards the end of the variational path, 𝜆 is of
the same magnitude as 𝛽 2 ∕c0 .
If c0 = 0, Eq. (39) immediately reduces to 𝜆2 = 𝛽 2 . In this case, the wavefunction
variation 𝛿Ψ in Eq. (26) can be regarded as linear in 𝛽 and the density variation 𝛿𝜌 =
𝛽 2 𝜌 is quadratic in 𝛽 as shown by Eq. (32). This immediately invalidates Fréchet
differentiability by destroying Eq. (26). Nonetheless, we are able to gain much deeper
understanding about the structure of Ψt through the following analysis.
Consider the case when c0 ≠ 0. Because of Eqs. (32), (35), and (36), we obtain
(⟨ ⟩ )
𝜆2 𝜌t + 2𝜆NRe Ψ0 ||Ψt N−1 = 𝛽 2 𝜌 . (43)

Substituting Eq. (39) into Eq. (43) and grouping terms according to the powers of 𝜆,
we have
[ 0
]
∑ (⟨ ⟩ )
2𝜆 c0 𝜌 − N |
ci Re Ψ0 |Ψi N−1 = 𝜆2 (𝜌t − 𝜌 ) , (44)
i

where the summation on the left-hand side (LHS) is only within the combined space
of  and Ψ0 ,
 0 ≡  ∪ Ψ0 . (45)

Separating the Ψ0 contribution from the summation, we can simplify Eq. (44) to


∑ (⟨ ⟩ ) ( )
2c0 (𝜌0 − 𝜌 ) + 2N ci Re Ψ0 ||Ψi N−1 = 𝜆 𝜌 − 𝜌t . (46)
i

Because 𝜌0 , 𝜌 , and all Ψi in Eq. (46) have no dependence on 𝜆, we must admit that
ci corresponding to any wavefunction in 0 must be a function of 𝜆.
Based on perturbation theory, ci (𝜆) can be expanded as

ci (𝜆) = c(0)
i
+ c(1)
i
𝜆 + h.o. , (47)

where “h.o.” represent higher-order terms in 𝜆. It can be readily shown that all the
c(0)
i
must be zero in order to satisfy Eq. (46) for any 𝜆 → 0, whereas c(1)
i
can be zero
or non-zero. Consequently, all {ci } for Ψi in 0 vary at least linearly in 𝜆. Some ci
340 P. Xiang and Y. A. Wang

may have a higher-than-linear dependence on 𝜆, but at least one ci is linear in 𝜆 for

Eq. (46) to be valid. At this point, we do not know the exact behavior of those {ci } for
Ψi in . Actually, in the above derivation, we only considered one constraint along
the variational path, namely, Eq. (36), but ignored the other constraint, ⟨Ψt |Ψt ⟩ = 1.
What will happen if both constraints are enforced concurrently?
By the minimization process, we can find the best set of coefficients {ci } that
delivers the lowest energy out of the following energy functional for the given 𝜌p (𝐫)
with a fixed 𝛽,

̃ H|
⟨Ψ| ̃ = ⟨Ψ0 + 𝜆Ψt |H|Ψ
̂ Ψ⟩ ̂ 0 + 𝜆Ψt ⟩
̂ 0 ⟩ + ⟨Ψ0 + 𝜆Ψt |H|𝜆Ψ
= ⟨Ψ0 + 𝜆Ψt |H|Ψ ̂ t⟩
̃ 0 ⟩ + ⟨Ψ0 |H|𝜆Ψ
= E0 ⟨Ψ|Ψ ̂ 2 ̂
t ⟩ + 𝜆 ⟨Ψt |H|Ψt ⟩
̃ Ψ⟩
= E0 ⟨Ψ| ̃ − E0 ⟨Ψ|𝜆Ψ
̃ 2 ̂
t ⟩ + E0 ⟨Ψ0 |𝜆Ψt ⟩ + 𝜆 ⟨Ψt |H|Ψt ⟩
̃ Ψ⟩
= E0 ⟨Ψ| ̃ − 𝜆2 E0 ⟨Ψt |Ψt ⟩ + 𝜆2 ⟨Ψt |H|Ψ
̂ t⟩
( )
2 2 ̂
= (1 + 𝛽 )E0 + 𝜆 ⟨Ψt |H|Ψt ⟩ − E0 . (48)
( )
̂ t ⟩ − E0 in Eq. (48)
To achieve this task, one only needs to minimize 𝜆2 ⟨Ψt |H|Ψ
under the following two constraints:
∞
∑
c2i = 1 , (49)
i=0

and
𝜌̃(𝐫) = 𝜌p (𝐫) . (50)

The second constraint, Eq. (50), is equivalent to the following equation based on our
previous analysis:
[ 
]
c0 (𝜌0 − 𝜌 ) ∑ (⟨ ⟩ )
2𝜆 + ci Re Ψ0 ||Ψi N−1
N i
( ∞
)
𝜌 ∑ ⟨ ⟩
=𝜆 2
− ci cj Ψi ||Ψj N−1 . (51)
N i,j

) to find the set of coefficients {ci }

We will use the Euler-Lagrange( multiplier method
̂ t ⟩ − E0 .
that minimizes the value of 𝜆2 ⟨Ψt |H|Ψ
Define
∞
∑
𝐀= c2i − 1 , (52)
i=0
Functional Derivatives and Differentiability in Density-Functional Theory 341
[ 
]
c0 (𝜌0 − 𝜌 ) ∑ (⟨ ⟩ )
𝐁 = 2𝜆 + ci Re Ψ0 ||Ψi N−1
N i
( ∞
)
2 𝜌 ∑ ⟨
|
⟩
−𝜆 − ci cj Ψi |Ψj N−1 , (53)
N i,j

and
( )
̂ t ⟩ − E0 − h𝐀 − ⟨g(𝐫) 𝐁⟩
𝛀 = 𝜆2 ⟨Ψt |H|Ψ
(∞ )
∑
= 𝜆2 ̂ j ⟩ − E0 − h𝐀 − ⟨g(𝐫) 𝐁⟩
ci cj ⟨Ψi |H|Ψ
i,j
( ∞
)
∑
2
=𝜆 ci cj Ej 𝛿ij − E0 − h𝐀 − ⟨g(𝐫) 𝐁⟩
i,j
[ ∞
]
∑
2
=𝜆 c2i (Ei − E0 ) − h𝐀 − ⟨g(𝐫) 𝐁⟩ , (54)
i=1

where h and g(𝐫) are the Lagrange multipliers corresponding to the two constraints
in Eqs. (52) and (53), and Eq. (49) has been used to derive the last expression. Min-
imizing Eq. (54) with respect to {ci }, one obtains
⟨[ ] ⟩
𝜌0 − 𝜌 ∑∞ (⟨ ⟩ )
𝜆 +𝜆 cj Re Ψ0 ||Ψj N−1 g(𝐫) = −hc0 , (55)
N j=0

and
⟨[ ] ⟩
(⟨ ⟩ ) ∞
∑ (⟨ ⟩ )
𝜆 Re Ψi || Ψ0 N−1 + 𝜆 cj Re Ψi ||Ψj N−1 g(𝐫)
j=0
[ 2
]
= 𝜆 (Ei − E0 ) − h ci (for i ≠ 0) . (56)

Because c0 is at least linear in 𝜆 as we showed above, we can readily infer from Eq.
(55) that g(𝐫) must take the following form,
∞
∑ g(k) (𝐫)
g𝜆 (𝐫) = g(0) (𝐫) + 𝜆k , (57)
k=1
k!

where g(0) can be zero depending on whether c0 is higher-than-linear in 𝜆 or not.

Substituting Eq. (57) back into Eq. (56) and ignoring the higher-order terms when 𝜆
approaches 0, we obtain a much simplified expression of ci for Ψi ∈ ,
⟨ (⟨ ⟩ ) ⟩
−hci = 𝜆 Re Ψi || Ψ0 N−1 g(0) (𝐫) + h.o. , (58)
342 P. Xiang and Y. A. Wang

where “h.o.” denotes higher-order terms in 𝜆. Obviously, we reach the same conclu-
sion as we derived before: {ci } for Ψi ∈  is at least linear in 𝜆 towards the end of the
variational path. For those {ci } for Ψi ∈ , where Ψi is order-1 strongly orthogonal
to Ψ0 , the first term in the square brackets on the LHS of Eq. (56) disappears, and
Eq. (56) reduces to
⟨ ⟩ ⟨ ⟩
∑0 (⟨ ⟩ ) ∑ (⟨ ⟩ )
𝜆2 cj Re Ψi ||Ψj N−1 g(𝐫) + 𝜆2 cj Re Ψi ||Ψj N−1 g(𝐫)
j j
2
= 𝜆 (Ei − E0 )ci − hci . (59)

For this equation to be valid at 𝜆 → 0, the LHS and the right-hand side (RHS) should
have the same dependence on 𝜆. On the RHS, the first term decays faster than the
second term, and the second term will dominate when 𝜆 approaches 0. Therefore,
we must match the magnitude of the second term on the RHS to the LHS. Of course,
we cannot match it with the second term on the LHS because doing so will lead to
self inconsistency. Then, the second term on the RHS must decay the same way as
the first term on the LHS. So, {ci } for Ψi ∈  are proportional to 𝜆3 or higher-than-
cubic terms in 𝜆. Unfortunately, such a behavior is contradictory to the normalization
∑
constraint in Eq. (49). Otherwise, i c2i will become 0 as 𝜆 → 0. Hence, we conclude
∑
that this contradiction must come from the initial assumption: Ψt = i ci Ψi , where
the expansion is over the complete set of eigenfunctions of H.̂
To resolve this contradiction, we have to modify our assumption about Ψt . We
∑
notice that if the summation i ci Ψi includes any wavefunction from 0 , the same
problem will persist. Thus, Ψt can only be expanded in ,


∑
Ψt = ci Ψi . (60)
i

In this case, Eq. (50) is equivalent to

𝜆2 𝜌t (𝐫) = 𝛽 2 𝜌 (𝐫) . (61)

Integrating both sides of Eq. (61) over the entire space of 𝐫, one finds

𝜆2 = 𝛽 2 , (62)

which further ensures that

𝜌t (𝐫) = 𝜌 (𝐫) . (63)

Now, the original minimization process is reduced to minimizing the following

term,
Functional Derivatives and Differentiability in Density-Functional Theory 343
( ) ( 
) ⟨( 
) ⟩
∑ ∑ 𝜌 ∑ ∗
2 2
𝚵= |ci | Ei − h |ci | − 1 − − ci cj ⟨Ψi |Ψj ⟩N−1 g(𝐫) ,
i i
N i,j
(64)

where all traces of 𝜆 (or 𝛽) are completely gone. The minimization will yield the
optimal set of expansion coefficients {̄ci }, which has no dependence on 𝜆 or 𝛽 from
the appearance of Eq. (64). At this stage, we are ready to test the condition for Fréchet
differentiability shown in Eq. (26):

⟨𝛿Ψ|Ĥ − E0 |𝛿Ψ⟩ ⟨𝛿Ψ|Ĥ − E0 |𝛿Ψ⟩

inf = inf
Ψ0 +𝛿Ψ→𝜌0 +𝛿𝜌 ||𝛿𝜌|| Ψ0 +𝛿Ψ→𝜌0 +𝛿𝜌 ||𝜌p − 𝜌0 ||
inf ⟨Ψ |Ĥ − E0 |Ψt ⟩
𝜆2 ⟨Ψt |Ĥ − E0 |Ψt ⟩ Ψt →𝜌 t
= inf =
Ψt →𝜌 ||𝛽 2 𝜌 || ||𝜌 ||
⟨ | | 
⟩ ( )
1 ∑ | |∑ 1 ∑
= | ̂ |
c̄ i Ψi | H − E0 | c̄ j Ψj = 2
|̄ci | Ei − E0
||𝜌 || | | j ||𝜌 ||
i | | i
( )
1 ∑
2
> |̄ci | E0 − E0 = 0 , (65)
||𝜌 || i

where Eq. (62) is used to simplify the expression after the second equal sign. Evi-
dently, Eq. (65) suggests that the condition for Fréchet differentiability is not fulfilled.
Since the above infimum approaches a nonzero constant towards the end of the
variational path, it is possible to combine the linear-order term of 𝛿𝜌 from the sec-
ond term with the first term in Eq. (25), and the new resulting residual term might
satisfy the condition for Fréchet differentiability. However, the corresponding path-
dependent functional derivative will be different from the one obtained by Lindgren
and Salomonson [8–10]. This is contradictory to the fact that the Fréchet derivative
is a global derivative, independent of variational path. Therefore, we have no choice
𝜔
but to conclude that the Levy-Lieb density functionals FLL are not Fréchet differ-
entiable at PS-v-representable densities. The places where Fréchet differentiability
breaks down (e.g., along our specially designed variational path in ) are exactly the
same locations where the Gâteaux derivative espouses different forms.
In the above analysis, we worked within unnormalized density domain. The situa-
tion in normalized density domain is almost identical. This can be straightforwardly
proven by normalizing the wavefunction after finishing the minimization process
in the unnormalized wavefunction space, simply because all the wavefunctions of
our concern are normalizable in Hilbert space and the normalization factor does not
affect the expectation values of observables. In Appendix 2, we have offered a much
more detailed but rather lengthy proof to further confirm our assessment here.
After showing the non-Fréchet differentiability of the Levy-Lieb functionals, we
are ready to examine the differentiability of the Lieb functionals, shown in Eq. (28).
344 P. Xiang and Y. A. Wang

Using similar derivations as shown in Sect. 2, we get the variation of the Lieb func-
tional:
⟨[ ] ⟩
∑ E0 ∑
𝛿FL [𝜌0 ] = sk − v(𝐫) 𝛿𝜌k (𝐫) + sk k inf ⟨𝛿Ψk |Ĥ − E0 |𝛿Ψk ⟩
k
N k
Ψ0 +𝛿Ψk →𝜌k0 +𝛿𝜌k
⟨[ ] ⟩
E0 ∑
= − v(𝐫) 𝛿𝜌0 (𝐫) + sk k inf ⟨𝛿Ψk |Ĥ − E0 |𝛿Ψk ⟩ , (66)
N k
Ψ0 +𝛿Ψk →𝜌k0 +𝛿𝜌k

based on Eq. (27), where the total density variation 𝛿𝜌 is a linear combination of
individual variation of 𝜌k0 ,
∑
𝛿𝜌 = sk 𝛿𝜌k , (67)
k

where sk is the same as that in Eq. (27). Then, the condition for the Lieb functional
to be Fréchet differentiable is

∑ ⟨𝛿Ψk |Ĥ − E0 |𝛿Ψk ⟩

sk inf → 0 , as ||𝛿𝜌|| → 0 . (68)
k
Ψk0 +𝛿Ψ →𝜌k0 +𝛿𝜌k
k ||𝛿𝜌||

For each PS-v-representable density 𝜌k0 in the expansion of the E-v-representable

density 𝜌0 , we use the same kind of variational path as shown before. Let the density
𝜌kp = 𝜌k0 + 𝛿𝜌k along the variational path be generated by the wavefunction:

Ψkp = Ψk0 + 𝛽Ψk , (69)

as 𝛽 approaches zero. The corresponding total density variation is

∑
𝛿𝜌 = 𝛽 2 sk 𝜌k . (70)
k

Along each variational path, let the trial wavefunction to be

̃ k = Ψ k + 𝜆k Ψ k .
Ψ (71)
0 t

By the same analysis, we can show that each scaling parameter 𝜆k must satisfy

𝜆2k = 𝛽 2 , (72)

and each trial wavefunction Ψkt can only be expanded in the corresponding  space,


∑
Ψkt = cki Ψki . (73)
i
Functional Derivatives and Differentiability in Density-Functional Theory 345

Consequently, we have

∑ ⟨𝛿Ψk |Ĥ − E0 |𝛿Ψk ⟩ ∑ 𝜆2k ⟨Ψkt |Ĥ − E0 |Ψk ⟩

sk inf = s inf ∑
k
Ψk0 +𝛿Ψ →𝜌k0 +𝛿𝜌k
k ||𝛿𝜌|| k
k k
𝜌t →𝜌k 𝛽 2 ||
k sk 𝛿𝜌 ||
k

∑ ⟨Ψkt |Ĥ − E0 |Ψkt ⟩

= sk kinf k ∑ , (74)
𝜌t →𝜌 ||
k sk 𝛿𝜌 ||
k
k

which is positive and cannot be zero for any positive semi-definite set of {sk } because

⟨Ψkt |Ĥ − E0 |Ψkt ⟩

inf k ∑ >0. (75)
𝜌t →𝜌 ||
k sk 𝛿𝜌 ||
k k

Therefore, the condition for Fréchet differentiability of the Lieb functional is not
fulfilled, and the Lieb functional is not Fréchet differentiable at any E-v-representable
densities.

4 Unconventional Density Variations

In this section, we will examine the implications of our above results.

Density variation in Hilbert space is defined as the difference of a trial density
𝜌(𝐫) from a GS density 𝜌0 (𝐫),

𝛿𝜌(𝐫) = 𝜌(𝐫) − 𝜌0 (𝐫) = 𝛽𝜂(𝐫) , (76)

where the scaling parameter 𝛽 can take any real value as long as the resultant density
𝜌(𝐫) is not negative,
𝜌(𝐫) = 𝜌0 (𝐫) + 𝛽𝜂(𝐫) ≥ 0 , (77)

everywhere in the entire space, for a given 𝜂(𝐫). As said before, if we let 𝜂(𝐫) to be
a density in , then 𝛽 must be in the range [0, 1] because of the convexity of .
Perdew and Levy proposed a kind of unconventional density variations (UDVs)
which satisfy [23]:
𝜂(𝐫)
√ ∉ 2 . (78)
𝜌0 (𝐫)

This class of UDVs gives rise to an unconventional energy variation, 𝛿Ev , of sub-
quadratic order of 𝛽 [25],

𝛿Ev [𝜌0 , 𝛿𝜌] = Ev [𝜌0 + 𝛿𝜌] − Ev [𝜌0 ] ∝ O(𝛽 b ), 1 < b < 2 , (79)

as opposed to the O(𝛽 2 ) behavior of conventional density variations (CDVs).

346 P. Xiang and Y. A. Wang

In general, the energy variation can be written as

⟨ ⟩
𝛿Ev ||
𝛿Ev [𝜌0 , 𝛿𝜌] = 𝛿𝜌(𝐫) + h.o. , (80)
𝛿𝜌(𝐫) ||𝜌0

where “h.o.” encompasses all higher-order terms in 𝛿𝜌(𝐫). Zhang and Wang [25]
recently discovered that Eq. (80) cannot be expanded to second order in 𝛿𝜌(𝐫) for
the UDVs of the first kind [23]. In other words, the second (or any higher-order)
functional derivative of the total energy functional may not exist for all allowed den-
sity variations within the current DFT framework. This imposes a serious problem
in DFT since the effort to interpret the second functional derivative as the chemical
hardness [1] will fail for the UDVs of the first kind [23]. Some modification of the
density variation domain must be in place to rescue the situation.
From a different perspective, we introduce a new definition of UDV based on
our discussion in Sect. 3. If a density variation, 𝛿𝜌(𝐫) = 𝜌(𝐫) − 𝜌0 (𝐫), comes from a
density 𝜌(𝐫) ∈ , then the density variation is an UDV. For these UDVs of the second
kind, their corresponding Gâteaux derivatives are different from the conventional
one [8–10], or simply, not Fréchet derivatives. √
Let us write the wavefunction in Eq. (31) differently as Ψ′p = Ψ0 + 𝛽Ψ , the
associated density variation will become linear in 𝛽,

𝛿𝜌′(𝐫) = 𝛽𝜌 (𝐫) . (81)

√
Because of Eqs. (31), (33), and (62), 𝛿Ψ′ = 𝛽Ψt , and Eq. (65) is invariant,
√ √
⟨𝛿Ψ′ |Ĥ − E0 |𝛿Ψ′ ⟩ ⟨ 𝛽Ψt |Ĥ − E0 | 𝛽Ψt ⟩
inf = inf
Ψ0 +𝛿Ψ′ →𝜌0 +𝛿𝜌′ ||𝛿𝜌′ || Ψt →𝜌 ||𝛽𝜌 (𝐫)||
̂
inf ⟨Ψt |H − E0 |Ψt ⟩
Ψt →𝜌
= = constant ≠ 0 . (82)
||𝜌 ||

Equation (25) then becomes

⟨[ ] ⟩
E0
𝜔
𝛿FLL = − v(𝐫) 𝛿𝜌′(𝐫) + inf ⟨𝛿Ψ′ |Ĥ − E0 |𝛿Ψ′ ⟩ . (83)
N Ψ0 +𝛿Ψ′ →𝜌0 +𝛿𝜌′

From Eq. (82), we know that the second term on the RHS of Eq. (83) is of linear
order in 𝛿𝜌′(𝐫) when the density is approaching 𝜌0 (𝐫). We can then split the second
term in Eq. (83) into a first-order term of 𝛿𝜌′(𝐫) and a higher-order residual:

inf ⟨𝛿Ψ′ |Ĥ − E0 |𝛿Ψ′ ⟩ = ⟨K(𝐫)𝛿𝜌′(𝐫)⟩ + R[𝜌0 , 𝛿𝜌′ ] , (84)

Ψ0 +𝛿Ψ′ →𝜌0 +𝛿𝜌′
Functional Derivatives and Differentiability in Density-Functional Theory 347

where K(𝐫) has no dependence on 𝛿𝜌′(𝐫) and R[𝜌0 , 𝛿𝜌′ ] satisfies:

R[𝜌0 , 𝛿𝜌′ ]
lim =0. (85)
𝛽→0 𝛽

Substituting Eq. (84) into Eq. (83), we obtain

⟨ 𝜔 |
⟩
𝜔
𝛿FLL |
𝛿FLL = | 𝛿𝜌 (𝐫) + R[𝜌0 , 𝛿𝜌′ ] ,
′
(86)
𝛿𝜌(𝐫) ||𝜌
0

where the new Gâteaux derivative is

𝜔 |
𝛿FLL | E0
| = − v(𝐫) + K(𝐫) . (87)
𝛿𝜌(𝐫) ||𝜌 N
0

If a CDV path is chosen instead, we will get the Gâteaux derivative previously
obtained by Lindgren and Salomonson [8–10]:
𝜔 |
𝛿FLL | E0
| = − v(𝐫) . (88)
𝛿𝜌(𝐫) ||𝜌 N
0

Equations (87) and (88) clearly indicate that the nonuniqueness of the Gâteaux
derivative along different paths. For the Lieb functionals FL𝜔 , it can be easily shown
that the same conclusion is still valid.
Because of the existence of UDVs of the first kind [23] and the second kind,
density variation domain has to be cleansed so that consistent results can be obtained.
However, there remains one problem: What is the relationship between these two
kinds of UDVs? In other words, does the set of Perdew and Levy’s UDVs contains
our UDVs or is the opposite true? Consider a noninteracting two-electron atom with
nuclear charge Z = 2, its ground state is 1s2 and the GS wavefunction is denoted by
Φ1s2 . If both electrons are excited to the 2s orbital, we would get a wavefunction Φ2s2
from the  space of Φ1s2 . The density variation,

𝜌(𝐫) = 𝜌0 (𝐫) + 𝛽𝛿𝜌(𝐫) = 𝜌1s2 (𝐫) + 𝛽𝜌2s2 (𝐫) , (89)

with 0 ≤ 𝛽 ≤ 1, would belong to the UDVs of the second kind. Due to the asymptotic
behavior of wavefunctions [1], we have
√
lim 𝜌1s2 (𝐫) ∼ e−2𝐫 2I1s2
(90)
𝐫→∞

and √
lim 𝜌2s2 (𝐫) ∼ e−2𝐫 2I2s2
, (91)
𝐫→∞
348 P. Xiang and Y. A. Wang

where I1s2 and I2s2 are first ionization potentials for 1s2 and 2s2 configurations,
respectively. Because the system concerned here is a noninteracting one, it can be
readily shown that I1s2 = 4I2s2 . In this case, we have

𝜂(𝐫) 𝜌 2 (𝐫) √ √
lim √ = lim √ 2s = lim e−𝐫 ( 8I2s2 − 2I1s2 ) = 1 . (92)
𝐫→∞
𝜌0 (𝐫) 𝐫→∞ 𝜌1s2 (𝐫) 𝐫→∞

Consequently, Perdew and Levy’s condition for the UDVs of the first kind, Eq. (78),
is satisfied and the density variation, Eq. (89), also belongs to the UDVs of the first
kind.
However, not all UDVs of the second kind are UDVs of the first kind. Here is
an example. Replacing the atom in the last example with a noninteracting 4-electron
hydrogen-like atom and letting 𝛿𝜌 to be from the wavefunction Φ1s2 3s2 , which is in
the  space of GS wavefunction Φ1s2 2s2 , then we have a new density variational path:

𝜌′ (𝐫) = 𝜌0 (𝐫) + 𝛽𝛿𝜌(𝐫) = 𝜌1s2 2s2 (𝐫) + 𝛽𝜌1s2 3s2 (𝐫) , (93)

with 0 ≤ 𝛽 ≤ 1. Then, we instead have

𝜂(𝐫) 𝜌 2 2 (𝐫) √ √
lim √ = lim √ 1s 3s = lim e−𝐫 ( 8I1s2 3s2 − 2I1s2 2s2 ) = 0 , (94)
𝐫→∞
𝜌0 (𝐫) 𝐫→∞ 𝜌1s2 2s2 (𝐫) 𝐫→∞

because the first ionization potentials of the two configurations, 1s2 2s2 and 1s2 3s2 ,
satisfy a different equation:

4I1s2 2s2 = 9I1s2 3s2 = I1s2 . (95)

Clearly, the condition for Perdew and Levy’s UDVs cannot be fulfilled in this case.
From the discussion above, we can see that the set of the UDVs of the second
kind is not enclosed in the set of the UDVs of the first kind. However, the question
of whether the UDVs of the second kind fully contain the UDVs of the first kind is
still left open. Most likely, these two sets of UDVs share some common elements,
but not mutually inclusive.
The current definition of density variation domain [25] is based on the pioneer
work of Lieb [5] and of Englisch and Englisch [6, 7, 24]. For wavefunctions in
Hilbert space, the density of concern belongs to the convex set of N-representable
densities, N . For wavefunctions in Fock space, the density domain is the direct sum
of N : ⨁
 ≡ N , (96)
N∈+

where + is the space of positive real numbers.

To ensure density functionals to be analytic through second order, we should
require the density to stay in the following modified variational domain (without
Functional Derivatives and Differentiability in Density-Functional Theory 349

the UDVs of the first kind [23]):

{ | }
| 𝜌 − 𝜌
N ≡ 𝜌 || 𝜌0 ∈ N , 𝜌 ∈ N , √ 0
∈ 2 , (97)
| 𝜌0
|
where N is the set of GS densities for an N-particle quantum system [5]. In addition,
to guarantee Fréchet differentiability of density functionals, the density should be
further restricted within
{ }
′N ≡ 𝜌 || 𝜌0 ∈ N , 𝜌 ∈ N , 𝜌 ∉ N , (98)

without the UDVs of the second kind, where N is defined in Appendix 2. In other
words, the density must be in the nexus of the above two restricted density domains:
{ | }
| 𝜌 − 𝜌
𝜌(𝐫) ∈ N′
≡ N ∩ ≡ 𝜌 || 𝜌0 ∈ N , 𝜌 ∈ N , √
′N 0 2
∈  , 𝜌 ∉ N ,
| 𝜌0
|
(99)
for wavefunctions in Hilbert space. Accordingly, we have to restrict the density for
wavefunctions in Fock space:
⨁
𝜌(𝐫) ∈  ′ ≡ N′ . (100)
N∈+

5 Conclusions

Within the current framework of DFT, the Levy-Lieb functionals are not Fréchet
differentiable at PS-v-representable densities and the Lieb functionals are not Fréchet
differentiable at E-v-representable densities. For the Levy-Lieb functionals, when
the density variation comes from a wavefunction in , the Gâteaux derivatives will
become path-dependent, taking a different form from the conventional ones. For the
Lieb functionals, when the variation of each individual PS-v-representable density
that comprises the total density comes from a wavefunction in its corresponding 
space, the Gâteaux derivatives will take a different form from the conventional ones.
Based on the analysis on UDVs, we have proposed necessary modifications on the
density variational domain on which density functionals are Fréchet differentiable
and possess the conventional analytic density expansion through second order in
density variation.

Acknowledgements Financial support for this project was provided by a grant from the Natural
Sciences and Engineering Research Council (NSERC) of Canada.
350 P. Xiang and Y. A. Wang

Appendix 1

Here, we will briefly introduce some mathematical concepts relevant to our discus-
sion in the main text. All the following content are adopted from an introductory
book on functional analysis [26].

Definition 1 A vector space  is a set of elements called vectors with two opera-
tions called addition and scalar multiplication, which satisfy the following axioms.

∙ Addition axioms: To every pair of vectors x, y ∈ , there corresponds a unique

vector x + y ∈ , the sum of x and y, such that
1. x + y = y + x;
2. (x + y) + z = x + (y + z);
3. there exists a unique zero vector 𝜃 ∈  such that x + 𝜃 = 𝜃 + x = x, ∀x ∈ ;
4. for every vector x there exists a unique vector (−x) ∈  such that x + (−x) = 𝜃.
∙ Scalar multiplication axioms: To every scalar 𝛼 and every vector x ∈  there cor-
responds a unique vector 𝛼x ∈  such that
1. 𝛼(𝛽x) = (𝛼𝛽)x for every scalar 𝛽;
2. 1x = x, 0x = 0, ∀x ∈ ;
3. 𝛼(x + y) = 𝛼x + 𝛼y and (𝛼 + 𝛽)x = 𝛼x + 𝛽x.

Definition 2 If x and y are two points of a vector space, then the line segment joining
them is the set of elements {𝛽x + (1 − 𝛽)y | 0 ≤ 𝛽 ≤ 1}. A subset S of a vector space
is convex if the line segment of joining any two points in S is contained in S.

Definition 3 Let  and  be two vector spaces with the same system of scalars.
Then a function (or mapping) that maps uniquely the elements of  onto elements
of  ,
T∶ → (101)

is called a linear transformation of  into  if

1. T(x + y) = Tx + Ty, ∀x, y ∈ ;
2. T(𝛼x) = 𝛼Tx, ∀x ∈  and for all scalars 𝛼.

Definition 4 A metric (or distance function) on a set S is a real-valued function

d(x, y) defined for all pairs of elements x and y in S and which satisfies the following
axioms:
1. d(x, y) > 0; d(x, y) = 0, if and only if x = y;
2. d(x, y) = d(y, x), ∀x, y ∈ S;
3. d(x, z) ≤ d(x, y) + d(y, z), ∀x, y, z ∈ S.
A metric space denoted by (S, d) consists of a set S and a metric d on S.
Functional Derivatives and Differentiability in Density-Functional Theory 351

Definition 5 Let T be an operator (mapping, transformation) whose domain Dom(T)

and range Ran(T) belong to metric spaces (X, dX ) and (Y, dY ), respectively. The oper-
ator T is continuous at point x0 ∈ Dom(T) if, for every 𝜖 > 0, there exists 𝛿 > 0 such
that
dY (Tx, Tx0 ) < 𝜖 (102)

whenever
dX (x, x0 ) < 𝛿 . (103)

Definition 6 A sequence {x(k) } in a metric space (S, d) is said to be a Cauchy

sequence if d(x(k) , x(l) ) → 0 as k, l → ∞. This means that for every 𝛿 > 0 there exists
N𝛿 such that d(x(k) , x(l) ) ≤ 𝛿 for any k, l ≥ N𝛿 .

Definition 7 A metric space (S, d) is said to be complete if every Cauchy sequence

in (S, d) has a limit in (S, d).

Definition 8 A norm (or length function) on a vector space  is a real-valued func-

tion, ||x||, defined for all vectors x ∈  and which satisfies the following axioms:
1. ||x|| > 0; ||x|| = 0 if and only if x = 𝜃;
2. ||x + y|| ≤ ||x|| + ||y||, ∀x, y ∈ ;
3. ||𝜆x|| = |𝜆| ⋅ ||x||, for an arbitrary scalar 𝜆.
A normed vector space, denoted by (, || ⋅ ||) consists of a vector space  and a
norm || ⋅ || on .

Definition 9 A complete (with respect to the norm) normed vector space is called
a Banach space.

Definition 10 Let T ∶  →  be a bounded linear transformation, that is,

||Tx|| ≤ K||x||. (104)

The smallest value of K which satisfies this inequality is denoted by ||T|| and called
the norm of T. It can be verified that this norm for operators satisfies the axioms for a
norm function and that we may therefore talk of the vector space of bounded linear
transformations T ∶  →  . This normed vector space is denoted by (,  ).

Definition 11 Consider an operator T ∶  →  where  is a vector space and 

is a normed vector space. Let the domain of the operator T, Dom(T) ⊂ , and s ∈ :
if the limit
T(x + 𝜆s) − T(x)
dT(x; s) = lim (105)
𝜆→0 𝜆

exists, it is called the Gâteaux differential of T at x in the direction s. The limit is

to be understood in the sense of convergence with respect to the norm in  . The
differential may exist for some s and fail to exist for others: if the differential exists
at x for all s we say that T is Gâteaux differentiable at x.
352 P. Xiang and Y. A. Wang

The Gâteaux differential is homogeneous in s in the sense that

dT(x; 𝛼s) = 𝛼dT(x; s) (106)

but is in general neither linear nor continuous in s. Nor does the existence of the
Gâteaux differential at x ensure continuity of T at x. For example,
{
𝜉13
(𝜉1 , 𝜉2 ≠ 0)
f (𝜉1 , 𝜉2 ) = 𝜉2 . (107)
0 (𝜉1 = 𝜉2 = 0)

At point (0, 0), it can be easily shown that the Gâteaux differential exists and it is
zero. Clearly, the Gâteaux differential is a continuous linear operator. However, f is
not continuous at (0,0). Therefore, we cannot relate the Gâteaux differentiability of
T to the continuity of T.
Let us go forward on the basis that  is also a normed vector space. Suppose
dT(x; s) is linear and continuous in s for some x ∈ , then we may write

T(x + 𝜆s) − T(x)

dT(x; s) = lim = TG′ (x)s . (108)
𝜆→0 𝜆

The operator TG′ is by definition, a mapping  →  and is linear and continuous:

we may conclude that
TG′ (x) ∈ (,  ) . (109)

This operator is called the Gâteaux or weak derivative of T at x. It is very important

to note that when speaking of the linearity and continuity of TG′ (x), we means those
properties in the operator sense with respect to a fixed s. TG′ itself may be a function of
x, but its continuity and linearity with respect to the variable x are complete different
things from the continuity and linearity we discussed here.
When TG′ (x) exists, it is certainly true that

T(x + 𝜆s) − T(x) = TG′ (x)𝜆s + 𝜖(x, s, 𝜆) , (110)

where 𝜖∕𝜆 → 0 as 𝜆 → 0 with x and s fixed. However, the convergence may not be
uniform with respect to s and in that case T cannot be approximated by a linear oper-
ator with uniform accuracy in the neighborhood of x. If we further demand uniform
convergence then we arrive at the strong derivative.

Definition 12 Let  and  be normed vector spaces. An operator T ∶  →  is

Fréchet differentiable at x ∈ Dom(T) ⊂  if there exists a continuous linear oper-
ator TF′ (x) ∈ (,  ) such that, for all s ∈ ,

T(x + s) − T(x) = TF′ (x)s + 𝜖(x; s) (111)

Functional Derivatives and Differentiability in Density-Functional Theory 353

with
||𝜖(x; s)||
lim =0. (112)
||s|| →0 ||s||

The operator TF′ (x) is called the Fréchet or strong derivative of T at x. The Fréchet
derivative at x is unique. It can be shown that the existence of the Fréchet derivative
of T at x implies continuity of T at x.

Theorem 1 If the Gâteaux derivative TG′ (x) exists in the neighborhood of x and is
continuous with respect to the norm in (,  ) at x, then the Fréchet derivative
TF′ (x) exists and is equal to TG′ (x).

Appendix 2

In this appendix, we show that the Fréchet derivative does not exist in the normalized
density domain, N .
Define a normalized path wavefunction,
√
Ψp = 1 − 𝛽 2 Ψ0 + 𝛽Ψ , (113)

where Ψ0 is the GS wavefunction for an N-electron quantum system, Ψ is a lin-

ear combination of eigenfunctions in  of Ψ0 , and 0 ≤ 𝛽 ≤ 1. Both Ψ0 and Ψ are
normalized to 1. The corresponding path density is

𝜌p (𝐫) = N⟨Ψp |Ψp ⟩N−1

= (1 − 𝛽 2 )N⟨Ψ0 |Ψ0 ⟩N−1 + 𝛽 2 N⟨Ψ |Ψ ⟩N−1
= (1 − 𝛽 2 )𝜌0 (𝐫) + 𝛽 2 𝜌 (𝐫) . (114)

When 𝛽 approaches 0, 𝜌p (𝐫) also approaches 𝜌0 (𝐫). Letting 𝛽 changes continuously

from 1 to 0, we obtain the desired density variational path. Equation (114) shows
that the path density is automatically normalized to N, therefore the density variation
stays within the normalized space. Clearly, 𝜌p (𝐫) lies in the neighborhood of 𝜌0 (𝐫)
within N . For convenience, we label N as the set of all legitimate N-representable
𝜌p (𝐫) defined for a given Ψ0 or 𝜌0 (𝐫) in Eq. (114).
A trial wavefunction is then assumed to yield the same path density:

√ √ ∞
∑
̃=
Ψ 1 − 𝛽 2 Ψ0 + 𝜆Ψt = 1 − 𝛽 2 Ψ0 + 𝜆 ci Ψi ⟼ 𝜌p (𝐫) , (115)
i=0

̂ ⟨Ψt |Ψt ⟩ = 1, and the expansion

where Ψi is the ith normalized eigenfunction of H,
coefficients {ci } are chosen to be real. The complete set of {Ψi } can be divided into
̃ takes the
three parts: Ψ0 , , and . The electron density (the trial density) for Ψ
354 P. Xiang and Y. A. Wang

following form:

̃ Ψ⟩
𝜌̃(𝐫) = N⟨Ψ| ̃ N−1
√ ( )
= (1 − 𝛽 2 )N⟨Ψ0 |Ψ0 ⟩N−1 + 𝜆2 N⟨Ψt |Ψt ⟩N−1 + 2𝜆 1 − 𝛽 2 NRe ⟨Ψ0 |Ψt ⟩N−1
√ ( )
= (1 − 𝛽 2 )𝜌0 (𝐫) + 𝜆2 𝜌t (𝐫) + 2𝜆 1 − 𝛽 2 NRe ⟨Ψ0 |Ψt ⟩N−1 . (116)

At any point, the trial density is identical to the path density to ensure that the density
variation is actually along the path we designed:

𝜌̃(𝐫) = 𝜌p (𝐫) → 𝜌0 (𝐫) . (117)

Therefore, we have ⟨ ⟩
𝜌(𝐫)⟩ = 𝜌p (𝐫) .
⟨̃ (118)

Substituting Eqs. (114) and (116) into Eq. (118) and simplifying the result, one
derives √
𝛽 2 = 𝜆2 + 2𝜆c0 1 − 𝛽 2 . (119)

At one specific point on the variational path, the value of 𝛽 is fixed, we can solve 𝜆
in terms of 𝛽 based on Eq. (119):
√ √
𝜆 = −c0 1 − 𝛽 2 ± c20 (1 − 𝛽 2 ) + 𝛽 2 . (120)

Near the end of the variational path, when 𝛽 → 0 and c0 ≠ 0,

√ [ √ ]
1 2
𝜆 → −c0 1 − 𝛽 2 ± c0 1 − 𝛽 2 + 𝛽 +⋯ . (121)
2c0

Again (see Appendix 3), the positive sign is chosen in Eq. (121), and we have

1 2
𝜆→ 𝛽 + ⋯ , as 𝛽 → 0 . (122)
2c0

Immediately, we can conclude that towards the end of variational path, 𝜆 is of the
same magnitude of 𝛽 2 ∕c0 . In other words, 𝜆 also approaches zero at nearly the same
rate as 𝛽 2 ∕c0 approaches zero.
Because of Eqs. (114), (116), and (117), we obtain
√ (⟨ ⟩ )
𝜆2 𝜌t + 2N𝜆 1 − 𝛽 2 Re Ψ0 ||Ψt N−1 = 𝛽 2 𝜌 . (123)
Functional Derivatives and Differentiability in Density-Functional Theory 355

Substituting Eq. (119) into Eq. (123) yields

[ 0
]
√ ∑ (⟨ ⟩ )
2 1 − 𝛽 2 c0 𝜌 − N ci Re Ψ0 ||Ψi N−1 = 𝜆(𝜌t − 𝜌 ) , (124)
i

where the summation on the LHS is only within 0 . At 𝛽 → 0, we find that the
coefficients {ci } for Ψi ∈ 0 are linear in 𝜆.
After knowing the property of {ci } for wavefunctions in 0 , we then investigate
other remaining {ci } for wavefunctions in . At one particular point on the varia-
tional path (𝛽 fixed), we optimize trial wavefunction to find out the set of coefficients
{ci } that yields the lowest energy for
⟨√ ⟩
̃ H| ̃ =
̂ Ψ⟩ | |√
⟨Ψ| 1 − 𝛽 2 Ψ0 + 𝜆Ψt |Ĥ | 1 − 𝛽 2 Ψ0 + 𝜆Ψt
| |
[ √ ]
= E0 − 𝛽 − 2𝜆c0 1 − 𝛽 2 E0 + 𝜆2 ⟨Ψt |H|Ψ
2 ̂ t⟩
( )
̂ t ⟩ = E0 + 𝜆2 ⟨Ψt |H|Ψ
= E0 − 𝜆2 E0 + 𝜆2 ⟨Ψt |H|Ψ ̂ t ⟩ − E0 , (125)

where Eq. (119) has been used to simplify the expression after the second equal sign.
Obviously, we only need to minimize the last term in Eq. (125) under the following
two constraints:
∞
∑
c2i = 1 , (126)
i=0

and
𝜌̃(𝐫) = 𝜌p (𝐫) . (127)

The density constraint, Eq. (127), is equivalent to the following identity based on
our previous analysis:
[ 
]
√ c0 (𝜌0 − 𝜌 ) ∑ (⟨ ⟩ )
2𝜆 1 − 𝛽2 + |
ci Re Ψ0 |Ψi N−1
N i
( )
𝜌 ∑∞ ⟨ ⟩
 |
= 𝜆2 − cj ci Ψj |Ψi . (128)
N | N−1
i,j

We will use the Euler-Lagrange (multiplier method

) to find the set of coefficients
̂ t ⟩ − E0 . Let
{ci } that minimizes the value of 𝜆2 ⟨Ψt |H|Ψ

∞
∑
𝐀= c2i − 1 , (129)
i=0
356 P. Xiang and Y. A. Wang
[ 
]
√ c (𝜌 − 𝜌 ) ∑ (⟨ ⟩ )
𝐁 = 2𝜆 1 − 𝛽 2 0 0 
+ ci Re Ψ0 ||Ψi N−1
N i
( ∞
)
2 𝜌 ∑ ⟨
|
⟩
−𝜆 − ci cj Ψi |Ψj N−1 , (130)
N i,j

and
( )
̂ t ⟩ − E0 − h𝐀 − ⟨g(𝐫) 𝐁⟩
𝛀 = 𝜆2 ⟨Ψt |H|Ψ
(∞ )
∑
= 𝜆2 ̂ j ⟩ − E0 − h𝐀 − ⟨g(𝐫) 𝐁⟩
ci cj ⟨Ψi |H|Ψ
i,j
( ∞
)
∑
2
=𝜆 ci cj Ej 𝛿ij − E0 − h𝐀 − ⟨g(𝐫) 𝐁⟩
i,j
[ ∞
]
∑
2
=𝜆 c2i (Ei − E0 ) − h𝐀 − ⟨g(𝐫) 𝐁⟩ , (131)
i=1

where h and g(𝐫) are the Lagrange multipliers corresponding to the two constraints
in Eqs. (129) and (130). Minimizing Eq. (131) with respect to {ci }, one obtains
⟨[ √ ] ⟩
1 − 𝛽 2 (𝜌0 − 𝜌 ) ∑∞ (⟨ ⟩ )
𝜆 +𝜆 cj Re Ψ0 ||Ψj N−1 g(𝐫) = −hc0 , (132)
N j=0

and
⟨[ ] ⟩
√ (⟨ ⟩ ) ∞
∑ (⟨ ⟩ )
𝜆 1 − 𝛽 2 Re Ψi || Ψ0 N−1 + 𝜆 cj Re Ψi ||Ψj N−1 g(𝐫)
j=0
[ ]
= 𝜆2 (Ei − E0 ) − h ci (for i ≠ 0) . (133)

Because c0 is linear in 𝜆 as we previously showed, we can readily infer from Eq.

(132) that g(𝐫) must take the following form:
∞
∑ g(k) (𝐫)
g𝜆 (𝐫) = g(0) (𝐫) + 𝜆k . (134)
k=1
k!

Substituting Eq. (134) into Eq. (133) and ignoring the higher-order terms as 𝜆 → 0,
we obtain an equation for Ψi ∈ ,
√ ⟨ (⟨ ⟩ ) ⟩
−hci = 𝜆 1 − 𝛽 2 Re Ψi || Ψ0 N−1 g(0) (𝐫) + h.o. , (135)
Functional Derivatives and Differentiability in Density-Functional Theory 357

where “h.o.” denotes higher-order terms in 𝜆. Therefore, we reach the same conclu-
sion as before: {ci } for Ψi ∈  is linear in 𝜆 towards the end of variational path.
For those {ci } for Ψi ∈ , utilizing the additional fact that Ψi is order-1 strongly
orthogonal to Ψ0 , we can further simplify Eq. (133) to
⟨ ⟩ ⟨ ⟩
∑0 (⟨ ⟩ ) ∑ (⟨ ⟩ )
𝜆2 cj Re Ψi ||Ψj N−1 g(𝐫) + 𝜆2 cj Re Ψi ||Ψj N−1 g(𝐫)
j j
2
= 𝜆 (Ei − E0 )ci − hci . (136)

For this equation to be valid at 𝜆 → 0, the LHS and the RHS must have the same
dependence on 𝜆. On the RHS, the first term decays faster than the second term, and
the second term will dominate when 𝜆 approaches 0. Therefore, we must match the
magnitude of the second term on the RHS to the LHS. Of course, we cannot match
it with the second term on the LHS because doing so will lead to self inconsistency.
Then, the second term on the RHS must decay in the same way as the first term
on the LHS. Thus, {ci } for Ψi ∈  are proportional to 𝜆3 . Unfortunately, such a
𝜆3 -behavior is contradictory to the normalization constraint in Eq. (126), because
∑ 2
i ci will become 0 as 𝜆 → ∑ 0. Hence, we conclude that this contradiction must come
from the assumption: Ψt = i ci Ψi , where the expansion is over the complete set of
eigenfunctions of H.̂
To resolve the contradiction, we have to modify our assumption about the expan-
∑
sion of Ψt . We notice that if the summation i ci Ψi includes any wavefunction from
0 , the same problem will persist. Therefore, Ψt can only be expanded in ,


∑
Ψt = ci Ψi . (137)
i

In this case, Eq. (127) is equivalent to

𝜆2 𝜌t (𝐫) = 𝛽 2 𝜌 (𝐫) . (138)

Integrating both sides of Eq. (138) over the entire space, one obtains

𝜆2 = 𝛽 2 , (139)

which further ensures that

𝜌t (𝐫) = 𝜌 (𝐫) . (140)

Now, the original minimization process is reduced to minimizing the following

term,
358 P. Xiang and Y. A. Wang
( ) ( 
) ⟨( 
) ⟩
∑ ∑ 𝜌 ∑ ∗
2 2
𝚵= |ci | Ei −h |ci | − 1 − − ci cj ⟨Ψi |Ψj ⟩N−1 g(𝐫) .
i i
N i,j
(141)

Suppose this minimization will yield the optimal set of expansion coefficients, {̄ci },
which have no dependence on 𝜆 and 𝛽 from the appearance of Eq. (141). Then, we
have

⟨𝛿Ψ|Ĥ − E0 |𝛿Ψ⟩ ̃ − Ψ0 |Ĥ − E0 |Ψ

⟨Ψ ̃ − Ψ0 ⟩
inf = inf
Ψ0 +𝛿Ψ0 →𝜌0 +𝛿𝜌 ||𝛿𝜌|| Ψ0 +𝛿Ψ→𝜌0 +𝛿𝜌 ||𝜌p − 𝜌0 ||
⟨√ √ ⟩
| |
( 1 − 𝛽 2 − 1)Ψ0 + 𝜆Ψt |Ĥ − E0 | ( 1 − 𝛽 2 − 1)Ψ0 + 𝜆Ψt
| |
= inf
Ψt →𝜌 ||𝛽 2 (𝜌 − 𝜌0 )||
inf ⟨Ψ |Ĥ − E0 |Ψt ⟩
𝜆2 ⟨Ψt |Ĥ − E0 |Ψt ⟩ Ψt →𝜌 t
= inf =
Ψt →𝜌 𝛽 2 ||𝜌 − 𝜌0 || ||𝜌 − 𝜌0 ||
⟨ | | 
⟩ [ ]
1 ∑ | |∑ 1 ∑
= | ̂ |
c̄ i Ψi | H − E0 | c̄ j Ψj = 2
|̄c | E − E0
||𝜌 − 𝜌0 || | | j ||𝜌 − 𝜌0 || i i i
i | |
[ ]
1 ∑
2
> |̄c | E − E0 = 0 , (142)
||𝜌 − 𝜌0 || i i 0

where Eq. (139) is used to simplify the expression after the third equal sign. Evi-
dently, Eq. (142) suggests that the condition for Fréchet differentiability proposed
by Lindgren and Salomonson [8–10] is not fulfilled. In other words, the Fréchet
derivative does not exist in the normalized density domain N either.

Appendix 3

In this appendix, we analyze the consequence of choosing the negative sign in Eqs.
(41) and (121). In the end, we will conclude that this particular choice is fully equiv-
alent to the more natural decision made in the main text and Appendix 2.
Let us start from a unified version of Eqs. (40) and (120):
√
𝜆 = −ac0 ± a2 c20 + 𝛽 2 , (143)

√
where constant a = 1 and 1 − 𝛽 2 in the main text and Appendix 2, respectively.
Obviously, if c0 = 0 or c0 → 0 as 𝛽 → 0, both 𝜆 and 𝛽 approach 0 concurrently near
the end of the variational path.
Functional Derivatives and Differentiability in Density-Functional Theory 359

We only need to further examine the situation when c0 ≠ 0 as 𝛽 → 0 with the

choice of the negative sign in Eq. (143):

𝜆 = −2ac0 − 𝜆′ , (144)

where the residual term 𝜆′ approaches 0 as 𝛽 → 0:

[ ]
′ 1 2 1 4
𝜆 = 𝛽 − 𝛽 +⋯ →0. (145)
2ac0 8a3 c30

Consequently, Eqs. (46) and (124) can be rewritten as

[ 
]
∑ (⟨ ⟩ ) ( )
2a c0 (𝜌0 − 𝜌t ) + N |
ci Re Ψ0 |Ψi N−1 = 𝜆′ 𝜌t − 𝜌 , (146)
i

which immediately suggests that as 𝛽 → 0, (𝜌0 − 𝜌t ) and the coefficients {ci } for
Ψi ∈  are linear in 𝜆′ . Because 𝜌t → 𝜌0 , Ψt → c0 Ψ0 with |c0 | → 1, as 𝛽 → 0.
Therefore, at the end of the variational path (𝛽 = 0 and 𝜆′ = 0), 𝜆 = −2ac0 , Ψt =
̃ = −aΨ0 .
c0 Ψ0 , |c0 | = 1, and Ψ
Evidently, the choice of the negative sign in Eqs. (41) and (121) yields a fully
equivalent, alternative trial wavefunction,

̃ ′ = −aΨ0 − 𝜆′ Ψt ,
Ψ (147)

where 𝜆′ → 0 as 𝛽 → 0. Then, we can carry out the discussion on the basis of 𝜆′ → 0

instead.

References

1. Parr RG, Yang W (1989) Density-functional theory of atoms and molecules. Oxford University
Press, New York
2. Hohenberg P, Kohn W (1964) Phys Rev 136:B864
3. Kohn W, Sham LJ (1965) Phys Rev 140:A1133
4. Wang YA, Xiang P (2013) In: Wesolowski TA, Wang YA (eds) Recent advances in orbital-free
density functional theory, Chap. 1. World Scientific, Singapore, pp 3–12
5. Lieb EH (1983) Int J Quantum Chem 24:243
6. Englisch H, Englisch R (1983) Phys Stat Sol 123:711
7. Englisch H, Englisch R (1984) Phys Stat Sol 124:373
8. Lindgren I, Salomonson S (2003) Phys Rev A 67:056501
9. Lindgren I, Salomonson S (2003) Adv Quantum Chem 43:95
10. Lindgren I, Salomonson S (2004) Phys Rev A 70:032509
11. Ekeland I, Temam R (1976) Convex analysis and variational problems. North-Holland, Ams-
terdam
12. Harris J, Jones RO (1974) J Phys F 4:1170
360 P. Xiang and Y. A. Wang

13. Harris J (1984) Phys Rev A 29:1648

14. Gunnarsson O, Lundqvist BI (1976) Phys Rev B 13:4274
15. Langreth DC, Perdew JP (1980) Phys Rev B 21:5469
16. Wang YA (1997) Phys Rev A 55:4589
17. Wang YA (1997) Phys Rev A 56:1646
18. Levy M (1979) Proc Natl Acad Sci USA 76:6062
19. Nesbet RK (2001) Phys Rev A 65:010502
20. Nesbet RK (2003) Adv Quantum Chem 43:1
21. Dreizler RM, Gross EKU (1990) Density functional theory. Springer, Berlin
22. Davidson ER (1976) Reduced density matrices in quantum chemistry. Academic, New York
23. Perdew JP, Levy M (1985) Phys Rev B 31:6264
24. Englisch H, Englisch R (1983) Physica A 121:253
25. Zhang YA, Wang YA (2009) Int J Quantum Chem 109:3199
26. Milne RD (1980) Applied functional analysis: an introductory treatment. Pitman Publishing,
UK
The Dirac Electron and Elementary
Interactions: The Gyromagnetic Factor,
Fine-Structure Constant,
and Gravitational Invariant:
Deviations from Whole Numbers

Jean Maruani

Abstract In previous papers, we revisited the Dirac equation and conjectured that
the electron can be viewed as a massless charge spinning at light speed, this internal
motion being responsible for the rest mass involved in external motions and
interactions. Implications of this concept on basic properties such as time, space,
electric charge, and magnetic moment were considered. The present paper inves-
tigates the deviations of the resulting gyromagnetic factor, fine-structure constant,
and gravitational invariant from their integer approximates, and their implication in
a better understanding of the electromagnetic, gravitational, and other interactions.

Keywords Dirac equation ⋅

Spin momentum ⋅
Magnetic moment
Matter antimatter ⋅Wave beat ⋅
Zitterbewegung ⋅
Light velocity
Compton diameter ⋅
Planck units ⋅
Catalan numbers ⋅
Casimir force
Nuclear forces ⋅ Gyromagnetic factor ⋅
Fine-structure constant
Gravitational invariant ⋅
Quantum electrodynamics ⋅
General relativity

1 The Heuristic Road to the Dirac Equation

In previous papers [1–4], we revisited the Dirac equation and developed a model for
the electron, making conjectures about its rest mass, spin motion, effective size, and
electric charge. This led us to investigate universal constants related to this model.
In the present paper, after recalling specificities related to the Dirac equation, we
elaborate on the light shed by peculiarities of these constants on the relations
between the electromagnetic and gravitational interactions.

J. Maruani (✉)
Laboratoire de Chimie Physique-Matière et Rayonnement,
CNRS & UPMC, 4 Place Jussieu, #32-42, 75005 Paris, France
e-mail: [email protected]

© Springer International Publishing AG, part of Springer Nature 2018 361

Y. A. Wang et al. (eds.), Concepts, Methods and Applications of Quantum Systems
in Chemistry and Physics, Progress in Theoretical Chemistry and Physics 31,
https://doi.org/10.1007/978-3-319-74582-4_19
362 J. Maruani

Today, the Dirac equation can be derived from more general theoretical frame-
works [5–8]. But the inductive derivation originally given by Dirac [9, 10] has shown
great heuristic value: it has explained the spin kinetic momentum and magnetic
moment, predicted antimatter, and set the ground for quantum electrodynamics.
In the 1920s, light appeared alternatively as geometric rays following Fermat’s
principle of least optical path (stemming from Snell-Descartes’ laws for reflection
and refraction), or as electromagnetic waves obeying Maxwell’s differential equa-
tions, or as massless particles following Planck-Einstein’s quantum relations:

E = hν = hc ̸ λ, p = E ̸ c; ð1Þ

whereas matter was considered as made of particles following Maupertuis’ prin-

ciple of least action integral (or Hamilton’s differential equations for position and
momentum), as well as Einstein-Poincaré’s relativistic relation:

E = mc2 . ð2Þ

Through a detailed analysis of the similarities between Fermat’s and Maupertuis’

principles and making use of the above relations, de Broglie came to the conjecture
that matter also is associated with waves, according to the formula [11]:

λ B = h ̸ p, p = mv. ð3Þ

It is to be noted that this de Broglie wavelength λB differs from the Compton

wavelength λC , introduced a year earlier in the theory of x-ray inelastic scattering
[12], in that the latter involves the rest mass and the speed of light:

λC = h ̸ m0 c; −λC ≡ λC ̸ 2π = ℏ ̸m0 c. ð4Þ

De Broglie’s matter-wave formula succeeded to explain Bohr’s quantization

rule as due to stationary waves and it led to predicting electron diffraction. That
brought Schrödinger to formulate his equation for Wave Mechanics. In the mean-
while, Heisenberg’s phenomenological approach had led to Matrix Mechanics, and
Dirac’s bra/ket approach to Operator Mechanics. Schrödinger eventually showed
that these three approaches are equivalent to a common pattern, nowadays known
as Quantum Mechanics.
However, although special relativity was originally involved in de Broglie’s
matter-wave derivation, further formulations of Quantum Mechanics used the
non-relativistic kinetic and potential energies:

T = p2 ̸2m, V = ke e2 ̸ r , ð5Þ

and were not Lorentz-invariant. This hybrid character was partly corrected by
adding ad hoc spin-symmetry conditions for the resulting eigenfunctions Ψ .
The Dirac Electron and Elementary Interactions … 363

A further step was taken by Klein and Gordon, who used as Hamiltonian H the
full energy: mc2 = (m20 c2 + p2)1/2 c, where p2 = p21 + p22 + p23 with pi = mvi along
xi, m = m0γ, γ = ð1 − v2 ̸c2 Þ1 ̸2 . This led to an equation that was relativistic (it had
time and space operators on the same footing), but not quantic (it was quadratic in
the time operator and did not allow to apply the probability principle). Dirac’s feat
was to design a road towards an equation that was both symmetric and linear in
time and space operators, by introducing 4-D matrices multiplying the momentum
operators.
Dimensionwise, an invariant ‘momentum’ p0 ≡ m0c can be defined for a par-
ticle at rest, and an overall ‘momentum’ p4 ≡ mc related to the time coordinate x4
≡ ct. With these notations, it can be written:

p4 2 = p0 2 + p1 2 + p2 2 + p3 2 . ð6Þ

This expression for the invariant (rest mass) ‘momentum’ p0 is similar to that for
the invariant (proper interval) ‘coordinate’ x0:

x4 2 = x0 2 + x1 2 + x2 2 + x3 2 . ð7Þ

Using the notations recalled in Eq. (8) (note that there is no coordinate derivative
associated with the invariant momentum p0), the Schrödinger (9), Klein-Gordon
(10), and Dirac (11) equations can be written as:

p1 ∼ − iℏ ∂ ̸ ∂x, p2 ∼ − iℏ ∂ ̸∂y, p3 ∼ − iℏ ∂ ̸∂z, p4 ∼ iℏ ∂ ̸∂ðctÞ, p0 ≡ m0 c, ð8Þ


 
p4 − p1 2 + p2 2 + p3 2 ̸2mc Ψ = 0. ð9Þ


p4 2 − p0 2 + p1 2 + p2 2 + p3 2 Ψ = 0. ð10Þ

½p4 − ðα0 p0 + α1 p1 + α2 p2 + α3 p3 ÞΨ = 0. ð11Þ

It can be seen that the Schrödinger equation, being linear in p4 but quadratic in
the pi’s, is rather a diffusion equation, while the Klein-Gordon equation reduces to a
wave equation for p0 = 0. In the Dirac equation, the αμ matrices are independent of
the p’s and x’s as well as Hermitian and normalized. For Eq. (11) to be equivalent
to Eq. (10), these matrices must also be four-dimensional and anticommutative.
There results that any vector representative of an eigenfunction Ψ must have
four components or, alternatively, that Ψ contains a variable that may take on four
values. Dirac explained that these are the well-known two components of the spin
(±½) and in addition positive and negative values for the mass energy (±mc2). The
existence of a spin kinetic momentum thus appears unseparable from negative-
energy states: both stem from the matrix linearization of a wave equation involving
a quadratic form for the energy.
364 J. Maruani

2 The Internal Motion of the Dirac Electron

Before the Dirac equation elucidated its origin, the electron spin had entered
quantum mechanics in two different ways [13]. (1.) A non-energetic, symmetry
requirement for systems of identical particles (Pauli 1925), antisymmetry of the
wave function, this implying an internal dynamical variable with two possible
values. (2.) The deflection of the trajectories of silver atoms by an inhomogeneous
magnetic field (Stern and Gerlach 1922) and the splitting of the spectral lines of
atoms by a magnetic field (Goodsmit and Uhlenbeck 1925), which implied an
intrinsic magnetic moment interacting with the field. The electron spin magnetic
moment is responsible for most of the macroscopic magnetism, from the oxygen
that we breathe to the hard disks of our computers.
To have the spin magnetic moment show up, Dirac made it interact with a
magnetic field. And to have its spin kinetic momentum appear, he had it combined
with an orbital kinetic momentum [10]. Equation (10) was thus extended to include
interactions with an electromagnetic potential ðA4 , AÞ:
h i
ðp4 + e A4 ̸ cÞ − α0 p0 − α ⋅ ðp + e AÞ Ψ = 0. ð12Þ

Note that the invariant momentum p0 is not affected by the external potential ðA4 , AÞ.
Writing H = m0c2 + H′, Dirac showed that, to first order:

H′ = p4 c − p0 c = − e A4 + ðp + e AÞ2 ̸2m0 + ðe ℏ ̸2m0 Þσ ⋅ B. ð13Þ

In addition to the classical potential and kinetic energies, there appears an extra term,
which he interpreted as due to the interaction of the magnetic field B with an intrinsic
magnetic moment: μs = − ðeℏ ̸ 2m0 Þσ = − μB σ, μB being the Bohr magneton.
The spin kinetic momentum does not give rise to any potential energy. To show
its existence, Dirac computed the angular momentum integrals for an electron
moving in a central electric field (e.g., that of a nucleus):

H = p4 c = − e A4 ðrÞ + c α0 p0 + c α ⋅ p. ð14Þ

For any component lk of the orbital momentum: l = − i ℏ r x ∇, Dirac obtained a

non-zero expression for iℏ ∂lk ̸∂t, and similarly for the component σ k of the Pauli
matrix vector used to build the Dirac matrices αμ ; thus, neither l nor σ was a
constant of the motion. But the sum was:

∂ lk ̸∂t + ðℏ ̸2Þ∂σ k ̸ ∂t = 0. ð15Þ

The Dirac Electron and Elementary Interactions … 365

Dirac interpreted this as meaning that the electron has a spin kinetic momentum:
s = ðℏ ̸ 2Þ σ, which is to be added to the orbital kinetic momentum l to get a constant
of the motion. The directions of s and μs being defined by the same matrix vector σ,
one has:

μs = − ðe ̸ m0 Þs. ð16Þ

The spin magnetic moment μs differs from the orbital magnetic moment μl by a
factor 2:

μl = − ðe ̸2m0 Þl, ð17Þ

as if the ‘loop’ described by the electron in its spin motion had half the length of its
orbital ‘loop’. The total magnetic moment can then be written:

μt = − ðμB ̸ℏÞ ðl + ge sÞ, ð18Þ

where we introduce the dimensionless factor ge, whose deviations from 2 (the value
given in Dirac’s theory) will be discussed in this paper. This factor distinguishes the
spin magnetic moment from the orbital magnetic moment derived classically. Being
the ratio of μs (in units of μB) to s (in units of ℏ), it is called gyromagnetic factor.
In another computation, Dirac used a field-free Hamiltonian to determine at
which velocity the electron ‘spins’ to acquire kinetic and magnetic momenta [10]:

H = cðα0 p0 + α1 p1 + α2 p2 + α3 p3 Þ. ð19Þ

Making use of the properties of the αk’s he obtained, for any component vk of the
electron velocity:

iℏ ∂ xk ̸∂t = ½xk , H = iℏ c αk → vk = ∂ xk ̸ ∂t = ±c. ð20Þ

The paradox of an electron moving at light velocity was elucidated by Schrödinger

[14] while investigating the Dirac velocity operators vk = cαk. He showed that:

iℏ ∂2 αk ̸ ∂t 2 = 2 ð∂ αk ̸ ∂tÞ H. ð21Þ

This differential equation can be integrated twice, yielding the explicit time
dependence of the velocity and then of the position. One first obtains:

vk = c αk = c2 pk H − 1 + ðiℏc ̸2Þγ 0k e − iωt H − 1 , ð22Þ

where ω = 2H ̸ℏ and γ 0k = ∂ αk ̸∂t at t = 0. As H = mc2, the first term is a constant

of the order of pk/m, the classical relation between momentum and velocity. But
here also there is an extra term, oscillating at the ‘Zitterbewegung’ frequency [14]:
366 J. Maruani

νe = 2 mc2 ̸ h. ð23Þ

The constant part gives the average velocity through a time interval larger than
ν−1
e , which is observed in practical measurements, while the oscillatory part explains
why the instantaneous velocity has eigenvalues ±c. Further integration yields the
time dependence of the electron coordinate xk, and it appears that the Zitterbewegung
amplitude is of the order of the Compton radius: rC =−λC ̸ 2 = λC ̸ 4π .
Thus, in addition to its external motion (e.g., an orbital motion around a
nucleus), governed by de Broglie’s wavelength λB, the electron is endowed with an
internal motion (Zitterbewegung), governed by Compton’s wavelength λC. Schrö-
dinger showed that Zitterbewegung vanishes when one takes expectation values
over wave packets made up entirely of positive or negative energy states. This was
understood by de Broglie [15] as it resulting from a wave beat between the two
coupled matter and antimatter energy states, the beat frequency νe being the dif-
ference of the two frequencies. This oscillation may be pictured as a Lissajous curve
Lπ ̸2, 1: 2 .
Then, the average mass of the vibrating entity can be considered as null,
departures from this value being allowed by Heisenberg’s uncertainty principle, i.e.:

2 m0 c ⋅ c τ0 ≈ ℏ → τ0 ≈ ℏ ̸ 2m0 c2 = 1 ̸ 2πνe ≈ 0.64 × 10 − 21 s.

To the rest mass ‘momentum’ m0 c ≡ p0 is associated an internal time ‘coordinate’

c τ0 ≡ x0 . One may then see Eq. (7) as involving three space dimensions and two
time dimensions, space ‘emerging’, so to say, from a Pythagorean substraction of
an ‘internal time’ from the ‘external time’.
Among the various authors who speculated on the electron internal motion [16],
Barut and coworkers [17] described the spin as the orbital momentum associated
with Zitterbewegung and the rest mass as the internal energy in the rest frame. In
previous papers [1–4], we conjectured that the rest mass observed in external
motions (inertia) and interactions (gravitation) essentially results from the kinetic
energy of a massless charge spinning at light speed. Thus, if mass is linked to spin,
spin to Zitterbewegung, and this latter to a wave beat between the particle and its
antiparticle, then there is no matter without antimatter, which together result in both
spin and mass.
Antimatter is thus around us and in us, like the two faces of a same coin. What
we call matter (an electron) would be the visible face of the coin, and what we call
antimatter (a positron) would result from a dephased oscillation between the two
energy states. In modern quantum field theories [8], the entanglement of matter and
antimatter appears in the necessity to include antiparticles to cope with infinities.
The Dirac Electron and Elementary Interactions … 367

3 The Gyromagnetic Factor ge

The value ½ of the spin of the electron (Eq. 15) and the related value 2 of its
gyromagnetic factor (Eq. 13) stem from the beat frequency νe of the positive and
negative energy waves being twice the electron mass-energy frequency (Eq. 23).
However, in Dirac’s theory, the electron interacts with an electromagnetic field
that fulfills relativistic but not quantum requirements. Further consistency was
reached by quantizing the electromagnetic field, which led to quantum electrody-
namics (QED) [7]. Resulting zero-point field (ZPF) oscillations entail ‘radiative
corrections’ that are responsible for the Lamb shift between the 2s1/2 and 2p3/2
levels of hydrogenoid atoms [13] and for the departure of ge from the Dirac integer
value 2. Several authors have used Feynman diagrams to compute increasingly
accurate corrections to ge, yielding the following expansion [18]:

εg ≡ ðge − 2Þ ̸2 ≈ 0.001159652181 ≈ 1 ̸ 2aπ + P ̸ 2ðaπÞ2 + L ̸2ðaπÞ3 + ⋯, ð24Þ

which is accurate within 1 ppb. Here a is the fine-structure constant inverse: a ≡

a−1 ≈ 137.036, and P and L are coefficients involving hyperlogarithms. It took half
a century to arrive at this expansion: the first term was obtained by Schwinger in
1948, the second one by Peterman in 1957, and the third one by Laporta in 1996.
But the deviation of the measured value of ge from 2 does not come only from
QED effects. Very accurate measurements [19] have shown that, after subtracting
these effects, the free electron ge value is still slightly larger than 2:

ge ðmeasuredÞ ≈ 2.002 319 304 362; ge ðcorrectedÞ ≈ 2.000 000 000 110 ð60Þ.

This was interpreted by endowing the Dirac ‘point charge’ with a tiny but finite
size: ρe ∼ 10−22 m, much smaller than the electron classical radius: r0 ≈ 2.82 ×
10−15 m, but larger than the Planck length: lP ≈ 1.62 × 10−35 m. It would be that
tiny charge which undergoes Zitterbewegung in a range set by the Compton radius:
rC ≈ 1.93 × 10−13 m.
This discussion deals solely with the free electron interacting with an applied
field. For electrons bound in paramagnetic systems, the measured values of ge are
effective values including, to second order, the orbital momentum and spin-orbit
coupling [20]. The effective factor measured may then be a tensor ge, whose
principal values and axes will depend on the anisotropy of the molecule or of the
crystal site bearing the unpaired electron. Then the extreme accuracy of the value
measured for ge on the free electron is a sign of the extreme isotropy of vacuum
fluctuations.
368 J. Maruani

4 The Fine-Structure Constant

The fine-structure constant a was surmised by several quantum physicists (Planck,

Haas, Bohr, …) before it was explicitly introduced by Sommerfeld in 1916 [21] to
express the relativistic line splittings in hydrogenoid atomic spectra:


 
Wðn, kÞ ≈ − R hc Z 2 ̸ n2 1 + α2 Z 2 ̸ n2 ðn ̸k − 3 ̸4Þ , ð25Þ

where R is Rydberg’s constant, n is the main quantum number, k = 1, 2, …, n, and:

α ≡ ke e2 ̸ℏ c ≡ 1 ̸ a ≈ 1 ̸ 137.0359991. ð26Þ

The dimensionless quantity a was identified with the ratio of the electron
velocity in the first Bohr orbit to that of light. Its inverse a was given a more general
significance by Eddington [22], who proposed the integer value 137 on speculative
grounds. In atomic units, a measures the velocity of light c; and in Planck units, the
electron charge e = qP a2. Then a can be seen as expressing the strength of the
electromagnetic interaction between charged particles.
The fine-structure constant α (or its inverse, the electric parameter a) shows up in
various domains in physics. For instance, it occurs (together with π) in the
expansion of εg given in Eq. (24). In earlier papers [1–4], we recalled that the
electron Compton diameter (or reduced wavelength) 2rC (−λC ) is the geometric
average of the classical electrostatic radius: r0 = kee2/m0c2, and the hydrogen Bohr
radius: a0 = ℏ2 ̸ ke me e2 , the ratio of this harmonic relation being α:

2rC ̸a0 = r0 ̸2rC = α. ð27Þ

It is commonly believed that Z = 137 sets a limit for the periodic table. This is
due to the fact that, when using the point-nucleus model, the energy expression
includes a factor: [1 − (aZ)2]1/2, which becomes imaginary for Z > 137. However,
this model is a poor approximation for superheavy elements. If one estimates the
1s orbital radius in these elements and compares it with actual nuclear sizes, it
displays a strong overlap with the nucleus, and the factor above appears as an
artifact [23]. There are several models of a finite nuclear charge distribution, none
of which gives this factor [24]. Therefore, there may in principle be elements with
Z > 137, although they might be too unstable to be observed. Nevertheless, the fact
that Z = 137 is a limit for point nuclei remains a peculiarity of a.
Another reservation about the number 137 is that a is close to this value only
when it is measured in our low-energy world. At a W boson energy (≈ 81 Gev),
a decreases to about 128; and at grand-unification energy, it merges with similar
constants defining the strong and weak nuclear forces: ae ≈ aw ≈ as [25]. But, here
again, the fact that a ≈ 137 at the limit of low energy remains a peculiarity of a
[26].
The Dirac Electron and Elementary Interactions … 369

For the gyromagnetic factor ge, both the integer value 2 resulting from the Dirac
equation and the shift of measured values from 2 have theoretical explanations: the
integer value is ultimately the expression of the duality matter/antimatter, and the
corrections listed in Eq. (24) express the interactions with the quantum vacuum. For
the electric parameter a, no such understanding is available, neither for the integer
value 137 nor for the measured shift of ∼ 0.26 ppt (Eq. 26). Feynman referred to
the constant a in these terms:
It has been a mystery ever since it was discovered more than fifty years ago […]. You
would like to know where this number for a coupling comes from: is it related to π or
perhaps to the base of natural logarithms? Nobody knows; it is one of the greatest damn
mysteries of physics: a magic number that comes to us with no understanding by man. You
might say the ‘hand of God’ wrote that number […]. We know what kind of a dance to do
experimentally to measure this number very accurately; but we don’t know what kind of
dance to do on the computer to make this number come out without putting it in secretly!
Richard Feynman, The Strange Theory of Light and Matter (Princeton UP, 1985).

Over the years, a kind of mystic has developed about the peculiar value a ≈ 137,
as illustrated in the quotation below:
The fine-structure constant holds a special place among cult numbers: [it] seduces otherwise
sedate people into seeking mystical truths and developing uncollaborated theories …. The
cult of 137 began with scientists who already had quite a reputation, including Pauli, Jung,
Heisenberg, and most notably Eddington […]. In more recent years, the tradition has spread
to the larger community of science theory hobbyists ….
Robert Munato, http://www.mrob.com/pub/num/n-b137_035.html.

However, although Eddington’s speculations about the integer 137 are dis-
putable, this number has, even more than 2, a number of mathematical properties
that make it rather unlikely that its physical occurrences be merely due to chance. In
addition to being the 33rd prime number, it is a superadditive prime in base 10
(1 + 3 + 7 = 11 and 1 + 1 = 2). It is also a Chen prime (twinned with 139), a
Stern prime (the 4th among 8), an Euler prime, an Eisenstein prime, a Pythagorean
prime, a binary quadratic prime, and an optimal primeval prime (the 10th of the 15
primes generated with its 3 digits).
It has been shown [27] that a and 137 are related to the conspicuous numbers π
and e (related themselves through the magic relation: eiπ = −1), and to the har-
monic sum 137/60 and golden ratio Φ (defined from the equation: Φ = Φ − 1 + 1),
through the relations (precise within ∼ 0.24 ppm, 0.15 ppm, 0.33 ppt and 0.15 ppt,
respectively):

 ̸ 60
a2 ≈ 1372 + π 2 ; Log a ̸ Log 137 ≈ a2 ̸ a2 − 1 ; 3 a ≈ 60 Φ4 ; Φ137 ≈ 3.

Sanchez [27] has also disclosed puzzling occurrences of a (or 137) in the mass
distributions of nucleotide bases and amino-acids in terms of the hydrogen mass
mH. For instance, every amino-acid being coded by three couples of nucleotide
bases (in pairs A-T and G-C), the average mass M C of all codons appears to be
(within <0.2%):
370 J. Maruani

M C ̸ mH ≈ 33 a! ð28Þ

The number 137 was known to the Egyptians, and 60 to the Chaldeans [4]. Both
occur in the 5th sum of the harmonic series: Σ n ð1 ̸ nÞ, i.e.: Σ 5 = 137/60. 137 is also
a central polygonal number within the series Cn = n(n + 1)/2 + 1. More precisely,
one has: C2 = 4, C4 = 11, C16 = 137, C60 = 1831, then 42 + 112 = 137 and [27]:

2
1372 + 18312 ≈ 36 π 10 ≈ 1836.122 ≈ mp ̸ me ð1836.152 Þ. ð29Þ

The number 137 also relates to the Catalan series built on the Mersenne numbers:
Mn = 2n − 1. The sequence of its terms is: M2 = 3, M3 = 7, M7 = 127, and (the
ending term) M127 ≈ 1.701411835 × 1038 [28]. Now M’7 ≡ M2 + M3 + M7 =
137 ≈ a (Eq. 26), and M’127 yields (within <0.6%) the Hubble radius RU of the
Universe in terms of the Compton radius rC of the electron [27]:
′
M127 ð4rC Þ ≈ 1.31403 × 1026 m ≈ 13.89 Gly ≈ RU ð13.81 GlyÞ. ð30Þ

Thus, the special integer 137 (and hence the electric constant a) is related to
quantities that play a major role in microphysics, biophysics, or astrophysics. This
has two main consequences: (1) There might be a deep connection between the
various levels of complexity, as had been surmised by Dirac, Schrödinger,
Eddington, and others. (2) Following Pythagoras’ conjecture that ‘everything
proceeds from numbers’, constants occurring in these fields might be determined
not by some cosmic ‘natural selection’ or ‘anthropic principle’ [29], but by
numerical properties of specific numbers.
We have looked empirically for an expansion of the relative shift of a from 137
similar to that of ge from 2 given by Eq. (24). The following expansion:

εa ≡ ða − 137Þ ̸137 ≈ 0.0002627671533 ≈ ð1 ̸ 2Þðπ ̸137Þ2 − ð9 ̸16Þðπ ̸137Þ4 + ⋯,

ð31Þ

which holds within 0.4 ppb, is surprisingly simpler and more accurate than
Eq. (24). A theoretical explanation is in progress.

5 The Gravitational Invariant

One of the major problems in modern physics is the relation between electro-
magnetism, governed by quantum electrodynamics [7], and gravitation / inertia,
governed by general relativity [30]. In this paper, we shall recall two unconven-
tional attempts to shed some light on this relation.
The Dirac equation for the electron [10], which was derived in the frame of
special relativity theory, introduced four-component spinors expressing quantum
The Dirac Electron and Elementary Interactions … 371

properties. On the other hand, Einstein’s general relativity theory [30] expressed
space-time curvature, entailed by local equivalence of acceleration and gravity, in
the language of Riemann tensors. Among others, Chapman and Leiter [31] made
the Dirac equation conform to general relativity by expanding the principle of
covariance to spinor transformations.
For Mendel Sachs [32], the formal structure of quantum theory, including the
Pauli exclusion principle, can be derived from a low-energy, linear approximation
of a generally covariant, nonlinear field theory of inertia based on the basic ideas of
general relativity. On the other hand, John Macken [33] starts with the high-energy,
nonlinear vacuum fluctuations of quantized space-time and then derives the basic
ideas of general relativity, including the equivalence of acceleration and gravity, as
a low-energy, smooth limit.
Expressing the distance r between two identical particles (e, m0) as a multiple
N of their Compton diameter (reduced Compton wavelength): 2rC =−λ ≡ ℏ ̸ m0 c,
and scaling the electrostatic force: Fe = ke ⋅ e2/r2, and the gravitational force:
Fg = G ⋅ m20 /r2, to Planck units [34]: FP = c4 ̸G, EP = ðℏc5 ̸GÞ1 ̸2 , Macken [33] has
managed to express the two widely different forces as simply two different powers
of the rest mass energy of the two particles, E0 = m0c2:

F e = α E 20 ̸ N 2 , F g = E 40 ̸ N 2 , ð32Þ

where F e ≡ Fe ̸ FP , F g ≡ Fg ̸ FP , and E 0 ≡ E0 ̸ EP are dimensionless quantities. The

fact that in Eq. (32) the larger F e appears as the square of the reduced rest mass E 0
while the smaller F e appears as its fourth power results from the fact that in E0 the
rest mass energy E0 is much smallet than the Planck energy EP.
For N = 1, the two particles being contiguous: r = 2rC, Eq. (32) yields a har-
monic relation similar to that between the Bohr radius and the classical radius,
Eq. (27):

a Fe ̸FP = Fg ̸a Fe = δ. ð33Þ

Here, the Compton diameter 2rC is replaced by the electric (quantum) force a Fe,
the larger Bohr radius a0 by the Planck force FP, and the smaller classical radius r0
by the gravific (relativistic) force Fg. According to Eq. (33), the gravific force is to
the electric force as the electric force is to the Planck force, the ratio of this relation
being a gravitational invariant: δ ≡ d − 1 , similar to the fine-structure constant:
α ≡ a − 1 , Eq. (26):

δ ≡ G m20 ̸ℏ c = E 20 = ðm0 ̸mP Þ2 ≈ 1 ̸5.7087 × 1044 , ð34Þ

where m0 is the electron rest mass and mP is the Planck limit mass: mP = (ħ c/G)1/2.
Equation (33) is an indication that the two forces are deeply related through the
Compton diameter and then, to the particle spin, this diameter defining the
amplitude of Zitterbewegung, responsible for the spin properties (§ 2).
372 J. Maruani

In previous papers [1, 2], we showed that 2rC is also the geometric average of
the space-time curvatures, defined from general relativity [30], ‘inside’ the electron:
2rG = 2(G/c2) m0, and ‘outside’ a volume of radius rQ: 2RG = 2r2Q /rG. For rQ =
rC, one obtains a harmonic relation similar to Eqs. (27) and (33):

2rC ̸ 2RG = 2rG ̸2rC = 2δ, ð35Þ

δ being defined in Eq. (34). Auxiliary relations resulting from Eqs. (27) and (35)
can be written:

rG ̸ r0 = δ ̸α, rG ̸2rC = δ, rG ̸ a0 = δ α. ð36Þ

As the fine-structure constant α (or its inverse a) is considered as defining the

electric force, the gravitational invariant δ (or its inverse d) can be viewed as
defining the gravific force. However, the constant introduced in Eq. (34) involves
the rest mass of the electron, a lepton, while for dealing with gravity it may be more
appropriate to use that of a baryon. Most authors defined d by using the proton mass
[29] or sometimes a cross product of the proton and the neutron (or proton and
hydrogen) masses [27].
In earlier papers [3, 4], we showed that the δn defined by using the cross product
of the proton and neutron masses and the δX defined from that of the electron and
Universe masses obey (within ∼ 0.4 ppt) the simple relation: δn × δX ≈ 1. It may
be interesting to compare the proton and neutron gravitational invariants (accurate
to the 4th decimal only due to the poor accuracy of the measurements of G). For the
proton, we obtain:

δp ≡ G m2p ̸ℏ c ≡ 1 ̸ dp ≈ 1 ̸1.69328 × 1038 . ð37Þ

From Eqs. (26, 37), canonical forms of the electric and gravific forces can be
drawn:

Fe = ℏc ̸ar 2 , Fg = ℏ c ̸dp r 2 . ð38Þ

The value of dp in Eq. (37) is very close to the Catalan sum M’127 occurring in
Eq. (30) [28]. We have looked empirically for a series expansion of the relative
shift of dp from M’127 similar to that of a from M’7 = 137 or of ge from 2, given in
Eqs. (31) and (24), respectively. The following expansions:

′  ′
εp ≡ M127 − dp ̸ M127 ≈ 0.00478021
ð39aÞ
≈ ð1 ̸ 3Þð2 ̸137Þ − ð2 ̸5Þð2 ̸ 137Þ2 + ⋯ ≈ ðge ̸ aπÞ + 6ðge ̸ aπÞ2 − ⋯,

which hold within ∼0.7 ppm and ∼0.6 ppm respectively, are surprisingly simpler
and more accurate than one would expect from the poor accuracy of the measured
value of G used in Eq. (37). It can be noticed that 2/137 ≈ ge/a within <0.1% and
that ge/a ≈ 6−3 (6 being the first ‘perfect’ number) within <0.5%.
The Dirac Electron and Elementary Interactions … 373

If now we use the neutron (instead of the proton) mass mn in Eq. (37), we obtain:
dn ≈ 1.68862 × 1038, and a resulting εn ≈ 0.00751719 which could be expanded
as:

εn ≈ ð1 ̸ 2Þð2 ̸137Þ + ð2 ̸2Þð2 ̸ 137Þ2 + ð3 ̸2Þð2 ̸ 137Þ3 + ⋯, ð39bÞ

accurate to ∼ 5 ppm with the first 2 terms (∼ 2 ppm if 2/137 is replaced by ge/a)
and to ∼ 0.1 ppm with the 3 terms. This expansion looks more consistent than that
of Eq. (39a).

6 Other Interactions and Force Constants

Nothing can be thought that doesn’t exist or coudn’t exist.

Parmenides, Peri Physeos (5th Century BC).

The gravitational and electromagnetic forces, which are the oldest known
(though rationalized only in modern times), thus share three features in common:
(1) they are long-distance and decrease as 1/r2 (Hooke-Newton and Coulomb laws);
(2) in Planck units, their magnitudes can be expressed as even powers of the
rest-mass energy (Eq. 32); (3) the parameters defining their strengths can be
expressed as Catalan integers of the Mersenne series: 2n − 1, with small shifts that
can be expanded, on the model of ge, as simple series involving a (or 137) and π
(Eqs. 24, 31, 39). However, these forces are governed by very different theories [7,
30], although various conciliation schemes have been attempted [e.g., 31–33].

6.1 The Casimir Force

Another force, of pure QED origin, is the Casimir force, which induces attraction
between two conducting plates separated by vacuum. It was proposed by Casimir in
1948 as a relativistically retarded van der Waals force. It has then been the subject
of various theoretical and experimental studies [35, 36], the first accurate mea-
surements being made not earlier than 1997. In the ideal case of perfect plates in
perfect vacuum at 0 °K, the force per unit area can be written as:

FC ̸ A = κ ℏ c ̸r 4 , κ = π 2 ̸ 240. ð40Þ

Contrary to those above, this force is very short ranged (<1 µm). However, three
interesting analogies can be drawn. (1) Similarly as empty space curvature (induced
by massive objects) is responsible for the gravity force [30], vacuum radiation
confinement (induced by conducting objects) is responsible for the Casimir force.
374 J. Maruani

0,0025

0,0020 Star : Casimir force

Round : Electric force
2
F (c / e) / E 0

0,0015

0,0010

0,0005

0,0000
1 2 3 4 5 6 7 8 9 10 11
Distance (number N of Compton diameters λ)

Fig. 1 Reduced Casimir force: F C ̸E 20 (Eq. 41 with A ∼ π−λ2 ̸ 4Þ, versus reduced electric force:
F e ̸ E 20 (Eq. 32), as functions of the number N of Compton diameters between two electrons. The
curves cross after N = 2. The value N = 7 corresponds to ∼ 5% of a Bohr radius (Eq. 27)

(2) Similarly as an effective inertia is endowed to light by confinement in an optical

resonator [33] or in a photonic crystal [37], an effective force is exerted over two
close objects by ZPF confinement between them. (3) If one uses the same scalings
as those that led to Eq. (32), the distance between the plates being expressed as an
integer number of Compton diameters: r = N−λ, one obtains:

F C ̸ A = κ E 20 ̸N 4−λ2 , κ ≅ 0.0411234. ð41Þ

This means that the reduced Casimir force F C exerted on a Compton-size square
(the size of an electron): A = −λ2 , has the same expression as F e but with the factor α
replaced by κ/N2. This factor can then be seen as a dimensionless parameter
characterizing the Casimir force. At a distance equal to a Compton diameter:
N = 1, the Casimir force is ∼ 6 times larger than the electric force. But it decreases
much faster with increasing N (Fig. 1). It may be worth investigating the role that a
Casimir attraction ‘inside’ the electron may play in its stability, as the strong force
in that of nuclei.

6.2 The Intranuclear Forces

These forces, which are responsible for nuclear stability, are also very short ranged.
The strong nuclear force overcomes the electrostatic repulsion between protons,
while the weak nuclear force allows neutrinos to bind to nucleons [38].
The Dirac Electron and Elementary Interactions … 375

A number of model potentials have been proposed for the strong force (Yukawa
1934, Woods-Saxon 1954, Reid 1968, …), most of which are central scalar
potentials involving some exponential decrease. Nowadays, in the frame of the
standard model (where protons and neutrons are seen as made up of tight-bonded
quarks), the strong force appears as a residual, dispersion-like force. Its mean
magnitude, which depends on various factors, is known with poor accuracy [38], and
its dimensionless coupling constant: αs ≈ 0.1185 [39], is ∼ 16 times larger than the
fine-structure constant αe. Although this force does not follow a 1/r2 decrease, its
magnitude with respect to the electric force is consistent with the first sum M’3 of the
combinatorial hierarchy [28]: M’7/M’3 = 137/10 ∼ 14 [40].
The weak nuclear force was introduced in 1933 by Fermi to explain β decay
[38]. It has a number of specific features: (1) contrary to the other interactions, it
does not create bound states, but it allows transformation of a neutron to a proton or
vice versa by changing a quark flavor, this inducing β decay or electron capture;
(2) it is the only interaction that can violate parity symmetry; (3) contrary to the
other interactions, mediated by massless bosons, it is mediated by three very heavy,
short-lived bosons, two charged and one neutral. Besides, it is expected to decrease
with distance even faster than the strong nuclear force. However, in spite of these
crucial differences with the electromagnetic force, it has been unified with it in an
electroweak theory.
The weak force coupling constant is usually expressed in terms of a Fermi mass:
mF ≈ 5.730073 × 105 me ≈ 5.219743 × 10−25 kg, i.e.: GF ≡ ℏ3 ̸c m2F , which is in
J.m3 while ħ is in J.s [27]. One often uses the reduced constant: GF ̸ ðℏ cÞ3 =
1 ̸ m2F c4 ≈ 0.454380 × 1015 J − 2 [39]. But this is not dimensionless, as were
α and δ. One could also use the more familiar form: GF ̸ ℏ c = ðℏ ̸ mF cÞ2 ≈
0.454164 × 10 − 36 m2 , which is a Fermi-mass Compton-like area.
In order to get a dimensionless constant, some authors [29] have used the ratio
αw of this area to that for the electron mass (or its inverse aw). However, since the
weak force acts at the nucleon level, a more relevant choice would be to use the
ratio βw (or its inverse bw) involving the neutron mass:

αw ≡ ðme ̸mF Þ2 ≈ 3.045648 × 10 − 12 ; βw ≡ ðmn ̸mF Þ2 ≈ 1.029660 × 10 − 5 . ð42Þ

The second ratio is ∼ 700 times smaller than the electric force constant α and
11,000 times smaller than the strong force constant αs , which is conform to
expected values.
Sanchez [27] has proposed the following relation (precise within 0.6 ppt)
between the gravitational, nucleoweak and electromagnetic force constants:

de5 ð≈ 6.063 × 10223 Þ ≈ a7w a67

e ð≈ 6.056 × 10
223
Þ, ð43Þ
376 J. Maruani

where the inverses of de, aw and ae are defined in Eqs. (34), (42) and (26),
respectively. It can be noted that the exponents in Eq. (43) are related to the
consecutive Catalan numbers 3, 7 and 127 by: 5 = (3 + 7)/2; 67 = (7 + 127)/2.

6.3 Hypothetical Infraelectric and Supragravific Forces

The question has often been raised as to whether the Newton and Coulomb forces
may not behave differently at very large or very short distances, or whether there
may not be other, undisclosed forces. Cosmological observations have led to the
conclusions that there is a dark energy and a dark matter constituting, respectively,
75% and 21% of the observable Universe. Could there also be forces still larger than
the electric and nuclear forces or smaller than the gravific force?
Looking back at the relation between Fe and Fg expressed by Eqs. (32, 33), one
may wonder whether the electric and gravific forces may not be just two of a series
of r−2-dependent forces acting at various levels, but with powers of E 0 other than 2
or 4. By extrapolation, one would then write:

F total = F a + Fe + Fg + F j + ⋯ = α2 + α E0 2 + E0 4 + α − 1 E0 6 + ⋯
ð44Þ
= α2 + α δp + δ2p + a − 1 δ3p + ⋯,

where the electric and gravific force terms are boldface. The divider N2, which is the
same for all r−2-dependent forces, has been omitted. Note that neither the Casimir
nor the nuclear forces, which are residual forces and decrease much faster than r−2,
fall in this frame. Each term in this series is derived from the previous one by
multiplication by α − 1 E 20 . These extra forces could then be expressed in terms of the
electric and gravific forces and, as a result of Eqs. (27) and (35), as ratios of particle
radii:

F a = F e 2 ̸F g , F j = F g 2 ̸F e . ð45Þ

Here it should be recalled that, following a stochastic electrodynamics approach

of Zitterbewegung, Haisch et al. [16] interpreted inertia as a resistance of ZPF to
spectral distorsion in an accelerated frame, and identified the Newtonian force with
the van der Waals force generated by this motion. The gravific force then appeared
as a kind of residue from the electromagnetic interaction, and the inertial and
gravific masses thus derived were equivalent. Similarly, according to Macken [33],
the highest-order term in Eq. (44) would amount to some relativistic residue from
the gravific force.
If extra forces do exist, they should have realistic strengths. Below are listed the
respective strengths (in Newtons) derived from Eq. (44) and from the definitions of
the reduced forces in terms of the Planck force: FP ≈ 1.21035 × 1045 N.
The Dirac Electron and Elementary Interactions … 377

F a − 1 ≈ 1.8779 × 104 → Fa ≈ 6.4450 × 1039 N;

F e − 1 ≈ 2.3204 × 1040 → Fe ≈ 5.2159 × 103 N;
F g − 1 ≈ 2.8672 × 1076 → Fg ≈ 4.2212 × 10 − 33 N;
F j − 1 ≈ 3.5428 × 10112 → Fj ≈ 3.4162 × 10 − 69 N.

The first force is 30 orders of magnitude larger than the strong nuclear force but,
as it is still 6 orders of magnitude smaller than the Planck force, it is not excluded
that it exists at a subnuclear level. The last force is 36 orders of magnitude smaller
than the gravity force and it appears much too small to show up in observable
effects, except maybe in the vicinity of neutron stars or black holes.
One may now turn back to the combinatorial hierarchy connection between
a and dp involved in Eqs. (31) and (39). Writing F x− 1 ≈ 2n − 1 (with 1 negligible
for n large) yields: na ≈ 14.18 ≈ 2 × 7+ (the increment to 7 being due to
137 > 127); ne ≈ 134.09 ≈ 127 + 7+; ng ≈ 253.99 ≈ 127 × 2, and nj ≈ 373.88
≈ 127 × 3 (from dp3) − 7+ (from a−1). However, none of the compound exponents
14 and 374 yields peculiar numbers comparable to those associated with a and dp.

7 Conclusions

In this paper, we first recalled the origins, features and main outcomes of the Dirac
equation: an underlying antimatter coupled with ordinary matter [9, 10], this
yielding a wave beat between the positive and negative energy states [15], resulting
in an internal motion at light speed within a Compton diameter [14]. Hence, the
spin kinetic momentum with quantum number s = ½ and magnetic moment with
gyromagnetic factor g0 = 2 [9, 10]. Various investigators had featured that the rest
mass would be related to the spin motion [16, 17], and we conjectured that the
electron can actually be seen as a massless charge spinning at light speed, the
observed rest mass stemming mainly from this very internal motion [1–4].
In this framework the Compton diameter−λ [12] plays a special role: as the range
of the internal motion, but also as the geometric average of the electron classical
radius r0 and the hydrogen Bohr radius a0, the ratio of this harmonic relation being
the fine-structure constant α = ke m20 ̸ℏc: −λ ̸ a0 = r0 ̸ −λ = α. It is also the geometric
average of the gravitational curvature diameters ‘inside’ and ‘outside’ the electron,
2rG and 2RG, the ratio of this harmonic relation being the gravitational invariant
2δe = 2 Gm20 ̸ ℏc: −λ ̸ 2RG = 2rG ̸−λ = 2δe .
Expressing the electric and gravific forces in Planck units, and the distance
between interacting particles as an integer number of Compton diameters, Macken
has shown [33] that they take the form of even powers of the rest mass energy. This
entails [2] a new harmonic relation, involving both α and δ, between the Planck,
electric and gravific forces: α − 1 Fe ̸ FP = Fg ̸ α − 1 Fe = δe . In this paper, we show that
a similar relation holds with the Casimir force but with α−1 replaced by a constant
378 J. Maruani

rapidly decreasing with distance. Other forces that would be expressed as similar
products of α and δp appear too large or too small to be observable.
Another link between the electric and gravific forces [27] results from the integer
parts of a and dp being the sums of Catalan numbers in the Mersenne series: 22 −
1 = 3, 23 − 1 = 7, 27 − 1 = 127, then M’7 = 3 + 7 + 127 = 137 ≈ a, and the
sum M’127 up to 2127 − 1 ≈ dp where dp ≡ δp− 1 and δp is defined as δe but with the
electron mass replaced by that of the proton (or the neutron). The magnitude of the
strong nuclear force, although it is a short distance one, seems consistent with the
sum M’3 = 10.
In this paper, we have discussed the deviations, from their integer approximates,
of the measured values of the free electron gyromagnetic factor ge, the fine-structure
constant inverse a, and the proton gravitational invariant inverse dp. After recalling
that the relative deviation of the measured ge from 2 (∼ 0.1%) was explained by
quantum field and size effects, we have tried to express that of a from 137 and that
of dp from M’127 as similar expansions.
Surprisingly, the relative deviation of measured a from 137 (< 0.03%) could
even better be expressed (with 0.4 ppb accuracy) as a simple 2-term series
involving even powers of π/137. Similarly, the relative deviation of measured dp
from M’127 (< 0.50%) could be expressed (with 0.6 ppm accuracy) as 2-term series
involving powers of either 2/137 ð ≈ ge ̸ aÞ or ge ̸ aπ ð ≈ 1 ̸ 63 Þ. For reasons yet
unexplained, the numbers ge, a, dp and π seem to be deeply related.

Acknowledgements I wish to thank the colleagues who helped me clarify these ideas at QSCP
meetings and elsewhere. Erkki Brändas, Uzi Kaldor, John Macken, Francis Sanchez, and Ivan
Todorov made especially useful comments. Thanks are due to my wife, Marja Rantanen, for
stimulating my speculations with inspiring piano playing.

References

1. Maruani J (2012) The Dirac electron: spin, Zitterbewegung, the Compton wavelength, and the
kinetic foundation of rest mass. Prog Theor Chem Phys B 26:23–46
2. Maruani J (2013) The Dirac electron as a massless charge spinning at light speed: implications
on some basic physical concepts. Prog Theor Chem Phys B 27:53–74
3. Maruani J (2015) The Dirac electron as a privileged road to the understanding of quantum
matter. Quantum Matter 4:3–11
4. Maruani J (2016) The Dirac electron: from quantum chemistry to holistic cosmology. J Chin
Chem Soc 63:33–48 and references therein
5. Thaller B (1992) The Dirac equation. Springer, Berlin
6. Sakurai J (1967) Advanced quantum mechanics. Addison-Wesley, Reading, MA, ch. 3
7. Feynman RP (1998) Quantum electrodynamics. Addison-Wesley, Reading, MA
8. Weinberg S (1995) The quantum theory of fields. Cambridge U P
9. Dirac PAM (1928) Quantum theory of the electron. Proc Roy Soc (London) A 117:610–624;
Quantised singularities in the electromagnetic field: ibid (1931) A 133:60–72; Theory of
electrons and positrons. Nobel lectures (1933) pp 320–325
10. Dirac PAM The principles of quantum mechanics. Clarendon Press, Oxford, 1st edn 1930,
4th edn 1958, chs 11–12
The Dirac Electron and Elementary Interactions … 379

11. de Broglie L (1924) Recherches sur la Théorie des Quanta. Thesis, Sorbonne, Paris; Ann Phys
10(III):22–128
12. Compton AH (1923) A quantum theory of the scattering of x-rays by light elements. Phys
Rev 21:483–502
13. Brandsden BH, Joachain CJ (1983) Physics of atoms and molecules. Longman, Harlow,
England, 1st edn, chs 1 and 5
14. Schrödinger E (1930) Über die kräftefreie Bewegung in der relativistischen Quantenmechanik:
Sitzungsber. Preuss Akad Wiss Berlin, Phys-Math Kl 24:418–428; Zur Quantendynamik des
Elecktrons: ibid (1931) 25:63–72
15. de Broglie L (1934) l’Electron Magnétique: Théorie de Dirac, Hermann, Paris, chs 9–22
16. Haisch B, Rueda A, Puthoff HE (1994) Inertia as a zero-point field Lorentz force. Phys Rev A
49:678–694 and references therein
17. Barut AO, Bracken AJ (1981) Zitterbewegung and the internal geometry of the electron. Phys
Rev D 23:2454–2463; D 24:3333–3342; Barut AO, Zanghi N (1984) Classical model of the
Dirac electron. Phys Rev Lett 52:2009–2012; Barut AO, Pavšič M (1987) Quantization of the
Zitterbewegung in the Schrödinger picture. Class Quant Grav 4:L131–L136
18. Todorov I (2015) Hyperlogarithms and periods in Feynman amplitudes. Opening lecture at
QSCP XX, Varna; published as CERN-TH-2016–042; and private communication
19. Dehmelt H (1993) Is the electron a composite particle? an experiment. Hyperfine Interact
81:1–3; updated and condensed version of the Nobel lecture in Les Prix Nobel 1989,
Stockholm, 1990, p 95
20. Maruani J (1980) Magnetic resonance and related techniques. In: Becker P (ed) Electron and
magnetization densities in molecules and crystals, NATO ASI series. Plenum, and references
therein
21. Sommerfeld A (1919) Atombau und Spektrallinien. Friedrich Vieweg, Braunschweig;
translated as Atomic structure and spectral lines. Methuen, London, 1923; and references therein
22. Eddington AS (1935) New pathways in science, Cambridge U P, ch. 11
23. Kaldor U, Eliav E, Landau A (2004) Relativistic electronic structure theory. Elsevier, ch. 2,
and private communication
24. Visscher L, Dyall KG (1997) At Data Nucl Data Tables 67:207
25. Kragh H (2003) Magic number—a partial history of the fine-structure constant. Arch Hist
Exact Sci 57:395–431
26. Gilson JG (1996) Calculating the fine-structure constant. Spec Sci Tech, 23 pp
27. Sanchez FM, Kotov VA, Bizouard C (2009) Evidence for a steady-state, holographic,
tachyonic, and supersymmetric cosmology. GED 20(3):43–53; Sanchez FM (2017) A
coherent resonant cosmology approach and its implications in microphysics and biophysics.
Prog Theor Chem Phys B 30:375–407, references therein, and private communication
28. Bastin T, Kilmister CW (1995) Combinatorial physics. World Scientific, Singapore
29. Carr BJ, Rees MJ (1979) The anthropic principle and the structure of the physical world.
Nature 278:605–612. See also: Barrow JD, Tipler FK (1986) The anthropic cosmological
principle. Oxford U P, and references therein
30. Einstein A (1916) Die Grundlage der allgemeinen Relativitätstheorie. Ann Phys (Leipzig)
49:769–823. See also: Eddington AS (1920) Space, time, and gravitation. Cambridge U P;
French edition completed with an updated mathematical presentation of general relativity.
Hermann, Paris, 1921. See also: Special issue commemorating Einstein A (2005). Ann Phys
(Leipzig) 14:1–204
31. Chapman TC, Leiter DJ (1976) On the generally covariant Dirac equation. Am J Phys
44:858–862
32. Sachs M (1986) Quantum mechanics from general relativity. Reidel, Dordrecht
33. Macken JA (2013) The universe is only spacetime. http://www.onlyspacetime/home; see also:
Spacetime-based foundation of quantum mechanics and general relativity. Prog Theor Chem
Phys A (2015) 29:219–245; and private communication
34. Pavšič M (2001) The landscape of theoretical physics: a global view. Kluwer, pp 347–352.
See also: http://en.wikipedia.org/wiki/planck_units
380 J. Maruani

35. Milton KA (2001) The Casimir effect: physical manifestations of zero-point energy. World
Scientific, Singapore
36. Genet C, Intravaia F, Lambrecht A, Reynaud S (2004) Electromagnetic vacuum fluctuations,
Casimir and van der Waals forces. Annales Fondation Louis de Broglie 29:311–328
37. André J-M, Jonnard Ph (2011) Effective mass of photons in a one-dimensional photonic
crystal. Phys Scr 84:035708 and references therein
38. Brown LM, Rechenberg H (1996) The origin of the concept of nuclear forces. Institute of
Physics Publishing, Bristol and Philadelphia
39. Beringer J et al (2012) Particle data group. Phys Rev D 86, 010001; for updates, see: http://
pdg.lbl.gov
40. It seems that Catalan numbers were known in the Antiquity. Not only the conspicuous 3 and 7
and their sum 10 (and various products and multiples), which occur in many traditions, but
also 127 (the number of provinces in Ahasuerus’ empire, cf Esther 1, 1) and the sum 137 (the
number of years Abraham’s elder son lived, cf Genesis 25, 17). The hypostyle room of
Ammon’s temple in Karnak, Egypt, has a total of 136 columns, the number that Eddington
initially proposed for the electric constant a [22]
A Simple Communication Hypothesis:
The Process of Evolution Reconsidered

Erkki J. Brändas

Consciousness is a transparent brain representation of the

world from a privileged egocentric perspective.
Arnold Trehub

Abstract The scientific basis of Darwinian evolution is reconsidered from the

recent progress in chemistry and physics. The idea, promoting a stochastic com-
munication hypothesis, reflects Kant’s famed insight that ‘space and time are the
two essential forms of human sensibility’, translated to modern practices of
quantum science. The formulation is commensurate with pioneering quantum
mechanics, yet extended to take account of dissipative dynamics of open systems
incorporating some fundamental features of special and general relativity. In par-
ticular we apply the idea to a class of Correlated Dissipative Structures, CDS, in
biology, construed to sanction fundamental processes in biological systems at finite
temperatures, ordering precise space-time scales of free energy configurations
subject to the Correlated Dissipative Ensemble, CDE. The modern scientific
approach is appraised and extended incorporating both the material- as well as the
immaterial parts of the Universe with significant inferences regarding processes
governed by an evolved program. The latter suggests a new understanding of the
controversy of molecular versus evolutionary biology. It is demonstrated by
numerous examples that such an all-inclusive description of Nature, including the
law of self-reference, widens the notion of evolution from the micro to the cosmic
rank of our Universe.

Keywords Density matrix ⋅

Space-time ⋅
Stochastic communication
Evolution ⋅
Correlated dissipative structure CDS ⋅ ⋅
Correlated dissipative
ensemble ⋅
CDE

Dedicated to Arnold Trehub for his great visions and intuitions displayed in his Retinoid Model
of vision.

E. J. Brändas (✉)
Department of Chemistry, Uppsala University, Uppsala, Sweden
e-mail: [email protected]

© Springer International Publishing AG, part of Springer Nature 2018 381

Y. A. Wang et al. (eds.), Concepts, Methods and Applications of Quantum Systems
in Chemistry and Physics, Progress in Theoretical Chemistry and Physics 31,
https://doi.org/10.1007/978-3-319-74582-4_20
382 E. J. Brändas

1 Introductory Remarks

A simple Communication Hypothesis is developed, explained and illustrated. It is

based on a quantum mechanical formulation of ensemble representable density
matrices qualifying the emergence of quantum- and thermal correlations, engrained in
long-range off-diagonal order, with precise thermal conditions satisfied at specific
temperatures. In analogy with condensed matter phenomena, like e.g. superconduc-
tivity, connectors occur between mechanical- and electromagnetic fluctuations. The
subsequent ensemble, the Correlated Dissipative Ensemble, CDE, subject to Pois-
sonian statistics, provides stochastic communication channels for cellular recognition
including a proliferated knowledge of neuron correlates. Since this document is
entrenched in a modus operandi built on the scientific evaluation and the conclusion of
the process of Darwinian evolution, some general comments are enclosed below as an
introduction.
It is quite remarkable that exacting correlations between modern science and
present philosophical enquiry are still largely wanting due, not only to the presumed
lack of interaction between the disciplines, but also owing to the disordered state of
commensurate agreements between the various scientific subjects. A satisfactory
elucidation of the situation should indeed be accomplished by clearing up these
issues from the appropriate knowledge of the state-of-the-art of physics, chemistry
and biology. While important suggestions to this challenge have been propounded
by e.g. Tegmark [1], Deutsch [2], Primas [3] and others, the conundrum still
remains. The general goal concerns the ultimate nature of reality; say our mathe-
matical universe [1], the consequences of the emergence of knowledge [2] or the
concrete need for a conceptual recasting of the logical and philosophical founda-
tions of physics as a result of the high degree of incommensurate specializations
that persist in isolation ignoring most other areas of pursuits [3].
In particular Primas’ critique, well-formulated and certainly thought provoking,
gives a personal characteristic that is rooted in his long career as chemistry pro-
fessor at ETH, starting with problem solving in physical- and theoretical chemistry
and ending up with issues in the cognitive sciences and the philosophy of mind [4].
He came to view the dual epistemic aspects of the mental and the material world to
be conceived as one underlying ontic1 reality, named dual aspect monism. In ref-
erence [3] there is an interesting foreword by the philosopher Paul Feyerabend,
where he places Hans Primas as a representative of the structural approach with ‘a
comprehensive picture of the world and man’s place in it’.
The notorious anarchist is widely known for his declaration ‘anything goes’ and his
book ‘Against Method’ [6], where he proposed the thesis: ‘the events, procedures and
results that constitute the sciences have no common structure’. Although Feyerabend

1
The ontic conceptualization is part of Heidegger’s neologisms [5]. His Being (Dasein) has a
pre-ontological-ontic signification, however, not to be understood as a biological human being.
This distinction is probably the reason why Heidegger takes a poetic turn as he investigates the
Question Concerning Technology and finds it to be but ‘a means to an end’.
A Simple Communication Hypothesis: The Process … 383

contributed his visions with a tongue in cheek, it should not give him absolute
immunity and exemption from liability. His attack on the claims of a philosophical
legislation for the exercise of science is stimulating and inspirational, yet it ignores the
foundation of abstracting science as an art of knowledge. To avoid the most bizarre
forms of philosophical relativism the scientific practice should focus on scientific
disciplines and their evolution rather than narrowing the perspective to a comparison
and review of various individual methodological rules and standards.
In the development of a scientific field, the historian of science, Mary Jo Nye [7]
identifies essentially six characteristics of a scientific discipline, i.e. genealogy, core
literature, practices, physical location, recognition and shared values. Within this
perspective it is clear that disciplines, as attractors for accumulated knowledge,
communicated between its members, focused on a particular sphere of Mother
Nature, do evolve with respect to styles and traditions compatible with existing
scientific paradigms and their inevitable changes. Hence scientific disciplines, as
examples of evolutionary lineages, imbedded in the human genetic code, do arise,
develop, evolve, die and diversify. For a detailed example of this process, see e.g.
the historical account of the emergence and evolution of quantum chemistry as a
sub-discipline and its importance in the province of science [8].
Recently the author advocated an evolutionary approach to represent and interpret
the origin and emergence of life, established by modern advances in chemical physics,
from unstable chemical states to biological evolution and order [9]. The Zero Energy
Universe Scenario (ZEUS) was elaborated commensurable with Darwin’s theory of
evolution, from the microscopic ranks to the cosmological domain.
In this contribution we will present an authentic stochastic communication
hypothesis, building on the results portrayed in [9], i.e. combining the novel formu-
lation of quantum mechanics for open systems, relativity theories, Gödel’s theorems,
Off-Diagonal Long-Range Order (ODLRO), the Correlated Dissipative Ensemble,
CDE, the Poisson distribution and associate encoding- decoding protocols for com-
munication between complex enough systems defining generic life forms. Some
examples of the various levels of organization, that abides by our Communication

Fig. 1 Communication
levels from the micro-to the COMMUNICATION
cosmic rank
function, homeostasis, information, ententional properties

Molecular level Cellular level Nervous system

DNA, RNA, Genes Stem- Somatic cells Cerebral cortex
Chromosome Neurons Brain
Amino acids Germ cells Spinal cord

Social level Ecological level Cosmological level

Mathematics Human signal General relativity
Semiotics Global warming Black holes
Mnemes Anthropogeny Gödelian self-reference
384 E. J. Brändas

Hypothesis, will be discussed in the final section, see also Fig. 1. We will also address
the longstanding disagreements between molecular- and evolutionary biologists [10]
on how natural selection acts, while supporting a trans-level semiotics that is
appropriate for geno-phenotypic translations including the role of somatic programs
in a Darwinian perspective. We will also compare and discuss our findings with other
results and strategies made recently in connection with the self-referential paradox
associated with the concept of consciousness. To make the presentation
self-contained, yet without bequeathing too many excruciating details, some brief
reviews are incorporated for the convenience of the prospective reader.

2 The Ontological Framework

Today Darwin’s theory of evolution, in spite of initially being a badly split field, has
reached a certain consensus and acquired a synthetic unification known as the evo-
lutionary synthesis [10]. Nevertheless it is still classed as a sort of unfinished business
as the materialistic Neo-Darwinism2 has recently been seriously critiqued and
appraised [11]. Our answer to the issues and questions rendered above, has been
considered conceptually as the Paradigm of Evolution, see e.g. Refs. [9, 11] and
references therein. The formulation derives from a general extension of standard
mathematical representations of stationary states in separation from the environment
to so-called open (dissipative) systems.3 The logical framework should be general
enough to include the axioms of quantum mechanics. Below, we will, as already
mentioned, give a short account, but for more details we refer to [9, 11, 13].
Before activating a reading of the specifics needed for a more basic portrait, we
must recognize already at the start that there are many controversial undertakings
instigated by the various practitioners of science and their views of what should be
considered as fundamental concepts in the scientific description of Mother Nature.
For instance Barbour [14] rhetorically asks whether time is ‘real’, while at the same
time advocating a timeless theory of the universe. Furthermore Primas, in his Mind
and Matter article [15] projects a holistic reality, where time is not taken to be an a
priori concept. These examples of deviation from standard practice are well
enunciated, yet they run the risk of drawing unwanted consequences that incur more
problems than it solves, e.g. the unavoidable emergence of internal times, the
imperative dependence on boundary conditions, or paraphrasing Primas: ‘the death
of natural laws’ by being trapped in a Gödelian self-reference.4 We will return to
this question further below.

2
For discussions on the somewhat controversial concept of Neo-Darwinism, see e.g. [10, 11].
3
The concept of dissipative systems and their irreversible processes has been the focus of Pri-
gogine and the Brussels–Austin School, [12].
4
Note that the Gödel (self-referential) paradox can be translated to a consistently formulated
mathematical singularity of a suitably extended logic [16].
A Simple Communication Hypothesis: The Process … 385

Since the dawn of civilization dichotomies like atom—void, realism—idealism,

reality—logos, body—soul, nature—spirit, res extensa—res cogitans, sein—zeit,
material—immaterial, objective—subjective etc., have been at the centre of atten-
tion for philosophers and naturalists alike. However, Kant’s scientific worldview
[17] engrained in the Enlightenment did set the terms for a metaphysics and phi-
losophy for the coming years with fundamental motifs and patterns surviving until
today. He told us, already more than two hundred years ago, that ‘space and time
are the two essential forms of human sensibility’, i.e. everything we, as evolved life
forms perceive and experience, concerns phenomena set in this perspective.
Of course it is impossible for us to comprehend what Kant and his contempo-
raries actually had in mind and meant with such a statement, nevertheless an
orthodox starting point for doing science would rightfully study material repre-
sentations in terms of energy and momentum as imagined and sensed by us in
Nature. Obviously energy-momentum constitutes the material part, while space and
time comprise the immaterial part of the world. Moreover, as any student of
quantum mechanics must learn today, it is imperative to realize that the two aspects
referred to are entwined through the non-commutative conjugate relationships of
their representations as linear operators.
This distinction is therefore mandatory in modern physics, however, not only in
connection with quantum physics, but also in preserving a fundamental interpre-
tation within classical physics, where key correspondences are discerned through
the celebrated Fourier Transform. It is perhaps somewhat surprising that this feature
extended to incorporate special and general relativity admits generalizations to open
system dynamics, see e.g. Ref. [18] and in the following section.

3 A Consistent Formulation of Energy-Momentum

and Space-Time

Based on the argument just given above it follows that once space and time are
specified, momentum and energy are inevitable consequences of the declaration—
and vice versa. For instance, if some object, defined by its energy and momentum,
permits a portrayal within the basic deductive axioms of science as practised at
present in physics and formulated by the most succinct language we know today,
mathematics, we cannot avoid deductio in domum of the immaterial degrees of
freedom as given by time and space. Likewise, if we are able to differentiate some
non-stationary entities in space, we are in principle capable to discuss their
energy-mass-momentum relationships. The obvious conclusion is thus: inducing
the question whether time (or space) is not fundamental in some general setting
invites situations where the conjugate relationship viz. time–energy (or space–
momentum) would be relaxed. Although such situations might occur, e.g. in a black
hole, it might still be preferable to keep the deductive structure of a physical theory
intact, in order to investigate and analyse the emergence of any violations as they
386 E. J. Brändas

might appear. In what follows we will see by a simple realization how such situ-
ations arise under the most trivial conditions.
In particular we will analyse what happens when a non-zero rest-mass particle,
satisfying the Klein–Gordon (wave) equation, via some physical process (pair
annihilation etc.) produces photons. The interest in this question is motivated by
what happens to the degrees of freedom, i.e. the loss of the longitudinal degree
exhibited by massless particles (photons). The starting point will be a consideration
of the abstract Dirac kets in terms of the coordinate vector x⃗ and the linear
momentum p⃗

jx⃗, ict ⟩, jp⃗, iE ̸c⟩

with the scalar product for a free particle given by

ψðx⃗, tjp⃗, EÞ = ð2πℏÞ − 2 eℏðp⃗ ⋅ x⃗ − EtÞ

i

Here the energy is given by the mass relation E = mc2 , m the mass, c the velocity
of light and t, τ the time variable (operator).
The Klein–Gordon equation for a non-zero rest-mass particle is given in obvious
notation by
2
E02 2 Eop
− = p⃗ − ð3:1Þ
c2 c2

where E0 = m0 c2 and p⃗ = mv⃗. Since we are interested in the immaterial degrees of

freedom we also write down the analogous relation for the conjugate variables
(operators) introducing the familiar eigentime expression τ0 given by

− c2 τ20 = x⃗2 − c2 τ2 ð3:2Þ

In order to analyse the intrinsic character of these two equations when the
non-zero rest-mass m0 goes to zero, we use Dirac’s trick to “take the square root of
the equation” by rewriting the observables in matrix form with Eqs. (3.1) and (3.2)
being their associated secular equation.
Hence our concern is the conjugate pair of observables usually represented in
operator form as (note that the time operator is trivially defined when the energy
interval is (–∞, +∞) as is also the time interval)

∂
Eop = iℏ ; p⃗op = − iℏ∇⃗x⃗ ð3:3Þ
∂t

and
A Simple Communication Hypothesis: The Process … 387

∂
τ = top = − iℏ ; x⃗op = iℏ∇⃗p⃗ ð3:4Þ
∂E

In writing down the ansatz, we recognize two5 fundamental characteristics of the

relativistic formulation. The first is due to Alexander S. Davydov, who asserted in
his celebrated text on quantum mechanics: “we show the inapplicability of the
concept of an essentially relativistic motion of a single particle” [19]. The second
does result from PerOlov Löwdin’s treatment of general binary products in Chap. 5
of his book Linear Algebra for Quantum Theory [20], where he discusses the
indefinite metric associated with the Minkowski space. It is easy to see that our open
system dynamics based on a complex symmetric matrix representation is com-
mensurate with an indefinite metric [18]. Hence we set forth the ansatz as (note that
the insertion of −i in the off-diagonal elements is just a convention in the con-
struction of a complex symmetric matrix)


Eop − ip⃗op c
ð3:5Þ
− ip⃗op c − Eop

Hence we recognize Eq. (3.1) as the secular equation of the matrix defined in
Eq. (3.5) with the two eigenvalues ±λ, given by λ2 = m20 c2 , with m0 ≠ 0. Since the
formulation works for both variables and for operators, we will not explicitly single
out which reading is made unless the situation calls for a preference.
Similarly one obtains for the conjugate operators in (3.4) that Eq. (3.2) becomes
the secular equation of the (operator) matrix


cτ − ix⃗
ð3:6Þ
− ix⃗ − cτ

with ±cτ0 being the associated eigenvalues exhibiting the two time directions and
the left- and right-handed coordinate systems connected by space inversion sym-
metry. From (3.5) follows directly the usual approximations and estimates that
result in the conventional time-dependent Schrödinger equation as restricted to the
non-relativistic domain. With this inherent conjugate structure in mind, we remind
the readers of the attempts, see e.g. Refs. [14, 15, 21] to derive pioneering quantum
mechanics from general algebraic structures asserting that time (or space-time) is
not a primary concept. Yet the present description, as consistently built above, is
fundamental and commensurate with the deductive nature of the Lorentz

5
In fact a third comment is due, viz. the use of operators in the matrix calling for logical exten-
sions, see more in the next section. Note also the juxtaposition of the “arrow” of time and the parity
of space.
388 E. J. Brändas

transformation, see e.g. Refs. [18, 22], since from Eqs. (3.1) and (3.2) one obtains
straightforwardly the relativistic space-time scales of special relativity
m0
m = pffiffiffiffiffiffiffiffiffiffiffiffi ð3:7Þ
1 − β2

with β = v ̸c = p ̸ mc, and

τ0 x0
τ = pffiffiffiffiffiffiffiffiffiffiffiffi ; x = pffiffiffiffiffiffiffiffiffiffiffiffi ð3:8Þ
1−β 2
1 − β2

In this setting Eqs. (3.7) and (3.8) are valid irrespective of whether we are
representing classical wave propagation, quantum matter waves or classical
particles.
It is now possible to investigate the limit m0 → 0, observing the connection with
Maxwell’s equation for scalar and vector fields, by placing m0 = 0 in Eq. (3.1).
Something extraordinary happens with the character of the matrices of Eqs. (3.5)
and (3.6), when inserting the relation E = pc, with the momentum p⃗ assumed to be
directed along the x-axis. The result is a complex symmetric degenerate matrix
(reduced to the dimension of mass)



p ̸c − ip ̸ c 1 −i
= p ̸c ð3:5′ Þ
− ip ̸ c − p ̸c −i −1

which is similar to the classical canonical form given by6


0 1
2p ̸ c ð3:5′′ Þ
0 0

In other words (3.5′) cannot be diagonalized since the two column vectors in the
matrix are linear dependent.
Analogously we obtain for (3.6) (since for the photon c2 = x⃗2 ̸ τ2 )




cτ − ix⃗ 1 −i 0 1
= cτ → 2cτ ð3:6′ Þ
− ix⃗ − cτ −i −1 0 0

We find to our surprise that that the complex symmetric matrices in Eqs. (3.5′)
and (3.6′) are nothing but a Jordan block of order 2, (a non-zero matrix, whose
square is zero) or, in more technical language, of Segrè characteristic 2. The linear
dependency leaves space-time with one less spatial dimension, i.e. along the x-axis.
Hence there is no longitudinal degree of freedom for a zero rest-mass particle with

6
The actual transformation is unitary, for details see e.g. Ref. [13].
A Simple Communication Hypothesis: The Process … 389

speed c, like the photon! As will be clear below, this implies that zero- and non-zero
rest-mass particles behave fundamentally different, which will become of crucial
importance in the case of the theory of general relativity to be reviewed below.
It is straightforward to extend our conjugate operator arrays above, to the general
case by the following modifications, see below, where μ is the gravitational radius,
G the gravitational constant, v = p ̸ c, r = jx⃗j, M a (usually large) spherically sym-
metric (non-rotating) mass,7 independent of m, i.e.


mð1 − κðr ÞÞ − iv
ð3:9Þ
− iv − mð1 − κðr ÞÞ

with8

mμ M
mκðr Þ = ; μ=G ⋅ 2 ð3:10Þ
r c

Note that as before, space-time and energy-momentum spaces associate the two
partitions into the material- and the immaterial sections of the Universe. Further-
more, as the area velocity multiplied by m is a constant of motion, one obtains for
local circular motion the boundary condition:

v = κ ðr Þc ð3:11Þ

to be incorporated in Eq. (3.9).

To be precise one must proceed by solving the corresponding secular equations
for the constituent partners of the representation of the Universe,9 as the space-time
background must be simultaneously incorporated with the energy-mass dynamics
so as to conform to background independence. Considering that Einstein’s law of
general relativity equates the matter-energy content with the metric of a curved
space-time, one encounters an interpretative difference in contrast to the present
formulation. While the equations of general relativity asserts that space-time are
regarded as the forms of existence of a real world, with Matter as its substance, our
material-immaterial partitioning imparts a conjugate structure that does not pre-
suppose one conjugate entity to be consigned to the other. This is an important
‘quality’ as we will see later.
Consequently one must address the problem of conjugate consistency and the
proper division of zero- and non-zero rest-mass particles. Though being not too
difficult, we skip the details, see e.g. [9, 11, 16, 18], and quote the result below.

7
Representing M by a ‘black hole’ shows the consistency between particle—antiparticle symmetry.
8
The matrix (3.9) has a direct link to Gödel’s self-referential paradox see e.g. [9].
9
Concepts like Nature, World, Universe, Cosmos, etc. are here used interchangeably.
390 E. J. Brändas

Leaving out the term r 2 dΩ2 the outcome becomes, not unexpectedly, the so-called
Schwarzschild10 line element ðm0 ≠ 0Þ:

− c2 ds2 = − c2 dτ2 ð1 − 2κ ðr ÞÞ + dr 2 ð1 − 2κðr ÞÞ − 1 ð3:12Þ

Rewriting Eq. (3.9) with the condition (3.11) one obtains


pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 
ð1 − κ ðr ÞÞ − iκðr Þ 1 − 2κðr Þ 0
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
m →m ð3:13Þ
− iκðr Þ − ð1 − κ ðr ÞÞ 0 − 1 − 2κðr Þ

trivially diagonalized, provided κðr Þ ≠ 1 ̸ 2. However at κðr Þ = 1 ̸ 2, which occurs at

the Schwartzschild radius r = 2μ, one encounters an old ‘friend’, i.e. a Jordan block
of order two (independent of m)



1 1 −i 0 1
m →m ð3:14Þ
2 −i −1 0 0

The singular behaviour occurs at the Schwarzschild radius signifying a bound-

ary, inside which the material particle, with m0 ≠ 0, becomes immaterial, defining a
black-hole-like object with at most rotational degrees of freedom.11
In analogy with the treatment of the special theory, we write for the line element,
see [18] for details



cds 0 cAdτ − iBdx⃗
= ð3:15Þ
0 − cds − iBdx⃗ − cAdτ

with
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
A = B−1 = 1 − 2κðr Þ ð3:16Þ

and observing that the conjugate operators are modified as follows

∂ ∂
iℏ = Eop ðt Þ; top = − iℏ ð3:16Þ
∂t ∂Et

∂ ∂
iℏ = Eop ðsÞ; sop = − iℏ ð3:17Þ
∂s ∂Es

with
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ∂s pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Et = Es 1 − 2κ ðr Þ; = 1 − 2κðr Þ ð3:18Þ
∂t

10
As it is usually projected today.
11
In retrospect the singularity shares the same self-referential conundrum as we associate with
Gödel’s incompleteness theorem(s) [9, 16].
A Simple Communication Hypothesis: The Process … 391

where Es and Et represents the energy at the space time-points s and t respec-
tively. With the notation Es = ms c2 , one might identify the energy at a space-time
point s with the mass ms . One can see that the matrix Eq. (3.15) is commensurate
with the metric given by Eq. (3.12). For the case of photons (zero-rest-mass) one
obtains for κ ðr Þ ≠ 1 ̸ 2 that ds = 0 and




cAdτ − iBdx⃗ 1 −i 0 1
= cAdτ → 2cAdτ ð3:19Þ
− iBdx⃗ − cAdτ −i −1 0 0

which is consistent with Eq. (3.6′). One notes that the energy formula for
photons becomes

mð1 − 2κ ðr ÞÞ = p ̸ c

in correspondence with the law of gravitational light bending. Furthermore at

κðr Þ = 1 ̸ 2 the metric is conditionally singular for both mass- and massless particles.
In summary we have demonstrated that the a priori choice of conjugate rela-
tionships, like energy-time, impart simple relations, commensurate with the theory
of relativity, including the emergence of mathematical singularities at the precise
physical conditions commensurate with constraints like the limit velocity of light,
the associated loss of longitudinal spatial dimension and the emergence of a black
hole entity trapping energetic particles. Although not emphasized here, see also
footnotes 8 and 11, the matrix to the left of Eq. (3.13) displays a direct analogy with
the self-referential enigma exhibited by the Gödel’s inconsistency theorem(s),12 see
e.g. Refs. [9, 11, 13, 16]. In the next section we will show how to extend the
degenerate formulation to any order including its interpretation in relation to the
Paradigm of Evolution.

4 The Correlated Dissipative Structure

As we have maintained above, time-energy and space-momentum provide a priori

observables of primary descriptions of the scientific discourse. In this portrayal of
processes in the natural world, one preferred option of a distinct constituent
involves by definition its conjugate companion. Within the theory of evolution,
time is the key variable with a fundamental spatial significance. To appreciate this
idea one needs to understand the emergence of spatio-temporal scales from the

12
The common inconsistency in the vernacular is also known as the liar’s paradox and it has
recently been discussed under the name of The Pinocchio Paradox [23].
392 E. J. Brändas

dynamics of their conjugate partners. The strategy will be to develop the appro-
priate thermodynamics, invoking temperature and entropy to derive apt scales in
concert with appropriate ensembles and to find out how complex enough systems
interact or rather communicate with each other. Since we have presented the various
derivations in earlier contributions [9, 11, 16], we will for the most part only state
the mathematical results, as they are required as well as their physical interpretation.
Though our formulation has a rigorous origin in the axioms of quantum theory
as a trace algebra, see Ref. [20], in terms of general system operators, various
ensembles etc., there are important generalizations to quantum logic as engrained in
the illustrious theorem due to Gleason [24]. Hence the present formulation does not
only refer to pioneering quantum mechanical interpretations, but it also covers
interpretations that go beyond classical Boolean structures.13
A convenient starting point is the second order, reduced for N fermions, char-
acterized by the space-spin coordinates14 xk normalized to the number of pairings
(for details see [9, 11, 16])
 0 0

Γð2Þ x1 , x2 jx1 , x2 =

Z  0 0  ð4:1Þ
N
Ψðx1 , x2 , x3 , . . . , xN ÞΨ* x1 , x2 , x3 , . . . , xN dx3 , . . . , dxN
2

with
n o
E = Tr H2 Γð2Þ ð4:2Þ

for a suitable reduced Hamiltonian—so far all in a standard setting of quantum

mechanics. An essentially wave-function representable two-matrix can be written
n
Γð2Þ = ∑ jhk ⟩γ kl ⟨hl j = jh⟩γ⟨hj ð4:3Þ
k, l = 1

where the n-dimensional matrix γ (in principle specified below) is represented in the
space of the preferred basis jh⟩, the latter referring to appropriate pairs of light
carriers, like electrons described by paired orbitals denoted as geminals localized at
various nuclear centers. In fact the density matrix should describe, together with the
nuclear motion, the full dynamics of the system,15 although we have here only

13
For instance the demonstrated connection with Gödel’s incompleteness theorem contains a
non-Boolean probability that is extended to describing the interactions and communications
between Complex Enough Systems, CES. This prompts the system to be density matrixdenoted as
a Gödelian network [25].
14
We will denote the spatial coordinates with a vector notation, i.e. x⃗k .
15
Usually one invokes the Born-Oppenheimer approximation, i.e. separating the nuclear motion
from the many electron quantum problem. However, this will not be adequate here as will be seen
further below.
A Simple Communication Hypothesis: The Process … 393

indicated the electronic variables. Examples of such representable density matrices

are those related to Yang’s celebrated notion of Off-Diagonal Long-Range Order,
ODLRO [26], see also an alternant derivation by Sasaki [27] and Coleman’s
concept of extreme states [28], of direct relevance for the understanding of
superconductivity and superfluidity. The extreme state corresponds to a degenerate
state with one large eigenvalue, λL , approaching (for large n) the number of pairs,
N/2, and other the (n – 1)-degenerate16 one, λS → 0, where the unitary matrix B will
be defined further below, i.e.
0 1
λL 0 0 0
B 0 λS ⋯
B 0 0 C
C
BγB − 1 = d = B ⋮ ⋱ ⋮ C ð4:4Þ
@ 0 0 λS 0 A
⋯
0 0 0 λS

Next we resume the exploration by confronting two major problems, i.e. dealing
with temperature dependences or accounting for the presence of quantum-thermal
correlations, and at the same time removing the notorious Born-Oppenheimer
approximation, in principle treating the nuclear degrees of freedom on an equal
footing with the electrons.
Let us treat the last problem first by exercising the so-called mirror theorem
employed by Carlson and Keller, [29], in connection with the reduced degrees of
freedom in connection with density matrix theory.17 In particular an application to
the entangled activities between the electronic motion and the movements of the
nuclei, implies that they are coupled through mirroring dynamics [31]. For instance
an electron orbital or geminal, hk projected around nucleus l, i.e. described locally
by the spatial coordinate x⃗l writes in Dirac notation

hk ðx⃗l Þ = ⟨x⃗l j hk ⟩

to be viewed as the kth electron orbital (geminal) as anticipated around the lth
nucleus, can also be interpreted as the scalar product between the lth nucleus
described by the Dirac ket,18 given by jx⃗l ⟩ and the electronic motion characterized
by jhk ⟩. Such scalar products suggest the key ingredients for the mapping between
the electronic and nuclear degrees of freedom. Hence system operators of the kind
Eq. (4.1) should in principle contain both electrons and nuclei, one of which,
electrons or nuclei, could be traced away in order to study the remaining dynamics
by suitable master equations, see e.g. Ref. [12]. The conclusion from the mirror

16
For fermionic systems there is also a 2n (n – 1) dimensional tail of unphysical pairings, which is
omitted.
17
The actual theorem goes back to Erhard Schmidt [30], the mathematician who was behind the
Gram-Schmidt orthogonalization.
18
The Dirac ket denotes a quantum state and its significance originates from the bra-c-ket form as a
scalar product.
394 E. J. Brändas

theorem is that the mappings between the electron- and the nuclear degrees of
freedom, and back, exhibits the same classical canonical form. This imparts the
possibility of a dual interpretation of the actual matrix representations of either the
electronic motion or the associated nuclear motion in principle bypassing the
Born-Oppenheimer approximation.
One may also view the entwined dynamics as a general scattering problem with
electronic particles scattered on a number of nuclear targets yielding consistent
scattering data, see e.g. Ref. [9]. This leads to the second problem of merging
quantum and thermal correlations at precise temperatures invoking associated time
scales and rigorous dissipative (open system) dynamics, which will be done in two
steps. First we need to match the relevant time scales of the system, via a proper
thermalization procedure obtaining the system operator for the open system at the
relevant temperature exhibiting authentic time scales for a realistic description of
non-equilibrium situations of relevance for, what we will denote, Complex Enough
Systems, CES, and their interpretation.
Consider an open system involving n bosonic or paired fermionic degrees of
freedom as an “incoming beam” impinging on a set of nuclear sites. The whole
scattering arrangement is characterized by a relaxation process19 with the time scale
τrel , assumed to be distinctly larger than the smaller thermal timescale τcorr = ℏ ̸kB T.
The protocol describes a process that one will, on the average, detect one quasi
particle degree of freedom in the differential solid-angle element dΩ during the
thermal timescale, e.g. with τcorr ≈2.46 × 10 − 14 s at 310 K. Straightforwardly one
obtains, with the incident flux, Ninc , being the number of particles/(degrees of
freedom) per unit area and time, Ns dΩ, the number of particles scattered into dΩ per
unit time being the standard relations between the differential- and the total cross
sections σ Ω and σ tot , the following formulas
Z Z
n dΩ Ns
Ninc = ; σ Ω dΩ = Ns dΩ = ; σ tot = σ Ω dΩ = dΩ ð4:5Þ
σ tot τrel τcorr Ninc

yielding the simple relationship between the two characteristic times, i.e.

n kB T τrel
= τrel = ð4:6Þ
4π ℏ τcorr

Assuming that the correlated cluster of nuclei perform harmonic oscillations

distributed over the various energies ℏτl− 1 , from the zero point energy, with
equidistant harmonic levels displaying a spectrum from the zero-point energy to
−1
ℏτcorr . Straightforward examination of the situation reveals
nτcorr
τrel = ðl − 1Þτl = τ2 = ; l = 2, 3, . . . , n ð4:6′ Þ
4π

19
Generally speaking a specific molecular process is considered as a local perturbation out of the
quantum state which then relaxes back to thermodynamic equilibrium after a certain time τrel .
A Simple Communication Hypothesis: The Process … 395

Utilizing the standard Heisenberg relation between life times and energy widths
of the state, i.e. εk = ℏ ̸2τk one may express Eq. (4.6) as (with β = 1 ̸kB T, T the
absolute temperature and kB the Boltzmann constant)

2πðl − 1Þ
βεl = ð4:7Þ
n

Relations (4.5)–(4.7) display the relation between the time scales, the tempera-
ture T and the relevant dimension n of the non-equilibrium dissipative system.
The thermalization procedure, extended to the density matrix subject to the
Bloch equation with a Hamiltonian, H, producing thermal fluctuations commen-
surate with a given temperature, gives (note that the usual condition of the eigen-
state thermalization hypothesis is not fulfilled due to ODLRO)

∂ϱ 1
− = LB ϱ; LB ϱ = ðHϱ + ϱH Þ ð4:8Þ
∂β 2

which together with (4.7) yields the surprising result, with γ → γterm

ϱ = e − βLB Γð2Þ = jh⟩γ term ⟨hj = jf ⟩B − 1 γ term B⟨f j = jf ⟩ωd ⟨f j ð4:9Þ

with
0 1
0 λS 0 λL
B0 0 ⋯
B 0 0 CC
ωd = B ⋮ ⋱ ⋮ C = λL J n − 1 + λS J ð4:10Þ
@ 0 0 0 λS A
⋯
0 0 0 0

with J denoting the standard nilpotent matrix with zeros everywhere except with
ones above the diagonal. Note that n → N ̸2 imparts, in contrast to n → ∞ that
λL → λS → 1. In the basis f the density matrix ϱ in Eq. (4.10) writes, while here
leaving out the second term20 above, since it will
n hereoonly play a minor role in the
time evolution, n ≈ N ̸2, ρ∝ϱ normalized to Tr ρρ† = 1

1 1 1 n−1
ρ = pffiffiffiffiffiffiffiffiffiffi jh⟩Q⟨hj = pffiffiffiffiffiffiffiffiffiffi jf ⟩J⟨f j = pffiffiffiffiffiffiffiffiffiffi ∑ jfk ⟩⟨fk + 1 j ð4:11Þ
n−1 n−1 n − 1 k=1

with

20
When n → ∞, λL → N ̸2 will dominate ρ∝ N2 jf1 ⟩⟨fn j activating e.g. large-scale coherent axonal
firing.
396 E. J. Brändas

jh⟩ = jf ⟩B and the complex symmetric

Jordan
 block given by
1 iπðk + l − 2Þ
Qkl = δkl − en ð4:12Þ
n

̸n
The unitary transformation B is finally given by21 with ω = eiπ
0 1
1 ω ω2 ... ωn − 1
1 B 1 ω3 ω6 ... ω3ðn − 1Þ C
B = pffiffiffi B C ð4:13Þ
n@⋮ ⋮ ⋮ ⋮ ⋮ A
1 ω2n − 1 ω2ð2n − 1Þ ... ω ðn − 1Þð2n − 1Þ

In summary we have derived an irreducible representation Eq. (4.11), which will

be denoted as a Correlated Dissipative Structure, CDS, that is commensurate with
the physical time scales τcorr and τrel related through Eqs. (4.6) and (4.7). The CDS
configuration depends on T releasing a thermal oscillation into the complex enough
system, CES. The transformation from the “local” preferred basis jh⟩ to the
canonical one jf ⟩ is given by B−1, or since it is unitary B† . Note that B also
transforms Γð2Þ to diagonal form, Eq. (4.4). Finally we observe an interesting factor
property of B, which will be of vital importance in the next section.

5 The Correlated Dissipative Ensemble

To illustrate the Gödelian Network as a Correlated Dissipative Structure we will

employ the CDS as base units for a “higher level” Liouville formulation based on
the Liouville generator

Lρ = ½Hρ − ρH

For instance applying L above to Eq. (4.11) gives to first order the sum of the
energy differences Ei − Ej . Since the thermalization, leading up to the successive
transitions in ρ, is brought about by the exchange of a thermal oscillation, due to the
energy super-operator in (4.8), the final outcome becomes the overall change
E1 − En , which is nothing but the thermal frequency associated with τcorr . Hence the
present Liouville picture describes the CDS by this frequency and with the char-
acteristic lifetime22 τrel .
As a result one obtains an entity, which we will call the Correlated Dissipative
Ensemble, CDE. The latter, by its construction, integrates a principal basis set of

21
This form was obtained in collaboration with C. E. Reid [32] see also [33].
22
A rigorous analytic continuation of L is given in [34]. While the real part of the eigenvalue
appear as energy differences, the imaginary parts add up here to be consistent with Eq. (4.6).
A Simple Communication Hypothesis: The Process … 397

CDS’s defined by Eq. (4.11), and denoted by jH ⟩, where each CDS is commen-
surate with the time scales τcorr and τrel related through Eqs. (4.6) and (4.7). In
terms of the frequency23 ω0 = ℏ ̸ kB T and τ = τrel each CDS is characterized by

n kT
= τrel = ω0 τ ð5:1Þ
4π ℏ
R
with Q = ω0 τdΩ = n serving as a quality factor for the CDE.24 Deriving the CDE
from CDS units, i.e. molecular aggregates, like the DNA and/or its protein over-
coats, in a biological system, will permit the definition of a particular cellular
quality value, Q. This is to some extent analogous to signal processing in com-
munication systems, where the quality aspects of cavity resonators, like musical
instruments etc., play a vital role in transmitting resonating qualities. For that reason
the quality value Q = n, for e.g. somatic cells in a multicellular organism, trans-
ferring vital details regarding their traits, provides cell recognition through the
process of “molecular communication”.
The CDE exhibits an analogical irreducible unit Q in the basis H of CDS base
units. A simplified analysis of the associated system operator shows that its diag-
onal elements can be determined by a general probability measure, let us say
pð AÞ = TrðρPð AÞÞ, (ρ a given density operator and P(A) a suitable self-adjoint
projection on Hilbert Space for the event A), and (1 – p(A))p(A) for any off-diagonal
one displaying ODLRO. An analogous analysis, cf. the CDS, leads through diag-
onalization and thermalization to a similar irreducible unit Q as in Eq. (4.11) but
generally with a different dimension m, in terms of H and the transformed F, cf. the
relations between h and f. The key difference is that τ = τrel play the role of the short
(relatively speaking) time scale with a longer scale emerging through Q see more
below. We can now write down the propagator corresponding as (with
I = ∑m k = 1 jFk ⟩⟨Fk j)

P = ðω0 τ − iÞI + iJ ð5:2Þ

m−1
J = ∑ jFk ⟩⟨Fk + 1 j ð5:3Þ
k=1

jH ⟩ = jF ⟩B ð5:4Þ

In analogy with classical dynamics we have the equivalent of a causal propa-

gator GðtÞ and a resolvent GR ðzÞ defined by

Gðt Þ = e − iP τ ; GR ðωÞ = ðωτI − P Þ − 1

t
ð5:5Þ

The notation ω0 = ℏ ̸ kB T should not be confused with ω = eiπ ̸n defined in Eq. (4.13).
23

The Q-value should not be confused with the complex symmetric matrix Q.
24
398 E. J. Brändas

which in the classical case are related through the standard Fourier transform. In the
present case the extension requires the separation of positive and negative times
focusing our interest on the retarded propagator.25 Inserting the Liouvillian,
Eq. (5.2) in (5.5), one obtains

m − 1 t  k
1 k
e − iP τ = e − iω0 t e − τ ∑
t t
J ð5:6Þ
k=0 τ k!
m
ðωτI − P Þ − 1 = ∑ ½ðω − ω0 Þτ + i − k ðiJÞðk − 1Þ ð5:7Þ
k=1

where one notes that the expansions are finite, limited by the dimension m. The
occurrence of higher order poles in Eq. (5.7) is reflected by the build up of a
polynomial in front of the decay factor in (5.6). As a result the usual microscopic
law of evolution dN ðt Þ = − ð1 ̸τÞN ðt Þdt modifies according to the highest power
m – 1 of J, see [13] for details
 t
dN ðt Þ = t m − 2 m − 1 − N ðt Þdt; dN ðt Þ > 0; t < ðm − 1Þτ ð5:8Þ
τ

The consequence of the irreducible perturbation in Eq. (5.2) is the emergence of

a new basic “communication” time scale τcom = ðm − 1Þτrel or mτrel adding the
decay time.
It might seem a misnomer to refer to Eqs. (5.5)–(5.7) as a CDE, i.e. a statistical
ensemble associated with a perturbation out of equilibrium. However it represents
assembled stochastic features construed for Complex Enough Systems with pro-
grammed timescales at precise temperatures. Yet the mystery of “molecular com-
munication” remains, i.e. how does transmission of crucial molecular traits extend
from microscopic levels generating significant cellular information? An answer is
given by comparing the two transformations

jh⟩ = jf ⟩B ð5:9Þ

or the CDS, relating n light carriers in a nuclear skeleton of n sites, treating the
nuclear and the electronic system on par, and

jH ⟩ = jF ⟩B ð5:10Þ

for the CDE, relating e.g. m cells in a certain organ localised at key positions in a
particular organ or organism. If m and n reveal no relation whatsoever, one might
not anticipate any affinity between the cells with non-commensurate Q-values.

25
A detailed exposition of the connection between the correlation function and the corresponding
spectrum, including analyticity requirement and appropriate integration contours have been dis-
cussed in the appendix of Ref. [35].
A Simple Communication Hypothesis: The Process … 399

However if they are the same—or as we will see containing many least common
multiples—they share the same time scale τ = τrel as well as displaying common
time factorizations as shown by the transformation B. Actually any perception,
depending on the rows of the matrix B, should amount to “translating” the cyclic
vectors of B via the transformation back to the “physical” basis h or H localized in
the cell and in the organism respectively. Hence one might envisage the translation
of compatible molecular information via the communication bearing transformation
B as authorized “messages” between the molecular-DNA-RNA-gene level and the
higher-level hierarchical organization leading to cellular function and biological
order.
On the condition that the Q-values in (5.9) and (5.10) are commensurate, it
suggests the possibility of a communication protocol given by the cyclic properties
and their nestings of the columns of B. For instance analysing the transformation in
Eq. (4.13), one observes the way the elements repeat themselves due to the
cyclicity of complex numbers. For instance choosing n = 12, computing the col-
umns of B, one may display the result as the simplified diagram below, in which
only the dimensions of the recurring vectors are indicated. Removing the first
column of one-dimensional units “1”, the resulting graph, containing 11 columns,
will look like

2
3 2 3
4 4
6 3 2 3 6
12 4 12 12 4 12 ð5:11Þ
6 3 2 3 6
4 2 4
3 3
2

where we do observe the obvious column symmetry of the table. From Eq. (5.6)
one finally realizes the typical behaviour of Poisson statistics that suggests a
communication concept that will share some analogy with the dynamics of a
telephone Call or Contact Center.26

6 Examples and Conclusions

In summary we have derived and suggested an explication of a possible programme

for the transmission of proper gens between two distinct levels of scientific for-
mulation, i.e. the molecular microscopic- and the cellular mesoscopic ones. Many
biologists flirt with quantum mechanics without coming to terms with its conse-
quences. This, in many cases, turn out to be quite confusing, since it is not easy to

26
This analogy has been carried out in some detail in relation to Trehub’s Retinoid Model of the
cognitive brain [36, 37]. The representation of neurons with a chemical synapse onto itself is of
particular significance for the Gödelian network [25].
400 E. J. Brändas

ascertain the Grand Master’s inner thoughts about the cornerstones of physics.
While Einstein did say that ‘time is an illusion’, Prigogine on the other hand
asserted that ‘we are the children of time’. As the prospective reader of this con-
tribution might have realized, energy-momentum and space-time are inevitably
coupled to each other as conjugate entities and that the present formulation, as a
consequence, invokes the characteristics of the Correlated Dissipative Ensemble,
CDE.
In previous studies we have emphasized the opportunity of cell recognition
suggested by the possibilities offered by the cellular Q-value prearranged by the
CDS as building blocks in every cell. This problem involves both inter- and
intra-cell communication. The lower- as well as the higher-level structures are
subject to Poisson statistics defining physical communication channels between and
within them. In particular we have focused on the CNS, the central nervous system,
in the presence of a special type of cells, i.e. neurons, with the dynamics charac-
terized by short time scale oscillations, building up pulses of light carriers (elec-
trons), i.e. spikes that correlate the basic dissipative systems e.g. cells/neurons, see
also footnote 21, and providing an irreducible coupling that arrange for the com-
munication between them. For more details regarding the usage of stochastic
backgrounds as objective physical communication channels, the former commen-
surate with the Poisson statistics, authorizing a direct call centre analogy, i.e. the
distribution of the rate of incoming phone calls received at each neuron, the
switchboard, see Refs. [9, 11, 25]. Note that the code for exchanging “communi-
cation” is given by the transformation B, Eq. (4.13), which applies in both limits:

N
λL → ; n→∞
2

and

N
λL → 1; λS → 1; n →
2

while not being contained in the actual “phone call”.

Special attention has been given to Arnold Trehub’s celebrated Retinoid Model
for vision [36, 37] constituted by a special type of neurons called the autapse, i.e.
with a chemical synapse linked onto itself. In particular we have utilized the
symmetry27 of the diagram (5.11), where the columns to the left of the centre are
minus one times the complex conjugate of the ones to the right.28 By projecting the
neural dynamics onto Trehub’s “3D egocentric space” one realizes a certain

27
It is important to realize that this is not a “biological” symmetry, but induced by the perception of
mirror neuron images produced by the autapse.
28
This is a crucial feature, since according to the Charge, Parity, Time reversal, CPT, theorem,
believed to be the fundamental symmetry of physical laws, see also Ref. [38] the brain perceive
time reversal symmetry as a space parity inversion, which explains the Necker cube illusion [25].
A Simple Communication Hypothesis: The Process … 401

mirroring symmetry that is instigated by the feedback loops of the autapse. Hence
every trajectory in egocentric space exhibits a time-reversed copy in agreement with
the above-mentioned symmetry. This yields a simple understanding of the Necker
Cube Illusion [25] and the fact that our eyes see everything upside-down. The
examples given above relate to communication between cells connected to the
central nervous system, CNS. In what follows, we will concentrate on the other
question, the trans level communication.
As our conclusion concerns open questions in the philosophy of biology, see
Ayala and Arp [39] for recent debates, there appear some highly interesting work,
where the communication between the cells plays a preferential role [40]. In his
formulation of the First Principles of Physiology, integrating the self-organizing
status of the unicellular state, see also Ref. [41], John Torday, asserts that biological
organisms exist far from equilibrium, ‘circumventing’ the Second Law of Ther-
modynamics generating negative entropy within them, sustained by chemiosmosis.
Moreover he finds, going through the life cycles of biological phenomena, that
evolution on this level is deterministic, while at the same time arguing that the
unicell “finds itself” in an ambiguous situation with respect to its environment.29
Along these lines of reasoning, while acquiring epigenetic marks, one discerns
Lamarckian traits emphasizing environmental input over genetic evolution advancing
the so-called Epigenetic Inheritance Systems, EIS, [42], suggesting alternative possi-
bilities for extending the boundaries in evolutionary biology. For instance, recent
studies on genomic imprinting in mammals, see the recent review by Li and Sasaki [43],
is redolent of mechanisms relating to the reprogramming of pluripotent stem cells. One
should emphasize, however, that these models have no direct relation to quantum
chemical processes inside a cell. Nevertheless the general question remains, i.e. how
does selection act, what sort of entities are selected? To answer this question, one needs
to go to the intracellular level translating the intercellular signals on the level of the
DNA-protein coding to the higher-order level of cell-interactions. For instance, Torday
demonstrated how the parathyroid hormone-related protein PTHrP and leptin signalling
mechanisms, do facilitate a somatic program of lung evolution, offering ontogenic/
phylogenic links. In consideration of the present idea, it would be intriguing to find a
way to transcend the intracellular level, translating the molecular signals from the
microcirculation of the alveolus to alveolar metabolism.
One may enquire what time scale τ would be appropriate for the CDE in general.
While the CDS engender a relaxation time commensurate with its process forming
dynamics, it might be interesting to investigate e.g. the important role of the primary
cilium in modulating neurogenesis. They consist of micro-tubule-based organelles,
the latter also being part of the neuronal structure, which has been particularly
emphasized by Hameroff and Penrose [44] in connection with their “Orch OR” theory
for the appearance of consciousness in the universe.30 In eukaryotes the structure of

29
This is an intriguing declaration, since this is precisely “communication” formulated in this
article and programmed by Gödel-like sentences.
30
See also the discussions following their article in the Physics of Life Reviews.
402 E. J. Brändas

microtubules are long cylinders of α- and β-tubulin dimers,31 each with a molecular
weight of about 50 kDa formed in parallel association of thirteen protofilaments. In
total we have a weight of around 1000 kDa, corresponding to about 106 protons. With
a nuclear average of approximately 10 Da, one might expect roughly 105 nuclear
degrees of freedom to be at work in each of the ca 109 microtubules. Using Eq. (4.6)
one finds cilia beating frequencies of the order τrel ≈ 25 ms at human body temper-
atures, which appears to be in the right range for motile cilia mediated pathways. It has
also been noted that defects in the structure of the cellular primary cilium leads to
various disorders and impairments [45]. Hence looking at recurrent frequencies,
around 10–40 Hz, one will conclude that obstructing cell organelle movements might
severely disturb the channel for stochastic communication. Despite the view that
perhaps 10–20,000 neurons are involved in a perception instigated by specific spike
trains, the present CDS concept, nevertheless, implies a more complex understanding
in that a Gödelian network must have all neurons ‘on alert’ in case of need.
Another critical challenge, while retaining quantum chemistry applications to
biological systems, is the temperature dependence, the thermoregulation, and the
associated problem of the necessary prohibition of short-time32 decoherences. The
development and appearances of Correlated Dissipative Structures, CDS, not only
suggests answers to this dilemma, it also incorporates realistic time scales and the
linked nested ‘code-encode bearing’ transformation B. In this connection it is also
interesting to observe that so-called poikilotherms, i.e. organisms whose internal
temperatures varies considerably, have several different enzyme systems operating
at different temperatures, hence representable by different CDE’s.
Finally, a reminder regarding our biophysical organisation, termed a complex
enough system, CES, subject to the CDE, viz. the interactions between the cells
commensurate with the dynamics inside the cell, derived from a CDS, is neither
deterministic nor probabilistic.33 This incurs no contradiction, even if the biological
process may appear deterministic at the mesoscopic level of understanding.
The CDE is an irreducible ensemble defined in terms of the CDS, the latter
exhibiting an analogous irreducible structure. Within this framework one might
finally be reminded of the words of Ernst Mayr [10, 46] regarding the character-
istics of biological processes: a teleonomic process or behaviour is one that owes
its goal-directedness to the influence of an evolved program. Our interpretation
suggests that the stochastic hypothesis depicted here, i.e. a “Communication

31
Since the microtubules have a distinct polarity one obtains a direct coupling between the
mechanical- and the electromagnetic oscillations generating the train of neuronal spikes.
32
The notion ‘short’ is here in relation to any other time scales discussed in this article.
33
It has been stated that the mathematical structure of cosmos, with the origin of the physical laws
being in pure numbers, may discard the theory of biological evolution as well as being incom-
mensurate with the non-deterministic interpretation of quantum mechanics [47]. It is clear, how-
ever, from the present work that numerical structures, like the formula 1 + 1 = 2, reveals
important and necessary facts about our universe that are essential for the communication between
life forms and contingent on evolution, see e.g. a recent scientific analysis of ‘Science and Music’
[48] focusing on musical transcriptions in the living world.
A Simple Communication Hypothesis: The Process … 403

Simpliciter”, derived from first principles, can be understood and generated as such
an authentic process.

Acknowledgements I am grateful to the Chair of QSCP-XXI, Prof. Yan Alexander Wang, and
the Cochair Prof. Jean Maruani, for generously allowing me to present this work in the present
proceedings from the Vancouver meeting. This work has, over time, been supported by the
Swedish Natural Science Research Council, Swedish Foundation for Strategic Research, the
European Commission, and the Nobel Foundation.

References

1. Tegmark M (2003) Parallel universes. Sci Am

2. Deutsch D (2011) The beginning of infinity. Viking, Penguin, New York
3. Primas H (1983) Chemistry, quantum mechanics and reductionism. Perspectives in theoretical
chemistry. Springer, Berlin
4. Atmanspacher H, Müller-Herold U (eds) (2016) From chemistry to consciousness. The legacy
of Hans Primas. Springer, Berlin
5. Heidegger M (1927) Sein und Zeit. Max Niemeyer Verlag, Tübingen
6. Feyerabend P (1975) Against method. New Left Books, London
7. Nye MJ (1993) From chemical philosophy to theoretical chemistry. University of California
Press, Berkeley
8. Brändas EJ (2017) Löwdin—Father of quantum chemistry. Mol Phys 115:1
9. Brändas EJ (2015) A zero energy universe scenario: from unstable chemical states to
biological evolution and cosmological order. In: Nascimento MAC, Maruani J, Brändas EJ,
Delgado-Barrio G (eds) Frontiers in quantum methods and applications in chemistry and
physics, vol 29. Springer, Dordrecht, p 247
10. Mayr E (1988) Towards a new philosophy of biology. Observations of an evolutionist.
Harvard University Press, Cambridge
11. Brändas EJ (2017) The origin and evolution of complex enough systems. In: Tadjer A,
Pavlov R, Maruani J, Brändas EJ, Delgado-Barrio G (eds) Quantum systems in physics,
chemistry, and biology, vol 30. Springer, Berlin, p 413
12. Prigogine I (1996) The end of certainty: time, chaos, and the new laws of nature. The Free
Press, New York
13. Brändas EJ (2012) Examining the limits of physical theory: analytical principles and logical
implications. In: Nicolaides CA, Brändas EJ (eds) Unstable states in the continuous spectra
Part II: Interpretation, theory, and applications. Advances in quantum chemistry, vol 63.
Elsevier, Amsterdam, p 33
14. Barbour JB (1999) The end of time the next revolution in physics. Oxford University Press,
Oxford
15. Primas H (2003) Time-entanglement between mind and matter. Mind and Matter 1(1):81
16. Brändas EJ (2011) Gödelian structures and self-organization in biological systems. Int J Quant
Chem 111:1321
17. Kant I (1781) Kritik der reinen Vernunft. Johann Friedrich Hartknoch, Riga
18. Brändas EJ (2016) A comment on background independence in quantum theory. J Chin Chem
Soc 63:11
19. Davydov AS (1965) Quantum mechanics. Pergamon Press, Oxford
20. Löwdin P-O (1998) Linear algebra for quantum theory. Wiley, New York
21. Baumgarten C (2015) Minkowski spacetime and QED from ontology of time.
arXiv:1409.5338v5[physics.hist-ph] 30 Nov
404 E. J. Brändas

22. Löwdin P-O (1998) Some comments on the foundations of physics. World Scientific,
Singapore
23. Eldridge-Smith P, Eldridge-Smith V (2010) The Pinocchio paradox. Analysis 70(2):212
24. Gleason AM (1957) Measures on the closed subspaces of a Hilbert space. J Math Mech 6:885
25. Brändas EJ (2015) ) Proposed explanation of the Phi phenomenon from a basic neural
viewpoint. Quant Biosyst 6(1):160
26. Yang CN (1962) Concept of off-diagonal long-range order and the quantum phases of liquid
helium and of superconductors. Rev Mod Phys 34:694
27. Sasaki F (1965) Eigenvalues of fermion density matrices. Phys Rev 138B:1338
28. Coleman AJ (1963) Structure of fermion density matrices. Rev Mod Phys 35:668
29. Carlson BC, Keller JM (1961) Eigenvalues of density matrices. Phys Rev 121:659
30. Schmidt E (1907) Math Ann 63:433
31. Brändas EJ, Hessmo B (1998) Indirect measurements and the mirror theorem. Lecture notes in
physics, vol 504, p 359
32. Reid CE, Brändas EJ (1989) On a theorem for complex symmetric matrices and its relevance
in the study of decay phenomena. Lecture notes in chemistry. Springer, Berlin, pp 325–475
33. Brändas EJ (2009) A theorem for complex symmetric matrices revisited. Int J Quant Chem
109:28960
34. Obcemea CH, Brändas EJ (1983) Analysis of Prigogine’s theory of subdynamics. Ann Phys
151:383
35. Brändas EJ (1997) Resonances and dilation analyticity in Liouville space. Adv Chem Phys
99:211
36. Trehub A (1991) The cognitive brain. MIT Press
37. Trehub A (2007) Space, self, and the theatre of consciousness. Conscious Cogn 16:310
38. Quack M (2011) Fundamental symmetries and symmetry violations from high resolution
spectroscopy. Handbook of high-resolution spectroscopy, vol 1, p 659
39. Ayala JF, Arp R (eds) (2010) Contemporary debates in philosophy of biology.
Wiley-Blackwell, Chichester, West Sussex
40. Torday JS, Miller WB Jr (2016) The unicellular state as a point source in a quantum biological
system. Biology 5(2):25
41. Torday JS (2016) Life is simple—biologic complexity is an epiphenomenon. Biology 5(2):17
42. Jablonka E, Lamb M (2005) Evolution in four dimension—genetic, epigenetic, behavioral,
and symbolic variation in the history of life. The MIT Press, Cambridge
43. Li Y, Sasaki H (2011) Genomic imprinting in mammals: its life cycle, molecular mechanisms
and reprogramming. Cell Res 21:466
44. Hameroff S, Penrose R (2014) Consciousness in the universe a review of the ‘Orch OR’
theory. Phys Life Rev Mar 11(1):39
45. Lee JH, Gleeson JG (2010) The role of primary cilia in neuronal function. Neurobiol Dis 38
(2):167
46. Mayr E (2004) What makes biology unique? Cambridge University Press, New York
47. Sanchez FM (2017) A coherent resonant cosmology approach and its implications in
microphysics and biophysics. In: Tadjer A, Pavlov R, Maruani J, Brändas EJ, Delgado-Barrio
(eds), Quantum systems in physics, chemistry, and biology, vol 30. Springer, Dordrecht,
p 379
48. Maruani J, Lefebvre R, Rantanen M (2003) Science and music: from the music of the depths
to the music of the spheres. In: Maruani J, Lefebvre R, Brändas EJ (eds), Advanced topics in
theoretical chemical physics and physics, vol 12. Kluwer, Dordrecht, p 479
Index

A Chemical semantics, 3–7, 9, 18, 21, 23, 25

Ab initio calculation, 119, 135, 194–196, 199, Chirality, 95–97, 100, 102, 103, 105
201, 253 Compton diameter, 368, 371, 374, 377
Ab initio potential energy surface, 138, 179, Correlated Dissipative Ensemble (CDE),
215 381–383, 396, 400–402
Ab initio study, 107, 135, 143, 194 Correlated Dissipative Structure (CDS), 381,
Acylphloroglucinols, 281, 282 396, 397, 400–402
Adducts with explicit water molecules, 285, Counterpoise correction, 44, 50
323 Covalent bonding, 203, 204, 206, 207, 209,
Alkali borate, 108, 120, 123, 124, 138 210
Alkaloids, 306 Crystal structure, 119, 120, 150, 154, 159, 173,
Annihilation momentum density, 246, 251 314
Antimalarials, 305, 306, 318, 322, 323, 325
Arzanol, 281–283, 287, 290, 293, 294, 296, D
297, 299–301 Debye plasmas, 56, 65
Atomic wave functions, 29 Density functional, 6, 16, 138, 193, 213, 214,
232, 244, 286, 306, 308, 331, 334, 335
B Density matrix, 244, 392, 393, 395
Basis set superposition error, 44, 194, 287 Density variation, 332, 337–339, 344–349, 353
Bond critical point, 46, 48 Diabatic atmospheric negative ion, 192, 193,
Bond dissociation energy, 44, 48 201
Bond energy, 44 Dirac equation, 98, 231, 232, 361–364, 370,
Bond order, 43, 44, 48, 50, 204 371
Borate, 107, 108, 112, 120, 128, 143, 144, 148,
159, 164, 165, 171, 174 E
Boric acid, 107, 108, 128, 130, 132, 143, 144, Electron-collisional processes, 66
148, 149, 151, 154, 155, 157, 158, Enantiomer, 95–97
164–166, 169–172, 174 Energy approach, 57, 60, 61, 66, 229, 231, 236,
BSSE, 44, 50, 194, 287, 288, 299, 300 237
Enhancement factors, 243, 244, 255, 257–259
C Evolution, 381–383, 391, 395, 401
Casimir force, 373, 374, 377 Explicitly-correlated atomic wave functions,
Catalan numbers, 376, 378 29–33
Chemical bonding rule, 204–206, 208, 210 Explosives, 44

© Springer International Publishing AG, part of Springer Nature 2018 405

Y. A. Wang et al. (eds.), Concepts, Methods and Applications of Quantum Systems
in Chemistry and Physics, Progress in Theoretical Chemistry and Physics 31,
https://doi.org/10.1007/978-3-319-74582-4
406 Index

Exponential correlation, 29, 32 N

Naphthyl-isoquinoline alkaloids, 306, 307,
F 318, 322, 325
Fine-structure constant, 361, 367–369, 371, Nitrate-nitric acid-water cluster, 194
375, 377 Noble-gas atoms, 243, 245, 249, 255, 257, 259
Functional derivative, 332, 335, 343, 346 Nuclear and radiative corrections, 71, 72
Functional differentiability, 331, 332 Nuclear forces, 368, 376

G P
General relativity, 370–372, 381, 385, 389 Parity violation, 96
Geometric phase, 265, 266, 268–270, 274, 275 Pionic atomic systems, 71, 72, 75
Gravitational invariant, 361, 371, 372, 377, 378 Planck units, 368, 371, 373, 377
Gyromagnetic factor, 365, 367, 369, 378 Polyborate, 108, 112, 120, 128, 144, 148
Population analysis, 112, 194
H Positron annihilation, 243–246, 248, 251–253,
Helium, 203, 204, 207–210, 253 255, 260
Homochirality, 95, 96 Proton transfer, 170, 179–182, 186, 188, 190,
Hydrogen, 30, 76, 82, 83, 100, 101, 108, 128, 191, 198, 200
130, 132, 135, 144, 151, 155–158, 164,
165, 169–171, 173, 190, 191, 193, 195, Q
198, 199, 201, 204, 206–210, 253, 255, Quantum chemistry, 3–8, 16, 72, 260, 383, 402
256, 265, 267, 268, 281, 289, 297, 305, Quantum electrodynamics, 362, 370
307, 309, 368, 369, 372, 377 Quantum theory of atoms in molecules, 46
Hydrogen bonding, 156, 164, 193, 201, 283,
309 R
Hylleraas-CI, 29–31, 33, 39 Radiation decay processes, 229
Hylleraas functions, 29, 30, 33 Relativistic energy approach, 57, 59, 66, 79,
231, 232
I Relativistic perturbation theory, 231, 234
Intramolecular hydrogen bond, 283, 305 Relativistic quantum chemistry, 72
Ionization potential, 65, 213, 224, 230, 247, Rhodium clusters, 213, 224
348 ROHF calculations, 223

J S
Jozimine A2, 305–307, 310–313, 316, 318, Semantic web, 3, 4, 7, 8, 18, 20, 26
320, 322, 324, 325 Solute-solvent interactions, 283
Space-time, 371, 372, 381, 387–389, 391, 400
K Spectroscopy of exotic atoms, 72
Kinetic-energy integrals, 29, 33, 39 Spin momentum, 98, 366
Stochastic communication, 382, 383, 402
L Structural indicator, 43, 44, 50, 53
Light velocity, 235, 365
T
M Transition metal, 215, 266
Magnetic moment, 88, 213, 222, 361, 364, 377 Trigger bond, 44–46, 50–53
Many-body theory, 243, 244, 259, 260 Trigger Bond Indicators (TBIs), 51, 52
Matter antimatter, 369
Molecular orbital, 119, 149, 204, 213, 215, 306 U
Multicharged ions, 55, 56, 66, 79 Ultracold chemistry, 265, 267
Multielectron atoms, 231 Ultracold collisions, 265
Mutual orientation, 283, 284, 288, 305, 308, Ultracold molecules, 267, 274
312, 314, 318
Index 407

V Web 3.0, 3
Vector spherical harmonics, 34, 35 Web of data, 3
Web of meaning, 3
W
Wave beat, 366, 377 Z
Weak bond, 43, 50–53 Zitterbewegung, 365–367, 371
Weakest bond, 43, 44, 50, 51, 53