Paulo Cesar Philippi

Thermodynamics
From Fundamentals to Multiphase
and Multicomponent Systems
Thermodynamics
Paulo Cesar Philippi

Thermodynamics
From Fundamentals to Multiphase
and Multicomponent Systems
Paulo Cesar Philippi
Mechanical Engineering Graduate Program
Pontifical Catholic University of Paraná
Curitiba, Paraná, Brazil

ISBN 978-3-031-49356-0 ISBN 978-3-031-49357-7 (eBook)

https://doi.org/10.1007/978-3-031-49357-7

Jointly published with Associação Paranaense de Cultura

The print edition is not for sale in China (Mainland). Customers from China (Mainland) please order the
print book from: Associação Paranaense de Cultura.

© Associação Paranaense de Cultura 2024

This work is subject to copyright. All rights are solely and exclusively licensed 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 publishers, 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 publishers nor the authors or
the editors give a warranty, expressed or implied, with respect to the material contained herein or for any
errors or omissions that may have been made. The publishers remain neutral with regard to jurisdictional
claims in published maps and institutional affiliations.

This Springer imprint is published by the registered company Springer Nature Switzerland AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland

Paper in this product is recyclable.

To Raquel
Foreword

This new book on thermodynamics, Thermodynamics: From Fundamentals to Multi-

phase and Multicomponent Systems came, originally, from a graduate/undergraduate
education which Paulo spoke to me during a stay at the Pontifical Catholic University
of Paraná in Curitiba in 2018.
But this book goes far beyond classic academic lessons!
What challenges first of all is a very high-quality iconography, with historical
images and many educational illustrations to implement the two fundamental prin-
ciples. This iconography also addresses thermodynamic systems in non-equilibrium
conditions, with the result of numerical simulations carried out by the author and his
colleagues. The book presents a wide variety of approaches with, on the one hand,
the formal aspects such as the introduction of the Legendre transform to specify the
different thermodynamic potentials and on the other hand the more physical aspects
with for example the Marangoni effect, going even toward chemistry with the notion
of affinity between species. A great variety of exercises stimulates the student to
consolidate and mature his learning.
The book is divided into three main parts: The Principles of Thermodynamics,
Thermodynamic Equilibrium, and an Introduction to Non-equilibrium Thermody-
namics. Hence a succession of chapters, from the First Principle of Thermodynamics
to Multicomponent Systems in non-equilibrium conditions, the book ends with a
chapter dedicated to the kinetic standpoint of non-equilibrium thermodynamics.
Of course, this mass of knowledge requires exposure time and Paulo Philippi’s
work includes nearly 350 ages and offers around 200 bibliographical references.
In summary, the book Thermodynamics: From Fundamentals to Multiphase and
Multicomponent Systems is a reference work for students of course, but also for
academic scholars and professionals, in the long tradition of books from Maxwell,
Callen, Huang, or Landau and Lifchitz.
Good reading!

August 2023 Francois Dubois

Conservatoire National des Arts et Métiers
Paris, France

vii
Preface

This book on the fundamentals of thermodynamics was written intended to be a

textbook for advanced undergraduate and graduate students, who are interested
in a more in-depth understanding of the influence of irreversibility on thermody-
namic processes between equilibrium states and to acquire a better knowledge of
the physical mechanisms underlying the macroscopic behavior of multiphase and
multicomponent systems in equilibrium and non-equilibrium states.
Part I deals with the laws of thermodynamics as developed by outstanding
nineteenth-century researchers and philosophers such as Benoît Clapeyron, Sadi
Carnot, James Prescott Joule, Rudolf Clausius, Willian Thomson, Hermann von
Helmholtz, and Max Planck. The first principle of thermodynamics was formulated
based on the idea that “heat is motion,” i.e., that “heat is nothing else than the mechan-
ical energy of the particles that form a body,” an atomistic concept that, at that time,
was not universally accepted by the scientific community.
The second principle is based on the idea that “nature has an arrow” or that in all
natural phenomena some “vis viva” is being converted into heat, and not the inverse.
This idea was first pictured by Carnot in his 1824 book Réflexions sur la puissance
motrice du feu and, as admitted by Willian Thomson, Lord Kelvin, in 1848, was the
main difficulty to be solved for the acceptance of Joule’s proposal of the equivalence
between heat and work. Carnot’s book is a chef d’oeuvre of scientific thought and had
a large influence on the scientific community in the first half of the nineteenth century.
In this book, Carnot bases his approach on the Lavoisier theory on the conservation
of the caloric and the caloric theory is more akin to the dissipative nature of the
physical processes than Joule´s equivalence principle.
It is important, for young students—supposed to be the actors who will be asked to
overcome new challenges in the future—to understand how all these old challenges
were overcome in the nineteenth century, giving rise to the recognition of the principle
of energy conservation and the birth of thermodynamics as a new science. It is also
important for their education that proper reference be made for the authors of the
knowledge they are acquiring and to be conscious of the amount of effort that was
employed to construct this knowledge.

ix
x Preface

Most of the development of thermodynamics in the nineteenth century was moti-

vated by the eighteenth-century steam atmospheric engines. The challenge was to
find the key to increasing the conversion efficiency of these power machines dissemi-
nated throughout Britain and being responsible for a true revolution in manufacturing
and transportation.
Nowadays—to mention only a few engineering applications—thermodynamics is
an invaluable tool for the design of power plants, refrigerating systems, heat pumps,
solar and photovoltaic plants, fuel cells and capillary pumps, and the heat pipes that
are, presently, wildly employed in computers.
Nevertheless, besides being born to solve engineering problems, thermodynamics
is, also, a unified science that tries to understand nature and the ubiquitous processes
around us, being fundamental to the scientific thought in several branches of human
knowledge such as geology, atmospheric science, chemistry, biomedical, and health
sciences.
Part I of this book has two chapters.
Chapter 1 deals with the first law and begins with a brief history of the science
of heat. Clapeyron’s equation is not only a suitable equation of state for elastic (or
ideal) gases—as frequently understood by students—but has a 200 years history (if
we start from the first Galileu’s experiments with the thermoscope) and means the
recognition of the absolute temperature and the first indication of the existence of an
unbridgeable absolute zero, firstly pictured by Gay-Lussac.
Chapter 2 is dedicated to the second law and begins with the meaning of irre-
versibility.
 The work delivered in irreversible expansions of gases cannot be written
as Pd V . In fact, it does not make any sense to talk about a single pressure for a
non-equilibrium thermodynamic system. Nevertheless, in certain simple situations,
the effective work can be computed and compared with the work that would be deliv-
ered in a reversible expansion. In this chapter, I also try to clarify the fundamental
role of James Joule in demonstrating the principle of energy conservation. Some
academic exercises are proposed at the end of each part.
The second law received different formulations by Clausius, Kelvin, and Planck.
Demonstrating the equivalence between different statements is important because
this equivalence shows that the underlying principle behind these statements is the
same. Indeed, all the formulations of the second law state that energy flows in one
direction, “from vis viva to heat.” In other words, dissipation is a common attribute
of all natural phenomena.
Classical mechanics, the groundwork of the mechanical theory that thermody-
namicists advocate for heat is, however, reversible. In effect, the previous states of
a given mechanical system of particles can always be retrieved when the current
state of this system is known. So, how do conciliate a reversible science—clas-
sical mechanics—with a science that has irreversibility as its one of its underlying
principles? The answer was given by Ludwig Boltzmann in the second half of the
nineteenth century and a brief summary of this outstanding development for human
knowledge is given at the end of Chap. 3.
Preface xi

Part II of this book is dedicated to the concept of thermodynamic equilibrium.

Chapter 3 deals with homogeneity, Legendre transforms (leading to the Helmholtz
and Gibbs energies and the concept of enthalpy) followed by Maxwell relations and
minimum principles and establishing the framework that is needed for the theoret-
ical analysis of thermodynamic systems in equilibrium. A thermodynamic system
in equilibrium is homogeneous, meaning that its intensive properties do not vary
from point to point inside the system. This property leads to an Euler’s relationship
between its extensive variables and the Gibbs–Duhem equation. This equation found,
simultaneously, by Josiah Willard Gibbs and Pierre Maurice Marie Duhem enables
a Lagrangian description of thermodynamic systems, of crucial importance in the
analysis of open systems in non-equilibrium states. Minimum principles state that
when a restriction that prevents a system from reaching equilibrium is removed, this
system will attain an equilibrium state given by the minimization of one of its ener-
gies (internal, Helmholtz, Gibbs, or enthalpy), in accordance with the thermodynamic
variables that are maintained constant during the equilibration process.
In Chap. 4, we begin by following the efforts of Maxwell to show that the attractive
forces between molecules were responsible for vapor condensation, at temperatures
below the critical temperature found by Andrews for CO2 . In his 1875 paper “On the
dynamical evidence of the molecular constitution of bodies,” Maxwell recognizes the
impact of the 1873 Doctoral Thesis of Johannes Diderik van der Waals for clearing
the role of the interaction of real molecules on phase transition and does not omit
to mention the important work of Willard Gibbs who, in Maxwell’s words, “has
given us a remarkably simple and thoroughly satisfactory method of representing
the relations of the different states of matter by means of a model.”
Chapter 5 is dedicated to multicomponent systems. In this kind of thermody-
namic system, there is a competition between the attraction forces between identical
molecules and the forces between molecules belonging to different species. When
the thermodynamic system is such that the former forces are dominant the mixture
will segregate leading to a system with distinct phases. When the latter forces are
dominant, the different components form a non-ideal homogeneous system, which is
the subject of this chapter. Equilibrium calculations are based on the Unifac method,
a semiempirical method for the prediction of non-electrolyte activity in non-ideal
mixtures that use the functional groups, present on the molecules that make up the
mixture to calculate the activity coefficients. The Unifac method was first published
in 1975 by Fredenslund, Jones, and Prausnitz, from the University of California.
In Chapter 6, we deal with fluid interfaces trying to give a deeper insight into
the molecular mechanisms that are responsible for the interface tension and for the
adhesion of liquids on solid surfaces.
In Part III, we give an introduction to non-equilibrium thermodynamics.
We start in Chapter 7 with the continuum mechanics hypothesis of local equi-
librium, when each point of the space is considered to have a characteristic length
much greater than the molecular dimensions and, at the same time, much smaller
than the characteristic length of the whole system. So, each point can be thought
of as an elementary system in local thermodynamic equilibrium, although the local
thermodynamic properties at this point differ from the ones at the neighboring points.
xii Preface

This local equilibrium hypothesis is the key to finding the mass conservation law and
the balance equations for the linear momentum and internal energy in their differ-
ential forms, enabling to write a closed set of differential equations for the density,
velocity, and temperature fields in non-equilibrium systems. This chapter ends with
the balance equation for entropy. When we consider a material volume, the rate with
which the entropy varies with time is dependent on the heat flow through its boundary
and on the internal production of entropy P(s), due to internal non-equilibrium and
related to the Clausius non-compensated heat. By the second law, P(s) cannot be
negative, P(s) ≥ 0, and is always positive in non-equilibrium systems.
In Chap. 8, we study single-component two-phase systems where two approaches
are presented: the singular and the diffuse interface approaches. In the singular inter-
face approach, the transport equations are solved in each phase and the interface is
considered as a boundary, where the field variables from the two phases are linked.
In the second approach, the interface is considered to be diffuse, with a continuous
variation of the thermodynamic properties from one phase to another. In the diffuse
interface approach, the law of increasing entropy, P(s) ≥ 0, is used to find the
influence of surface forces on the pressure tensor and heat flux.
The source of entropy is written as a product of fluxes and forces, and this form
is used throughout Chap. 9 as a method for obtaining a closed system of differen-
tial equations for non-equilibrium thermodynamic systems, especially for systems
involving non-ideal fluids and their mixtures. For multicomponent systems, this is the
key to relating the macroscopic properties of the system with the molecular sources
of segregation. Indeed, miscibility is not only an intrinsic property of a pair of liquids
but is dependent on the thermodynamic state of their mixture. Two liquids that can
mix in all proportions forming homogeneous solutions at a given temperature can
partially segregate when the temperature is decreased. The amount of solute that
remains in the solution at this lower temperature is a measure of the solubility, and
this property is dependent on the temperature and pressure. In this chapter, the phys-
ical underlying processes that are at the origin of mixing and segregation in non-ideal
mixtures of partially miscible fluids are investigated.
The book ends with a chapter dedicated to exposing the kinetic standpoint of non-
equilibrium thermodynamics. The kinetic theory is the branch of science to which
the mechanical theory of heat of Clausius, Maxwell, and Boltzmann has evolved
after a system of particles ruled by deterministic mechanics was considered from a
probabilistic point of view. Furthermore, the complex behavior of non-ideal fluids
and solutions requires downscaling to the molecular scale. The Boltzmann equation is
nevertheless restricted to rarefied gases, and this restriction requires the introduction
of kinetic models that, in the reverse sense, make use of the available information
in the macroscopic scale. Kinetic models for the Boltzmann collision term and for
non-ideal fluids and their mixtures are presented in this chapter.
The numerical simulation of kinetic models for thermodynamic systems in non-
equilibrium conditions has had a rapid growth after 1990, and most of the simulation
examples presented throughout the book were performed with discrete forms of
kinetic models in the lattice Boltzmann framework.
Preface xiii

I wish to give my thanks to Pablo Javier Blanco from the National Labora-
tory of Scientific Computation (LNCC) for his careful revision of the text and to
François Dubois from the Conservatoire National des Arts et Métiers de Paris for
his stimulating words in the foreword. Most of the material presented in this book
resulted from my almost 30 years of research work and my thermodynamics grad-
uate/undergraduate courses and I thank my colleagues and my former graduate and
doctoral students with whom I had, during these years, many fruitful discussions and
a very friendly relationship.

Curitiba, Brazil Paulo Cesar Philippi

September 2023
Contents

Part I The Principles of Thermodynamics

1 The First Principle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.1 The Laws of Gases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.1.1 Pressure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.1.2 Temperature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.2 From Stahl to Lavoisier . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.2.1 Gay-Lussac and Clapeyron . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.3 The Steam Engines and Carnot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.4 The Principle of Energy Conservation . . . . . . . . . . . . . . . . . . . . . . . 15
1.4.1 Joseph Black . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
1.4.2 Joule’s Experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
1.4.3 Exercise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
1.4.4 The Principle of Energy Conservation . . . . . . . . . . . . . . . 22
1.4.5 The Joule Expansion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
1.4.6 Specific or Molar Heats . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
1.4.7 Ideal Gases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
1.5 Equilibrium and Non-equilibrium States . . . . . . . . . . . . . . . . . . . . . 27
2 The Second Principle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.1 Reversible and Irreversible Processes . . . . . . . . . . . . . . . . . . . . . . . . 29
2.1.1 An Abrupt Isothermal Expansion . . . . . . . . . . . . . . . . . . . 29
2.1.2 Expansion Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
2.1.3 Quasi-static Isothermal Expansion . . . . . . . . . . . . . . . . . . 34
2.1.4 Inverse Process. The Cycle . . . . . . . . . . . . . . . . . . . . . . . . . 36
2.1.5 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
2.1.6 Work and Heat Conversion. The Role
of Dissipation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
2.1.7 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
2.2 The Second Principle of Thermodynamics . . . . . . . . . . . . . . . . . . . 47
2.2.1 Statements of the Second Principle . . . . . . . . . . . . . . . . . . 50
2.2.2 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54

xv
xvi Contents

2.3 Sadi Carnot and Émile Clapeyron . . . . . . . . . . . . . . . . . . . . . . . . . . . 55

2.3.1 Thermodynamic Scale of Temperature . . . . . . . . . . . . . . . 59
2.3.2 Using the Ideal Gas Properties for Calculating
the Conversion Efficiency for the Carnot’s Cycle . . . . . . 61
2.3.3 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
2.4 Reversible Cycles Other Than Carnot’s Cycle . . . . . . . . . . . . . . . . 64
2.4.1 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
2.5 Clausius Inequality and Entropy . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
2.5.1 External and Internal Sources of Entropy . . . . . . . . . . . . . 72
2.5.2 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
2.6 The Meaning of Entropy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79

Part II Thermodynamic Equilibrium

3 The Equilibrium State . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
3.2 Open Systems and Gibbs Potential . . . . . . . . . . . . . . . . . . . . . . . . . . 90
3.3 Intensive and Extensive Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
3.4 Homogeneity of the Equilibrium State . . . . . . . . . . . . . . . . . . . . . . . 92
3.4.1 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
3.5 Legendre Transforms of the Internal Energy . . . . . . . . . . . . . . . . . . 95
3.5.1 Helmholtz Energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
3.5.2 Enthalpy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
3.5.3 Gibbs Energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
3.5.4 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
3.6 Maxwell Relations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
3.6.1 The Entropy as a Function of the Temperature
and Volume . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
3.6.2 The Internal Energy as a Function
of the Temperature and Volume . . . . . . . . . . . . . . . . . . . . . 108
3.6.3 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
3.7 Minimum Principles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
3.7.1 Minimum Principle for the Internal Energy . . . . . . . . . . . 112
3.7.2 Minimum Principle for Enthalpy . . . . . . . . . . . . . . . . . . . . 114
3.7.3 Minimum Principle for the Helmholtz Energy . . . . . . . . . 115
3.7.4 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
4 Phase Equilibrium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
4.1 Andrews, van der Waals and Maxwell . . . . . . . . . . . . . . . . . . . . . . . 121
4.1.1 Exercise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
4.2 Gibbs Potential to Find the Transition Pressure . . . . . . . . . . . . . . . 126
4.2.1 Maxwell Area Rule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
4.2.2 Exercise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
Contents xvii

4.3 The Transition State . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130

4.3.1 Exercise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
4.4 The Liquid-Vapor Phase Diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
4.5 Clausius-Clapeyron Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
4.5.1 Exercise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
4.6 The Law of Corresponding States . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
4.7 Acentric Factor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
4.7.1 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
4.8 Some Popular Cubic Equations of State . . . . . . . . . . . . . . . . . . . . . 140
4.8.1 Redlich-Kwong . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
4.8.2 Soave . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
4.8.3 Peng-Robinson . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
4.8.4 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
4.9 The Virial Equation of State . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
4.9.1 Second Virial Coefficient . . . . . . . . . . . . . . . . . . . . . . . . . . 143
4.9.2 Third Virial Coefficient . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
4.9.3 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
4.10 Calculation of State Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146
4.10.1 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
5 Non-ideal Mixtures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
5.1 Partial Volumes and the Amagat’s Law . . . . . . . . . . . . . . . . . . . . . . 151
5.1.1 Exercise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
5.2 Gibbs–Duhem Equation for Mixtures . . . . . . . . . . . . . . . . . . . . . . . 156
5.3 Ideal Gases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
5.3.1 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
5.4 The Lewis Concept of Fugacity . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
5.4.1 Calculation of the Fugacity Coefficient for a Virial
Equation of State . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164
5.4.2 Calculation of the Fugacity Coefficient for a Cubic
Equation of State . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
5.4.3 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
5.5 Ideal Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
5.5.1 Raoult’s Law . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
5.5.2 Exercise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
5.6 Non-ideal Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
5.6.1 Gibbs Energy in Excess . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
5.6.2 Margules Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175
5.6.3 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176
5.7 Activity Coefficients from Molecular Groups . . . . . . . . . . . . . . . . . 177
5.7.1 Combinatorial Activity Coefficient . . . . . . . . . . . . . . . . . . 178
5.7.2 Residual Activity Coefficient . . . . . . . . . . . . . . . . . . . . . . . 180
5.7.3 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188
xviii Contents

5.8 Phase Equilibrium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188

5.8.1 Phase Equilibrium at Low Pressures . . . . . . . . . . . . . . . . . 190
5.8.2 Phase-Equilibrium Using the Virial Equation
of State for the Vapor Phase . . . . . . . . . . . . . . . . . . . . . . . . 193
5.8.3 Phase Equilibrium Using a Cubic Equation
of State for the Vapor Phase . . . . . . . . . . . . . . . . . . . . . . . . 195
5.9 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200
6 Surface Physics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203
6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203
6.2 Laplace . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206
6.2.1 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 208
6.3 The Form of a Sessile Drop . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
6.4 Surface Tension and the Helmholtz Energy . . . . . . . . . . . . . . . . . . . 213
6.4.1 Exercise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215
6.5 A Deeper Insight into the Physics of Interfaces . . . . . . . . . . . . . . . 215
6.5.1 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
6.5.2 Density Profile . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
6.5.3 Interface Thickness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
6.5.4 Surface Tension . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222
6.5.5 Sample Case: Water . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
6.5.6 Exercise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 226
6.6 Interaction Between Fluids and Solid Surfaces . . . . . . . . . . . . . . . . 226
6.6.1 Exercise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 228
6.6.2 Dynamic Contact Angle . . . . . . . . . . . . . . . . . . . . . . . . . . . 229
6.7 Surface Energies: Cahn’s Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . 233
6.7.1 Hydrophilic Surfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237
6.7.2 Hydrophobic Surfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 240
6.7.3 Contact Angle and Adhesion Energy . . . . . . . . . . . . . . . . 240
6.7.4 Exercise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243
6.8 Emulsions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243

Part III Introduction to Non-equilibrium Thermodynamics

7 Non-equilibrium States . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 249
7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 249
7.2 The Local Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 251
7.3 Mass Conservation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 253
7.3.1 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 255
7.4 Momentum Balance Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 256
7.4.1 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 257
7.5 Energy Conservation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 258
7.6 The Heated Cavity Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 263
7.7 The Law of Increasing Entropy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 270
7.8 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 274
Contents xix

8 Multiphase Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277

8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277
8.2 The Singular Interface Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . 279
8.2.1 Mass Conservation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279
8.2.2 Momentum Balance Equation . . . . . . . . . . . . . . . . . . . . . . 282
8.2.3 Particular Cases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 285
8.2.4 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 286
8.2.5 Numerical Methods for the Singular Interface
Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 287
8.3 The Diffuse Interface Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . 287
8.3.1 Surface Forces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 287
8.3.2 Momentum Balance Equation . . . . . . . . . . . . . . . . . . . . . . 293
8.3.3 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 293
8.3.4 Exercise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 294
8.3.5 Internal Energy Balance Equation . . . . . . . . . . . . . . . . . . . 295
8.3.6 Numerical Methods for the Diffuse Interface
Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 297
9 Multicomponent Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 301
9.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 301
9.2 Mixtures Without Segregation Effects . . . . . . . . . . . . . . . . . . . . . . . 303
9.2.1 Transport Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 308
9.2.2 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 309
9.3 Segregation Effects in Non-ideal Mixtures . . . . . . . . . . . . . . . . . . . 309
9.3.1 Pressure Tensor and the Equation of State . . . . . . . . . . . . 315
9.3.2 Interface Physics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 319
9.3.3 Spinodal Decomposition in Multicomponent
Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 322
10 Non-equilibrium Thermodynamics from a Kinetic Standpoint . . . . 325
10.1 Scales of Investigation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 325
10.2 The Boltzmann Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 327
10.2.1 The Liouville Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 328
10.2.2 The Long-Range Term. Mean Field Theory . . . . . . . . . . . 330
10.2.3 The Short-Range Collision Term . . . . . . . . . . . . . . . . . . . . 331
10.2.4 Elementary Properties of the Collision Term . . . . . . . . . . 335
10.2.5 The H-Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 337
10.2.6 Macroscopic Transport Equations . . . . . . . . . . . . . . . . . . . 341
10.2.7 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 342
10.3 Kinetic Models for the Collision Term . . . . . . . . . . . . . . . . . . . . . . . 345
10.3.1 The BGK Collision Model . . . . . . . . . . . . . . . . . . . . . . . . . 346
10.3.2 Beyond BGK . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 347
10.3.3 The Stokes Hypothesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . 352
10.3.4 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 353
xx Contents

10.4 Non-ideal Fluids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 354

10.4.1 Two Phase Liquid-Vapor Systems . . . . . . . . . . . . . . . . . . . 356
10.4.2 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 360
10.5 Multicomponent Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 360
10.5.1 Macroscopic Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 363
10.5.2 Using a Reverse Standpoint to Find
the Microscopic Parameters and the Intermolecular
Force . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 365
10.5.3 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 366

Appendix: The Euler–Lagrange Equations . . . . . . . . . . . . . . . . . . . . . . . . . . 369

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 372
List of Figures

Fig. 1.1 Measurement of the atmospheric pressure by Evangelista

Torricelli . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
Fig. 1.2 Pascal’s drawing showing the law of pressure for liquids . . . . . . 5
Fig. 1.3 Front page of the book New experiments
Physico-Mechanical touching the spring of the air and its
effects, published by Boyle in 1660 . . . . . . . . . . . . . . . . . . . . . . . . 6
Fig. 1.4 A replica of Galileo’s thermoscope exposed in Galileo’s
museum (Florence) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
Fig. 1.5 a Gay-Lussac experiment for measuring the dilation
of gases with the temperature. b Results found for the molar
volume dependence on the temperature at atmospheric
pressure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
Fig. 1.6 Clapeyron’s equation published at page 164 of the Mémoire
sur la Puissance Motrice de la Chaleur . . . . . . . . . . . . . . . . . . . . 12
Fig. 1.7 Newcomen steam engine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
Fig. 1.8 First stage of the James Watt steam engine . . . . . . . . . . . . . . . . . . 13
Fig. 1.9 Second stage of the James Watt steam engine . . . . . . . . . . . . . . . . 14
Fig. 1.10 Joule apparatus for measuring the mechanical equivalent
of heat . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
Fig. 1.11 Joule’s experiment on gas expansion . . . . . . . . . . . . . . . . . . . . . . . 24
Fig. 1.12 Ice calorimeter created by Lavoisier and Laplace based
on Black’s idea of latent heat . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
Fig. 1.13 Sketch for the specific or molar heat capacity measurement
a at constant volume and b at constant pressure . . . . . . . . . . . . . . 25
Fig. 2.1 Isothermal expansion of an ideal gas . . . . . . . . . . . . . . . . . . . . . . . 30
Fig. 2.2 Isotherms in an intermediate time of the abrupt expansion . . . . . 31
Fig. 2.3 Time evolution of the external and internal pressure just
above and below the base plate of the piston . . . . . . . . . . . . . . . . 32
Fig. 2.4 Quasi-static isothermal expansion . . . . . . . . . . . . . . . . . . . . . . . . . 35

xxi
xxii List of Figures

Fig. 2.5 Work delivered by the isothermal gas expansion

when a certain number of plumb spheres with mass m
are removed, one after the other, from the top of the piston
in Fig. 2.4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
Fig. 2.6 Compression work when sequentially adding plumb
spheres of mass m to the top of the piston . . . . . . . . . . . . . . . . . 37
Fig. 2.7 P-V diagram of the entire cycle 1-2-1. For the abrupt
process, only the states 1 and 2 are equilibrium states.
For the quasi-static process, when m → 0, in addition
to these terminal states, all the intermediate states are
equilibrium states . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
Fig. 2.8 Isovolumetric heating . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
Fig. 2.9 Cloud cells formation by Rayleigh-Bénard convection
over the Pacific Ocean. Each cell has 10–15 km
characteristic length (from https://earthobservatory.nasa.
gov/images/2387/closed-small-cell-clouds-in-the-south-
pacific). Image courtesy NASA/GSFC/LaRC/JPL, MISR
Team . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
Fig. 2.10 Isothermal abrupt compression . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
Fig. 2.11 Energy balance for the isothermal expansion of section . . . . . . . . 41
Fig. 2.12 The role of dissipation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
Fig. 2.13 An irreversible adiabatic expansion . . . . . . . . . . . . . . . . . . . . . . . . 43
Fig. 2.14 Adiabatic expansion produced by the removal of balance
weights of mass m . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
Fig. 2.15 Using the first principle of thermodynamics to show
that the work W1−3 delivered in a reversible adiabatic
expansion is greater (in absolute value) than the work W1−2 . . . . 45
Fig. 2.16 A Rayleigh-Bénard cellular convection heats the Helium
in the chamber B. Nevertheless, the adiabatic compression
of the Helium in the chamber A is a very slow process
and supposing this compression to be reversible can be
a fair hypothesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
Fig. 2.17 Using the first principle of thermodynamics to show
that the work W1−3 delivered in a reversible adiabatic
compression is smaller (in absolute value) than the work
W1−2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
Fig. 2.18 A thermodynamic system composed of two chambers
A and B in thermal contact with a thermal reservoir . . . . . . . . . . 47
Fig. 2.19 Papin steam engine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
Fig. 2.20 A simplified scheme for a thermal engine . . . . . . . . . . . . . . . . . . . 48
List of Figures xxiii

Fig. 2.21 Demonstration that it is impossible to conceive a heat

engine that entirely converts heat into mechanical work.
In a it is supposed that a thermal engine with η = 1 can
be used to drive a heat pump operating between the same
thermal reservoirs. In b an energy balance is performed
for the composite system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
Fig. 2.22 Clausius principle in graphical form. Heat can pass
from a warmer to a colder body (a), but, in the words
of Clausius, cannot by itself pass from a colder to a warmer
body (b) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
Fig. 2.23 The natural arrow of heat . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
Fig. 2.24 Graphical formulation of the Planck statement . . . . . . . . . . . . . . . 52
Fig. 2.25 Kelvin statement of the second principle . . . . . . . . . . . . . . . . . . . . 53
Fig. 2.26 Caratheodory principle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
Fig. 2.27 Adiabatic and reversible lines do not intercept . . . . . . . . . . . . . . . 55
Fig. 2.28 Perpetual motion engine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
Fig. 2.29 a Front page of the Clapeyron 1834 article Mémoire sur la
puissance motrice de la chaleur; b graphical representation
by Clapeyron of the Carnot reversible cycle at page 190 . . . . . . . 57
Fig. 2.30 The Carnot cycle as presented by Clapeyron in its 1834
Mémoire for a power engine (a) and for a heat pump (b) . . . . . . 58
Fig. 2.31 If a non-reversible power engine is more efficient
than a Carnot reversible engine when both engines are
operating between the same heat reservoirs as in (a), then
the reversible engine can be reverted and pump the rejected
heat Q H to the non-reversible engine (b) . . . . . . . . . . . . . . . . . . . 59
Fig. 2.32 The thermodynamic scale of temperature . . . . . . . . . . . . . . . . . . . 60
Fig. 2.33 A Carnot cycle where the individual processes fail to be
reversible . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
Fig. 2.34 a An isobaric expansion of a gas is never reversible: the gas
is heated from below giving rise to Rayleigh-Bénard
convection cells due to the gas density decrease close
to the hot liquid bath. During the heating process,
the gas temperature is T2 close to the heat reservoir,
but smaller than T2 in the remaining space of the gas
chamber. b From a theoretical perspective, it is possible
to conceive an isobaric reversible expansion by considering
it as the limit of a sequence of isothermal expansions
and adiabatic compressions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
Fig. 2.35 a A reversible cycle (Brayton cycle) composed of two
isobaric and two adiabatic processes may be decomposed
in a set of Carnot cycles. b Each Carnot engine operates
in a temperature range that is a fraction of the entire
temperature range between TL and TH . . . . . . . . . . . . . . . . . . . . . 66
xxiv List of Figures

Fig. 2.36 Reversible processes between two arbitrary equilibrium

states . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
Fig. 2.37 Another reversible cycle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
Fig. 2.38 Clausius entropy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
Fig. 2.39 Irreversible cycle of an ideal gas in contact with a heat
reservoir: Heat and work balances for the expansion
and compression processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
Fig. 2.40 Isobaric expansion of an ideal gas in contact with a heat
reservoir . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
Fig. 2.41 Isochoric heating of an ideal gas in contact with a heat
reservoir . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
Fig. 2.42 Isothermal expansion against vacuum . . . . . . . . . . . . . . . . . . . . . . 80
Fig. 2.43 Front page of Bernoulli’s Hydrodynamica . . . . . . . . . . . . . . . . . . 81
Fig. 3.1 The heat and work that are exchanged are dependent
on mass m, but the state variables in equilibrium states 1
and 2 are always the same . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
Fig. 3.2 Fluxes and forces in a reversible process . . . . . . . . . . . . . . . . . . . . 89
Fig. 3.3 Open systems can change their internal energy by mass
transfer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
Fig. 3.4 Thermodynamic systems in equilibrium are homogeneous.
For a given partition, the extensive properties are divided
by the number of parts, whereas the intensive properties
remain the same throughout the whole system . . . . . . . . . . . . . . . 93
Fig. 3.5 Lagrangian description. Volume V(t) is a material volume
and although it is deformed along the flow, the mass
inside it is constant . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
Fig. 3.6 Given a function Y = Y(X), the Legendre transform
L = L(P) preserves all the information given
by the representation Y = Y(X) . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
Fig. 3.7 Energy balance for the air compressor of a turbojet . . . . . . . . . . . 100
Fig. 3.8 Considering that chambers A and B are in equilibrium
with a thermal bath at temperature T , show
that the Helmholtz energy f = F/N is the same in the two
chambers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
Fig. 3.9 Application of the Schwarz theorem in deriving differential
relationships for the thermodynamic properties . . . . . . . . . . . . . . 106
Fig. 3.10 When the internal energy U = const. the internal
non-equilibrium will increase the entropy . . . . . . . . . . . . . . . . . . . 111
Fig. 3.11 Isentropic expansion for showing the minimum principle
for the internal energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
Fig. 3.12 Minimum principle for enthalpy . . . . . . . . . . . . . . . . . . . . . . . . . . 114
Fig. 3.13 Minimum principle for the Helmholtz energy . . . . . . . . . . . . . . . . 116
Fig. 3.14 Minimum principle for the Gibbs energy . . . . . . . . . . . . . . . . . . . 118
Fig. 3.15 Osmotic pressure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
List of Figures xxv

Fig. 4.1 Andrews experimental results for carbon dioxide

at page 583 of Ref. [5] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
Fig. 4.2 a Van der Waals equation of state has 3 roots for each
pressure; b Maxwell’s area rule . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
Fig. 4.3 a Calculation of the transition pressure using a worksheet.
a P − v diagram corresponding to a single isotherm.
b Gibbs energy for different pressures . . . . . . . . . . . . . . . . . . . . . . 127
Fig. 4.4 Maxwell area rule derived from the equality of the Gibbs
potentials μ and μv . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
Fig. 4.5 During the phase transition at pressure Ps and temperature
T , the molar volume of vapor and liquid do not change
and are vvs and vs respectively . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
Fig. 4.6 Phase diagram for NH3 showing the isotherms of 360
K (blue), 390 K (black) and 400 K (green) modeled
with the Peng-Robinson equation of state. The acentric
factor was taken as ω = 0.25. Pressure transition
was found using the Maxwell area rule for each one
of these isotherms. It is seen that the molar volume
variation vv = vvs − vs between the saturated liquid
and saturated vapor decreases with the temperature
and reaches the zero value at the critical point. Above
the critical temperature, there is no phase transition.
For NH3 the critical temperature is Tc = 405.55K
and the critical pressure is Pc = 112.78 bar . . . . . . . . . . . . . . . . . 133
Fig. 4.7 The Clausius-Clapeyron equation . . . . . . . . . . . . . . . . . . . . . . . . . 134
Fig. 4.8 Compressibility factor in terms of the reduced pressure
and temperature for the van der Waals equation of state . . . . . . . 136
Fig. 4.9 Estimation of the acentric factor based on experimental
data for the saturation pressure . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138
Fig. 4.10 log10 Ps∗ in terms of the inverse of the reduced temperature
T ∗ for water . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
Fig. 4.11 PVT experimental data allows the determination
of the virial coefficients B and C for each isotherm.
Virial parameter B (T) is found from the interception
of the isotherm T with the vertical axis. Parameter C(T) is
given by the slope of the isotherm T when n = 1/v̄ → 0 . . . . . . 143
Fig. 4.12 Calculation of state variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146
Fig. 5.1 If oxygen in the atmospheric air is separated from nitrogen
it will occupy a volume of 21 L for every 100 L of air
at atmospheric pressure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
Fig. 5.2 A theoretical process where a mixture of ideal gases is
formed by adding r − 1 components to component i
without changing the temperature and total pressure . . . . . . . . . . 158
xxvi List of Figures

Fig. 5.3 Oxygen will flow from chamber B to chamber A because its
chemical potential in chamber B is greater than its
chemical potential in chamber A . . . . . . . . . . . . . . . . . . . . . . . . . . 161
Fig. 5.4 The concept of fugacity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
Fig. 5.5 Raoult’s law: a the liquid phase is considered as an ideal
solution and, the gas phase, a mixture of ideal gases; b in its
saturated condition the fugacity of a liquid is its saturation
pressure when the vapor is considered as an ideal gas . . . . . . . . . 170
Fig. 5.6 Raoult’s law: for a given pressure P the amount
of component A in the liquid solution will be given by its
molar fraction, x A , solution of Eq. (5.37) and the amount
of this component in the vapor phase by y A , solution of Eq.
(5.38) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
Fig. 5.7 Raoult’s law . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
Fig. 5.8 Volume and area group contributions for the combinatorial
activity coefficient. Reproduced from Table 8.23 of Ref.
[137] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
Fig. 5.9 The molecule of n-butane . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
Fig. 5.10 Normal-pentane and acetone molecules . . . . . . . . . . . . . . . . . . . . 181
Fig. 5.11 Interaction parameters amn in Kelvin. Reproduced
from Table 8.24 of Poling et al. [137] . . . . . . . . . . . . . . . . . . . . . . 182
Fig. 5.12 Method for calculating the gamma factors in the reference
solutions [137] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
Fig. 5.13 Water–methanol activity coefficients at 25 ◦ C in terms
of the molar fraction of methanol . . . . . . . . . . . . . . . . . . . . . . . . . . 191
Fig. 5.14 Water–methanol bubble and dew lines at 25 ◦ C calculated
from functional groups and compared with experimental
values [89] and with Raoult’s law . . . . . . . . . . . . . . . . . . . . . . . . . 192
Fig. 6.1 Figure 6 of Laplace’s supplement Sur l’action capillaire . . . . . . 207
Fig. 6.2 Derivation of Laplace’s law . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207
Fig. 6.3 Capillary rise inside a cylindrical tube . . . . . . . . . . . . . . . . . . . . . . 208
Fig. 6.4 A sessile drop and its meridional and transverse plans . . . . . . . . . 210
Fig. 6.5 Measurement of the interface tension by the pendant drop
method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212
Fig. 6.6 Helmholtz excess energy in the liquid–vapor interface,
calculated with Eq. (6.10) for different reduced
temperatures. The van der Waals equation of state was used
for the calculation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
Fig. 6.7 Figure shows the density profile n ∗ and the pressure
P ∗ − Ps∗ along the interface thickness . . . . . . . . . . . . . . . . . . . . . 221
Fig. 6.8 Definition of the interface thickness . . . . . . . . . . . . . . . . . . . . . . . 221
List of Figures xxvii

Fig. 6.9 Comparison of the continuum (PR) and molecular

dynamics (MD) approach with ellipsometry experimental
results for estimating the interface thickness of water.
MD simulation results are from Matsumoto and Kataoka
[112] and Taylor et al. [169]. Experimental results are
from Kinosita and Yokota [85] . . . . . . . . . . . . . . . . . . . . . . . . . . . . 225
Fig. 6.10 Contact angle between a fluid and a solid surface . . . . . . . . . . . . 226
Fig. 6.11 a Force balance on a surface element e of a liquid droplet.
b Potential energy related to the intermolecular interaction . . . . . 229
Fig. 6.12 Sketch of the variation of the dynamic contact angle
in accordance with the flow direction (a, b); dependence
of the contact angle with the capillary number
(c, d) streamlines showing the formation of vortices
near the triple line [43] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231
Fig. 6.13 First steps of the meniscus formation in capillary rise
between parallel plates simulated with a lattice-Boltzmann
model. h ∗ = 2R h
, x ∗ = 2R x
, t ∗ = tt0 are dimensionless
variables [187] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232
Fig. 6.14 Time evolution of the dynamic contact angle
during the capillary rise between two parallel plates
when the equilibrium contact angle is 44◦ [187] . . . . . . . . . . . . . . 233
Fig. 6.15 The electrostatic field produced by a solid surface induces
polarization on non-polar molecules. These polarized
molecules induce dipoles in the next layers until several
layers are built up . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234
Fig. 6.16 Density distribution at the fluid–solid interface:
a hydrophilic surface, φ > 0, b hydrophobic surface, φ < 0 . . . . 235
Fig. 6.17 Dimensionless excess function for water at T* = 0.8
calculated using the van der Waals EOS plotted against
the reduced density. The information on the solid–fluid
interaction is given by the dimensionless specific free
energy φ ∗ considered in this case as φ ∗ = 0.4 (hydrophilic
surface). In accordance with Cahn’s theory, the water
reduced density at the solid surface will be ρs∗ = 0.49337
and ρs∗ = 2.14093 corresponding, respectively, to the vapor
and liquid in equilibrium with the solid . . . . . . . . . . . . . . . . . . . . . 238
Fig. 6.18 Hydrophobic surface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 240
Fig. 6.19 Adhesion energy and contact angle in terms of parameter φ ∗ . . . 242
∗ ∗
Fig. 6.20 Vapor ρs−v and liquid ρs− densities at the solid surface
in terms of the equilibrium contact angle θ for water
at T ∗ = 0.8, when a van der Waals equation of state is used . . . . 242
Fig. 6.21 The molecule of cetyl alcohol CH3 (CH2 )15 OH . . . . . . . . . . . . . . 243
xxviii List of Figures

Fig. 6.22 Due to its amphiphilic nature, when surfactants

are added to an oil–water system their molecules
migrate to the interface. a When oil is fragmented
and dispersed in a continuous water phase the polar
heads of the emulsifier at the interface attract the water
molecules in their neighborhoods preventing the oil
drops from coalescing. b When water is dispersed in oil,
the long tails of the emulsifier prevent their polar heads
to attract the water molecules in the contiguous water drop,
because they are beyond the interaction length of these
intermolecular polar forces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244
Fig. 6.23 When a surfactant is added to an oil–water system,
the polar heads in the interface will create an additional
force from the surface that will reduce the force imbalance
responsible for the interface tension . . . . . . . . . . . . . . . . . . . . . . . 245
Fig. 6.24 Effect of the addition of cetyl alcohol on the interface
tension of a water–nujol system at 30 ◦ C. Experimental
results from Wolf [186] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 246
Fig. 7.1 Material or Lagrangian volume V (t) moving inside a fluid
has its volume and shape affected by pressure and viscous
forces, but the mass m is always the same . . . . . . . . . . . . . . . . . . . 254
Fig. 7.2 A two-dimensional channel flow . . . . . . . . . . . . . . . . . . . . . . . . . . 258
Fig. 7.3 Helmholtz in 1848 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259
Fig. 7.4 Pressure work and friction are responsible for transferring
the kinetic energy of the compressor blades to the air . . . . . . . . . 263
Fig. 7.5 The heated cavity problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 263
Fig. 7.6 a Temperature field and b streamlines during the heating
time of the cavity (tˆ = 0.008). Dimensionless parameters
are Pr = 1, Ra = 106 , Ec = 10−4 [73] . . . . . . . . . . . . . . . . . . . . . . 267
Fig. 7.7 a Temperature field and b streamlines at the dissipation
period tˆ = 0.026. Dimensionless parameters are Pr = 1,
Ra = 106 , Ec = 10−4 [73] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 268
Fig. 7.8 A constant heat flow q0 into a cavity filled with an ideal
gas increases its temperature and produces natural
convection currents. After heating is ceased these currents
are dissipated, and their kinetic energy is transformed
into internal energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269
Fig. 7.9 Dimensionless viscous dissipation ˆ as a function of time
[73] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269
Fig. 7.10 Isothermal expansion of an ideal gas . . . . . . . . . . . . . . . . . . . . . . . 273
Fig. 7.11 Channel flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 275
Fig. 7.12 Bénard cellular convection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 275
Fig. 8.1 The Plateau–Rayleigh instability . . . . . . . . . . . . . . . . . . . . . . . . . . 278
Fig. 8.2 The first stages of spinodal decomposition in liquid–vapor
systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279
List of Figures xxix

Fig. 8.3 The singular interface approach . . . . . . . . . . . . . . . . . . . . . . . . . . . 280

Fig. 8.4 The Rayleigh–Taylor instability . . . . . . . . . . . . . . . . . . . . . . . . . . . 286
Fig. 8.5 Excerpt of van der Waals doctoral thesis [181] . . . . . . . . . . . . . . . 289
Fig. 8.6 Coalescence between two liquid droplets in a vapor
atmosphere in different time steps δ [157] . . . . . . . . . . . . . . . . . . 299
Fig. 8.7 Vapor condensation in the absence of gravity simulated
with the lattice-Boltzmann method . . . . . . . . . . . . . . . . . . . . . . . . 300
Fig. 9.1 Dodecane-ethanol phase diagram (from Dortmund Data
Bank) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 302
Fig. 9.2 Lattice Boltzmann simulation of the spinodal
decomposition. The mixture forms a homogeneous
solution at the beginning of the simulation (a).
Intermolecular attraction forces between the molecules
of the same species segregate the two components forming
isolated islands of one fluid inside the other one (b),
till the equilibrium is reached in (c) . . . . . . . . . . . . . . . . . . . . . . . . 303
Fig. 9.3 Time evolution of the interface length [133] . . . . . . . . . . . . . . . . . 322
Fig. 9.4 When the temperature is below a critical value
for a homogeneous mixture of two components,
the segregating forces will lead to a two-phase system
by spinodal decomposition. Snapshots are related to
500 (a), 1000 (b), 5000 (c), and 10,000 (d) time steps
for a reduced temperature Tr = 0.74 [133] . . . . . . . . . . . . . . . . . . 323
Fig. 10.1 The kinetic scale as a conceptual bridge between molecular
physics and the macroscopic domain . . . . . . . . . . . . . . . . . . . . . . . 326
Fig. 10.2 Atraction-repulsion field around a single molecule . . . . . . . . . . . 327
Fig. 10.3 Bullet particle trajectory inside the σ -sphere of the target
particle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 333
Fig. 10.4 Elementary volume in cylindrical coordinates . . . . . . . . . . . . . . . 334
Fig. 10.5 Interchanging ξ and ξ 1 invert γ and changes the sign of α
but θ remains unchanged . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 336
Fig. 10.6 Attractive potential related to the interaction of all
molecules 2 with the single molecule 1 . . . . . . . . . . . . . . . . . . . . . 355
List of Tables

Table 2.1 Initial and final equilibrium states . . . . . . . . . . . . . . . . . . . . . . . . 30

Table 3.1 Differential equations for the thermodynamic functions . . . . . . . 107
Table 4.1 Sample worksheet for the calculation of thermodynamic
functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
Table 4.2 Acentric factors found by Pitzer . . . . . . . . . . . . . . . . . . . . . . . . . . 139
Table 4.3 Saturation pressure for Ethane . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
Table 5.1 Activity coefficients γ for a water–methanol solution
at T = 50 ◦ C . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
Table 5.2 Calculation of the geometric parameters r, q, and 
for the molecules of n-pentane and acetone . . . . . . . . . . . . . . . . . 181
Table 5.3 Combinatorial activity coefficient for n-pentane
and acetone, based on data given in Table 5.2,
when the molar fraction of acetone is xacetone = 0.25 . . . . . . . . . 181
Table 5.4 Functional groups in the acetone and n-pentane solution . . . . . . 182
Table 5.5 Peng–Robinson parameters, compressibility factor,
and fugacity coefficient for pure components methanol
and water in the saturated liquid state at 50 ◦ C . . . . . . . . . . . . . . 198
Table 6.1 Surface properties of water . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 224
Table 6.2 Surface properties of water with the van der Waals
equation of state . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 239
Table 6.3 Reduced densities at the solid surface, interface tensions
and equilibrium contact angle for several solid surface
parameters φ ∗ when T ∗ = 0.8 and the van der Waals
equation of state . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241
Table 8.1 Single and two-phase flows . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 294

xxxi
 Part I
The Principles of Thermodynamics
Chapter 1
The First Principle

Abstract The first principle of thermodynamics is considered to be one of the great-

est achievements of scientific thought when heat, or the internal energy of a thermo-
dynamic system, was recognized as a kind of mechanical energy. In this chapter, we
will try to understand the main reasons that led to this outstanding discovery. We start
with the main seventeenth and eighteenth century findings, from the Mariotte dis-
covery of atmospheric pressure to James Watt steam engines. While the seventeenth
century was the century of the discovery of vacuum by Pascal, the thermoscope of
Galileu, the gas thermometer of the Grand Duke of Tuscany, Ferdinand II, and the
discovery of the properties of elastic gases by Robert Boyle, the eighteenth cen-
tury saw the born of the steam engines and the industrial revolution they produced,
atmospheric gases were isolated and the Stahl phlogiston theory was overthrown
by Lavoisier’s mass conservation principle. At the end of the eighteenth century,
Lavoisier, controversially, also proposed the conservation of caloric principle, in
the reverse sense with the principle of energy conservation sustained by important
philosophers and further postulated by Helmholtz, based on the pioneer experimental
works of Benjamin Thompson, Joule, and Mayer.

1.1 The Laws of Gases

1.1.1 Pressure

The term ‘pressure’ as the normal force per unit area acting on a given surface in
contact with a fluid was introduced by Pascal. But, pressure is not only a force per
unit area acting on a given surface, but the result of the momentum variation of the
molecules that are reflected by the wall. Indeed, in accordance with Newton’s second
law, the force .F that a single molecule of mass .m exerts on a surface is the rate of
change of its momentum before and after the collision of this molecule with the
surface and during the time interval .Δt this molecule is in contact with the surface

mvin − mvout
. = F.
Δt
© Associação Paranaense de Cultura 2024 3
P. C. Philippi, Thermodynamics, https://doi.org/10.1007/978-3-031-49357-7_1
4 1 The First Principle

Fig. 1.1 Measurement of the atmospheric pressure by Evangelista Torricelli

Considering the molecules to be material points, this collision will be elastic, and
the tangential components of .vin and .vout will be identical in modulus and opposed.
So, the resulting force will be normal to the surface.
Pressure is not only a normal force but it is, also, isotropic, in the sense that, in a
given point.x, it has the same value independently of the surface normal direction with
which the fluid (a gas or a liquid) is in contact, at this point .x. Therefore the pressure
inside a spherical steel vessel filled with vapor is the result of the collisions of the
vapor molecules with the inner surface of the vessel shell and the liquid pressure
inside a pressure cooker is the pressure of the vapor above it.
In 1594 Galileo obtained the patent for a machine to pump water from a river for
the irrigation of land. Galileo found that 10 m was the limit to which the water would
rise in the suction pump, but had no explanation for this phenomenon. Knowing this
curious fact, Galileo encouraged his former pupil Evangelista Torricelli to investigate
this supposed limitation. Torricelli reasoned, that this limitation was the result of the
pressure exerted on the liquid by the surrounding air.
To prove this theory, he filled a long glass tube, sealed at one end, with mercury
and upended it into a dish also containing mercury. The column of mercury invariably
fell to about 760 mm, leaving an empty space above its level. Torricelli attributed the
cause of the phenomenon to a force on the surface of the earth, without knowing,
where it came from. He also concluded that the space on top of the tube is empty,
that nothing is in there, and called it ‘vacuum’ (Fig. 1.1).
From about 1645 Blaise Pascal (1623–1662) heard about the experiments of Tor-
ricelli and was searching for the reasons for Galileo’s and Torricelli’s findings. He
came to the conviction that the force, that keeps the column at 760 mm, is the weight
of the air above. Thus, on a mountain, the force must be reduced by the weight of
1.1 The Laws of Gases 5

Fig. 1.2 Pascal’s drawing

showing the law of pressure
for liquids

the air between the valley and the mountain. He predicted that the height of the col-
umn would decrease which he proved with his experiments at the mountain Puy de
Dome in central France. From the decrease, he could calculate the weight of the air.
Pascal also formulated that this force, he called it ‘pressure’, is acting uniformly in
all directions.1 Descartes visited Pascal on 23 September. His visit only lasted two
days and the two argued about the vacuum which Descartes did not believe in. In
August of 1648 Pascal observed that the pressure of the atmosphere decreases with
height and deduced that a vacuum existed above the atmosphere.2
From May 1653 Pascal worked on mathematics and physics and wrote the book
Traité de l’ équilibre des liqueurs3 in which he explains Pascal’s law of pressure
(Fig. 1.2). This treatise is a complete outline of a system of hydrostatics, the first in
the history of science, it embodies his most distinctive and important contribution to
physical theory.
Robert Boyle (1627–1691) was an alchemist, but he is most known by its important
work on elastic gases, showing the constancy of the product of the pressure by the
volume,

. P V = const.

published in his 1660 book4 (Fig. 1.3) and meaning that the volume of a fixed amount
of an elastic gas is inversely proportional to the pressure. This finding was also
discovered by Edme Mariotte (1620–1684) and published in his 1679 book.5

1 B. Pascal (1647). Expériences nouvelles touchant le vide. O’Euvres de Blaise Pascal, II. La bib-
liothèque libre.
2 B. Pascal (1648) Récit de la grande expérience de l’équilibre des liqueurs. Bibliothèque Nationale

de France. Available in http://gallica.bnf.fr/ark:/12148/bpt6k105083f/f1.image.

3 B. Pascal (1653) Traité de l’équilibre des liqueurs. A new edition with comments was produced

by Arvensa Éditions. Paris.

4 Boyle R (1660) New experiments Physico-Mechanical touching the spring of the air and its effects.

H. Hall, Oxford, [19].

5 E. Mariotte, Essais de Physique, ou mémoires pour servir á la science des choses naturelles. Paris:

E. Michallet, 1679 [111].

6 1 The First Principle

Fig. 1.3 Front page of the

book New experiments
Physico-Mechanical
touching the spring of the air
and its effects, published by
Boyle in 1660

1.1.2 Temperature

Galileo Galilei in the end of the sixteenth century was able to gauge the relative
“coldness” or “hotness” of air, using a rudimentary thermoscope (Fig. 1.4), but the
first thermometer was invented by Santorio Santorio in 1612. Santorio’s instrument
was an air thermometer. Its accuracy was poor as the effects of varying air pressure
on the thermometer were not understood at that time. The sealed liquid-in-glass
thermometer was first produced in 1654 by the Grand Duke of Tuscany, Ferdinand II
(1610–1670). His thermometer had an alcohol filling. Although this was a significant
development, his thermometer was inaccurate and there was no standardized scale
in use. Gabriel Fahrenheit (1686–1736) was the first person to make a thermometer
using mercury in 1714. The more predictable expansion of mercury combined with
improved glass working techniques led to a much more accurate thermometer. In 1742
a Swedish scientist named Anders Celsius (1701–1744) devised a thermometer scale
dividing the freezing and boiling points of water into 100.◦ C. Celsius chose 0.◦ C for
the boiling point of water, and 100.◦ C for the freezing point [119]. One year later,
Jean Pierre Cristin (1683–1755) inverted the Celsius scale to produce the Centigrade
scale used today (freezing point 0.◦ C, boiling point 100.◦ C).
1.2 From Stahl to Lavoisier 7

Fig. 1.4 A replica of

Galileo’s thermoscope
exposed in Galileo’s
museum (Florence)

1.2 From Stahl to Lavoisier

There have been few, if any, revolutions in science so great, so sudden, and so general, as
the prevalence of what is now usually termed the new system of chemistry or that of the
Antiphlogistians, over the doctrine of Stahl, which was at one time thought to have been
the greatest discovery that had ever been made in science. I remember hearing Mr. Peter
Woulfe, whose knowledge of chemistry will not be questioned, say, that there had hardly
been anything that deserved to be called a discovery subsequent to it. Though there had
been some who occasionally expressed doubts of the existence of such a principle as that
of phlogiston, nothing had been advanced that could have laid the foundation of another
system before the labors of Mr. Lavoisier and his friends, from whom this new system is
often called that of the French. (Joseph Priestley, 17966 )

It is almost impossible to fully understand the history of thermodynamics without the

help of some knowledge of the main achievements in chemistry during the eighteenth
century, when the chemical elements were discovered, establishing the framework
for the atomistic theory.
In 1667, Johann Joachim Becher published his book Physica subterranea, in
which was the first mention of what would become the phlogiston theory. Tradition-
ally, alchemists considered that there were four classical elements: fire, water, air, and
earth. In his book, Becher eliminated fire, water, and air from the classical element
model and replaced them with three forms of earth: terra lapidea, terra fluida, and
terra pinguis. Becher believed that terra pinguis was a key feature of combustibles
and was released when combustible substances were burned.

6Priestley J (1796) Considerations on the doctrine of phlogiston and the decomposition of water.
Thomas Dobson, Philadelphia, [139].
8 1 The First Principle

In 1703 Georg Ernst Stahl, professor of medicine and chemistry at Halle, proposed
a variant of the theory which he renamed Becher’s terra pinguis to phlogiston, and
it was in this form that the theory probably had its greatest influence.
Carbon dioxide was the first gas to be described as a discrete substance. In about
1640, the Flemish chemist Jan Baptist van Helmont observed that when he burned
charcoal in a closed vessel, the mass of the resulting ash was much less than that
of the original charcoal. His interpretation was that the rest of the charcoal had
been transmuted into an invisible substance he termed a gas or wild spirit (spiritus
sylvestre).
The properties of carbon dioxide were studied more thoroughly in the 1750s by
the Scottish physician Joseph Black. He found that limestone (calcium carbonate)
could be heated or treated with acids to yield a gas he called fixed air. He observed
that the fixed air was denser than air and supported neither flame nor animal life.
Black also discovered the magnesium and the latent and specific heat.
In his article On Fractious airs, published in 1766,7 Henry Cavendish (1731–
1810) reports the discovery of hydrogen (flammable air). He speculated that the
flammable air was in fact identical to the hypothetical substance called phlogiston
and in a further finding in 1781, he discovered that the gas produces water when
burned. In 1783, Antoine Lavoisier gave the element the name hydrogen (from the
Greek hydro meaning water and genes meaning generator) when he and Laplace
reproduced Cavendish’s finding that water is produced when hydrogen is burned.
The discovery of oxygen is attributed to Joseph Priestley (1733–1804) in 1774.
Priestley was an eighteenth century English theologian, dissenting clergyman, natural
philosopher, chemist, educator, and a liberal political theorist who published over
150 works.
During his lifetime, Priestley’s considerable scientific reputation rested on his
invention of soda water, his writings on electricity, and his discovery of several airs
(gases), the most famous being what Priestley dubbed dephlogisticated air (oxygen).
Many of the friends of Priestley were members of the Lunar Society, a group of man-
ufacturers, inventors, and natural philosophers who assembled monthly to discuss
their work. The core of the group included men such as the manufacturer Matthew
Boulton, the chemist and geologist James Keir, the inventor and engineer James Watt,
and the botanist, chemist, and geologist William Withering. Priestley was asked to
join this unique society and contributed much to the work of its members. However,
Priestley’s determination to defend the phlogiston theory and to reject what would
become the chemical revolution eventually left him isolated within the scientific
community.
Priestley’s science was integral to his theology, and he consistently tried to fuse
Enlightenment rationalism with Christian theism. In his metaphysical texts, Priestley
attempted to combine theism, materialism, and determinism. Priestley, who strongly
believed in the free and open exchange of ideas, advocated toleration and equal rights
for religious Dissenters, which also led him to help found Unitarianism in England.

7Cavendish H (1766) Three Papers, Containing Experiments on Factitious Air, by the Hon. Henry
Cavendish, F. R. S. Philos Trans R Soc London 56:141–184 [28].
1.2 From Stahl to Lavoisier 9

The controversial nature of Priestley’s publications combined with his outspoken

support of the French Revolution aroused public and governmental suspicion; he
was eventually forced to flee, in 1791, first to London, and then to the United States
after his home and church were burned. He spent the last ten years of his life living
in Northumberland County, Pennsylvania.
Antoine Laurent de Lavoisier was an outstanding chemist, considered to be the
father of modern chemistry. Lavoisier’s chemical research between 1772 and 1778
was largely concerned with developing his own new theory of combustion.
In 1777, Antoine Lavoisier wrote the Mémoire sur la combustion en général, the
first of what proved to be a series of attacks on phlogiston theory8 ,9 ; it was against
these attacks that Priestley responded in 1796. While Priestley accepted parts of
Lavoisier’s theory, he was unprepared to assent to the major revolutions Lavoisier
proposed: the overthrow of phlogiston, a chemistry-based conceptually on elements
and compounds, and a new chemical nomenclature. Priestley’s original experiments
on dephlogisticated air, combustion, and water provided Lavoisier with the data he
needed to construct much of his system; yet Priestley never accepted Lavoisier’s
new theories and continued to defend the phlogiston theory for the rest of his life.
Lavoisier’s system was based largely on the quantitative concept that mass is neither
created nor destroyed in chemical reactions (i.e., the conservation of mass).
Mais si tout s’explique en chimie d’une manière satisfaisante sans le secours du phlogis-
tique, il est par cela seul infiniment probable que ce principe n’existe pas; que c’est un être
hypothétique, une supposition gratuite; et, en effet, il est dans les principes d’une bonne
logique de ne point multiplier les êtres sans nécessité. Peut-être aurais-je pu m’en tenir à ces
preuves négatives, et me contenter d’avoir prouvé qu’on rend mieux compte des phénomènes
sans phlogistique qu’avec le phlogistique; mais il est temps que je m’explique d’une manière
plus précise et plus formelle sur une opinion que je regarde comme une erreur funeste à la
chimie, et qui me paraît en avoir retardé. (Antoine Lavoisier, 1783, [101])

1.2.1 Gay-Lussac and Clapeyron

Louis Joseph Gay-Lussac (1778–1850) measured the dilation of ideal gases as

affected by the temperature in 1802.10 His apparatus is sketched in Fig. 1.5. It includes
a balloon full of a dried gas, isolated from the atmosphere through a mercury stopper
placed in a graduated capillary tube. It is submitted at melting ice and the temperature
is increased till boiling water temperature. The volume of the gas was measured by
reading the mercury drop position.
One of Gay-Lussac’s conclusions was: “For the permanent gases, the increase
of volume received by each of them between the temperature of melting ice and that

8 Lavoisier A (1777) Mémoire sur la combustion en général. Mémoires l’Académie des Sci 592–
600.
9 Lavoisier A (1783) Réflexions sur le phlogistique, pour servir de suite á la théorie de la combustion

et de la calcination publiée en 1777. Mémoires l’Académie des Sci 505–538, [101].

10 Gay-Lussac LJ (1802) L’expansion des gaz par la chaleur. Ann Chim 43:137, [58].
10 1 The First Principle

Fig. 1.5 a Gay-Lussac

experiment for measuring
the dilation of gases with the
temperature. b Results found
for the molar volume
dependence on the
temperature at atmospheric
pressure

of boiling water is equal to (...) 100/266.66 of the same volume for the centigrade
thermometer”.
So, he found that for all ‘permanent’, or ideal gases, the relative variation of the
volume . Δv
v0
between the melting ice and boiling water was constant,

Δv 100
. = .
v0 267

For an arbitrary temperature .θ in .◦ C, this relationship can be written as

v (θ) θ + 267
. = or v (θ) = const. × (θ + 267)
v0 267

and, although Gay-Lussac did not report this finding, the value .θ = −267 ◦ C was
the first estimation of the zero absolute temperature.11

11 Presently, the absolute zero is considered to be .θ = −273.16 ◦ C.

1.2 From Stahl to Lavoisier 11

The law of gas expansion is presently known as the Charles and Gay-Lussac
law. Indeed, in his 1802 paper, Gay-Lussac wrote: “Before going further, I must
jump ahead. Although I had recognized on many occasions that the gases oxygen,
nitrogen, hydrogen, carbonic acid, and atmospheric air all expand identically, citizen
Charles had noticed the same property in these gases 15 years ago; however, since
he never published his results, it is only by great luck that I knew it”.
An important conclusion in his article was that the law of gas expansion could
not be applied to vapors—or condensable gases—except when these vapors were
in the ‘elastic state’: “Since all gases are equally expansible by heat and equally
compressible, and since these two properties depend on each other, as I shall show
in another place, the vapors which are equally expansible with the gases should also
be equally compressible, but I may mention that this last conclusion cannot be true
except so long as the compressed vapors remain entirely in the elastic state, and this
requires that their temperature shall be sufficiently elevated to enable them to resist
the pressure which tends to make them assume the liquid state”.
The equation of state for ideal gases as it is presently known was only enunciated
in 1834, 32 years after Gay-Lussac, in a historical article by Benoît émile Clapeyron
(1799–1864), published in the Journal de l’Ecole Royale Polytechnique12 (Fig. 1.6).
In this article, Clapeyron combined the Gay-Lussac’s law with Boyle-Mariotte’s law,
reasoning that when the temperature increases from .θ0 to .θ at a constant pressure . P0 ,
then the molar volume .v (θ, P0 ) will follow

v (θ, P0 ) θ + 267
. = .
v (θ0 , P0 ) θ0 + 267

Therefore, by inserting this relationship into Boyle-Mariotte’s law

p0 v (θ0 , P0 )
. pv (θ, P) = p0 v (θ, P0 ) = (θ + 267) ,
θ0 + 267
or

pv (θ, P) p0 v (θ0 , P0 )
. = = cte = R,
θ + 267 θ0 + 267

where .R is the well-known universal gas constant nowadays.

12Clapeyron BE (1834) Mémoire sur la Puissance Motrice de la Chaleur. J l’Ecole R Polytech

23:153–190, [33].
12 1 The First Principle

Fig. 1.6 Clapeyron’s equation published at page 164 of the Mémoire sur la Puissance Motrice de
la Chaleur

1.3 The Steam Engines and Carnot

L’étude de ces machines est du plus haut intérêt, leur importance est immense, leur emploi
s’accroît tous les jours. Elles paraissent destinées à produire une grande révolution dans le
monde civilisé. Déjá la machine à feu exploite nos mines, fait mouvoir nos navires, creuse
nos ports et nos rivières, forge le fer, façonne les bois, écrase les grains, file et ourdit nos
étoffes, transporte les plus pesans fardeaux, etc. Elle semble devoir un jour servir de moteur
universel et obtenir la préférence sur la force des animaux, les chutes d’eau et les courants
d’air. Elle a, sur le premier de ces moteurs, l’avantage de l’économie; sur les deux autres,
l’avantage inappréciable de pouvoir s’employer en tous temps et en tous lieux, et de ne
jamais souffrir d’interruption dans son travail. (Sadi Carnot, 1824, [27])

Readers are referred to Robert Henry Thurston13 book ‘A History of the Growth
of the Steam-Engine’, [175], for a complete and detailed history of steam engines
beginning with Hero in Greece. This book was re-edited several times and a pdf is
available for downloading.
The steam engine was first used to pump water out of coal mines. In 1712 Thomas
Newcomen’s atmospheric engine became the first commercially successful engine
using the principle of the piston and cylinder, which was the fundamental type of
steam engine used until the early twentieth century (Fig. 1.7).
Steam produced by boiling water pushes the piston upward. When the piston is at
the top of the cylinder, a valve opens allowing a spray of cold water to be admitted
into the cylinder, condensing the steam and, so, creating a partial vacuum below
the piston. The piston is, then, forced to move downward, completing the cycle, by
the action of the atmospheric pressure above it. In the Newcomen engine, the steam
generator and the condenser were, then, in the same place.
While working as an instrument maker at the University of Glasgow, James Watt
became interested in the technology of steam engines. He realized that contempo-
rary engine designs wasted a great deal of energy by repeatedly cooling and reheat-
ing the cylinder. Watt introduced a design enhancement, the separate condenser,
which avoided this waste of energy and radically improved the power, efficiency,

13
Robert Henry Thurston (1839–1903) was an American engineer, and the first Professor of
Mechanical Engineering at Stevens Institute of Technology.
1.3 The Steam Engines and Carnot 13

Fig. 1.7 Newcomen steam

engine

Fig. 1.8 First stage of the

James Watt steam engine

and cost-effectiveness of steam engines. He, also, invented the double-acting cylin-
der (Figs. 1.8 and 1.9). These improvements more than doubled the efficiency of the
engine with respect to Newcomen’s engine.
In Watt’s words,14 “I had gone to take a walk on a fine Sabbath afternoon. I had
entered the Green by the gate at the foot of Charlotte Street and had passed the old

14 Reminiscences of James Watt, Robert Hart, Transactions of the Glasgow Arch Theological Soci-
ety, l709.
14 1 The First Principle

Fig. 1.9 Second stage of the

James Watt steam engine

washing house. I was thinking upon the engine at the time and had gone as far as
the herd’s house, when the idea came into my mind that, as steam was an elastic
body, it would rush into a vacuum, and, if communication were made between the
cylinder and an exhausted vessel, it would rush into it and might be there condensed
without cooling the cylinder. I then saw that I must get rid of the condensed steam
and injection water if I used a jet, as in Newcomen’s engine. Two ways of doing this
occurred to me: First, the water might be run off by a descending pipe, if an offlet
could be got at the depth of 35 or 36 feet, and any air might be extracted by a small
pump. The second was, to make the pump large enough to extract both water and air.
I had not walked farther than the Golf house when the whole thing was arranged in
my mind”
Watt attempted to commercialize his invention but experienced great financial
difficulties until he entered a partnership with Matthew Boulton in 1775. The new
firm of Boulton and Watt was highly successful and Watt became a wealthy man.
Watt’s invention was a key point in the Industrial Revolution.
Both Newcomen and Watt steam engines depend on a source of energy—the heat
of combustion of coal—and on a heat sink—the cold water.
When Sadi Carnot [27], began working on his ‘Réflexions sur la puissance motrice
du feu’, published in 1824, steam engines had achieved widely recognized economic
and industrial importance, but there had been no real scientific study of them. Engi-
neers in Carnot time had tried to use highly pressurized steam to improve the effi-
ciency of engines. In these early stages of engine development, the efficiency of a
typical engine—the useful work it was able to do when a given quantity of fuel was
burned—was only about 3%.
In the first part of his RéflexionsCarnot advocates that everywhere where we have
a difference in temperature it is possible to generate motive power, a central idea in
1.4 The Principle of Energy Conservation 15

thermodynamics. A corollary of this idea is that it is impossible to produce motive

power unless we dispose, simultaneously, of a hot and a cold body, which is a first
enunciate of the second principle of thermodynamics. Carnot based his réflexions on
Lavoisier’s caloric theory, comparing the flow of caloric from the high to the low
temperature to the flow of water through a hydraulic machine, producing work in
its passage: the higher the height of the waterfall the higher the delivered work. He
concluded: “The production of motive power is, therefore, due in steam engines not
to actual consumption of caloric but to its transportation from a warm body to a cold
body”.

1.4 The Principle of Energy Conservation

Believing that the power to destroy belongs to the Creator alone I affirm…that any theory
which, when carried out, demands the annihilation of force is necessarily erroneous. (James
Prescott Joule, 1843)

1.4.1 Joseph Black

The concept of ‘heat’ or the internal energy of a thermodynamic system as the result
of the mechanical energy of its individual molecules is very old and was supported by
philosophers and scientists. “Heat is motion” in the words of Francis Bacon (around
1600). “Heat is nothing else but the motion of the particles that form a body” (Thomas
Hooke, around 1650). Nevertheless, the caloric theory and mechanics evolved as
independent sciences since the invention of the thermoscope by Galileu in 1600, the
built-up of the mercury thermometer and the establishment of measurement methods
for the specific and latent heat in terms of heat unity, the calorie (the amount of heat
that raises the temperature of 1 g of water by 1.◦ C).
Joseph Black (1728—1799) was a Scottish physicist and chemist, known for
his discoveries of magnesium and carbon dioxide, latent and specific heat. He was
Professor of Anatomy and Chemistry at the University of Glasgow for 10 years from
1756, and then Professor of Medicine and Chemistry at the University of Edinburgh
from 1766, teaching and lecturing there for more than 30 years.
Black was the first to illustrate that heat and temperature are different things.
Although thermometers existed, there were several different ideas as to what they
were measuring. One way Black demonstrated the distinction was to take a bucket
of ice and monitor its temperature constantly. The ice gradually melted, yet the
temperature stayed the same. This finding credited Black with the discovery of latent
heat. From this Black developed the idea of’specific heat’ by measuring the amount
of heat required to raise the temperature of a substance by a specified number of
degrees.
16 1 The First Principle

In his Lectures on the Elements of Chemistry delivered at the University of Edin-

burgh [12] Black gives a specific example involving the measurement of the specific
heat of mercury (or quicksilver, as mercury was known at that time): “To make this
plainer by an example in numbers, let us suppose the water to be at the 100th degree
of heat, and that an equal measure of warm quicksilver at the 150th degree, is sud-
denly mixed and agitated with it. We know that the middle temperature between 100
and 150 is 125, and we know that this middle temperature would be produced by
mixing the cold water at 100 with an equal measure of warm water at 150; the heat
of the warm water being lowered by 25 degrees, while that of the cold is raised just
as much. But when warm quicksilver is used in place of warm water, the temperature
of the mixture turns out 120 degrees only instead of 125. The quicksilver, therefore,
is become less warm by 30 degrees, while the water has become warmer by twenty
degrees only; and yet the quantity of heat that the water has gained is the very same
quantity that the quicksilver has lost”.
In [12] Black proposed in his lecture ‘On the General Effects of Heat’: (i) to
ascertain what he means by the word heat; (ii) to explain the meaning of the term
cold, and ascertain the real difference between cold and heat; (iii) to mention some
of the attempts that have been made to discover the nature of heat, or to form an
idea of what may be the immediate cause of it; (iv) to describe the sensible effects
produced by heat on the bodies to which it is communicated.
He writes in page 25 of [12] “When we consider the communication of heat from
hot bodies to colder ones, the first question that may naturally occur to our mind is,
in what manner have these two bodies acted, the one on the other, on this occasion?
Has one of them lost something, which the other has gained? The vulgar opinion is,
that the hot body has lost something which has been added to the other. And those
who have attempted to reason more profoundly on the nature of heat, have agreed
with the multitude in this point and have supposed that heat is a positive quality
and depends either upon an exceedingly subtle and active matter, introduced into the
pores of bodies, or upon a tremor or vibration excited among their particles”.
Therefore, although Black does not deny the idea of heat as a subtle fluid, the
concept of heat or internal energy as the mechanical energy of the particles that form
a body is present in Black’s writings.

1.4.1.1 Lavoisier and the Caloric Theory

In Sect. 1.2 we saw that Lavoisier showed the inconsistency of Stahl’s phlogiston
theory of combustion with his own experimental results leading him to postulate the
principle of mass conservation. Lavoisier did early research in physical chemistry
and thermodynamics in joint experiments with Laplace and, controversially, also
postulated the principle of caloric conservation. Caloric was considered by him as a
kind of weightless fluid and, according to this theory, the quantity of this substance
is constant throughout the universe and flows from warmer to colder bodies.
He writes in his Réflexions sur le phlogistique: on the nature of heat [101]:
1.4 The Principle of Energy Conservation 17

When an ordinary body—solid or fluid—is heated, that body increases in dimensions in all
directions it occupies a larger and larger volume. If the cause of heating ceases, the body
retreats through the same degrees of extension that it reached, at the same rate as it cools.
Finally, if it is returned to the same temperature that it had at the first instant, it will clearly
return to the same volume as it had before. Hence the corpuscles of matter do not touch
each other, there exists between them a distance that heat increases and that cold decreases.
One can scarcely conceive of these phenomena except by admitting the existence of a subtle
fluid, the accumulation of which is the cause of heat and the absence of which is the cause of
coldness. No doubt it is this fluid that lodges between the particles of matter, which spreads
them apart and which occupies the space left between them. Along with the majority of
physicists, I name this fluid (whatever it is) igneous fluid, the matter of heat and fire.
I do not deny that the existence of this fluid is, up to a certain point, hypothetical. But
supposing it is a hypothesis (that it is not rigorously proven), it is the only one that I am
obliged to posit. The partisans of the phlogiston doctrine are no more sophisticated than me
on this point and, if the existence of the igneous fluid is a hypothesis, it is common to both
their system and my own.
We see that in such a state of affairs, the corpuscles would not have any connection with
each other, that there would be no solid bodies if they were not held together by another
force—by attraction—which is a general principle of nature to which all matter seems to
submit, whatever the cause.
According to this first glimpse, all natural bodies obey two forces: igneous fluid (the matter of
fire), which tends to continually separate corpuscles, and the attraction that counter-balances
this force. So long as the latter force—attraction—is victorious, bodies remain in a solid
state. When these two forces are in a state of equilibrium, the substance becomes liquid.
Finally, when the expansive force of heat carries it off, the substance passes into the aeriform
state. But if only these two forces existed, at the moment the bodies ceased to be in the
solid state the slightest increase in heat that they received would be sufficient to vaporize
them. Not only would they pass quickly to the aeriform state, but their corpuscles would
also be carried away endlessly further and further. But there is a third force, which stops
this effect from taking place—this is the weight of the atmosphere. Without this pressure,
at the moment when water ceases to be ice (at zero degrees) it would become an aeriform
fluid, whereas in fact this effect only occurs at a temperature of 100 C, at a pressure of one
atmosphere.

In present days, the conservation of caloric principle would be written as

δ Q = dU,
. (1.1)

meaning that the amount of caloric .U of a body—or the mechanical energy of their
molecules, as we consider today—can only be changed by the addition of a certain
amount of heat.δ Q from an external source of caloric. For an isolated thermodynamic
system

U = constant.
.

Therefore, following Lavoisier, the total amount of caloric .U in the universe was
constant and was not to be confused with force, which was the designation of energy
at the time, but should be considered as a subtle fluid that was transferred from warm
to cold bodies.
18 1 The First Principle

1.4.2 Joule’s Experiments

The text in the following is based on the notes about Joule by Ingo Muller [119] and
on the Wikipedia article15 (https://en.wikipedia.org/wiki/James_Prescott_Joule).
The son of a wealthy brewer, Joule was tutored as a young man by the famous
scientist John Dalton16 and was strongly influenced by chemist William Henry17
and Manchester engineers Peter Ewart and Eaton Hodgkinson. He was fascinated by
electricity. As an adult, Joule managed the brewery. Science was merely a serious
hobby. Sometime around 1840, he started to investigate the feasibility of replacing
the brewery’s steam engines with the newly invented electric motor. He discovered
Joule’s first law in 1841, that the heat which is evolved by the proper action of any
voltaic current is proportional to the square of the intensity of that current, multiplied
by the resistance to conduction which it experiences.18
Joule’s interest diverted from the narrow financial question to that of how much
work could be extracted from a given source, leading him to speculate about the
convertibility of energy. In 1843 he published results of experiments showing that
the heating effect he had quantified in 1841 was due to the generation of heat in
the conductor. This was a direct challenge to the caloric theory which held that heat
could neither be created nor destroyed. The caloric theory has dominated thinking
in the science of heat since introduced by Antoine Lavoisier in 1783. Lavoisier’s
prestige and the practical success of Sadi Carnot’s caloric theory of the heat engine

15 James P. Joule received several biographies. See for instance:

Bottomley, J. T. (1882).James Prescott Joule. Nature 26 (678): 617–620. Bib-
code:1882Natur..26..617B. https://doi.org/10.1038/026617a0.
Cardwell, D. S. L. (1991). James Joule: A Biography. Manchester University Press. ISBN 0-
7190-3479-5.
Forrester, J. (1975). Chemistry and the Conservation of Energy: The Work of James Prescott
Joule. Studies in the History and Philosophy of Science 6 (4): 273–313. https://doi.org/10.1016/
0039-3681(75)90025-4.
Fox, R, James Prescott Joule, 1818–1889, in North, J. (1969). Mid-nineteenth-century scientists.
Elsevier. pp. 72–103. ISBN 0-7190-3479-5.
Reynolds, Osbourne (1892). Memoir of James Prescott Joule. Manchester, England: Manchester
Literary and Philosophical Society.
16 The most important of all Dalton’s investigations are those concerned with the beginning of

atomic theory in chemistry. The main points of Dalton’s atomic theory were: (i) elements are made
of extremely small particles called atoms; (ii) atoms of a given element are identical in size, mass,
and other properties; atoms of different elements differ in size, mass, and other properties; (iii)
atoms cannot be subdivided, created, or destroyed; (iv) atoms of different elements combine in
simple whole-number ratios to form chemical compounds; (v) in chemical reactions, atoms are
combined, separated, or rearranged.
17 Henry W (1803) Experiments on the Quantity of Gases Absorbed by Water, at Different Temper-

atures, and under Different Pressures. Philos Trans R Soc London 93:29–274. https://doi.org/10.
1098/rstl.1803.0004 [69].
18 Joule JP (1841) On the heat evolved by metallic conductors of electricity, and in the

cells of a battery during electrolysis. Philos Mag Ser 3 19:260–277. https://doi.org/10.1080/

14786444108650416 [79].
1.4 The Principle of Energy Conservation 19

since 1824 ensured that the young Joule, working outside either academia or the
engineering profession, had a difficult road ahead.
Further experiments and measurements led Joule to estimate the mechanical
equivalent of heat as 4.1868 J per calorie of work to raise the temperature of a pound
of water by one degree Fahrenheit. He announced his results at a meeting of the
chemical section of the British Association for the Advancement of Science in Cork
in August 1843 and was met by silence.
Joule was undaunted and started to seek a purely mechanical demonstration of
the conversion of work into heat. By forcing water through a perforated cylinder,
he could measure the slight viscous heating of the fluid. He obtained a mechanical
equivalent of 770 ft.lbf/Btu (4.14 J/cal). The fact that the values obtained both by
electrical and purely mechanical means were in agreement to at least one order of
magnitude was, to Joule, compelling evidence of the reality of the convertibility of
work into heat.
Joule now tried a third route. He measured the heat generated against the work
done in compressing a gas. He obtained a mechanical equivalent of 798 ft.lbf/Btu
(4.29 J/cal). In many ways, this experiment offered the easiest target for Joule’s critics
but Joule disposed of the anticipated objections by clever experimentation. Joule read
his paper to the Royal Society on 20 June 1844, however, his paper was rejected for
publishing by the Royal Society and he had to be content with publishing in the
Philosophical Magazine in 1845.19
In June 1845, Joule read his paper On the Mechanical Equivalent of Heat to
the British Association meeting in Cambridge. In this work, he reported his best-
known experiment, involving the use of a falling weight, in which gravity does
the mechanical work, to spin a paddle wheel in an insulated barrel of water which
increased the temperature (Fig. 1.10). He now estimated a mechanical equivalent of
819 ft.lbf/Btu (4.41 J/cal). He wrote a letter to the Philosophical Magazine, published
in September 1845 describing his experiment.20
In 1850, Joule published a refined measurement of 772.692 ft.lbf/Btu
(4.159 J/cal).21
Joule was proposing a mechanical theory of heat, and this required a conceptual
leap: if heat was a form of molecular motion, why the motion of the molecules did
not gradually die out? Joule’s ideas required one to believe that the collisions of
molecules were perfectly elastic. We should also remember that the very existence
of atoms and molecules was not widely accepted for another 50 years.
Although it may be hard today to understand the allure of the caloric theory, at
the time it seemed to have some clear advantages. Carnot’s successful theory of heat
engines had also been based on the caloric assumption, and only later was it proved

19 Joule JP (1845) On the changes of temperature produced by the rarefaction and condensation of
air. Philos Mag Ser 3 26:369–383. https://doi.org/10.1080/14786444508645153 [78].
20 Joule, J.P. (1845) On the Mechanical Equivalent of Heat, Brit. Assoc. Rep., trans. Chemical Sect,

p.31, read before the British Association at Cambridge, June 1845.

21 Joule, J.P. (1850). On the Mechanical Equivalent of Heat. Philosophical Transactions of the Royal

Society of London 140: 61–82. https://doi.org/10.1098/rstl.1850.0004 [80].

20 1 The First Principle

Fig. 1.10 Joule apparatus for measuring the mechanical equivalent of heat

by Lord Kelvin that Carnot’s mathematics were equally valid without assuming a
caloric fluid.
However, in Germany, Hermann Helmholtz became aware both of Joule’s work
and the similar 1842 work of Julius Robert von Mayer. Though both men had been
neglected since their respective publications, Helmholtz’s definitive 1847 declaration
of the conservation of energy credited them both.22
Also in 1847, another of Joule’s presentations at the British Association in
Oxford was attended by George Gabriel Stokes, Michael Faraday, and the preco-
cious and maverick William Thomson, later to become Lord Kelvin, who had just
been appointed professor of natural philosophy at the University of Glasgow. Stokes
was inclined to be a Joulite and Faraday was much struck with it though he harbored
doubts. Thomson was intrigued but skeptical.
Though Thomson felt that Joule’s results demanded a theoretical explanation, he
retreated into a spirited defense of the Carnot-Clapeyron school. In 1848, Thomson
wrote that the conversion of heat (or caloric) into mechanical effect is probably
impossible, certainly undiscovered—but a footnote signaled his first doubts about
the caloric theory, referring to Joule’s very remarkable discoveries. Surprisingly,
Thomson did not send Joule a copy of his paper but when Joule eventually read
it he wrote to Thomson on 6 October, claiming that his studies had demonstrated
conversion of heat into work but that he was planning further experiments. Thomson
replied on the 27th, revealing that he was planning his own experiments and hoping for
a reconciliation of their two views. Though Thomson conducted no new experiments,

22 von Helmholtz, HLF. Über die Erhaltung der Kraft. Reimer, Berlin 1847 [68].
1.4 The Principle of Energy Conservation 21

over the next two years he became increasingly dissatisfied with Carnot’s theory and
convinced of Joule’s. In his 1851 paper, Thomson was willing to go no further than
a compromise and declared the whole theory of the motive power of heat is founded
on two propositions,23 due respectively to Joule, and to Carnot and Clausius.
As soon as Joule read the paper, he wrote to Thomson with his comments and
questions. Thus began a fruitful, though largely epistolary, collaboration between
the two men, Joule conducting experiments, Thomson analyzing the results and
suggesting further experiments. The collaboration lasted from 1852 to 1856, and its
discoveries including the Joule-Thomson effect and the published results did much
to bring about general acceptance of Joule’s work and the kinetic theory.

1.4.3 Exercise

In his 1850 paper [80], Joule wrote: “For a long time it had been a favorite
hypothesis that heat consists of a force or power belonging to bodies, but it was
reserved for Count Rumford to make the first experiments decidedly in favor of
that view. That justly celebrated natural philosopher demonstrated by his ingenious
experiments that the very great quantity of heat excited by the boring of cannon could
not be ascribed to a change taking place in the calorific capacity of the metal; and he
therefore concluded that the motion of the borer was communicated to the particles of
metal, thus producing the phenomena of heat. It appears to me, he remarks, extremely
difficult, if not quite impossible, to form any distinct idea of anything, capable of being
excited and communicated, in the manner the heat was excited and communicated
in these experiments, except it be a renovation of the most important parts of Count
Rumford’s paper, though one to which little attention has hitherto been paid, is that in
which he makes an estimate of the quantity of mechanical force required to produce
a certain amount of heat. Referring to his third experiment, he remarks that the
total quantity of ice-cold water which, with the heat actually generated by friction,
accumulated in 2h30, might have been heated 180.◦ F, or made to boil 26.58 lbs. In
the next page, he states that the machinery used in the experiment could easily be
carried round by the force of one horse (though to render the work lighter, two horses
were actually employed in doing it). Now the power of a horse is estimated by Watt
as 33,000 foot-pounds per minute.”
Using the data from Count Rumford and the Watt estimation of the power of a
horse, find the mechanical equivalent of heat, and compare it with the 772 lbf.foot/btu
finding of Joule.

23 The first and second principle of thermodynamics.

22 1 The First Principle

1.4.4 The Principle of Energy Conservation

The principle of mechanical energy conservation was first evidenced by Galileo with
his experiments on falling bodies and on the inclined plane. Gottfried Leibniz (1646–
1716) extended the work of his mentor Huygens on the importance of the product
of mass times velocity squared which had been originally investigated by Huygens
and which Leibniz called vis viva, the living force. He believed the vis viva to be
the real measure of force, as opposed to Descartes’s force of motion (equivalent to
mass .× velocity, or momentum ). That vis viva was the most fundamental conserved
quantity comes extremely close to an early statement of the principle of conservation
of energy in mechanics. Since, however, the conservation of momentum had become
one of the pillars of Cartesian natural philosophy, his vis viva was seen as rivaling
the conservation of momentum championed by Newton in England and by Descartes
in France; hence academics in those countries tended to neglect Leibniz’s idea. In
reality, both energy, and momentum are conserved, so the two approaches are equally
valid.
In present days, the principle of mechanical energy conservation would be written
as

.δW = d E c + d E p , (1.2)

meaning that the total mechanical energy . E c + E p of a system of particles is con-

served unless it is affected by the work performed by an external force. So, if we
neglect the air friction and consider a single body at rest, under the action of gravity
and suspended at an height .h from the ground level, the velocity .v this body will
acquire at the ground level when falling from a height .h will be found by solving

1 2
. mv + 0
,,,, = 0
,,,, + mgh .
,,,,
,2 ,, , E p at ground level E c at heigth h E p at heigth h
E c at ground level

√
The solution of this equation gives .v = 2gh, which was the expression for the
velocity found by Galileu for falling bodies: the velocity of a falling body is indepen-
dent of its weight. This was an important finding, because contrary to Aristotle’s early
idea that heavy objects fall faster than lighter ones, in direct proportion to weight.
Drawing on the earlier work of Sadi Carnot, Benoît Paul émile Clapeyron, James
Prescott Joule, and Faraday, Helmholtz [68] postulated a relationship between
mechanics, heat, light, electricity and magnetism by treating them all as manifesta-
tions of a single force: energy in modern terms.
When heat is recognized as energy, the principle of energy conservation postulated
by Helmholtz would be written, in a purely mechanical context, as the sum of Eqs.
(1.1) and (1.2)

δ Q + δW = dU + d E c + d E p .
. (1.3)
1.4 The Principle of Energy Conservation 23

Therefore, when a thermodynamic system performs a cycle, starting from and

returning to the same thermodynamic state

∮
. (δ Q + δW ) = 0.

Considering in Fig. 1.10 the liquid inside the vessel as a thermodynamic system,
this equation summarizes Joule’s ideas for the conversion of work into heat and vice-
versa. Indeed, Joule’s principle of the mechanical equivalent of heat can be written
as
∮ ∮
. δW = − δ Q,

where the l.h.s. of the above equation describes the friction work performed on the
liquid by the paddles, with the weight fall, and the r.h.s. the heat that must be removed
from the liquid for recovering its initial temperature.
In addition to its kinetic (. E c ) and potential energy (. E p ) a thermodynamic system
has the internal energy, U (or the mechanical energy of its molecules) as a third kind
of energy. The total energy of this system can be exchanged with its environment by
heat or work. This is the principle of energy conservation

.δ Q + δW = dU + d E c + d E p .

1.4.5 The Joule Expansion

In 1843, Joule did the experiment shown in Fig. 1.11. In the left glass sphere, he
introduced an elastic gas (as the ideal gases were known at the time), while the right
sphere was under vacuum. The set was insulated from the ambient. When the tap
was suddenly opened, the gas in the left sphere flowed without resistance into the
right sphere. No work was performed, no heat was transferred and, in spite of the
abrupt gas expansion, the gas temperature remained constant.
This was a surprising result since we would expect that the gas would cool after
an adiabatic expansion. Nevertheless, by considering the first principle of thermo-
dynamics for this problem

. δ Q + δW = dU, (1.4)

it is easily seen that under an adiabatic free expansion both .δ Q and .δW are null, so,
the internal energy remains unchanged in this expansion.
24 1 The First Principle

Fig. 1.11 Joule’s

experiment on gas expansion

The conclusion is that the internal energy of ideal gases depends only on the
temperature .U = U (T ).

1.4.6 Specific or Molar Heats

Fig. 1.12 shows an ice calorimeter developed by Lavoisier and Laplace to measure the
heat that is released by chemical reactions. Ice calorimeters were based on Black’s
idea of latent heat.
The addition of a certain amount of heat .δ Q to a constant volume of a gas will
elevate its temperature by .ΔT (Fig. 1.13). The specific heat capacity at constant
volume is defined as

( )
δQ
c = lim
. v ,
ΔT →0 mΔT V =const

whereas the molar heat capacity at constant volume as,

( )
δQ
c = lim
. v ,
ΔT →0 N ΔT V =const

where .m and . N are, respectively, the mass and the molar amount of the gas in the
chamber.
The same definitions apply to the specific heat capacity at a constant pressure,
but, in this case, is the pressure that is maintained constant during the heating process
(Fig. 1.13b).
From the first principle of thermodynamics, Eq. (1.4), when heat is added at
constant volume the amount .δ Q = dU , because the volume is constant and there is
1.4 The Principle of Energy Conservation 25

Fig. 1.12 Ice calorimeter

created by Lavoisier and
Laplace based on Black’s
idea of latent heat

Fig. 1.13 Sketch for the

specific or molar heat
capacity measurement a at
constant volume and b at
constant pressure
26 1 The First Principle

no expansion work during the heating. So,

dU = N cv dT or dU = mcv dT.
.

On the other hand, by recalling that the enthalpy is defined as . H = U + P V the

amount .δ Q = d H when this heat is added at constant pressure (Fig. 1.13b), because
the related expansion work is .−Pd V . So

.d H = N c p dT or d H = mc p dT.

1.4.7 Ideal Gases

The molar, .cv , and specific, .cv , heat capacities depend on both the temperature and
pressure. Nevertheless, for ideal gases Joule’s expansion experiments showed that
they are only dependent on the temperature .cv = .cv (T ) and .cv = cv (T ).
In addition, since for ideal gases . H = U + N RT , we are led to the conclusion
that

c = cv + R
. p or c p = cv + R

where . R =.R/M, . M being the molecular mass of the gas.

Specific or molar heat capacities measure the ability of the molecules to store
thermal energy. It is thus an intrinsic molecular property and is related to the degrees
of freedom of each molecule,

c̄ = c̄v,trans + c̄v,r ot + c̄v,vib ,

. v

where .c̄v,trans , c̄v,r ot , c̄v,vib , are respectively the translational, rotational, and vibra-
tional molar heat capacities of the molecules. Monatomic molecules have only trans-
lational degrees of freedom and the kinetic theory shows that the translational molar
heat capacity is constant and given by

3
c̄
. v,trans = R.
2
On the other hand, polyatomic molecules have also rotational an vibrational
degrees of freedom and these are both dependent on the temperature. Many text-
books and thermodynamic tables show the specific or molar heat capacities of ideal
1.5 Equilibrium and Non-equilibrium States 27

gases as the result of numerical interpolation on experimental data. A frequently used

expression is

c̄ = A + BT + C/T 2 ,
. v

where . A, . B, .C are interpolation constants.

1.5 Equilibrium and Non-equilibrium States

We can consider a thermodynamic system as a portion of matter with a well-defined

frontier. This frontier can be deformed and the total volume of this system can change
with time. When the system does not exchange mass with the external environment,
the system is said to be closed, otherwise it is called an open system.
For the moment, let us limit ourselves to closed systems in equilibrium and we
understand an equilibrium state as a state where all the intensive variables are the
same everywhere, in all points of this system. We designate intensive variables as
the ones which do not depend on the mass of the system such as the temperature, the
pressure, and the density. In other words, intensive variables are the ones that do not
change when you move from one point to another inside a system in equilibrium.
For closed systems that change their equilibrium state by exchanging heat and/or
work with its external environment, the first principle of thermodynamics reads as,

. Q 1−2 + W1−2 = U2 − U1 ,

where .U is the thermal or internal energy of the thermodynamic system, . Q 1−2 is the
heat exchanged between the equilibrium states 1 and 2, and.W1−2 is the corresponding
work exchanged between these two equilibrium states.
Nevertheless, some important points must be cleared before going to the Second
Principle:
(a) the first principle as above written gives an energy balance between two equi-
librium states; both states 1 and 2 are equilibrium states and the heat and work
exchanged are the result of an energy balance between these two equilibrium
states;
(b) when a system moves from an equilibrium state to another it will necessarily
experience non-equilibrium in its intensive variables; when, e.g., heat is added
to a cold system, this addition will first heat the frontier and the equilibrium state
will be only reached after the heat is transferred from the hot frontier to the inner
parts of the system.
28 1 The First Principle

Therefore, in real processes, thermodynamic systems always move through non-

equilibrium states. These processes are called irreversible because they always
involve the transformation of a certain amount of mechanical energy into heat. In
other words, real processes are always accompanied by dissipation.
In the following chapter, we will try to better understand the role of energy dissi-
pation in real processes of a thermodynamic system.
Chapter 2
The Second Principle

Abstract For producing motive power in a cyclic operation, a heat engine is required
to operate between two thermal reservoirs at different temperatures. What is nowa-
days known as Planck’s statement of the second principle of thermodynamics was
first perceived by Carnot and was the main difficulty for the acceptance of the equiv-
alence between heat and work and of the principle of energy conservation. In the
reverse sense, as stated by Clausius, heat cannot be transferred from a cold to a hot
body by itself. Although with different words and worries, the underlying meaning
behind both Plank and Clausius’s statements is the same: nature has an arrow. We
start this chapter by presenting the role of dissipation when reversible and irreversible
processes are compared. This subject is followed by the presentation of the state-
ments of the second principle and the demonstration of their equivalence and by an
analysis of the works of Carnot and Clapeyron and of reversible cycles other than
Carnot’s cycles. Real processes are irreversible and irreversible processes are always
dissipative and sources of entropy, as presented in Sect. 2.5. The chapter ends with
a section on the meaning of entropy following the outstanding findings of Clausius,
Maxwell, and Boltzmann.

2.1 Reversible and Irreversible Processes

2.1.1 An Abrupt Isothermal Expansion

Consider Fig. 2.1. One mole of an ideal gas is inside a cylinder with a 10 .cm2 straight
section, under the action of the atmospheric pressure and three 50 .kg balance weights
and in thermal contact with a heat reservoir at 300 .K. In these conditions, considering
the atmospheric pressure to be . Patm = 101325 .Pa, the initial equilibrium state of the
gas is given in Table 2.1.
Now, we are interested in knowing what happens with this thermodynamic system,
when a single balance weight is removed from the top of the piston, here supposed to
have negligible weight. Since the gas is in thermal contact with a bath at a constant
temperature and under the action of the two remaining balance weights, the final
equilibrium state will be established at .T = 300 .K and . P = 1.0823 .MPa (Table 2.1).

© Associação Paranaense de Cultura 2024 29

P. C. Philippi, Thermodynamics, https://doi.org/10.1007/978-3-031-49357-7_2
30 2 The Second Principle

Fig. 2.1 Isothermal

expansion of an ideal gas

Table 2.1 Initial and final equilibrium states

Thermodynamic variables T (K) P (MPa) V (.l)
Initial equilibrium state (1) 300 1.5728 1.5858
Final equilibrium state (2) 300 1.0823 2.3045

Since the gas is ideal, it follows the Clapeyron equation of state and the final volume
is, then,
N RT
.V = = 2.3045l.
P
Nevertheless, the sudden removal of a 50 .kg weight will produce an appreciable
non-equilibrium in the internal fields of temperature, pressure and density, with a
rarefaction wave moving downward from the top of the gas chamber, with the speed
of sound.
Consider, for instance, the temperature field produced by gas expansion. During
the sudden expansion, the gas will cool below the temperature .T = 300 .K that was
imposed on it in the beginning. In fact, the immediate neighboring of the cylinder
walls is the only place where it is possible to assert that the temperature remains at
300 .K in the course of the gas expansion.
2.1 Reversible and Irreversible Processes 31

Fig. 2.2 Isotherms in an

intermediate time of the
abrupt expansion

Figure 2.2 shows a sketch of the isotherms in an intermediate time of the abrupt
expansion. Temperature can reach very low values near the center of the gas chamber
and the transfer of heat from the wall surface to the inner core is dependent on the gas
thermal conductivity and on the advection currents that are formed with the piston
displacement. Since the thermal conductivity of gases is very small, the attainment
of thermal equilibrium can be a very slow process.
Let us try to get a deeper understanding of what happens between the initial and
the final equilibrium states, by sketching the time evolution of the pressure that would
be measured in two points, just above and below the piston base (Fig. 2.3). So, let. Pext
be the external pressure just above the piston base and . Pint be the internal pressure
just below this base.
The external pressure . Pext falls, instantaneously, from . P1 , the initial pressure in
the gas chamber to . P2 , the final pressure. Nevertheless, the internal pressure . Pint is
not able to follow . Pext , because when the pressure falls just below the piston base, the
remaining gas in the chamber will move upward, trying to restore the pressure. The
time evolution of . Pext and . Pint are pictured in Fig. 2.3. This pressure delay creates
a pressure difference between the outer and inner faces of the piston base. So, the
32 2 The Second Principle

Fig. 2.3 Time evolution of the external and internal pressure just above and below the base plate
of the piston

piston will be accelerated upward when . Pint > Pext . This is the case of point . M in
the figure. When . Pint = Pext , the net force on the piston is null, but the piston has a
finite velocity and will continue to move upward due to its inertia (the Newton’s First
principle of Mechanics). After that, the piston is braked, because . Pint < Pext until
it has a null displacement velocity at point .m, returning back to its final equilibrium
position with an accelerated motion.
It is thus easy to conclude that the piston will have an oscillatory motion around its
final equilibrium position. These oscillations will be damped by two kinds of physical
mechanisms: (a) friction between the piston and the cylinder wall and (b) viscous
forces.
The final equilibrium state is reached when: (a) the heat transferred from the liquid
bath restores the initial temperature.T = 300.K in the whole gas phase; (b) the viscous
forces inside the gas dissipate the advection flows that appear during the expansion,
transforming kinetic energy into internal energy. So, the time that the gas needs to
reach the final equilibrium state is dependent on both its thermal conductivity and
viscosity.

2.1.2 Expansion Work

In classical mechanics, the work .δW that is required for displacing a particle from a
point .x to a point .x + dl is defined as

δW = F · dl,
.
2.1 Reversible and Irreversible Processes 33

where .F is the force that is required for this displacement.

In the present case, this force . F is given by the internal pressure . Pint depicted in
Fig. 2.3 multiplied by the area . A of the cylinder cross-section. On the other hand the
displacement .dl is given by. dl = vdt, where .v is the piston velocity. Both . Pint and
.v are dependent on the time, .t. So, the total work that is released by the gas between
its initial and final equilibrium states will be given by
∫∞
. W1−2 = − A Pint (t) v (t) dt. (2.1)
0

The minus sign means that this work is being released by the thermodynamic
system in the expansion process. In compression processes, when work is added to
the system, it would be positive.
The above expression is a simplified form of the general expression for the transfer
rate of mechanical work, .Ẇ , through a surface . A, moving with a velocity .v (x, t)
∫
. Ẇ = − Pv · nd A,
A

where, in this case, the pressure . P is supposed to vary along the boundary . A and
n (x, t) is the unitary vector normal to the surface at point.x ∈A. In present conditions,
.
the velocity .v(t) in Eq. (2.1) can be found in terms of . Pint (t) by solving the initial
value problem,
dv
m
. = (Pint − Pext ) A, v = 0 for t = 0,
dt
where .m is the mass of the two balance weights placed on the upper part of the
weightless piston (Fig. 2.3).
Finding the time evolution of the internal pressure . Pint (t) is, however, a difficult
task since it requires solving the fundamental thermo-hydrodynamic equations in
their differential form, for a compressible flow problem with a moving boundary.
Nevertheless, it can be argued that what is important in this problem is to take into
account the effective work .W1−2 released by the gas, when elevating the two balance
weights to a level .h = (V2 − V1 ) /A. For displacing the two balance weights, the
thermodynamic system must overcome a resistance force that is given by . P2 A—the
external pressure during the whole expansion multiplied by the area of the cylinder
cross section—and by the friction force between the piston and the cylinder wall.
Friction work is transformed into heat and, in this case, this heat is absorbed by the
heat reservoir without any consequence for the problem that is here being treated.
So we are led to the conclusion that the effective work released by the thermody-
namic system is
. W1−2 = − P2 × (V2 − V1 ) = − 778 J
34 2 The Second Principle

and this is the work needed for elevating the two balance weights and the atmosphere
by a height .h. After the final equilibrium state is reached, the change in the potential
energy of the two balance weights and the atmosphere will be given by

( )
2 × 50 × g
ΔE p = (2 × 50 × g + Patm × A) h =
. + Patm (V2 − V1 )
A
, ,, ,
=P2

So, if we ignore all the complex physical phenomena that happen in the course of
the gas expansion and consider only the net balance of energy.— from the initial to the
final equilibrium states of the gas.— what it is important to take into account in this
expansion process is the change of the potential energy of the external system (the two
balance weights and the atmosphere) due to the work released by the thermodynamic
system.

2.1.3 Quasi-static Isothermal Expansion

Consider now that the three balance weights of Fig. 2.1 are replaced by a great number
of plumb spheres with mass .Δm as shown in Fig. 2.4. The total mass of these spheres
is 150 .kg and the gas expansion is produced by removing these plumb spheres one
after the other. The final equilibrium state is reached when the residual mass on the
top of the piston is 100 .kg.
The initial and final equilibrium states are the same as before and given in Table 2.1
independently of the mass .Δm of each plumb sphere. In the same way, the total
expansion work delivered by the gas can be found by summing all the elementary
work .δW that is delivered when each sphere is removed from the top plate. This
elementary work can be found using the same procedure of Sect. 2.1.2.
It is easy to conclude that, although the initial and final equilibrium states remain
the same as before, the delivered work will be dependent on the mass .Δm and that
it increases in absolute value, approaching an asymptotic value when .Δm becomes
smaller (Fig. 2.5 ).
Consider now the theoretical limiting case when .Δm → 0. In this case, the
removal of a single plumb sphere reduces the external pressure . Pext , but the internal
pressure . Pint will follow this reduction without any meaningful time delay. In addi-
tion, it is possible to assert that this removal will not unbalance the pressure field in
the gas phase in such a manner that the pressure . P in the whole gas phase can be
considered equal to . Pint . So, each elementary work .δW can be calculated as

δW = − Pd V,
.

and the total work .W1−2 as the integral

2.1 Reversible and Irreversible Processes 35

Fig. 2.4 Quasi-static isothermal expansion

Fig. 2.5 Work delivered by the isothermal gas expansion when a certain number of plumb spheres
with mass .Δm are removed, one after the other, from the top of the piston in Fig. 2.4
36 2 The Second Principle

∫V2
. W1−2 = − Pd V.
V1

Since . P is the thermodynamic pressure, it can be replaced by the equation of state

of an ideal gas

N RT
. P= ,
V
resulting in the following expression for the total work
V2
. W1−2 = − N RT ln .
V1

In the present case, using the data given in Table 2.1, the total work can be cal-
culated as .W1−2 = − 932 .J. So, since the delivered work .W1−2 = − 930 .J when the
mass .Δm of each plumb sphere is .0.5 .kg, it is possible to say that with this .Δm, we
are very close to the asymptotic case .Δm → 0.
What is more important to be retained from this section is that the expansion work
can be only calculated as

∫V2
. W1−2 = − Pd V.
V1

when all the states between .V1 and .V2 are equilibrium states, because we are only
authorized to replace the external pressure . Pext by the gas pressure . P when the gas
is in thermodynamic equilibrium.

2.1.4 Inverse Process. The Cycle

Consider, now, the inverse process, when the plumb spheres are repositioned, one
after the other, on the top of the piston till the initial equilibrium state is restored.
When .Δm is finite, the gas will suffer a compression process with an appreciable
unbalance in their fields of temperature, pressure and density and we will now have
a compression wave traveling from the top of the gas chamber, with the speed of
sound.
When .Δm = 50 .kg, the gas will be compressed with a constant external pressure
. P1 =1.5728 .MPa and, by using the same reasoning of Sect. 2.1.2, the corresponding
compression work that is added to the gas will be calculated as
2.1 Reversible and Irreversible Processes 37

Fig. 2.6 Compression work when sequentially adding plumb spheres of mass .Δm to the top of the
piston

. W2−1 = − P1 × (V1 − V2 ) = 1130 J.

The compression work .W2−1 can be found for different .Δm in the same way as it
was done for the expansion work. Figure 2.6 shows that .W2−1 tends asymptotically
to the quasi-static work .W2−1 = 932 .J when .Δm → 0.
Consider now the entire cycle 1-2-1 when the two limiting cases are compared.
Figure 2.7 shows a P-V equilibrium diagram representing the two processes along the
300 .K isotherm. The abrupt process is represented by a dashed line because only the
states 1 and 2 are equilibrium states. All the intermediate states are non-equilibrium
states and cannot be represented in this equilibrium diagram. When .Δm → 0, in

Fig. 2.7 P-V diagram of the entire cycle 1-2-1. For the abrupt process, only the states 1 and 2 are
equilibrium states. For the quasi-static process, when .Δm → 0, in addition to these terminal states,
all the intermediate states are equilibrium states
38 2 The Second Principle

addition to these terminal states, all the intermediate states are equilibrium states.
So, quasi-static processes are represented by a continuous line.
Let us now calculate the net balance of work when the gas completes each full
cycle 1-2-1.
In the quasi-static cycle, both the expansion and compression work are the same
in absolute value, .|W1−2 | = 932 .J. So the net balance of work will be
∮
. δWquasi−static = −932
, ,, , + ,,,,
932 = 0.
expansion compression

On the other hand, for the limiting abrupt process, when a mass .Δm = 50kg is
removed from the top and repositioned on the top of the piston,
∮
. δWabr upt = ,−778
,, , + 1130
, ,, , = 352 J.
expansion compression

Therefore, while each quasi-static cycle can be completed without any expense
of work, the abrupt cycle requires an expense of 352 .J of work to be completed. This
is a general rule for thermodynamic cycles that are run in thermal contact with a
constant temperature heat reservoir,
∮
. δW ≥ 0,

meaning that, independently of the mass .Δm that is used for performing the cycle,
a certain amount of work must always be added to the system to complete the cycle.
This added work will be only null in the limit .Δm → 0.
So, while quasi-static processes are reversible, abrupt processes are irreversible
and by irreversible we mean a process that requires a certain amount of work to be
reversed—and not that such a process cannot be reversed.
Irreversible processes are easy to recognize because they always present a spatial
imbalance in their thermodynamic fields. So, real processes are always irreversible,
because the transition of a given state to another are always accompanied by the
formation of internal non-equilibrium fields.

2.1.5 Exercises

1. Show that in every gas cycle, 1-2-1 ran in contact with a thermal pond at constant
temperature the compression work .W2−1 is always greater in absolute value than
the expansion work .W1−2 .
2. Considering that the liquid phase will remain at 90.◦ C, due to its high thermal
capacity, explain the physical mechanisms that will heat the gas from 20–90.◦ C
2.1 Reversible and Irreversible Processes 39

in Fig. 2.8, after the insulation plate separating the liquid from the gas phase is
removed. Is this heating process a reversible or an irreversible process? Why?

Fig. 2.8 Isovolumetric heating

Further reading: Rayleigh–Bénard convection is a type of natural convection,

occurring in a plane horizontal layer of fluid heated from below, in which the fluid
develops a regular pattern of convection cells known as Bénard cells. Rayleigh–
Bénard convection is one of the most commonly studied convection phenomena
because of its analytical and experimental accessibility. The convection patterns
are the most carefully examined example of self-organizing nonlinear systems.
There are several papers and internet articles written about the Rayleigh–Bénard
convection. The Landsat image in Fig. 2.9 shows a pattern of stratocumulus cloud
cells, each about 10–15 km over the Pacific Ocean. As warmer moist air rises in
convection cells over the ocean and cools, condensing the water vapor into cloud
droplets—which usually coalesce to form clouds—cold air then sinks around the
sides of the cells (from https://earthobservatory.nasa.gov/images/2387/closed-
small-cell-clouds-in-the-south-pacific).

Fig. 2.9 Cloud cells formation by Rayleigh-Bénard convection over the Pacific Ocean. Each cell
has 10–15 km characteristic length (from https://earthobservatory.nasa.gov/images/2387/closed-
small-cell-clouds-in-the-south-pacific). Image courtesy NASA/GSFC/LaRC/JPL, MISR Team
40 2 The Second Principle

3. One mole of an ideal gas is in equilibrium inside a cylinder with a 10 .cm2 straight
section (Fig. 2.10). The system is under the action of atmospheric pressure and
one 50 .kg balance weight and in thermal contact with a heat reservoir at 300 .K.
Consider now that we place two others 50 .kg balance weights on the top plate
of the piston. (a) Find the final equilibrium state of the gas; (b) try to explain
what happens with . Pext and . Pint in the course of time and detail the physical
mechanisms that lead the gas to a final equilibrium state; (c) calculate the effective
compression work between the initial and final equilibrium states.

Fig. 2.10 Isothermal abrupt

compression

2.1.6 Work and Heat Conversion. The Role of Dissipation

It is now time to go back to Sect. 1.4 and to the first principle of thermodynamics.
When a gas expands it cools, but if it is in thermal contact with a heat reservoir at a
constant temperature, this reservoir will supply the heat that is needed to restore its
equilibrium temperature.
If the gas is an ideal one, its internal energy is only dependent on the temperature
and so, by integrating the first principle of thermodynamics, Eq. (1.4), between the
equilibrium states 1 and 2,

. Q 1−2 + W1−2 = U2 − U1 = 0.

It was found in Sect. 2.1.2 that, when a weight of .50 .kg is retired from the top plate
of the piston, the work in the irreversible expansion was .W1−2 = − 778 .J, the heat
absorbed by the gas will be . Q 1−2 = 778 .J. The conclusion is that, in this expansion
process, heat is entirely converted into work, i.e., that the work delivered by the gas
2.1 Reversible and Irreversible Processes 41

results from the withdraw of the equivalent amount of heat from the reservoir. On the
other hand, when the mass of the balance weight .Δm → 0 and the gas expansion is
reversible the heat absorbed by the gas and transformed into work is . Q 1−2 = 932 J.
The question we must answer is why the amount of heat that is absorbed by the
gas in an irreversible expansion is .158J smaller than the heat that is absorbed in a
reversible expansion? In other words, why in irreversible gas expansions in contact
with a thermal pond the gas requires less amount of heat to reestablish its equilibrium
temperature?
An irreversible expansion such as the one of Sect. 2.1 is always accompanied by
advection currents with the formation of a velocity field, in addition to the temperature
field sketched in Figure 2.2, picturing a complex non-equilibrium state. For restoring
its final equilibrium state these velocity and temperature fields must be destroyed.
This is achieved by the heat transfer from the wall and, also, by viscous dissipation
which transforms the kinetic energy of the advection currents into internal energy.
Viscous dissipation is an intrinsically irreversible process and is always responsible
for the transformation of mechanical energy into heat (or internal energy).
The result is that in irreversible expansions the thermodynamic system does not
require the same amount of heat from the thermal pond to attain its final equilibrium
state when compared to a reversible expansion. In this example, this also means that
the .158 J above mentioned corresponds to the mechanical energy of the advection
currents that is dissipated in the course of the irreversible expansion.
On the other hand, considering the inverse compression process .2 − 1, the com-
pression work is now .W2−1 = 1130 .J, when the compression is irreversible. So the
gas needs to deliver . Q 2−1 = − 1130 .J to the reservoir to restore its equilibrium tem-
perature, an amount that is.198J greater in absolute values than the one corresponding
to the reversible compression. In compression processes, work is converted into heat,
and the surplus of .198J corresponds to the mechanical energy of the advection cur-
rents that is dissipated, i.e., transformed into heat and delivered to the thermal pond.
Figures 2.11 and 2.12 summarizes the energy balance for this isothermal cycle
when a weight of .50 .kg is retired and put it back again on the top plate of the piston.
From Figs. 2.1 and 2.10 and related comments in Sect. 2.1, an important conclusion
is that, in isothermal cycles,

Fig. 2.11 Energy balance

for the isothermal expansion
of section
42 2 The Second Principle

Fig. 2.12 The role of

dissipation

∮ ∮
. δW ≥ 0 and δ Q ≤ 0,

independently of the way the gas expansion and compression are performed, the
equality being restricted to reversible isothermal cycles.
Therefore, when a thermodynamic system follows an irreversible cycle and returns
to its initial state at the same temperature, we will always have the conversion of work
into heat, and not the inverse.
Indeed, the natural tendency of physical phenomena to transform mechanical
energy into heat (a dissipation process) was a major objection to the equivalence
principle of work and heat. This question will be seen in more detail in the following
sections.
In isothermal cycles, there is, always, a conversion of work into heat and, never,
the inverse.

2.1.7 Exercises

1. One mole of an ideal monatomic gas is in equilibrium inside a cylinder with a 10

2
.cm straight section (Fig. 2.13). The system is under the action of the atmospheric
pressure and three 50 .kg balance weights. Thermal insulation separates the gas
from the outer ambient. The initial temperature is 300 .K. Consider now that
we remove one of the three 50 .kg balance weights from the top plate of the
piston. (a) Considering the friction work between the piston and the cylinder
wall, is it possible to find the final equilibrium state of the gas? (b) try to explain
what happens with . Pext and . Pint in the course of time and detail the physical
mechanisms that lead the gas to a final equilibrium state; (c) calculate the final
temperature of the gas, when the friction work can be neglected; (d) what would
be the influence of the friction work on the final temperature? (e) what would
be the maximum temperature at the final equilibrium state? (f) calculate the
expansion work between the initial and final equilibrium states, when the friction
work is neglected.
2.1 Reversible and Irreversible Processes 43

Fig. 2.13 An irreversible

adiabatic expansion

2. In the previous exercise, consider now that we have replaced the three 50 .kg
balance weights by 150 .kg of plumb spheres (Fig. 2.14). Consider also that the
mass of each sphere .Δm is, alternatively, 10, 5, 1, and 0.5 .kg. By neglecting the
friction work between the piston and the cylinder wall, calculate the final equi-
librium temperature of the gas and the delivered work when 50 .kg are removed
from the piston top plate. (a) Plot the expansion work in terms of the mass of the
plumb spheres and calculate the expansion work when .Δm → 0. What would
be the friction work in this last case?
3. In Exercise 2 what is the minimum temperature that is attained by the gas after the
adiabatic expansion? Why? In adiabatic expansions, the delivered work is given
by the internal energy variation between the initial and final equilibrium states.
Try to explain why the internal energy is not entirely converted into useful work
in irreversible gas expansions. The internal energy is often called the free—or
available—energy in adiabatic expansions. Try to explain the reasons for this
designation.
4. Calculate the temperature and the expansion work for .Δm → 0 in Exercise 2,
when the gas is pure nitrogen. The molar heat capacity of Nitrogen was found
as
( )
J
.c̄v = 20.586 − 0.1571 × 10−2 T + 0.8081 × 10−5 T 2 − 2.8730 × 10−9 T 3 ,
mol K
44 2 The Second Principle

Fig. 2.14 Adiabatic

expansion produced by the
removal of balance weights
of mass .Δm

in the range 273–1800.K,1 with an average deviation with respect to experimental

data of 0.34%. What would be these values if the molar heat capacity of nitrogen
at 273 .K, .c̄v = 20.701 . J/ (mol K), was used in the calculations?
5. Consider that the gas in Exercise 2 is in its final equilibrium state, after a reversible
adiabatic expansion .(Δm → 0). By neglecting the friction work, calculate the
final temperature and the compression work when its initial pressure is restored
by successively adding balance weights with 10, 5, 1, and 0.5 .kg.
6. The first principle is usually written as
∮
. (δ Q + δW ) = 0.

Consider the two adiabatic expansions of Fig. 2.15. Both expansions start from
the same initial equilibrium state and finish on the same final volume .V2 . Expan-
sion 1-2 is irreversible while expansion 1-3 is reversible. Use the first principle
as given above to show that the work delivered in the reversible expansion is
necessarily greater than in the irreversible expansion when their absolute values
are compared, .|W1−3 | ≥ |W1−2 |.

7. Show that, when the molar heat capacities .c̄ p and .c̄v are considered constants,

. Pvγ = const.,

1 B. G. Kyle, Chemical and Process Thermodynamics (Englewood Cliffs, NJ: Prentice-Hall, 1984).
2.1 Reversible and Irreversible Processes 45

Fig. 2.15 Using the first

principle of thermodynamics
to show that the work .W1−3
delivered in a reversible
adiabatic expansion is
greater (in absolute value)
than the work .W1−2

in adiabatic and reversible expansions of ideal gases. Parameter .γ is given by

γ = c p /cv . In the sequence, show that the adiabatic reversible lines and the
.
isotherms are both hyperboles, but the adiabatic lines are more inclined than the
isotherms.
8. A gas is compressed adiabatically from an initial equilibrium state 1 to a final
equilibrium state 2 (Fig. 2.17). Show that, independently of the way the com-
pression is performed, the compression work .|W1−2 |, in absolute value, will be
always greater than the compression work .|W1−3 | in a reversible process, with
the same compression rate. (Hint: consider an isotherm starting from state 2 and
crossing the extension of the reversible line . Pvγ = const. that joins states 1 and
3 and use the first principle).
9. A cylinder is divided into two equal-volume chambers A and B. Each chamber
contains 1 mole of Helium initially at 20.◦ C (Fig. 2.16). Chambers A and B are
separated by a thermally insulated plate, free to move inside the cylinder, in such
a manner as to equalize the pressures in the two chambers. The base plate of
the cylinder is a thin thermal contact plate. Consider now that this base plate is
put in contact with a thermal reservoir at 200.◦ C. The Helium in chamber B will
expand moving the dividing plate upward.

Fig. 2.16 A Rayleigh-Bénard cellular convection heats the Helium in the chamber B. Nevertheless,
the adiabatic compression of the Helium in the chamber A is a very slow process and supposing
this compression to be reversible can be a fair hypothesis
46 2 The Second Principle

Fig. 2.17 Using the first principle of thermodynamics to show that the work .W1−3 delivered in a
reversible adiabatic compression is smaller (in absolute value) than the work .W1−2

Try to picture the heating process of the Helium in chamber B. A Rayleigh-

Bénard cellular convection heats the Helium in chamber B and this is clearly an
irreversible process. Nevertheless, the adiabatic compression of the Helium in
chamber A is a very slow process in such a way that supposing this compression
to be reversible can be a fair hypothesis. With this assumption: (a) Find the final
equilibrium state of the system; (b) calculate the work released by the Helium
in chamber B to compress the Helium in chamber A; (c) calculate the heat that
is withdrawn from the thermal reservoir at 200.◦ C between the initial and final
equilibrium states of the Helium in the two chambers.
10. A thermodynamic system is composed of two chambers A and B in thermal
contact with a thermal reservoir at 300 K. The two chambers are separated by a
metallic plate free to move. Each one of these chambers is filled with 1 mole of
an ideal gas, but the pressure in A is, initially, greater than the pressure in B due
to two pins that maintain the separating plate at the position shown in Fig. 2.18.
Consider now that the pins are removed: (a) Find the final equilibrium state of the
gas in chambers A and B; (b) try to describe the physical phenomena that happen
in this process; in particular, try to picture the movement of the separating plate
with the time; (c) since the gas in chamber A suffers an expansion process, it
will receive heat from the heat reservoir, . Q A−R and from chamber B, . Q A−B ; in
the inverse sense the gas in chamber B will be compressed and will release heat
to the reservoir, . Q B−R and chamber A, . Q B−A . We know that . Q A−B = − Q B−A ,
but what can we say about . Q A−R and . Q B−R ? d) in this process, is it possible
to calculate the expansion work .W A−B that is delivered to compress the gas in
chamber B?
2.2 The Second Principle of Thermodynamics 47

Fig. 2.18 A thermodynamic

system composed of two
chambers A and B in thermal
contact with a thermal
reservoir

2.2 The Second Principle of Thermodynamics

La production de la puissance motrice est donc due, dans les machines à vapeur, non à une
consommation réelle du calorique, mais à son transport d’un corps chaud à un corps froid
(....) D’après ce principe, il ne suffit pas, pour donner naissance à la puissance motrice, de
produire de la chaleur: il faut encore se procurer du froid; sans lui la chaleur serait inutile.
Et en effet, si l’on ne rencontrait autour de soi que des corps aussi chauds que nos foyers,
comment parviendrait-on à condenser la vapeur?…. (Sadi Carnot, 1824 [27])

Denis Papin (1647—1712) was a French physicist, mathematician and inventor,

best known for his pioneering invention of the first prototype of a steam engine in
1691. Papin’s engine, Fig. 2.19, used the vapor, produced by heating water, to move
a piston upward inside a cylindrical chamber. When the piston was in the top of this
chamber, cold water was admitted into the jacket around the chamber condensing
the vapor. The vacuum created by vapor condensation pulled the piston back to its
lower position. The refrigerating water was then stopped from entering the jacket,
allowing the cycle to be restarted.
Thermal machines such as Papin’s engine, working in cycles can be represented
by using a simplified scheme like the one shown in Fig. 2.20.
Although a very simplified sketch of a real thermal engine, this sketch represents
the main processes that may concern us when performing a thermodynamic analysis
of such engines. The first thing that it is important to observe is that a source of
heat—the heat from the combustion of fuel—and a heat sink—the cold water—are
both needed if one wants the engine to work in a cycle. In this sketch, the heat source
is represented by a heat reservoir at a temperature .TH , related to the combustion
temperature where the combustion gases are sensed to be after the fuel reacts with the
atmospheric air. The heat sink is, here, represented by a heat reservoir at a temperature
. TL , related to the temperature of the cold water that was used to condense the vapor.

In each cycle, the combustion process releases an amount of energy . Q H to the

cycle, which is the amount of energy that is needed to produce vapor and to heat this
48 2 The Second Principle

Fig. 2.19 Papin steam

engine

Fig. 2.20 A simplified

scheme for a thermal engine

vapor until the vapor pressure inside the chamber is high enough to move the piston
upward.
After the principle of energy conservation was established in the first half of the
nineteenth century and heat was recognized as a kind of energy,2 what was considered
to happen was the partial conversion of . Q H into mechanical work .W , a part of . Q H
being used to cool the vapor bringing it to its initial state. This amount . Q L represents
the heat that must, necessarily, be lost if one wants the engine to work in a cycle.
The total amount of energy that is exchanged by the thermodynamic system, i.e.,
by the water inside the chamber, must be null after each cycle. This is the principle
of energy conservation—the first principle of thermodynamics—and means that the
output work .W must be, in absolute value, smaller than . Q H . So the thermodynamic
efficiency .η of the cycle must, necessarily, be smaller than 1,
|W | W
η=
. =− < 1, (2.2)
|Q H | QH

2 In the early years of thermodynamics it was thought that a heat engine was like a hydraulic turbine,

the flow of caloric from the heat source to the heat sink being responsible to put the engine to work
as water makes a hydraulic turbine to work when it flows from a higher to a lower level.
2.2 The Second Principle of Thermodynamics 49

Fig. 2.21 Demonstration that it is impossible to conceive a heat engine that entirely converts heat
into mechanical work. In a it is supposed that a thermal engine with .η = 1 can be used to drive a
heat pump operating between the same thermal reservoirs. In b an energy balance is performed for
the composite system

because, from the first principle,

∮
. (δ Q + δW ) = Q H + Q L + ,,,,
W = 0,
,,,, ,,,,
>0 <0 <0

and the heat loss . Q L is, necessarily, different from zero.

Is there any thermal engine with an energy conversion efficiency .η = 1?
The answer was given by Rudolf Clausius in 1850 [34]: If it is possible to conceive
a thermal engine, operating in a cycle, that entirely converts, heat into mechanical
power, without any loss, it would, also, be possible to transfer heat from a cold to a
hot body without any supply of mechanical work.
Indeed, if it is possible to conceive a thermal engine with a conversion efficiency
equal to 1, this thermal engine can be used to drive a heat pump operating between the
same thermal reservoirs, as sketched in Fig. 2.21a. Following the operating principle
of all heat pumps, this heat pump withdraws the amount . Q L from the cold reservoir
and delivers . Q L + W to the hot reservoir.
Consider now the composite system, i.e., the set ‘thermal engine + heat pump’.
Since the work .W is delivered by the heat engine to the heat pump, it becomes an
internal transfer of energy of the composite system and consequently does not appear
in the energy balance of Fig. 2.21b. Now, the work .W has the same absolute value of
. Q H , because the conversion efficiency of the power engine was supposed as .η = 1.
Therefore the heat that will be delivered to the hot reservoir by the composite system
is given by

. |Q L | + |W | − |Q H | = |Q L |

This result is shown in Fig. 2.21b and demonstrates the impossibility of conceiving
a thermal engine with unitary conversion efficiency .η = 1, since heat cannot be
transferred from a cold to a hot body without any supply of work.
50 2 The Second Principle

2.2.1 Statements of the Second Principle

2.2.1.1 Clausius Formulation

The German scientist Rudolf Clausius (1822–1888) laid the foundation for the second
principle of thermodynamics in 18503 by examining the relation between heat trans-
fer and work. In this 1850 paper, he showed that there was a contradiction between
Carnot’s principle and the concept of conservation of energy. Clausius restated the
two principles of thermodynamics to overcome this contradiction. His version of the
second principle, which was published in German in 18544 reads as follows:
Heat can never pass from a colder to a warmer body without some other change, connected
therewith, occurring at the same time.

He explains this principle: “Everything we know concerning the interchange of

heat between two bodies of different temperature confirms this; for heat everywhere
manifests a tendency to equalize differences of temperature, and therefore to pass,
from warm to cold bodies. Without further explanation, therefore, the truth of this
principle will be granted. The principle may be more briefly expressed thus: Heat
cannot by itself pass from a colder to a warmer body; the words by itself, however,
here requires explanation. Their meaning will, it is true, be rendered sufficiently
clear by the exposition contained in the present memoir, nevertheless, it appears
desirable to add a few words here in order to leave no doubt as to the signification
and comprehensiveness of the principle.
The Clausius principle can be stated in a graphical form as shown in Fig. 2.22.
The Clausius principle can be better understood from the sketch given in Fig. 2.23.
The natural arrow of heat is from the hot to the cold bodies and one wants to pump
heat from a cold to a hot body, using a thermodynamic system working in a cycle.
For this purpose, one needs to add the work .W taken, for instance, from a heat engine
that works, e.g., using the hot body as its source of heat . Q 'H and the cold body as its
heat sink to which it will release the heat . Q 'L . We do not know what the values . Q 'H
and . Q 'L are to be, individually, but we must assure . Q 'H + Q 'L = − W . Consider now
a composite system, i.e., a system composed of the heat pump and the power engine.
Since the mechanical work that is released by the power engine will be used as a shaft

3 Clausius R (1850) Ueber die bewegende Kraft der Wärme und die Gesetze, welche sich daraus
für die Wärmelehre selbst ableiten lassen. Ann der Phys und Chemie 155:500–524. https://doi.org/
10.1002/andp.18501550403 [34]Translated into English: Clausius, R. (July 1851). On the Moving
Force of Heat, and the principles regarding the Nature of Heat itself which are deducible therefrom.
London, Edinburgh and Dublin Philosophical Magazine and Journal of Science. 4th 2 (VIII): 1–21;
102–119.
4 Clausius R (1854) Ueber eine veränderte Form des zweiten Hauptsatzes der mechanis-

chen Wärmetheorie. Ann der Phys und Chemie 169:481–506. https://doi.org/10.1002/andp.

18541691202 [35]. Translated into English: Clausius, R. (July 1856). On a Modified Form of the
Second Fundamental Theorem in the Mechanical Theory of Heat. London, Edinburgh and Dublin
Philosophical Magazine and Journal of Science. 4th 2: 86. Reprinted in: Clausius, R. (1867). The
Mechanical Theory of Heat—with its Applications to the Steam Engine and to Physical Properties
of Bodies. London: John van Voorst [36].
2.2 The Second Principle of Thermodynamics 51

Fig. 2.22 Clausius principle

in graphical form. Heat can
pass from a warmer to a
colder body (a), but, in the
words of Clausius, cannot by
itself pass from a colder to a
warmer body (b)

Fig. 2.23 The natural arrow

of heat

work to the heat pump compressor, the net effect will |be the | flow of heat following
its natural arrow from the hot to the cold body, when .| Q 'H | ≥ |Q H |, otherwise, we
would break a fundamental
| | principle, since heat must flow from a hot to a cold body.
The limiting case .| Q 'H | = |Q H | means that the amount of heat that is to be delivered
to the heat engine can be furnished by the heat pump itself, This is only possible for
theoretical engines operating under reversible cycles.
Therefore, when a heat pump is built to transfer heat from a colder to a warmer
body, what can be concluded from Fig. 2.23 is that the net effect will be a transfer of
heat from the warmer to the colder body.

2.2.1.2 Planck Formulation

Max Karl Ernst Ludwig Planck (1858–1947) was a German theoretical physicist
whose work on quantum theory won him the Nobel Prize in Physics in 1918. Planck
stated the following proposition as derived directly from experience. This is often
considered as his statement of the second principle, but he regarded it as a starting
point for the derivation of the second principle.5
It is impossible to construct an engine that will work in a complete cycle, and produce no
effect except the raising of a weight and cooling of a heat reservoir.

For a better understanding of the Planck statement of the second principle, consider
Fig. 2.24 which represents a thermal engine which works in a cycle in contact with
a single heat reservoir.

5 Planck, M. (1903). Treatise on Thermodynamics. Published in German in 1897. Translated to

English by Alexander Ogg. London: Longmans, Green & Co. This book was also edited by Dover
[136].
52 2 The Second Principle

Fig. 2.24 Graphical

formulation of the Planck
statement

In fact, although the full conversion of work into heat is always possible
(Fig. 2.24a, the sketch shown in Fig. 2.24b represents a thermal engine with a full
conversion of heat into work, i.e., with a thermal efficiency .η = 1, without any other
effect except the raising of a weight (work) and cooling of a (single) heat reservoir.
As we have seen in Sect. 2.1.6, this statement can also be written as
∮ ∮
. δW ≥ 0 and δ Q ≤ 0,

for a thermodynamic system that performs a cycle in contact with a single heat
reservoir.

2.2.1.3 Kelvin Formulation

Before talking about Kelvin’s statement for the second principle, it is perhaps inter-
esting to read a text from William Thomson, the Lord Kelvin, extracted from its 1851
famous paper ‘On the dynamical theory of heat; with numerical results deduced from
Mr. Joule’s equivalent of a thermal unit and M. Regnault’s observations on steam’.6
In his words: I believe the tendency in the material world is for the motion to
become diffused, and that as a whole the reverse of concentration is gradually going
on—I believe that no physical action can ever restore the heat emitted from the sun,
and this source is not inexhaustible; also that the motions of the earth and other
planets are losing vis viva which is converted into heat; and that although some vis
viva may be restored for instance to the earth by heat received from the sun, or by
other means, that the loss cannot be precisely compensated and I think it probable
that it is under compensated.

6 Thomson, W. (1851). On the Dynamical Theory of Heat, with numerical results deduced from Mr
Joule’s equivalent of a Thermal Unit, and M. Regnault’s Observations on Steam. Transactions of
the Royal Society of Edinburgh XX (part II): 261–288 [172].
2.2 The Second Principle of Thermodynamics 53

Thomson was reluctant to accept Joule’s ideas of the equivalence between heat
and work. In an 1848 paper7 he wrote that the conversion of heat (or caloric) into
mechanical effect is probably impossible, certainly undiscovered.
In his 1851 publication, Thomson retreated from a radical departure and declared
the whole theory of the motive power of heat is founded on two propositions, due
respectively to Joule, and to Carnot and Clausius. Thomson went on to estate a form
of the second principle:
It is impossible, by means of inanimate material agency, to derive mechanical effect from
any portion of matter by cooling it below the temperature of the coldest of the surrounding
objects.

This statement is followed by a footnote at the bottom of page 265 of his paper:
If this axiom be denied for all temperatures, it would have to be admitted that a self-acting
machine might be set to work and produce a mechanical effect by cooling the sea or earth,
with no limit but the total loss of heat from the earth and sea or, in reality, from the whole
material world.

Kelvin’s statement is considered to be analogous to the one enunciated by Planck

and, in some textbooks, the Planck statement is called as ‘Kelvin-Planck statement’.
Figure 2.25 helps to understand this analogy, based on the footnote of Kelvin’s prin-
ciple: if it would be possible for a boat to use the heat pumped from the ocean to
generate the vapor for powering the boat steam engine, the composite system ‘steam
engine + heat pump’ would be able to fully convert heat into mechanical work as
shown in the figure and this is in contradiction with Planck’s statement of the second
principle.

Fig. 2.25 Kelvin statement

of the second principle

7 Thomson W (1848) On an Absolute Thermometric Scale founded on Carnot’s Theory of the

Motive Power of Heat, and calculated from Regnault’s Observations. Proc. Camb. Philol. Soc.
[171]. Reprinted in Thomson, William (1882). Mathematical and Physical Papers. Cambridge,
England: Cambridge University Press. pp. 100–106 [173].
54 2 The Second Principle

2.2.1.4 The Arrow of Natural Phenomena

The Clausius and Plank statements can be shown to be equivalent (see Exercise 1
in Sect. 2.2.2). Two assertions are equivalent when the second can be proved from
the first and vice-versa. Nevertheless, in Clausius statement the focus is the arrow of
heat, from warmer to colder bodies and not from colder to warmer bodies, and the
Planck statement is focused on the direction of the energy conversion: work to heat
and not heat to work.
What do Clausius and Planck’s statements have in common?
Figures 2.22 and 2.24 try to clear this question. Without any supply of mechanical
work, heat will flow from warmer to colder bodies and never from colder to warmer
bodies. When a thermodynamic system performs a cycle in contact with a single heat
reservoir it will convert work into heat and never heat into work.
Nature has an arrow and this arrow is what the two statements have in common.

2.2.2 Exercises

1. Show that “If it is possible to transfer heat from a colder to a warm body without
any supply of work, then it is also possible to conceive a heat engine working in
a cycle that fully converts heat into work”. Showing this assertion is the same as
showing that when the Clausius statement of the second principle is denied, then
the Planck statement is also denied or, in other words, that the Planck statement
is a particular case of the Clausius statement.
In Fig. 2.21 it is shown that when the Planck statement of the second principle is
denied, then the Clausius statement is also denied or that If it is possible to conceive
a thermal engine, operating in a cycle, that entirely convert, heat into mechanical
power, without any loss, it would, also, be possible to transfer heat from a cold
to a hot body without any supply of mechanical work. This last assertion means
that the Clausius statement is a particular case of the Planck statement. When two
assertions A and B are shown to be particular cases one of each other, i.e., A.⊆B
and B.⊆A, they are equivalent or A.≡B.
2. Constantin Carathéodory (1873–1950) formulated thermodynamics on a purely
mathematical axiomatic foundation. His statement of the second principle is
known as the Principle of Carathéodory, which may be formulated as follows8 :
In every neighborhood of any state S of an adiabatically enclosed system there
are states inaccessible from S.
In Fig. 2.26, show that if the state . I is accessible from state . S by an adiabatic
expansion, then the second principle will be contradicted.

8Carathéodory, C. Untersuchungen Über die Grundlagen der Thermodynamik. Mathematische

Annalen 67 (3): 355–386. [26].
2.3 Sadi Carnot and Émile Clapeyron 55

Fig. 2.26 Caratheodory

principle

Fig. 2.27 Adiabatic and

reversible lines do not
intercept

3. In Fig. 2.27, show that two different lines representing an adiabatic reversible
process in a P-V equilibrium diagram will never have an equilibrium state in
common (Hint: Suppose that these two lines have a common point and show that
this assumption contradicts the second principle).
4. A perpetual motion engine of the second kind is an engine that although does not
contradict the principle of energy conservation, it denies the second principle of
thermodynamics. Is the thermodynamic system shown in Fig. 2.28 a ‘perpetual
motion engine of the second kind’?

2.3 Sadi Carnot and Émile Clapeyron

Sadi Carnot (1796–1832) was a French engineer who lived in a time when the steam
engine was founding rapid growth in Europe, after the revolutionary steam engine of
James Watt, patented in 1769. Carnot’s ideas about the maximum motive power that
is possible to extract from a heat engine were described in his book Reflexions sur
la puissance motrice du feu et sur les machine propes á développer cette puissance
published in 1824. Carnot died of cholera eight years later and, one year after his
death, his book attracted the attention of Émile Clapeyron who made it known to
56 2 The Second Principle

Fig. 2.28 Perpetual motion engine

the scientific community. In 1834, Émile Clapeyron reproduced the main ideas of
Carnot in the Journal of the l’École Polytechnique [33].
In the page 154 of his ‘Mémoire’ Clapeyron writes:
Je citerai enfin parmi les travaux qui ont paru sur la théorie de la chaleur, un ouvrage de M.
S. Carnot , publié en 1824, sous le titre de Réflexions sur la puissance motrice du feu. L’idée
qui sert de base à ses recherches me paraît féconde et incontestable, ses démonstrations sont
fondées sur l’absurdité qu’ily aurait à admettre la possibilité de créer de toutes pièces de la
force motrice ou de la chaleur.
Voici l’énoncé de plusieurs théorèmes auxquels le conduit cette méthode nouvelle de raison-
nement.
1- Lorsqu’un gaz passe, sans changer de température, d’un volume et d ’une pression déter-
minés, á un autre volume et á une autre pression également déterminés, la quantité de
calorique absorbée ou abandonnée est toujours la même, quelle que soit la nature du gaz
choisi comme sujet d’expérience.
2- La diférence entre la chaleur spécifique sous pression constante, et la chaleur spécifique
sous’volume constant, est la même pour tous les gaz.
3- Lorsqu’un gaz varie de volume sans changer de température les quantités de chaleur
absorbées ou dégagées par ce gaz sont en progression arithmétique, si les accroissements ou
réductions de volume sont en progression géométrique.

In his paper, Clapeyron gave a graphical representation of Carnot’s reversible

cycle in a PV diagram, shown in Fig. 2.29.
Carnot reversible cycle deals with a gas—an ideal gas that follows the Clapeyron
equation of state—and is composed of two isothermal and two adiabatic processes
(Fig. 2.30a). For an isothermal expansion to be reversible it must be quasi-static and
the gas and the heat source must have the same temperature, say .TH . Otherwise, it
2.3 Sadi Carnot and Émile Clapeyron 57

Fig. 2.29 a Front page of

the Clapeyron 1834 article
Mémoire sur la puissance
motrice de la chaleur; b
graphical representation by
Clapeyron of the Carnot
reversible cycle at page 190

will be irreversible (see Sect. 2.1). During this expansion, the gas absorbs the amount
Q H from the heat source. The next process is a quasi-static adiabatic expansion that
.

will cool the gas from .TH to the temperature of the cold reservoir .TL . The heat . Q L
is then delivered to the cold reservoir during the quasi-static isothermal compression
of the gas, in thermal contact with this reservoir at .TL . Finally, the gas is compressed
in a quasi-static adiabatic process and is heated from .TL to .TH .
The cycle of reversible engines can be reverted without changing the the absolute
value of the heat and work that are exchanged.
In Carnot’s ideas:
Motive power can be produced wherever there exists a difference of temperature and the
maximum motive power can be extracted from the heat (caloric) that flows from the hot to
the cold reservoir, when the power cycle used for that is reversible.

This is the statement of the Carnot’s first theorem.

A reversible engine is one that can be reverted to work as a heat pump without
affecting the amounts of heat and work that are exchanged (Fig. 2.30b).
Carnot used the caloric theory to explain that a reversible engine must produce
the maximum amount of work.
Consider that there is an engine operating in accordance with a power cycle that is
more efficient than another engine that we suppose is capable of following a Carnot
reversible engine, the two engines operating between the same heat reservoirs at .TH
and .TL . Since we want to compare efficiencies and not power, we will suppose that
the heat delivered to both engines is the same . Q H (2.31a).
58 2 The Second Principle

Fig. 2.30 The Carnot cycle

as presented by Clapeyron in
its 1834 Mémoire for a
power engine (a) and for a
heat pump (b)

The Carnot engine was supposed to be less|efficient, | so it produces a smaller shaft

work .|Wc | < |W | and has a greater heat loss .| Q L ,c | > |Q L |.
Nevertheless, Carnot’s engine is reversible, so the cycle can be reverted to make
the engine operate as a heat pump delivering the same amount of heat . Q H to the heat
reservoir .TH . Remember that reversible cycles can be reverted without changing the
absolute values of the heat and work that are exchanged. This amount . Q H can be
used to supply the heat that is required by the more efficient power engine, and the
heat reservoir .TH can, thus, be discarded (Fig. 2.31b).
Finally, under the light of the first principle of thermodynamics and by making
an energy balance for the composite system (Fig. 2.31b), we see that this composite
system contradicts the Planck statement of the| second | principle of thermodynamics
because it entirely converts the delivered heat .| Q L ,c | − |Q L | into work .|W | − |Wc |.
In Carnot’s words: ‘This would be not only perpetual motion, but an unlimited
creation of motive power, without the consumption of either caloric or any other
agent whatever.’
Figure 2.31b shows what is now know as Carnot’s first theorem.
Carnot’s second theorem
Carnot’s second theorem may be enunciated as follows:
Two reversible engines operating in accordance with Carnot’s cycle between the same heat
reservoirs, will have the same efficiency.

The demonstration of this assertion follows the same reasoning as given in

(Fig. 2.31) and is left as an exercise (see Exercise 2.3.3)
2.3 Sadi Carnot and Émile Clapeyron 59

Fig. 2.31 If a non-reversible

power engine is more
efficient than a Carnot
reversible engine when both
engines are operating
between the same heat
reservoirs as in (a), then the
reversible engine can be
reverted and pump the
rejected heat . Q H to the
non-reversible engine (b)

2.3.1 Thermodynamic Scale of Temperature

The relation between motive power and heat, as established by Carnot, is such that quantities
of heat, and intervals of temperature, are involved as the sole elements in the expression for
the amount of mechanical effect to be obtained through the agency of heat; and since we
have, independently, a definite system for the measurement of quantities of heat, we are thus
furnished with a measure for intervals according to which absolute differences in temperature
may be estimated. (William Thomson, 18489 )

Following Carnot’ s second theorem, the thermodynamic efficiency .η of a

reversible engine operating in accordance with a Carnot’ s cycle will be only depen-
dent on the temperatures of the heat source and sink.
Since
QL
.η =1+ ,
QH

and supposing that we are using the symbol .θ to represent, e.g., the Celsius scale,
this would mean that,

9 William Thomson. On an Absolute Thermometric Scale founded on Carnot’s Theory of the Motive
Power of Heat, and calculated from Regnault’s Observations. Philosophical Magazine October 1848
[from Sir William Thomson, Mathematical and Physical Papers, vol. 1 (Cambridge University Press,
1882), pp. 100–106.].
60 2 The Second Principle

Fig. 2.32 The

thermodynamic scale of
temperature

QL
. = f (θ L , θ H ) ,
QH

where .θ L , θ H designate the temperatures that a Celsius thermometer would measure

for, respectively, the heat sink and source.
Consider now that a first Carnot’ s engine operates between temperatures .θ H and
an intermediate temperature .θ I , while a second Carnot’ s engine operates between
.θ I and .θ L (Fig. 2.32). Following Carnot’ s second theorem

Q L ,I QL
. = f (θ I , θ H ) , ( ) = f (θ L , θ I ) . (2.3)
QH −Q L ,I

From Carnot’ s second theorem, remark that, for a given . Q H , the heat loss . Q L
is only dependent on .θ H and .θ L and that when these temperatures are also given,
the amount . Q L ,I is only dependent on the intermediate
( ) temperature .θ I : . Q L ,I will
approach . Q L when .θ I becomes close to .θ L and . −Q L ,I will approach . Q H when .θ I
becomes close to.θ H This was the main idea of Kelvin: ’ since we have, independently,
a definite system for the measurement of quantities of heat, we are thus furnished
with a measure for intervals according to which absolute differences of temperature
may be estimated’.
Therefore, from Eq. (2.3)
QL Q L Q L ,I
. = = − f (θ I , θ H ) f (θ L , θ I ) ,
QH Q L ,I Q H

and since
QL
. = f (θ L , θ H ) ,
QH

we see that

. f (θ L , θ H ) = − f (θ I , θ H ) f (θ L , θ I ) .
2.3 Sadi Carnot and Émile Clapeyron 61

The above equation can be only satisfied when its r.h.s. becomes independent of
θ . A possible choice for eliminating this dependence is to write
. I

T (θ I ) T (θ L )
. f (θ L , θ H ) = − .
T (θ H ) T (θ I )

The conclusion is that in a Carnot’ s cycle the ratio between the heat loss . Q L and
(θ L )
the heat gain . Q H is only dependent on the ratio . TT (θ H)

QL T (θ L )
. =− .
QH T (θ H )

For a given .θ H the thermodynamic efficiency of a Carnot’ s cycle approaches 1

when .T (θ L ) → 0 and approaches 0 when .θ L → θ H . Between these two limits, the
thermodynamic efficiency will depend on .θ L . Nevertheless, the main idea of Kelvin
was to measure the absolute temperature .T by measuring the quantity of heat . Q (T ).

2.3.2 Using the Ideal Gas Properties for Calculating

the Conversion Efficiency for the Carnot’s Cycle

Following the results of Kelvin, the thermodynamic efficiency of a Carnot’ s engine

is given by
T (θ L )
. η =1− .
T (θ H )

In this section, we will show that the Kelvin’ s relationship .T (θ ) of the previous
section can be found by using the thermodynamic properties of an ideal gas.
An important consequence of Carnot’s second theorem is that the thermodynamic
efficiency of a Carnot engine is independent of the gas that is used as the working
fluid. So let us suppose that this working fluid is an ideal and monatomic gas, with
a constant molar heat .c̄v = 23 R.
The conversion efficiency of any power engine is given as in Eq. (2.2) by
W QL
η=−
. =1+ . (2.4)
QH QH

In the Carnot’s reversible cycle the heat, . Q H is delivered during the isothermal
expansion 1-2 of Fig. 2.30a. From the first principle

δ Q + δW = dU.
.

Since the internal energy of an ideal gas is only dependent on the temperature,
dU = 0 and
.
62 2 The Second Principle

δ Q = − δW = Pd V.
.

So
∫V2
V2
. Q H = Q 1−2 = Pd V = N RTH ln .
V1
V1

In the same way, the rejected heat can be calculated as

∫V4
V3
. Q L = Q 3−4 = Pd V = − N RTL ln .
V4
V3

Therefore, replacing. Q H and. Q L into Eq. (2.4), the conversion efficiency becomes
V3
TL ln V4
. η =1− V2
. (2.5)
TH ln V1

When the gas suffers a reversible adiabatic expansion from .TH to .TL , pictured
as process 2-3 in Fig. 2.30a, the delivered work will be equal to the variation in its
internal energy or

δW = dU,
.

or, since we are dealing with an ideal monatomic gas

3
. − Pd v̄ = RdT.
2

Now, from Clapeyron’s equation . P v̄ = RT and so

d v̄ 3 dT
. =− .
v̄ 2 T
By integrating the above equation between states 2 and 3 of Fig. 2.30a it can be
concluded that
( ) 23
v̄3 TH
. = .
v̄2 TL

Similarly, using the same reasoning for the adiabatic compression 3-4,
( ) 23
v̄4 TH
. = .
v̄1 TL
2.3 Sadi Carnot and Émile Clapeyron 63

So
v̄3 v̄2
. = ,
v̄4 v̄1

and the conversion efficiency, Eq. (2.5) of the Carnot’s cycle becomes

TL
η =1−
. . (2.6)
TH

In the above equation,.TL and.TH are the absolute ideal gas temperatures, following
the discoveries of Gay-Lussac and Clapeyron (see Sect. 1.2.1).
Three important conclusions can be derived from this equation:
• The conversion efficiency is only dependent on the ratio between the low and high
absolute temperatures .TL /TH .
• The conversion efficiency increases when the temperature .TH of the hot source
is increased with respect to .TL . In steam power engines this implies increasing
the vapor pressure in the steam generator above the pressure of the James Watt,
eighteenth century, atmospheric steam engines. Also, to produce, in the condenser,
the vacuum necessary to condense the vapor at reduced pressures (decreasing the
temperature .TL ).
• The gas absolute temperature .T is never negative, otherwise it would be possible
to run a cycle with conversion efficiency .η greater than 1.

2.3.3 Exercises

1. Two Carnot’s engines A and B operate between the same heat reservoirs at .TH
and .TL . (a) Suppose that the conversion efficiency of engine A, .η A , is greater
than the conversion efficiency of engine B, .η B . Reverting engine B and using it to
pump the rejected heat from engine A, show that this hypothesis leads to denying
the second principle of thermodynamics and that, therefore, .η B ≥ η A . (b) Use the
same reasoning, but now supposing .η B > η A for showing that .η B ≤ η A . (c) What
is the single solution of the inequalities .η B ≥ η A and .η B ≤ η A ?
2. Suppose that the isothermal and adiabatic processes in a Carnot cycle are irre-
versible, the irreversible cycle being represented by 1-2-3.'-4.'-1 as in Fig. 2.33.
Using the analysis method of Sect. 2.1, show that:
(a) after the irreversible adiabatic expansion from state 2, the final equilibrium state
3.' at .TL will necessarily have a larger volume than .V3 ;
(b) to complete the cycle reaching state 1 the irreversible adiabatic compression
must start | a state 4.' at the left of state 4 in the isotherm .TL ;
| from
(c) the heat.| Q 'H | absorbed by the gas during the isothermal irreversible expansion is
necessarily smaller than the heat .|Q H | absorbed during the reversible expansion;
64 2 The Second Principle

Fig. 2.33 A Carnot cycle

where the individual
processes fail to be reversible

| |
(d) the heat .| Q 'L | rejected by the gas during the isothermal irreversible compression
is necessarily larger than the heat .|Q L | rejected during the reversible compres-
sion;
(e) the conversion efficiency .ηirr of the irreversible cycle is necessarily smaller than
the one corresponding to the reversible cycle .ηr ev .

2.4 Reversible Cycles Other Than Carnot’s Cycle

The main condition for the Carnot first and second theorem to be demonstrated is
that a reversible power engine can be reverted and work as a heat pump. The next
question to be answered is why the Carnot cycle is the one with the highest efficiency
and not any other reversible cycle? Since all reversible cycles can be reverted they
would satisfy the condition that was imposed when demonstrating these theorems.
Consider the isobaric expansion represented in Fig. 2.34a when a gas at temper-
ature .T1 is put in contact with a hot liquid bath. It is hard to conceive an isobaric
expansion of a gas as a reversible process. Indeed, the gas is heated from below
giving rise to Rayleigh-Bénard convection cells due to the gas density decrease close
to the hot liquid bath. The heated gas will lift upward and cool, transferring heat to
the cold gas above it, while the cold gas goes down due to its higher density.
Nevertheless, from a theoretical point of view it is possible to conceive an isobaric
expansion as a sequence of very small isothermal expansions followed by adiabatic
compressions, as shown in Fig. 2.34b. Each isothermal expansion is produced by
removing a small balance weight with mass .Δm. After the equilibrium is reached
the gas chamber is insulated and this balance weight is put back on the piston head,
retrieving the original pressure at a higher temperature .T1 + ΔT . The gas is, then,
put in contact with a heat reservoir at the same temperature.T1 + ΔT , and the balance
weight is removed a second time. This sequence is repeated till the gas attains tem-
perature .T2 . All the states represented as circles along the isobaric line in Fig. 2.34b
are equilibrium states and, in the limit, when the mass .Δm → 0, this isobaric line
will represent a reversible isobaric expansion.
2.4 Reversible Cycles Other Than Carnot’s Cycle 65

Fig. 2.34 a An isobaric expansion of a gas is never reversible: the gas is heated from below giving
rise to Rayleigh-Bénard convection cells due to the gas density decrease close to the hot liquid
bath. During the heating process, the gas temperature is .T2 close to the heat reservoir, but smaller
than .T2 in the remaining space of the gas chamber. b From a theoretical perspective, it is possible to
conceive an isobaric reversible expansion by considering it as the limit of a sequence of isothermal
expansions and adiabatic compressions

Consider now a reversible cycle, composed of two isobaric and two adiabatic
process such as cycle 1-2-3-4-1 in Fig. 2.35a. This cycle is known as the Brayton
cycle.10 The isobaric expansion 1-2 is as above: a sequence of reversible isothermal
expansions followed by adiabatic compressions in the limit .Δm → 0. The isobaric
compression 3-4 is a sequence of reversible isothermal compressions followed by
adiabatic expansions, in the same limit.
We can conceive the Brayton cycle as a set of . N Carnot cycles as shown in
Fig. 2.35a, with . N → ∞. The first Carnot cycle operates between temperatures .T1
and .T4 , the lowest temperature of the cycle, .T4 = TL . The last Carnot cycle, cycle . N ,
operates between temperatures .T2 .— the highest temperature of the cycle, .T2 = TH
.— and . T3 .
Since the adiabatic expansion work of each Carnot cycle .i will be canceled by the
adiabatic compression work of Carnot cycle .i + 1, the net adiabatic expansion work
will be the area under line 2-3 and the net adiabatic compression work will be the
area under the line 4-1. Also, the total heat received by the gas during the isobaric
expansion 1-2 will be the sum

10 The cycle is named after George Brayton (1830–1892), the American engineer who developed
it originally for use in piston engines. It is also sometimes known as the Joule cycle. There are two
types of Brayton cycles, open to the atmosphere and using internal combustion chamber or closed
and using a heat exchanger.
66 2 The Second Principle

Fig. 2.35 a A reversible

cycle (Brayton cycle)
composed of two isobaric
and two adiabatic processes
may be decomposed in a set
of Carnot cycles. b Each
Carnot engine operates in a
temperature range that is a
fraction of the entire
temperature range between
. TL and . TH

∑
N
. QH = ΔQ H,i ,
i=1

and each.ΔQ H,i is the heat absorbed by the gas during the small isothermal expansion
at temperature .Ti . It can be shown .— see Exercise 1 in this section .— that in the
limit .ΔT = Ti+1 − Ti → 0, this absorbed heat will be .ΔQ H,i = c p (Ti ) N ΔT for
an ideal gas,11 and this is the heat that is absorbed by the gas during the isobaric
expansion from the equilibrium state .i to .i + 1.
The same reasoning can be made for the total heat that is rejected during the
isobaric compression 3-4,

∑
N
. QL = ΔQ L ,i .
i=1

Therefore, the present reversible cycle can be considered as a set of Carnot cycles
as shown in Fig. 2.35b, with the same heat and work exchanged as the original cycle.
Let us now calculate the conversion efficiency of this reversible cycle starting
from Eq. (2.4)

11Since .ΔT → 0 and the molar heats .c p and .cv depend only on the temperature, they can be
considered as constants.
2.4 Reversible Cycles Other Than Carnot’s Cycle 67

∑N
QL ΔQ L ,i
.η = 1 + = 1 + ∑ i=1 . (2.7)
N
i=1 ΔQ H,i
QH

For each Carnot engine .i,

( )
ΔQ L ,i = ΔQ H,i ηc,i − 1 ,
.

where .ηc,i is the conversion efficiency of each Carnot engine .i. So,
∑N ( ) ∑N
i=1 ΔQ H,i ηc,i − 1 i=1 ηc,i ΔQ H,i
.η = 1 + ∑N = ∑ . (2.8)
N
i=1 ΔQ H,i i=1 ΔQ H,i

Nevertheless, using the concept of the mean value for the conversion efficiency
∑N
i=1 ηc,i ΔQ H,i
. ⟨ηc ⟩ = ∑N ,
i=1 ΔQ H,i

it results

η = ⟨ηc ⟩
.

So, the conversion efficiency of the present reversible cycle is the mean value of
the conversion efficiencies of the set of . N Carnot cycles in the limit when . N → ∞.
Since each Carnot cycle operates in a temperature interval which is a fraction of
. TH − TL , we conclude that .η < 1 −
TL
TH
.
Indeed, during the (reversible) isobaric expansion heat is supplied to the gas
at temperatures that vary from .T1 to .T2 = TH , the highest temperature of the gas,
contrary to Carnot’s cycle, where heat is supplied at the highest temperature in the
cycle; in the reverse direction, during the (reversible) isobaric compression heat
is rejected by the gas at temperatures that vary from .T3 to .T4 = TL , the lowest
temperature of the gas, contrary to the Carnot’s cycle, where heat is rejected at the
lowest temperature in the cycle.
In Brayton and other reversible cycles, heat is supplied and rejected at variable temperatures,
in contrast with the Carnot’ s cycle, where heat is supplied at the highest and rejected at the
lowest temperatures of the cycle. This is the reason why their thermodynamic efficiency is
smaller than the one of Carnot’ s cycle.

2.4.1 Exercises

1. The conversion efficiency of the Brayton cycle shown in Fig. 2.35 can be exactly
found as
68 2 The Second Principle

Fig. 2.36 Reversible

processes between two
arbitrary equilibrium states

Fig. 2.37 Another reversible

cycle

TL
.η =1−ζ .
TH

Find .ζ when the gas is an ideal, monatomic12 gas and show that .ζ > 1.
2. How many reversible lines it is possible to draw between the equilibrium states
1 and 2 shown in Fig. 2.36? Is it possible to join points 1 and 2 using a straight
line that represents a reversible process?
3. Show that every reversible cycle can be decomposed into Carnot cycles.
4. Show that every reversible cycle can be decomposed into Carnot cycles. Con-
sider, in particular, a reversible Stirling13 cycle 1-2-3-4-1, composed of two
isothermic and two isochoric processes as pictured in Fig. 2.37 and show in detail
that this cycle can effectively be decomposed into small Carnot’s cycles. Calcu-
late the conversion efficiency of this cycle when the gas is an ideal, monatomic
gas. Show that this conversion efficiency is smaller than the one of a Carnot’ s
cycle operating between .TL and .TH .

12For monatomic gases .c̄v = 23 R .

13The Stirling cycle is a reversible cycle that theoretically describes the external combustion Stirling
engines, invented, developed, and patented in 1816 by Robert Stirling.
2.5 Clausius Inequality and Entropy 69

2.5 Clausius Inequality and Entropy

Die Energie der Welt ist constant. Die Entropie der Welt strebt einem Maximum zu. (Rudolf
Clausius, 1867 [36])

For a Carnot cycle between temperatures.TL and.TH it was seen that the conversion
efficiency is given by
QL TL
.η =1+ =1− .
QH TH

So, for a Carnot cycle,

QH QL
. + = 0, (2.9)
TH TL

meaning that the absorbed and rejected heat are, respectively, proportional to the
temperatures of the hot and cold heat reservoirs.
Consider now an arbitrary reversible cycle such as the one shown in Fig. 2.35.
Since, as was shown in the previous section, all reversible cycles can be considered
as a set of . N Carnot cycles, and so,
N (
∑ )
ΔQ H,i ΔQ L ,i
. + = 0.
i=1
TH,i TL ,i

which can also be written as

∑N )
ΔQ i
. = 0,
i=1
Ti r ev

and when . N → ∞,
∮ )
δQ
. =0 (2.10)
T r ev

meaning that in reversible cycles, the heat exchanged by a system in thermodynamic

equilibrium is proportional to the temperature at which this heat is exchanged.
Since this integral is zero for all reversible cycles, the integrand must be a state
variable. In the words of Clausius14 “…whenever a body, starting from any initial
condition, returns thereto after its passage through any other conditions, then the
expression . δTQ under the sign of integration must be the complete differential of a

14 R. Clausius (1867), The mechanical theory of heat, with its applications to the steam-engine and

to the physical properties of bodies, Edited by T.A. Hirst, John van Voorst, London [36]. Ninth
Memoir: On several convenient forms of the fundamental equations of the mechanical theory of
heat, pp. 327–365.
70 2 The Second Principle

magnitude that depends only on the present existing condition of the body, and not
upon the way by which it reached the latter. Denoting this magnitude by S, we can
write .d S = δTQ .”
The name “entropy is justified at the page 357 of its Ninth Memoir: “We might
call S the transformational content of the body, just as we termed the magnitude
U its thermal and ergonal15 content. But as I hold it to be better to borrow terms
for important magnitudes from the ancient languages, so that they may be adopted
unchanged in all modern languages, I propose to call the magnitude S the entropy of
the body, from the Greek word .τρoπ η, transformation. I have intentionally formed
the word entropy so as to be as similar as possible to the word energy; for the
two magnitudes to be denoted by these words are so nearly allied in their physical
meanings, that a certain similarity in designation appears to be desirable”.
We rewrite the Clausius definition of entropy as
)
δQ
.d S = , (2.11)
T r ev

meaning that the entropy variation between any two states can be only measured by
considering a reversible path joining these states.
Consider, now, a Carnot cycle where heat is supplied and rejected along irre-
versible isothermal processes such as the ones shown | by| the dashed lines in Fig. 2.33.
As it was shown in Sect. 2.1 the absorbed heat .| Q 'H | will always be smaller than
the corresponding absorbed heat .|Q H | if the expansion was reversible; indeed, in
irreversible isothermal expansions.— the ones due to the removal of a finite balance
weight with mass .Δm .— the delivered work decreases with respect to the work that
would be delivered
| | in an isothermal reversible expansion. On the other hand, the
rejected heat .| Q 'L | during the irreversible compression is always greater than .|Q L |,
i.e., the heat that would be rejected during an isothermal reversible compression.
Therefore, in this ‘irreversible Carnot cycle’, Eq. (2.9) giving the proportionality
between the exchanged heat and the temperature must be replaced by
QH QL
. + ≤ 0, (2.12)
TH TL

15 The term ergon is defined at page 253 of ‘The mechanical theory of heat’, Sixt Memoir as what
would, presently, be designated as ‘work’: “Now, in the mechanical theory of heat, after admitting
that heat can be transformed into work and work into heat, in other words, that either of these
may replace the other, it becomes frequently necessary to form a magnitude of which heat and
work are constituent parts. (....) Let heat and work continue to be measured each according to
its most convenient unit, that is to say, heat according to the thermal unit, and work according to
the mechanical one. But besides the work measured according to the mechanical unit, let another
magnitude be introduced denoting the work measured according to the thermal unit, that is to say,
the numerical value of the work when the unit of work is that which is equivalent to the thermal
unit. For the work thus expressed a particular name is requisite. I propose to adopt for it the Greek
word (.∈ργ oν) ergon”.
2.5 Clausius Inequality and Entropy 71

Fig. 2.38 Clausius entropy

meaning that, in absolute values, the heat gains are smaller, and the heat losses are
greater than the ones corresponding to a theoretical Carnot reversible cycle. The
equality sign only applies when the mass .Δm → 0 and the entire cycle is reversible.
For arbitrary cycles, the above inequality can be easily generalized to give
∮
δQ
. ≤ 0.
T

This inequality received the name of ‘Clausius inequality’ or ‘Clausius theorem’

and is described on page 133 of Clausius Fourth Memoir ‘On a modified form of the
second fundamental theorem in the mechanical theory of heat’.16
Consider now that a cycle is composed of an arbitrary process (reversible or non-
reversible) leading a thermodynamic system from an equilibrium state 1 to another
equilibrium state 2. It is always possible to return back from state 2 to the initial
equilibrium state 1, by taking one of the infinite reversible processes from 2 to 1
(Fig. 2.38).
Therefore, from the Clausius theorem,

∮ ∫2 ∫1 )
δQ δQ δQ
. = + ≤ 0,
T T T r ev
1 2

and, since the returning process, 2-1 is reversible

∫1 )
δQ
. = S1 − S2 .
T r ev
2

So, we are led to the conclusion that

16 R. Clausius (1865), The mechanical theory of heat, with its applications to the steam engine and
to the physical properties of bodies, Edited by T.A. Hirst, John van Voorst, London [36]. Fourth
Memoir: On a modified form of the second fundamental theorem in the mechanical theory of heat,
pp.111–135.
72 2 The Second Principle

∫2
δQ
S − S1 ≥
. 2 . (2.13)
T
1

or, that the integral of the ratio between the heat exchanged and the temperature at
which it is exchanged is always smaller than the related variation of entropy between
any two equilibrium states. The equality only holds for reversible processes.
For an isolated system the heat exchanged is null, so, any internal process will
increase its entropy

d S ≥ 0.
. (2.14)

This equation can be seen as the mathematical expression of the arrow of natural
processes since all internal processes that result from a non-equilibrium in the fields
of temperature, density, pressure, or local velocity will produce an increase in entropy.

2.5.1 External and Internal Sources of Entropy

Irreversible processes are always escorted by the appearance of non-equilibrium in

their local fields. So, when a thermodynamic system is subjected to an irreversible
process, its temperature field is unbalanced and it is impossible to access its value at a
given point without a local thermodynamic non-equilibrium analysis. Nevertheless,
all the heat that is exchanged by a thermodynamic system flows through the boundary
that separates it from the external world and Eq. (2.13) is better written as

∫2 )
δQ
S − S1 ≥
. 2 . (2.15)
T bound.
1

In fact, the elementary variation of entropy of a thermodynamic system can be

written as17

d S = de S + di S,
.

where
)
δQ
d S=
. e ,
T bound.

is the variation of entropy of this system due to the heat that flows through its boundary
and the term .di S is the source of entropy due to internal non-equilibrium. From the

17 I. Prigogine, Introduction to Thermodynamics of Irreversible Processes (Interscience, 1968).

2.5 Clausius Inequality and Entropy 73

Fig. 2.39 Irreversible cycle

of an ideal gas in contact
with a heat reservoir: Heat
and work balances for the
expansion and compression
processes

principle of increasing entropy for isolated systems, Eq. (2.14) it is clear that this
term must be non-negative,

d S ≥ 0.
. i

It is also clear that the elementary variation of entropy .de S can be positive or
negative in accordance with the heat that is flowing through the boundary: when heat
is flowing into the system .de S > 0, otherwise it is negative.
Example
Consider the irreversible isothermal cycle of Fig. 2.1, when a balance weight of 50 .kg
is removed from the piston top plate above a gas chamber, producing an irreversible
expansion (Fig. 2.39).
The initial and final equilibrium states are given in Table 2.1.
It is possible to calculate the entropy variation between the equilibrium states 1
and 2, by integrating Eq. (2.11), following a reversible path between states 1 and 2.
The most obvious reversible path to follow is the isothermal expansion 1-2. In this
case, the heat absorbed by the gas was calculated as 930 .J (see Sect. 2.1). So,
Q 1−2 )r ev 930 J
S − S1 =
. 2 = = 3.10 .
T 300 K
On the other hand, the entropy flow from the heat reservoir to the gas will be

Q 1−2 778 J
. (S2 − S1 )e = = = 2.5933 ,
T 300 K

This ratio becomes closer to .3.10 . KJ when the gas expansion is produced by
sequentially removing lighter balance weights with, e.g., 10, 5, 2, 1, and 0.5 .kg,
but in all these expansion processes the entropy variation of the gas is always the
same because entropy is a state variable that “depends only on the present existing
condition of the body, and not upon the way by which it reached this state”.
74 2 The Second Principle

The entropy that is generated due to internal non-equilibrium produced by the

removal of a balance weight of 50 .kg from the piston head can be calculated as

. (S2 − S1 )i = (S2 − S1 ) − (S2 − S1 )e

J
= 0.5067 .
K
Gas compression
Let us now repeat the same calculation for the compression process when the 50 .kg
balance weight is placed back on the piston plate. In this case
J
. 1 S − S2 = − 3.10 ,
K
and the heat that is delivered from the gas to the heat reservoir during the compression
is . Q 2−1 = − 1130 .J. So
1130 J
. (S1 − S2 )e = − = − 3.77 .
300 K
On the other hand, the entropy production will be given by

J
. (S1 − S2 )i = 0.67 .
K
Cycle
It was seen that the heat and work exchanged by the system in an isothermal cycle
must satisfy
∮ ∮
. δW = − δ Q ≥ 0,

otherwise, the second principle would be denied.

The Clausius inequality means that
∮
δQ J
. = 2.5933 − 3.77 = − 1.1767 ,
T K

is negative for all irreversible cycles or that, in absolute value, the heat gains are
smaller and the heat losses are greater than the corresponding heat gains and losses
in a reversible cycle.
Since entropy is a state variable,
∮
. d S = 0.
2.5 Clausius Inequality and Entropy 75

Nevertheless, there is a net production of entropy after each irreversible cycle

∮
J
. d Si = 0.5067 + 0.67 = 1.1767 .
K

Also,
∮
J
. d Se = 2.5933 − 3.77 = − 1.176 7 ,
K

meaning that all the entropy that is internally produced flows to the heat reservoir
after each cycle.
Reservoir
Consider now the heat reservoir as our thermodynamic system. Since the reservoir
is a system that, theoretically, is always in equilibrium, its entropy variation will be
only due to the flow of heat. So, after the gas expansion
778 J
. (S2 − S1 ) R = − = − 2.5933 ,
300 K
and after the gas compression

1130 J
. (S1 − S2 ) R = = 3.77 ,
300 K
For the entire cycle
∮
J
. d S R = 1.176 7 ,
K

as would be expected, since each cycle that is performed by the gas system produces
a flow of entropy from the gas system to the reservoir.
Whole system gas + reservoir
The whole system gas + reservoir is an isolated system and the entropy variation
must be positive for both the expansion and compression processes. So, after the gas
expansion
J
. (S2 − S1 ) R+G = 0.5067 ,
K
and after the gas compression
J
. (S1 − S2 ) R+G = 0.67 .
K
For the entire cycle
76 2 The Second Principle
∮
J
. d S R+G = 1.176 7 .
K

During each cycle, work is converted into heat (and not the inverse) and this is a
general rule for all isolated systems. The arrow of nature is clearly pictured in this
rule and is translated by a cumulative increase of entropy,
In concluding this section it is perhaps important to read again William Thomson
in his 1851 famous paper ‘On the dynamical theory of heat’: “I believe the tendency
in the material world is for motion to become diffused, and that as a whole the reverse
of concentration is gradually going on - I believe that no physical action can ever
restore the heat emitted from the sun, and that this source is not inexhaustible; also
that the motions of the earth and other planets are losing vis viva which is converted
into heat; and that although some vis viva may be restored for instance to the earth
by heat received from the sun, or by other means, that the loss cannot be precisely
compensated and I think it probable that it is under-compensated.

2.5.2 Exercises

1. A single mole of Helium initially at .T1 =300 .K and . P = 1 .MPa, is put in contact
with a heat reservoir at.T2 =800.K as shown in Fig. 2.40. In consequence, it has an
isobaric expansion till equilibrium is reached at 800.K. Helium can be considered
as satisfying the Clapeyron equation of state in this range of temperatures.
Calculate: (a) the entropy variation . S2 − S1 of the gas ; (b) the heat . Q 1−2 that is
transferred to the gas during the expansion; (c) the increase in the gas entropy
.(S2 − S1 )e due to the flow of heat from the reservoir; (d) the entropy that is

generated inside the gas phase .(S2 − S1 )i ; (e) the entropy variation of the heat

Fig. 2.40 Isobaric

expansion of an ideal gas in
contact with a heat reservoir
2.5 Clausius Inequality and Entropy 77

Fig. 2.41 Isochoric heating

of an ideal gas in contact
with a heat reservoir

reservoir after the gas expansion; (f) the entropy variation of the whole system
gas+reservoir .(S2 − S1 ) R+G ; (g) comment on the results that were found.
Repeat the same calculations, considering now that the Helium is replaced by
Nitrogen. The molar thermal capacity at a constant volume of Nitrogen is given
in terms of the temperature by
( )
J
.c̄v = 20.586 − 0.1571 × 10−2 T + 0.8081 × 10−5 T 2 − 2.873 × 10−9 T 3 .
mol K

2. One mole of air initially at .20 .◦ C is put in thermal contact with a heat reservoir
at .T2 =90 .◦ C as shown in Fig. 2.41 and suffers iso-volumetric heating till it
attains the final equilibrium state at 90 .◦ C. For this problem, the specific thermal
capacity of the air can be considered as a constant and given by .cv = 0.718 . kg·KkJ
.
Air can be considered as satisfying the Clapeyron equation of state in this range
of temperatures.
Calculate: (a) the entropy variation . S2 − S1 of the gas ; (b) the heat . Q 1−2 that
is transferred to the gas during the heating; (c) the increase in the gas entropy
.(S2 − S1 )e due to the flow of heat from the reservoir; (d) the entropy that is
generated inside the gas phase .(S2 − S1 )i ; (e) the entropy variation of the heat
reservoir after the gas expansion; (f) the entropy variation of the whole system
gas+reservoir .(S2 − S1 ) R+G . Comment on the results that were found.
3. A cylinder is divided into two equal-volume chambers A and B. Each chamber
contains 1 mole of Helium initially at 20.◦ C (Fig. 2.16). Chambers A and B are
separated by a thermal insulated plate, free to move inside the cylinder, in such
a manner as to equalize the pressures in the two chambers. The base plate of
the cylinder is a thin thermal contact plate. Consider now that this base plate is
put in contact with a thermal reservoir at 200.◦ C. The Helium in chamber B will
expand moving the dividing plate upward.
Since the adiabatic compression of the Helium in chamber A is a very slow
process the compression process of He in chamber B can be supposed to be
78 2 The Second Principle

reversible. Calculate: (a) the entropy variation .(S2 − S1 ) A+B of the system A+B,
(b) the entropy variation of the system A+B due to the flow of heat from the
reservoir, .(S2 − S1 )e,A+B ; (c) the entropy that is internally produced in system
A + B, .(S2 − S1 )i,A+B .
Suppose, now, that the compression of the Helium in chamber A is an irreversible
process. In this case, it is not possible to know what would be the final equilibrium
state without a local non-equilibrium analysis. Nevertheless, it is possible to
establish some inequalities.
(d) When He is irreversibly compressed in chamber A the only equilibrium states
that are possible to be attained by this gas are the ones with a higher temperature
for the same pressure that was reached after the reversible adiabatic compression.
So the final pressure in this second process cannot be the same as in the first
when He is in the chamber A was supposed to be reversibly compressed, since
the total volume did not change. What is it possible to say about the final pressure
. P2,irr in this second process, it will be greater or smaller than . P2,r ev , i.e., the
final pressure in the first process? Try to answer the same question for the final
A
temperature .T2,irr. of the He in chamber A.
(e) The heat absorbed from the reservoir is used for heating the gas in chamber
B and to expand it against chamber A. In the second process, it is not possible
to calculate the compression work delivered to chamber A, without a local non-
equilibrium analysis. But what about the value of the total heat . Q 1−2 that is
absorbed from the heat reservoir? What it is possible to say about . Q 1−2 when
the two processes are compared?
4. After the expansion of an ideal gas from the equilibrium state 1 to 2, in thermal
contact with a heat reservoir supposed to be at a constant temperature18
∮
. di S R = 0,

and all the entropy variation of the reservoir is due to the heat that it exchanges
with the gas
∮ ∮ )
δQ
. de S R = ,
T bound.

This is only true for idealized heat reservoirs. In the real world, the net transfer of
heat after each cycle of the gas system produces non-equilibrium in the temper-
ature field of the liquid bath that is being used with the role of heat reservoir and
the internal generation of entropy will be different from zero after each cycle.
The analysis of this kind of system requires, nevertheless, a local approach, the
entropy variation of the gas is .(S2 − S1 ) and the entropy variation of the heat
reservoir is .(S2 − S1 ) R .

18In this exercise the heat reservoir is supposed to be in thermal equilibrium during the entire cycle
1-2-1. Therefore the entropy generated in the reservoir is null.
2.6 The Meaning of Entropy 79

(i) What is the physical meaning of.(S2 − S1 ) + (S2 − S1 ) R ? What is necessary

for this sum be zero?
(ii) Consider now the isothermal cycle 1-2-1. What alternatives below are the
correct ones? Explain your answer.
∮ ∮ ∮ ∮
(a) . d S > 0, (b) . d S = 0, (c) . d S R = 0, (d) . d (S + S R ) ≥ 0.
5. A single mole of an ideal gas is in chamber A with 1.l of Fig. 2.42. This chamber
is separated from an evacuated chamber B with 9.l by a plate . p that is fixed on
the cylinder wall by two pins. The pins are then removed and the gas has a free
expansion occupying all the remaining volume. Both chambers are in perfect
thermal contact with an idealized heat reservoir.
Find:
(a) the entropy variation . S2 − S1 of the gas between the initial, 1 and final, 2,
equilibrium states. What are the fractions of this entropy variation that are
due to heat flow, .(S2 − S1 )e , and to internal generation .(S2 − S1 )i ;
(b) the entropy variation of the heat reservoir .(S2 − S1 ) R ;
Suppose now that the initial pressure . P1 is suddenly restored above the plate
. p and the gas returns to its initial state after suffering an isobaric compression

Find:
(c) the work .W2−1 and heat . Q 2−1 exchanged;
(d) the entropy variation of the gas between the equilibrium states 2 and 1 that is
due to the heat flow .(S1 − S2 )e and the one that is due to internal generation;
(e) the entropy variation of the reservoir between the equilibrium states 2 and 1
Finally, consider the entire cycle 1-2-1.
Find:
∮ ∮
(f) .∮ δ Q and . δW for the gas;
(g) .∮ de S for the gas and reservoir;
(h) .∮ di S for the gas and reservoir;
(i) . d S for the global system gas+reservoir.
(j) What it can be concluded from the above calculations?

2.6 The Meaning of Entropy

Philosophy gets on my nerves. If we analyze the ultimate ground of everything, then every-
thing finally falls into nothingness. But I have decided to resume my lectures again and look
the Hydra of doubt straight into the eye. (Ludwig Boltzmann)

The statement of the law of energy conservation in 1847 by Helmholtz meant the
acceptance of the corpuscular nature of the matter by the scientific world.
80 2 The Second Principle

Fig. 2.42 Isothermal

expansion against vacuum

References to the concept of atomism and atoms are found in ancient India and
ancient Greece. In India the Jain, Ajivika and Carvaka19 schools of atomism may date
back to the fourth century BC.20 The Nyaya and Vaisheshika schools later developed
theories on how atoms combined into more complex objects.21 In Greece, atomism
emerged in the fifth century BC with Leucippus and Democritus.22
The particles of chemical matter for which chemists and other natural philosophers
of the early nineteenth century found experimental evidence were thought to be
indivisible, and therefore were given the name ‘atom’, long used by the atomist
philosophy.
The concept of ‘heat’ or ‘internal energy of a thermodynamic system as the result
of the mechanical energy of its individual molecules was long advocated by philoso-
phers and scientists such as Francis Bacon (1561–1626), John Locke (1632–1704),
and Thomas Hooke (1635–1703), .
In his ‘Hydrodynamica’ (Fig. 2.43), Bernoulli (1700–1782) proposed that gases
consist of great numbers of molecules moving in all directions, that their impact on a
surface causes the gas pressure and that heat transfer occurs from regions with high
energetic molecules to those with low energetic molecules.
James Joule equivalence principle and the nineteenth century mechanical theory
of heat were highly influenced by the atomistic philosophy.

19 M. Gangopadhyaya (1981). Indian Atomism: History and Sources. Atlantic Highlands, New
Jersey: Humanities Press. ISBN 0-391-02177-X. OCLC 10916778.
20 T. McEvilley (2002), The Shape of Ancient Thought: Comparative Studies in Greek and Indian

Philosophies ISBN 1-58115-203-5, Allwarth Press.

21 R. King, Indian philosophy: an introduction to Hindu and Buddhist thought, Edinburgh University

Press, 1999, ISBN 0-7486-0954-7.

22 C.C.W. Taylor The atomists, Leucippus and Democritus: fragments, a text, and translation with

a commentary, University of Toronto Press Incorporated 1999, ISBN 0-8020-4390-9.

2.6 The Meaning of Entropy 81

Fig. 2.43 Front page of

Bernoulli’s Hydrodynamica

But how to conciliate the dissipative second principle of thermodynamics with the
conservative laws of mechanics? Indeed, in an isolated universe of particles satisfying
Newton’s laws, the mechanical energy is preserved and this system is reversible: the
previous states can always be attained by reversing the time in the equations of
motion.
So, if heat has a mechanical foundation how to explain irreversibility based on
the mechanical laws of motion?
The answer was only given by Ludwig Boltzmann23 in the second half of the
nineteenth century, by introducing the concept of probability in mechanics.
When the exact position and velocity of each particle of a mechanical system
cannot be ascertained without errors, the best it is possible to do is to try to prescribe
the probability that this particle will have a certain velocity when it is in a position
.x at time t.
As written by Max Planck in the Preface of his book24 : “The full content of the
second principle can only be understood if we look for its foundation in the known

23 L. Boltzmann. Vorlesungen uber Gas Theory, .(1896) . J.A. Barth, Leipzig [15]. L. Boltzmann.
Vorlesungen uber Gas Theory .(1898) J.A. Barth. Leipzig [16].These books were translated in
English by Stephen Brush and edited by University of California Press. Dover editions are presently
available .
24 Planck, M. (1903). Treatise on Thermodynamics. Published in German in 1897. Translated to

English by Alexander Ogg. London: Longmans, Green & Co. [136].

82 2 The Second Principle

laws of the theory of probability as they were laid down by Clausius and Maxwell,
and further extended by Boltzmann”.
Nevertheless, much of the physics establishment did not share Boltzmann’s belief
in the reality of atoms and molecules, a belief that was however, shared by most
chemists since the discoveries of John Dalton in 1808. He had a long-running dis-
pute with the editor of the prominent German physics journal of his day, who refused
to let Boltzmann refer to atoms and molecules as anything other than convenient
theoretical constructs. Only a couple of years after Boltzmann’s death, Perrin’s stud-
ies of colloidal suspensions (1908–1909), based on Einstein’s theoretical studies of
1905, confirmed the values of Avogadro’s number and Boltzmann’s constant, and
convinced the world that the tiny particles really exist.
All the early versions of the mechanical theory of heat assumed that heat was
some kind of motion but they were unclear on what kind. In 1857, Clausius made
explicit the idea that this motion was a molecular motion. He claimed “ to have
held this all along but avoided mentioning these conceptions because I wished to
separate the conclusions which are deducible from certain general principles from
those which presuppose a particular kind of motion”. On this basis, he developed a
dynamic theory of gases. He argued that gases are made up of molecules that move
in every direction and at every speed, showing how known properties of the gas, like
pressure and temperature, could be related to the mean motion of the molecules. He
made calculations about the average speeds and the average distances that molecules
would travel before striking another molecule.
Maxwell25 read these papers and began to play with the model. Maxwell made a
fully statistical model of gases, in which the velocities were distributed according to
the same formula as the errors are distributed in observations (the normal distribu-
tion). In 1873 Maxwell wrote26 : “The modern atomists have adopted a method which
is, I believe, new in mathematical physics, though it has long been in use in Statistics.
The data of the statistical method as applied to molecular science are the sums of
large numbers of molecular quantities. In studying the relation between quantities
of this kind, we meet with a new kind of regularity, the regularity of averages (...) If
a great many equal spherical particles were in motion in a perfectly elastic vessel,
collisions would take place among the particles and their velocities would be altered
at every collision, so that after a certain time the vis viva will be divided among
the particles according to some regular law, the average number of particles whose
velocity lies between certain limits being ascertainable, though the velocity of each
particle changes at every collision”.
In 1877, Boltzmann was in Graz, as a professor of experimental physics. He sub-
mitted two memoirs to the Academy of Sciences in Vienna. In the first of these,
27
presented on January, 1877, On Some Problems of the ∫ Mechanical Theory of Heat,
Boltzmann made an incisive analysis of the formula . δ Q/T ≤ 0 and argued that its

25 Maxwell JC (1867) On the dynamical theory of gases. Philos Trans R Soc 157:49–88 [115].
26 Maxwell, James Clerk (1873). ’Molecules’, Nature: 437–441 [116].
27 Ludwig Boltzmann (1878). On some problems of the mechanical theory of heat Philosophical

Magazine. Series 5. Vol. 6, pp. 236–237. Reprinted from [14].

2.6 The Meaning of Entropy 83

validity was not based on the inherent laws of nature but rather on the choice of the
initial conditions. According to him, one had to assume on the basis of the kinetic
theory of gases that the perfectly elastic balls representing the molecules always tend
to change their actual positions. Consequently, any configuration, however improba-
ble, could conceivably occur as time went on: “The calculus of probabilities teaches
us precisely this: any non-uniform distribution, unlikely as it may be, is not strictly
speaking impossible”.
It now remained to give this conclusion a rigorous derivation and quantitative
applicability, which Boltzmann did in his memoir submitted to the academy on
October 1877, On the Relation between the Second Law of the Mechanical Theory
of Heat and the Probability Calculus with respect to the Propositions about Heat-
Equivalence. He concluded that the entropy of a state is proportional to the probability
of the configuration of its component particles. In this way, entropy became a purely
statistical attribute of the overall system. As the entropy increases, the configuration
of the system becomes more probable.
Following Boltzmann’s reasoning, let us consider a mechanical system with a
great number of particles. Let . f (x, ξ, t)ΔxΔξ be the expected number of particles
with velocities between .ξ and .ξ + Δξ that are found at time .t in the elementary
volume .Δx between .x and .x + Δx, Boltzmann was able to show that the volumetric
integral of the quantity
∫
. H (x, t) = f ln f dξ, (2.16)

always decreases with time for isolated systems, in such a manner that the. H -function
could be identified with the Clausius local entropy (per particle) by

.s = − k H,

where .k is the Boltzmann constant that can be written as the ratio between the
universal gas constant .R and the Avogadro number .N .
R
k=
. .
N
Readers are suggested to read Chapter 10 for further details.
The Boltzmann statistical concept of entropy has been extended to discrete sys-
tems in Information Theory by Shanon.28 In this case, the entropy is defined as
∑
.s=− pi ln pi
i

where . pi is the probability of event .i.

28 Shannon, Claude E. (1948). A Mathematical Theory of Communication. Bell System Technical

Journal 27 (3): 379–423. https://doi.org/10.1002/j.1538-7305.1948.tb01338.x. Ref. [155].

84 2 The Second Principle

Recent developments in statistical mechanics based on the concept of Tsallis

entropy, have intensified the interest in investigating a possible extension of Shan-
non’s entropy to Information Theory [176].
 Part II
Thermodynamic Equilibrium
Chapter 3
The Equilibrium State

Abstract This chapter deals with homogeneity, Legendre transforms—leading to

the Helmholtz and Gibbs energies and the concept of enthalpy—followed by Maxwell
relations and minimum principles and establishing the framework that is needed for
the theoretical analysis of thermodynamic systems in equilibrium. A thermodynamic
system in equilibrium is homogeneous, meaning that its intensive properties do not
vary from point to point inside the system. This property leads to an Euler’s relation-
ship between its extensive variables and the Gibbs-Duhem equation. This equation
found, simultaneously, by Josiah Willard Gibbs and Pierre Maurice Marie Duhem
enables a Lagrangian description of thermodynamic systems, of crucial importance
in the analysis of open systems in non-equilibrium states. Minimum principles state
that when a restriction that prevents a system from reaching equilibrium is removed,
this system will attain an equilibrium state given by the minimization of one of its
energies—internal, Helmholtz, Gibbs, or enthalpy—in accordance with the thermo-
dynamic variables that are maintained constant during the equilibration process.

3.1 Introduction

In the first part of this book, the first law of thermodynamics was stated as

.δ Q + δW = dU, (3.1)

meaning that the internal energy of a thermodynamic system—or the mechanical

energy of its molecules—is affected both by the addition of heat and by the com-
pression/expansion work that is exchanged between the system and the external
environment.
In this second part, we focus on equilibrium states and not on the process that leads
from one equilibrium state to another. To be clear, consider an isothermal expansion
of an ideal gas in contact with a heat reservoir, between any two given pressures . P1
and . P2 . This expansion is promoted by successively removing balance weights of
mass .Δm till pressure . P2 is reached (Fig. 3.1). We know that both the work .W1−2
and the heat . Q 1−2 are dependent on the mass .Δm. Nevertheless, when the process

© Associação Paranaense de Cultura 2024 87

P. C. Philippi, Thermodynamics, https://doi.org/10.1007/978-3-031-49357-7_3
88 3 The Equilibrium State

T -940

-920

Expansion work (J)

Patm -900

-880
-860

-840

-820
-800
Ideal gas
-780
-760
0.1 1 10 100

Thermal reservoir m (kg)

Fig. 3.1 The heat and work that are exchanged are dependent on mass .Δm, but the state variables
in equilibrium states 1 and 2 are always the same

between states 1 and 2 is reversible, the entropy variation between the initial and
final equilibrium state will be always the same and will be given by

∫2 )
δQ
S − S1 =
. 2 ,
T r ev
1

independently of the reversible path we take for going from state 1 to state 2.
Therefore, for calculating the state variables for the different equilibrium states
of a thermodynamic system we will only consider reversible processes, for which,

δ Q = T d S,
.

δW = −Pd V.

So, the first law of thermodynamics, Eq. (3.1), becomes a differential equation

. dU = T d S − Pd V, (3.2)

relating the state variable .U with the state variables . S and .V .

In this representation, we can consider . S and .V as independent variables and .U
as the dependent variable. In addition, the absolute temperature .T can be written as
)
∂U
.T = , (3.3)
∂S V
3.1 Introduction 89

Fig. 3.2 Fluxes and forces a potential for the transfer of entropy
in a reversible process
entropy flux

dU=TdS+(-P)dV
volume flux
a potential for the transfer of volume

and the pressure . P as

)
∂U
. P=− , (3.4)
∂V S
)
where . ∂∂ XZ Y means that the partial derivative . ∂∂ XZ is being performed with .Y being
kept constant. This notation is necessary because, in thermodynamics, contrary to
elementary differential calculus, we have a great number of state variables, but only
a small number of them are independent.
Before going further let us try to give a physical interpretation to the terms that con-
tribute to the variation of the internal energy in Eq. (3.2) in the reversible world. When
a thermodynamic system changes its internal energy, this change can be attributed
to two sources. The first one is the heat .δ Q = T d S it receives, or the flux of entropy
through its boundary. In reversible processes there are no internal sources of entropy
and .δ Q is the only source of entropy. On the other hand, the temperature .T is the
potential (or force) for entropy transfer, the entropy flux taking place from higher to
smaller temperatures (Fig. 3.2). The second source of internal energy is the compres-
sion work. Compression work can be only performed by the transfer of volume and
the pressure is the potential for this transfer. Nevertheless, when a system increases
its volume because its pressure is higher than the pressure of its neighbor’s surround-
ings, the volume is always transferred from places with lower pressures to places
with higher pressure. This explains the negative sign before . Pd V in Eq. (3.2).
The temperature .T and the pressure . P can be also reinterpreted based on Eqs.
(3.3) and (3.4), respectively. Temperature is a measure of the kinetic energy of the
molecules, but, based on Eq. (3.3) it can be reinterpreted as the measure of the internal
energy variation .ΔU of a system when this system increases its entropy by .ΔS, in a
process where the volume is kept constant. Similarly, the pressure . P, which is related
to the momentum of the molecules, can be also reinterpreted based on Eq. (3.4).
90 3 The Equilibrium State

3.2 Open Systems and Gibbs Potential

It is universally recognized that Gibbs work published by the Connecticut Academy was an
event of the first importance in the history of chemistry. Nevertheless, it was a number of
years before its value was generally known, this delay was due largely to the fact that its
mathematical form and rigorous deductive processes make it difficult to read for anyone,
especially for students of experimental chemistry whom it most concerns. J. J. O’Connor
and E. F. Robertson, 19971

Lorsqu’il s’était déterminé à faire imprimer les résultats de l’une de ses études, Gibbs ne
cherchait nullement à leur assurer une diffusion rapide et étendue; les grands recueils scien-
tifiques, grâce auxquels une pensée est bientôt communiquée à tous ceux qui sont capables
d’en faire bon usage, n’eurent pas à transmettre les découvertes du professeur du Yale Col-
lège; ces découvertes, il les confia presque toujours à des recueils américains d’une moindre
vogue, comme s’il eût éprouvé quelque regret de leur avoir donné libre vol ; ses premiers
mémoires de Thermodynamique et, en particulier, son grand travail Sur l´ équilibre des
substances hétérogènes parurent dans les Transactions, bien peu répandues, que publiait,
depuis peu, à New-Haven, l’Académie des Arts et des Sciences du Connecticut. Il semble
parfois qu’en publiant ses travaux, Gibbs eût été possédé du désir de les voir passer inaperçus;
s’il en fut ainsi, il fut bien souvent servi à souhait; bien souvent, ses idées demeurèrent
ignorées de ceux-là mêmes qui auraient eu le plus grand intérêt à les connaitre. (Pierre
Duhem, 18862 )

Open systems can change their internal energy by mass transfer. Consider
Fig. 3.3a. Chamber A is filled with a liquid and is separated from an evacuated
chamber B by a metallic plate. When this plate is removed the liquid will evapo-
rate till the vapor pressure . P in chamber B reaches the saturation pressure . Ps (T )
(Fig. 3.3b).
Consider now Fig. 3.3c. Each molecule that reaches the liquid surface is able to
migrate to the vapor phase. Nevertheless, all the remaining molecules in the liquid
phase will try to keep it in this phase with attraction forces that are essentially
electrostatic forces.3 The molecule shown in the sketch has a kinetic energy . E c ,
which is positive and related to the temperature T. However its potential energy .Φ is
negative and related to the net attraction force .F that is pulling it back to the liquid
phase, in accordance with

.F = −∇Φ.

Therefore, this molecule will only jump to the vapor phase when its mechanical
energy . E c + Φ is greater than zero. In other words, when its kinetic energy is high

1 O’Connor, John J.; Robertson, Edmund F. (1997). Josiah Willard Gibbs. The MacTutor History
of Mathematics archive. University of St Andrews, Scotland. School of Mathematics and Statistics.
2 Duhem, P., Josiah Willard Gibbs, à propos de l publication de ses mémoires scientifiques, A.

Hermann, Paris, 1908 [46].

3 The book of Adamson [2] is suggested for readers interested in better understanding the nature

of these intermolecular forces.

3.2 Open Systems and Gibbs Potential 91

T P

B B
vacuum
evaporation
plate

liquid liquid intermolecular forces

A A
(a) (b) (c)

Fig. 3.3 Open systems can change their internal energy by mass transfer

enough to counterbalance the potential energy .Φ that is due to its interaction with its
neighboring liquid molecules. If this is the case its residual energy will be transferred
from the liquid to the vapor phase.
From a macroscopic point of view and considering the reasoning given at the
beginning of this section, the variation in internal energy in the vapor phase due to
the flux of .d N moles from the liquid phase would be written as

dUv = μ̄v d N ,
.

where .μv is a sort of potential for mass transfer in the vapor phase—the Gibbs or
chemical potential.4 In the same way, the variation of internal energy in the liquid
phase due to this transfer will be

.dUl = −μ̄l d N .

So, the liquid will evaporate till the Gibbs potentials .μv and .μl are identical. In
this condition, the vapor pressure reaches the saturation value . Ps (T ).

4 Josiah Willard Gibbs (1839–1903) was an American scientist who made important theoretical

contributions to physics, chemistry, and mathematics. His work on the applications of thermody-
namics was fundamental for transforming physical chemistry into a rigorous deductive science.
Together with James Clerk Maxwell and Ludwig Boltzmann, he created statistical mechanics (a
term that he coined), explaining the laws of thermodynamics as consequences of the statistical
properties of ensembles of the possible states of a physical system composed of many particles. He
extended the thermodynamic analysis to multi-phase chemical systems and considered a variety
of concrete applications. He described that research in a monograph titled ’On the Equilibrium
of Heterogeneous Substances’, published by the Connecticut Academy in two parts that appeared
respectively in 1875 [59] and in 1877 [60].
92 3 The Equilibrium State

3.3 Intensive and Extensive Variables

For open systems, Eq. (3.2) must be rewritten as

dU = T d S − Pd V + μ̄d N ,
. (3.5)

to take the mass transfer into account as a source of internal energy.

This equation expresses the extensive state variable .U as a function of the exten-
sible variables . S, .V , and . N . Extensive variables are dependent on the number of
moles . N —or on the mass .m—of the system. On the other hand, intensive variables
can be written as the ratio of extensive variables and, thus, are independent of the
mass. This is the case for the temperature in Eq. (3.3), pressure in Eq. (3.4) and is
also the case for the Gibbs potential

)
∂U
.μ̄ = .
∂N S,V

When using the mass .m of the thermodynamic system—instead of the number of

moles—Eq. (3.5) would be written as

dU = T d S − Pd V + μdm,
.

where

)
∂U
μ=
. .
∂m S,V

3.4 Homogeneity of the Equilibrium State

Thermodynamic systems in equilibrium are homogeneous. So, by partitioning such

a system into equal parts, the extensive properties of each part is a corresponding
fraction of the total, while the intensive properties remain the same throughout the
whole system (Fig. 3.4).
So if .λ is the inverse of the number of parts by which the system was partitioned

U (λS, λV, λN ) = λU (S, V, N ) .

. (3.6)
3.4 Homogeneity of the Equilibrium State 93

Fig. 3.4 Thermodynamic U(S,V,N)

systems in equilibrium are
homogeneous. For a given
partition, the extensive S/4
properties are divided by the V/4
number of parts, whereas the N/4
intensive properties remain U/4
the same throughout the
whole system

Consider now that .λ is variable, while the system remains the same for every
partitioning. By differentiating Eq. (3.6) with respect to .λ

∂U (λS, λV, λN ) ∂U (λS, λV, λN ) ∂U (λS, λV, λN )

. S +V +N = U.
∂ (λS) ∂ (λV ) ∂ (λN )
, ,, , , ,, , , ,, ,
=T =−P =μ̄

Each derivative in the above equation is an intensive property related to the par-
titioned part where the internal energy is a fraction .λU of the total, the entropy is
.λS, the volume is .λV and the number of moles (or mass) is .λN . Therefore, for a
thermodynamic system in equilibrium

U = T S − P V + μ̄N .
. (3.7)

The above equation is a general consequence of homogeneity expressed by the

Euler relation, Eq. (3.6). When a thermodynamic system is doubled all the extensive
variables (entropy, volume and number of moles) are doubled, including the internal
energy, but the intensive variables remain with the same values.
Consider now a process from a given equilibrium state to another one. By differ-
entiating Eq. (3.7) and considering Eq. (3.5), we obtain

0 = SdT − V d P + N d μ̄.
.

So, if we divide the above equation by the number of moles . N , we get the Gibbs-
Duhem equation,5 relating the Gibbs potential with the temperature and pressure

5Pierre Maurice Marie Duhem (1861–1916) was a French physicist, mathematician, historian, and
philosopher of science. He is best known for his work on chemical thermodynamics, for his philo-
94 3 The Equilibrium State

.d μ̄ = −s̄dT + v̄d P, (3.8)

where .s̄ is the molar entropy and .v̄ is the molar volume. The Gibbs-Duhem equation
enables us to find the chemical potential in terms of the temperature and pressure
when the molar entropy .s̄ and the molar volume .v̄ are given in terms of .T and . P.
Consider now the homogeneity relation Eq. (3.7). By dividing this equation by
.N

. ū = T s̄ − P v̄ + μ̄,

and so

d ū = T d s̄ − Pd v̄+s̄dT − v̄d P + d μ̄,

.
, ,, ,
=0

Therefore
d ū = T d s̄ − Pd v̄.
. (3.9)

The above equation is a differential equation that can be solved for expressing
the dependence of the intensive variable, the molar internal energy .ū in terms of the
molar entropy .s̄ and the molar volume .v̄. Although both Eqs. (3.5) and (3.9) were
derived for open systems, the former equation means an Eulerian description and
the second one, a Lagrangian description. In other words, while Eq. (3.5) applies to
control volumes, accounting for the influence of mass transfer on the total internal
energy of a fixed volume, Eq. (3.9) describes the variation of the internal energy of a
fixed mass of fluid—a single mole, in this case—when this single mole is displaced
along the flow lines (Fig. 3.5).

3.4.1 Exercises

1. What is the physical meaning of the Euler homogeneity relation Eq. (3.7)? Try
to translate this relationship into words. What happens with the intensive and
extensive variables of a thermodynamic system in equilibrium when this thermo-
dynamic system is doubled?
2. Write Eq. (3.9) in terms of the specific entropy .s, specific internal energy .u and
mass density .ρ. This kind of representation is usual in non-equilibrium thermo-
dynamics when dealing with material derivatives .dρ/dt, .du/dt, and .ds/dt.

sophical writings on the indeterminacy of experimental criteria, and for his historical research into
the science of the European Middle Ages. As a scientist, Duhem also contributed to hydrodynamics
and to the theory of elasticity. His thermodynamics book was written in 1886: Duhem P (1886) Le
potentiel thermodynamique et ses applications. A. Hermann, Paris [45].
3.5 Legendre Transforms of the Internal Energy 95

V(t+ t) A(t+ t)

V(t)

A(t)

Fig. 3.5 Lagrangian description. Volume V(t) is a material volume and although it is deformed
along the flow, the mass inside it is constant

3.5 Legendre Transforms of the Internal Energy

The Legendre transforms6 are commonly used in classical mechanics to derive the
Hamiltonian formalism out of Lagrangian formalism and in thermodynamics to
derive the thermodynamic potentials.
Consider the function .Y = Y (X ) shown in Fig. 3.6. We seek for a transformation
.Y → L that preserves all the information given by .Y (X ), in such a way that .Y can
be fully retrieved by the inverse transform .L → Y . Figure 3.6 shows that this can
be accomplished if, for each point . X of the curve .Y (X ), we take the derivative
'
.Y (X ) = P and the ordinate where the tangent to .Y at point . X crosses the .Y -axis,
or .L. So, from,

Y −L
. P= ,
X
the Legendre transform .L = L (P) of .Y = Y (X ) will be given by

.L = Y − P X.

6Adrien-Marie Legendre (1787). Mémoire sur l’intégration de quelques équations aux différences
partielles. Mémoires de l’ Academie des Sciences. pp. 309–351. [103].
96 3 The Equilibrium State

Fig. 3.6 Given a function Y

= Y(X), the Legendre
transform .L = .L(P) preserves
all the information given by
the representation Y = Y(X)

Therefore, in the transformed representation the dependent variable is .L and

the independent variable is . P = dY/d X . This concept can be easily generalized to
functions with several independent variables.

3.5.1 Helmholtz Energy

The Helmholtz energy was developed by Hermann von Helmholtz and is usually
denoted by the letter A (from the German “Arbeit” or work), or the letter F. In
physics, the Helmholtz energy is sometimes referred to as the Helmholtz function,
Helmholtz free energy, or simply free energy. The Helmholtz energy . F is defined
as the Legendre transform of the internal energy that replaces the entropy as an
independent variable by the temperature .T = (∂U/∂ S)V,N ,

. F = U − T S. (3.10)

From Eq. (3.5),

d F = −SdT − Pd V + μ̄d N .
. (3.11)

So, in the Helmholtz energy representation, the independent variables are .T , .V ,

and . N instead of . S, .V , and . N . From Eqs. (3.10) and (3.7)
3.5 Legendre Transforms of the Internal Energy 97

. F = −P V + μ̄N . (3.12)

In terms of intensive properties, the above equation can be written as

. f¯ = −P v̄ + μ̄,

and so

d f¯ = −Pd v̄ − v̄d P + d μ̄,

.

or, from the Gibbs-Duhem equation, Eq. (3.8),

d f¯ = −s̄dT − Pd v̄,
. (3.13)

which is the differential equation for the Helmholtz energy in its Lagrangian differ-
ential form.

3.5.1.1 The Meaning of the Helmholtz Energy

Consider an isothermal reversible expansion of a gas as pictured in Fig. 2.4. From

the first law

δ Q + δW = dU = d F + T d S,
.

or, since .δ Q = T d S because the expansion is reversible

δW = d F,
.

and

. W1−2 = F2 − F1 .

Since this expansion is reversible, the work delivered by the gas in its isothermal
expansion is maximum and this maximum work corresponds to the variation of its
Helmholtz energy between the initial and final equilibrium states.
Therefore, since the expansion work.W1−2 is negative, when the isothermal expan-
sion is arbitrary,

. |W1−2 | ≤ |F2 − F1 | ,

or,
δW ≥ d F,
. (3.14)
98 3 The Equilibrium State

the equality sign applying solely for reversible expansions.

While internal energy can be considered as the available work in adiabatic expan-
sions, the Helmholtz energy can thus be seen as the work that is available (or free)
in isothermal expansions.
The same reasoning can be used for isothermal compression, leading to the same
inequality, Eq. (3.14), which is a general rule in isothermal processes.
Considering the entire cycle
∮
. δW ≥ 0,

which is the same result presented in Chap. 2 of this book: In isothermal cycles the
absolute value of the compression work is always greater than the absolute value of
the expansion work.
Beyond its meaning as an available work, Helmholtz energy is widely used in
surface physics and is directly related to the surface tension of liquids. Readers are
invited to read Chap. 6 on surface physics for acquiring a better familiarity with this
important thermodynamic concept.

3.5.2 Enthalpy

The word enthalpy is based on the ancient Greek verb enthalpein, which means ‘to
warm in’. The name enthalpy is often incorrectly attributed to Benoît Paul émile
Clapeyron7 and Rudolf Clausius8 through the 1850 publication of their Clausius—
Clapeyron relation. However, neither the concept, the word, nor the symbol for
enthalpy existed until well after Clapeyron’s death.
The earliest writings to contain the concept of enthalpy did not appear until 1875,
when Josiah Willard Gibbs introduced “a heat function for constant pressure”. How-
ever, Gibbs did not use the word enthalpy in his writings [59, 60];
The actual word first appeared in the scientific literature in a 1909 publication by
J. P. Dalton.9 Proceedings of the Royal Netherlands Academy of Arts and Sciences
11: 863–873 [37]. According to that publication, Heike Kamerlingh Onnes (1853–
1926)10 actually coined the word.11

7 Clapeyron, M. C. (1834). Mémoire sur la puissance motrice de la chaleur. Journal de l’École

polytechnique 23: 153–190. [33].
8 Clausius, R. (1850). Ueber die bewegende Kraft der Wärme und die Gesetze, welche sich daraus

für die Wärmelehre selbst ableiten lassen. Annalen der Physik 155: 500–524. https://doi.org/10.
1002/andp.18501550403 [34].
9 J. P. Dalton, (1909). Researches on the Joule—Kelvin effect, especially at low temperatures. I.

Calculations for hydrogen.

10 H. Kamerlingh Onnes, Dutch physicist who first liquefied helium in 1908, discovered supercon-

ductivity in 1911, and won the Nobel Prize for physics in 1913.
11 In accordance with Hendrick C. Van Ness [182], Kamerlingh Onnes seems never to have used

the word enthalpy in any publication. A message from the Kamerlingh Onnes Laboratory at Leiden
3.5 Legendre Transforms of the Internal Energy 99

Over the years, many different symbols were used to denote enthalpy. In accor-
dance with Irmgard Howard [71] it was not until 1922 that Alfred W. Porter12 pro-
posed the symbol ‘H’ as the accepted standard, thus finalizing the terminology still
in use today.
The enthalpy is the Legendre transform of the internal energy that replaces the
volume with the pressure as the independent variable

. H = U − (−P V ) . (3.15)

So, from Eq. (3.5),

.d H = T d S + V d P + μ̄d N . (3.16)

In the enthalpy representation, the independent variables are . S, . P, and . N instead

of . S, .V , and . N . From Eqs. (3.15) and (3.7)

. H = T S + μ̄N .

In terms of intensive properties, the above equation can be written as

. h̄ = T s̄ + μ̄,

and so

d h̄ = T d s̄ + s̄dT + d μ̄,
.

or, from the Gibbs-Duhem equation, Eq. (3.8),

. d h̄ = T d s̄ + v̄d P, (3.17)

which shows the enthalpy in its Lagrangian differential form.

3.5.2.1 The Meaning of Enthalpy

Consider the isobaric expansion shown in Fig. 2.40. From the first law of thermody-
namics,

suggests that he likely proposed the word orally at the first meeting of the Institute of Refrigeration
in Paris in 1908. However, nothing appears in the Comptes rendus du Congres international du
froid, 1908.
12 Alfred W Porter was President of the Faraday Society in 1920–1922.
100 3 The Equilibrium State

.δ Q + δW = dU.

If there is no imbalance in the pressure field of the gas, a single internal pressure
is set up. In addition, the internal and external pressure are identical during the entire
process. So

dU = d H − Pd V − ,V,,
. d P, and δW = −Pd V.
=0

Consequently,

δ Q = d H.
.

Therefore, while the heat that is supplied to a constant volume system corresponds
to the variation of its internal energy, the variation of the enthalpy measures the
supplied heat when the pressure is maintained constant, and so it is ‘a heat function
for constant pressure’.
The concept of enthalpy is widely used in engineering, for solving energy balance
problems. In his book, ‘Heat and Thermodynamics’, Zemansky13 writes at page 275
of its 5.th Edition that enthalpy is a measurement of energy in a thermodynamic
system that “includes the internal energy and the amount of energy required to make
room for it by displacing its environment and establishing its volume and pressure”.
In effect, consider the air compressor of a turbojet, shown in Fig. 3.7. The control
volume V comprises the air that is being compressed and the compressor blades
enabling a global analysis by considering the power .Ẇsha f t that is transferred to the
compressor by the turbine shaft.

. air exit
Q turbine
air intake

V .
Wshaft

n Combustion nozzle
A chambers

Fig. 3.7 Energy balance for the air compressor of a turbojet

13 Zemansky, Mark W. (1968). Heat and Thermodynamics (5th ed.). New York, NY: McGraw-Hill.
3.5 Legendre Transforms of the Internal Energy 101

The total energy of this thermodynamic system is given by

∫ ( )
v2
. E= ρ u+ + g Z d V.
2
V

The time rate variation of the total energy .d E/dt is due to the following sources:
(a) the flow of energy across its boundary carried into the system with the fluid
∫ ( )
v2
.− ρ u+ + g Z v·ndA, (3.18)
2
A

where. A refers to the boundary surface,.n is the unitary vector normal to the boundary
surface and oriented toward its exterior;
(b) the pressure work or “the amount of energy required to make room for the fluid
when entering or exiting out of the control volume” is given by
∫
.− Pv·ndA; (3.19)
A

(c) the amount of heat that is exchanged between the system and its environment
∫
. Q̇ = − q·ndA; (3.20)
A

(d) the shaft power that must be taken from the turbine for compressing the air
. Ẇsha f t .

By assembling Eqs. (3.18–3.20) into a single equation

∫ ( ) ∫
dE v2
. =− ρ h+ + g Z v·ndA − q·ndA+Ẇsha f t , (3.21)
dt 2
A A

where the internal energy per unit mass .u and the pressure work . Pv = P/ρ were
assembled into a single energy, the enthalpy, through

ρh = ρu + P
.

Equation (3.21) can be further simplified in steady-state conditions, by supposing

the potential energy to be negligible and that the compression process is adiabatic.
It results
102 3 The Equilibrium State

∫ ( )
v2
. Ẇsha f t = ρ h+ v·ndA
2
A
[ ( 2 )]
ve v2
= ṁ (h e − h i ) + − i (3.22)
2 2

where symbols .e and .i refer, respectively, to the compressor exit and inlet sections,
h v for.s = i, e are average values of the enthalpy and velocity, respectively. Finally,
. s s

∫ ∫
.ṁ = − ρv·ndA = ρv·ndA,
Ai Ae

is the mass flow rate through the compressor in steady-state operation.

Equation (3.22) means that the power that must be delivered to the air compressor
from the turbine is the one that is necessary to increase the enthalpy and kinetic
energy of the air that is flowing through the compressor. Since the air can be usually
considered as an ideal gas, the enthalpy (per unit mass) can be written as

P
. h=u+ = u + RT,
ρ

in such a way that

.dh = du + RdT = c p dT.

3.5.3 Gibbs Energy

The Gibbs energy, sometimes called ‘available’ or ‘free’ Gibbs energy, was developed
in the 1870s by Josiah Willard Gibbs. In 1873, Gibbs described this energy as “the
greatest amount of mechanical work that can be obtained from a given quantity of
a certain substance in a given initial state, without increasing its total volume or
3.5 Legendre Transforms of the Internal Energy 103

allowing heat to pass to or from external bodies, except such as at the close of the
processes are left in their initial condition”.14
The Gibbs energy .G is the Legendre transform of the internal energy that replaces
the entropy and the volume, as independent variables, by, respectively the temperature
and pressure.

. G = U − (−P V ) − T S = H − T S.

So, from Eq. (3.5)

dG = −SdT + V d P + μ̄d N .
.

So, when a thermodynamic system is in a state of homogeneous equilibrium,

. G = μ̄N ,

or, in terms of intensive variables

. g = μ̄,

and

.dg = −s̄dT + v̄d P.

3.5.4 Exercises

1. Show that for an ideal gas, the specific entropy can be written as

dT dP
ds = c p
. −R ,
T P

where . R = R/M. With this information, find the entropy s an ideal gas would
have with respect to the entropy .s0 of a reference state given by .T0 , . P0 . Find,
also, the change in the chemical potential .μ (T, P) − μ0 (T0 , P0 ) with respect
to this reference state.

14 J. W. Gibbs (1873), A Method of Geometrical Representation of the Thermodynamic Properties

of Substances by Means of Surfaces Transactions of the Connecticut Academy of Arts and Sciences
2, pp. 382–404 [61].
104 3 The Equilibrium State

Fig. 3.8 Considering that

chambers A and B are in T
equilibrium with a thermal
bath at temperature .T , show
that the Helmholtz energy
. f = F/N is the same in the
two chambers
A
1 mol of N2

B piston free to move

2 moles of N2

2. In Fig. 3.8, chamber B has the double of the volume of chamber A and both
chambers contain . N2 . Considering that the two chambers are in equilibrium with
a thermal bath at a temperature .T , show that the Helmholtz energy . f = F/N is
the same in the two chambers.
What would be your answer if the . N2 is replaced by . O2 in chamber A?…
3. The thrust of a turbojet is its main design variable and gives the propellant force
of the turbojet. This force can be derived from Newton’s second law as

. F = ṁ nozzle exit vnozzle exit − ṁ compr essor inlet vcompr essor inlet

(a) Suppose that the compression ratio in the compressor is 8:1 and that the air
enters the compressor inlet at the atmospheric pressure and.T = 250 C. What
would be the air temperature at the compressor outlet, if the compression is
supposed to be isentropic?
(b) The Boeing 777 has a GE-90 turbojet with a thrust of.4.168 × 105 . N in static
conditions. Considering that the fuel is kerosene with a fuel-air mass ratio
of 0.0137, what should be the air mass flow rate when the intake velocity
is 40 .m/s and the nozzle exit velocity cannot exceed 340 .m/s (the speed of
sound)? What should be, in your calculations, the minimum inlet area of
3.6 Maxwell Relations 105

the GE-90 turbojet for enabling such air mass flow rate (the diameter of the
GE-90 turbo-jet is 7.29 m).
(c) The take-off speed of a Boeing 777 is about 100 .m/s. Consider that a Boeing
777 with two GE-90 turbojets has a maximum take-off weight of 405,000
Kg. Neglecting friction, what is the minimum take-off length of the Boeing
777?
(d) The compressor exit area of the GE-90 is about 40 inlet area. Estimate the
shaft power that the turbojet turbine must deliver to the compressor.

3.6 Maxwell Relations

The Maxwell relations follow directly from the fact that the order of differentiation
of an analytic function of two variables is irrelevant (Schwarz theorem). In the case
of Maxwell relations the function to be considered is a thermodynamic variable Z
and X and Y are two different independent variables for that function,
(( ) ) (( ) )
∂ ∂Z ∂ ∂Z
. = . (3.23)
∂Y ∂X Y X ∂Y ∂X Y X

Now, we rewrite the main differential equations that were derived in this chapter,
beginning with Eq. (3.9)
.du = T ds − Pdv, (3.24)

that expresses the specific internal energy .u in terms of the specific entropy .s and
specific volume .v, .u = u (s, v). We have
( ) ( )
∂u ∂u
. T (s, v) = , P (s, v) = − .
∂s v ∂v s

The specific Helmholtz energy . f creplaces the entropy .s, as independent variable
by the temperature .T , resulting in the differential equation

d f = −sdT − Pdv,
. (3.25)

expressing the Helmholtz energy. f in terms of the temperature.T and specific volume
v,. f = f (T, v). In this representation the entropy and the pressure are also dependent
.
on .T and .v, ( ) ( )
∂f ∂f
.s (T, v) = − , P (T, v) = − . (3.26)
∂T v ∂v T

For the enthalpy .h (s, P)

106 3 The Equilibrium State

Fig. 3.9 Application of the

Schwarz theorem in deriving
differential relationships for
the thermodynamic
properties

dh = T ds + vd P,
. (3.27)

with
( ) ( )
∂h ∂h
. T (s, P) = , v (s, P) = .
∂s P ∂P s

Finally, the differential equation for the Gibbs energy .g (T, P)

dg = −sdT + vd P,
. (3.28)

with
( ) ( )
∂g ∂g
s (T, P) = −
. , v (T, P) = . (3.29)
∂T P ∂P T

Now, let us take the derivative of the entropy in Eq. (3.26) with respect to the
specific volume
( ) ( ( ))
∂s (T, v) ∂ ∂f
. =− .
∂v T ∂v ∂T v T

By changing the independent variables in accordance with the Schwarz theorem

( ) ( ( ) )
∂s (T, v) ∂ ∂f
. =− .
∂v T ∂T ∂v T v

But, in accordance with Eq. (3.26)

( )
∂f
. = −P,
∂v T

and so
( ) ( )
∂s ∂P
. = . (3.30)
∂v T ∂T v
3.6 Maxwell Relations 107

Table 3.1 Differential equations for the thermodynamic functions

Thermodynamic function Differential equation
Internal energy .du = T ds − Pdv
Enthalpy .dh = T ds + vd P
Helmholtz energy .d f = − sdT − Pdv

Gibbs energy .dg = − sdT + vd P

This is an important relationship since it enables to calculate the entropy .s (T, v)

in terms of the temperature and volume when the equation of state of the thermody-
namic system . P (T, v) is known. Figure 3.9 illustrates the use of Schwarz theorem
in deriving differential relationships for the thermodynamic properties. Table 3.1
summarizes the differential equations for the main thermodynamic functions.

3.6.1 The Entropy as a Function of the Temperature

and Volume

Consider that we want to express the entropy in terms of the temperature and specific
volume .s (T, v).
So, ( ) ( )
∂s ∂s
.ds = dT + dv. (3.31)
∂T v ∂v T

In the first part of this book, we saw that the specific heat capacity at constant
volume can be written as
( )
δQ
.cv = lim .
ΔT →0 mΔT N ,V

Nevertheless, when a thermodynamic system is heated at a constant volume and

mass, the expansion work is null and

. δ Q = dU,

and, so, .cv can be written as

( )
∂u
c =
. v ,
∂T v

or, using Eq. (3.24) and considering that the measurement of the heat capacity must
be performed with .ΔT → 0,
108 3 The Equilibrium State
( )
∂s
c =T
. v .
∂T v

So, replacing the above equation and Eq. (3.30) into Eq. (3.31) it results
( )
cv ∂P
.ds = dT + dv. (3.32)
T ∂T v

Therefore if we know the heat capacity .cv in terms of .T and .v, and the equation
of state, it is possible to integrate the above equation and find the entropy .s for a
thermodynamic state given by its intensive variables, the temperature .T and specific
volume .v.

3.6.2 The Internal Energy as a Function of the Temperature

and Volume

From Eqs. (3.24) and (3.32) a differential equation can be derived for the internal
energy, considering the temperature .T and volume .v as the independent variables
( ( ) )
∂P
. du = cv dT + T − P dv. (3.33)
∂T v

This equation can be used to write the internal energy .u in terms of .T and .v, when
we know the equation of state for the fluid and .cv in terms of .T and .v.
Let us consider, for instance, that the fluid is an ideal gas. In this case,
( )
∂P
. T − P = 0.
∂T v

So, for an ideal gas,

du = cv dT.
.

The above equation does not mean that the internal energy of an ideal gas is only
dependent on the temperature. In fact, the heat capacity .cv is in general a function of
two state variables, such as .T and .v or .T and . P or . P and .v. In the present case, the
independent variables are .T and .v, and if we want to show that the internal energy
of an ideal gas is only dependent on the temperature we must prove that

( )
∂cv
. = 0.
∂v T

See Exercise 2 of the Exercises list.

3.6 Maxwell Relations 109

3.6.3 Exercises

1. Show that when the molar entropy .s̄ is expressed in terms of the temperature .T
and pressure . P, its differential equation will be written, on a molar basis, as
( )
c̄ p ∂ v̄
.d s̄ = dT − d P.
T ∂T P

This equation is frequently used in expansion or compression devices involving

an ideal gas when .(∂ v̄/∂ T ) P = R/P and .c̄ p is only dependent on the tempera-
ture. ( v)
2. Show that . ∂c
∂v T
= 0 for an ideal gas, which equation of state is . Pv = RT .
Show also that the heat capacity .c p does not depend on the pressure for an ideal
gas.
3. Using the Maxwell relations, show that
( ( ) )
dT ∂v
dh = c p
. + v−T d P.
T ∂T P

4. Mathematical skills. Consider a function . Z = f (X, Y ). By considering

) )
∂Z ∂Z
dZ =
. dX + dY,
∂X Y ∂Y X

show that when this function has an inverse .Y = f (X, Z ) then, in each plane
where . Z = const.,
) ∂Z
)
∂Y ∂X Y
. =− ∂Z
) .
∂X Z ∂Y X

This relationship is of great importance for deriving thermodynamic relations

between state variables.
In the same way, the following relations can, almost always, be used for ther-
modynamic variables

)
∂Y 1
. = ∂X
) (inversible functions) ,
∂X Z ∂Y Z
) )
∂Y ∂Y ∂ W
= (chain rule) .
∂X Z ∂W ∂ X Z

5. Consider an ideal gas and show why

110 3 The Equilibrium State
) ) )
∂T ∂T ∂T
. , , ,
∂P v ∂P s ∂P h

have different meanings. Try to illustrate these meanings using words instead of
equations.
6. The speed of sound is given by
√ )
∂P
c =
. s .
∂ρ s

Find the speed of sound for an ideal gas.

7. Reducing the pressure can have a refrigerating effect on real gases. What happens
with the temperature, when the pressure is reduced following an isenthalpic
process? Calculate
)
∂T
. ,
∂P h

and base your analysis on the van der Waals equation of state

RT a
. P= − 2.
v̄ − b v̄

3.7 Minimum Principles

As we saw in Chap. 2 of Part I of this book, the entropy measures the tendency of a
thermodynamic system to reach the equilibrium state. It attains its maximum value
when this equilibrium state is reached; furthermore, we are sure that if this system
does not exchange heat with its environment, all the internal processes resulting
from internal non-equilibrium in its intensive variables will contribute to increase
the entropy.
We can also say that the entropy will not decrease if the system is isolated and, so,
does not exchange mass, heat and work with its environment. Therefore, when the
internal energy .U is constant, all the internal processes will contribute to an increase
in the entropy

d S ≥ 0 when U = const.
.

We can illustrate this principle using a very simple example. Consider a ther-
mally insulated system composed of two chambers A and B (Fig. 3.10). These two
3.7 Minimum Principles 111

Fig. 3.10 When the internal

energy .U = const. the
internal non-equilibrium will
increase the entropy

chambers are separated by a thin plate that does not offer any resistance to the flow
of heat between chambers A and B. So, we can consider that, in equilibrium states
of chambers A and B, .T A = TB = T . Nevertheless, this plate is fixed to the walls
by pins, and the pressures . PA and . PB will be different. By letting the plate free, by
supposing that we can do that in an infinitesimal process, the volumes .V A and .VB
will change by .d V A and .d VB = −d V A , respectively, the two chambers, individu-
ally, reaching other equilibrium states. The same comment for the internal energy,
because .d (U A + U B ) = 0.
Since, in equilibrium states, . S A = S A (V A , U A , N A ) and . S B = S B (VB , U B , N B ),
the respective changes in the entropy will be

∂ SA ∂ SA ∂ SA
. d SA = d VA + dU A + d N A.
∂ VA ∂U ∂N
,,,, , ,,A, , A,, ,
=
PA
= T1 =0
T

In the same way

∂ SB ∂ SB ∂ SB
d SB =
. d VB + dU B + d NB .
∂ VB ∂U ∂N
,,,, , ,,B, , B,, ,
=
PB
= T1 =0
T

Therefore, the change of entropy of the whole system between these two equilib-
rium states will be
(PA − PB )
.d S = d S A + d SB = d V A + T (dU A + dU B )
T , ,, ,
=0

So, if . PA > PB , .d V A must be positive for assuring .d S > 0. In addition, .d S = 0 if

and only if . PA = PB , or when the whole system A+B is in equilibrium.
112 3 The Equilibrium State

Fig. 3.11 Isentropic

expansion for showing the
minimum principle for the
internal energy

weightless
B

Thermal insulation
A

3.7.1 Minimum Principle for the Internal Energy

Consider now the inverse problem pictured in Fig. 3.11. A gas in chamber A is
maintained at a high pressure . PA by a certain amount of very small balance weights
with mass .Δm. The piston is considered to be weightless and the two chambers are
separated by a very thin plate that does not offer any resistance to the flow of heat
between chambers A and B, in such a way that in equilibrium states of chambers A
and B, .T A = TB = T .
When a single balance weight is removed from the upper piston plate, the gas
in chamber A has an expansion .d V A . Since .Δm is very small, we can consider this
expansion to be reversible and, so, isentropic since the whole system is separated
from the external environment. In equilibrium states .U A = U A .(V A , S A , N A ) and,
so,

∂U A ∂U A ∂U A
. dU A = d SA + d VA + d N A.
∂S ∂V ∂N
, ,,A, , ,,A, , A,, ,
=T =−PA =0

In the same way

∂U B ∂U B ∂U B
.dU B = d SB + d VB + d NB .
∂S ∂V ∂N
, ,,B, , ,,B, , B,, ,
=T =−PB =0
3.7 Minimum Principles 113

Summing the two above equations

dU = T (d S A + d S B ) − (PA − PB ) d V A
.
, ,, ,
=0

Since. PA > PB ,.d V A will be greater than zero and.dU will decrease. Furthermore,
dU = 0, if and only if . PA = PB , which means the whole system A+B has attained
.
its equilibrium state.
A more theoretical demonstration can perhaps help to convince ourselves that
internal non-equilibrium between parts of a thermodynamic system that are them-
selves in equilibrium, contribute to reducing the internal energy in isentropic pro-
cesses. In fact, since the internal energy .U of the whole system will be dependent on
the volume of the chamber . A
( )
( ) ∂S
∂U ∂ V A U,V
.dU = d VA = − ( ∂ S ) d VA (3.34)
∂ V A S,N ∂U V ,V A

( ∂S ) ( )
Since . ∂U is the temperature of the whole system, the sign of . ∂∂U will
V A ,V V A S,V
( )
be the inverse of . ∂∂VSA and we know that in processes where the internal energy
U,V
U is a constant . ∂∂VSA ≥ 0.
.
When the system attains the full equilibrium state, in such a way that . PA = PB ,
we can consider the second derivative of Eq. (3.34)

( ) ( )( )
( ) ∂2 S
T− 1 ∂S
∂ U 2 ∂ V A2 U,V
∂U
∂T
∂ VA U,V
. =− >0
∂ V A2 S,V
T2
( )
because in the equilibrium state, the entropy is a maximum, . ∂∂VSA = 0 and
( 2 ) U,V
∂ S
.
∂V 2
< 0.
A U,V
In conclusion:
If a thermodynamic homogeneous system is in equilibrium, its entropy will be a maximum
when compared to other states with the same internal energy

Reversely:

The equilibrium value of any unconstrained internal parameter .— in this case, the volume
.V A — in a system is the one that minimizes the internal energy at constant volume and
entropy.
114 3 The Equilibrium State

Fig. 3.12 Minimum

principle for enthalpy P

Thermal insulation
B

A R

3.7.2 Minimum Principle for Enthalpy

To illustrate that enthalpy has also a minimum principle consider Fig. 3.12. One
mole of a gas is contained in each chamber A and B of a thermodynamic system.
Both chambers are insulated as well as the reservoir with which the system can
exchange work, but not heat. This “pressure reservoir” is sketched by imagining that
all pressure fluctuations will be absorbed by the reservoir. In the same way, the upper
and lower plates separating the system from the pressure reservoir and chamber A
from the chamber B are considered to be free to move along the inner cylinder wall.
So, the pressure . P is assured to be the same everywhere.
Nevertheless temperature .TB is, initially, greater than .T A and this is the reason
the volume of chamber B is greater than the volume of chamber A. Now, suppose
that we are capable of removing the thermal insulation separating chamber A from
chamber B and, quickly, replace it again. During this process, heat will flow from
the chamber B to chamber A, increasing its volume and this will produce a small
pressure perturbation, absorbed by the reservoir.
What we can say is that the enthalpy variation between the two equilibrium states
of the whole system, at constant pressure, is given by

d H = d H A + d H B = T A d S A + TB d S B ,
.

Now, the entropy variation of the thermodynamic system will be

3.7 Minimum Principles 115

. d S = d S A + d SB ,

and so

d H = (T A − TB ) d S A + TB d S.
.

Consider that the process is isentropic in such a way that .d S = 0. We know that

d SA ≥ 0
.

because heat will flow from the hot chamber B to the cold chamber A.
When these conditions are satisfied

.d H ≤ 0,

and the principle of minimum enthalpy can be enunciated as: “The equilibrium value
of any unconstrained internal parameter .— in this case, the entropy . S A — in a
system in contact with a pressure reservoir is the one that minimizes the enthalpy at
constant pressure and total entropy. ”

3.7.3 Minimum Principle for the Helmholtz Energy

It is difficult to make use of the minimum principle for internal energy or enthalpy in
practical applications. In fact, isentropic processes are rather theoretical processes.
On the other hand, all we need to perform an isothermal process is a thermal bath.
So, consider the system pictured in Fig. 3.13.
When the pins are removed, enabling chamber A to have a small expansion, we
can say that

.d (S + S R ) ≥ 0, d (U + U R ) = 0,

where . S, .U are the entropy and internal energy of the thermodynamic system—the
two chambers A and B—and R refers to the thermal reservoir.
Since the temperature remains constant, the variation of the Helmholtz energy of
the system between the two equilibrium states is

d F = dU − T d S,
.

or
116 3 The Equilibrium State

Fig. 3.13 Minimum

principle for the Helmholtz
T
energy

A B

Thermal reservoir

d F = − dU R − T d S.
.

Now

dU R = T d S R ,
.

related to the net heat that is exchanged between the system and the heat reservoir.
So,

d F = −T (d S R + d S) ≤ 0
.

and .d F = 0 when and only when .d S + d S R = 0, or when the whole system has
reached its equilibrium state at the reservoir temperature.
The minimum principle for the Helmholtz energy can be also demonstrated by
considering that, after the pins are removed and replaced again, the Helmholtz energy
of partitions A and B will change by

.d FA = −PA d V A , d FB = −PB d VB

and since .d VB = d V − d V A ,

d F = (PB − PA ) d V A − PB d V
.

in such way that if .d V = 0, .d F is necessarily non-positive, .d F ≤ 0, because

3.7 Minimum Principles 117

if PB ≤ PA then d V A ≥ 0
.

if PB ≥ PA then d V A ≤ 0

In conclusion, what we have demonstrated is that:

The equilibrium value of any unconstrained internal parameter—in this case, the volume
. V A —ina system is one that minimizes the Helmholtz energy at a constant volume and
temperature.

Thermodynamic systems are subjected to local non-equilibrium, sketched in this

section by using two complimentary partitions A and B that are themselves in equilib-
rium but at different pressures. When the pins are removed and replaced after a small
volume variation—supposing that this is possible—the two partitions attain differ-
ent equilibrium states at different pressures with respect to the ones they had at the
beginning. The use of partitions to demonstrate that in the equilibrium state—when
the pressures are the same—the Helmholtz energy has its lowest value was necessary
because in equilibrium thermodynamics, we are limited to equilibrium states, i.e., to
deal with states where all the intensive variables do not have any spatial variation.
This limitation can be avoided with the help of the methods that were developed for
describing non-equilibrium systems (Part III of this book).
Nevertheless, the above assertion can be rewritten—or generalized—for homo-
geneous system, i.e., without any partitions, in the following form:

If a thermodynamic homogeneous system is in equilibrium, its Helmholtz energy will be a

minimum when compared to other (non-equilibrium) states with the same temperature and
total volume

3.7.4 Exercises

1. Minimum principle for the Gibbs energy. A thermodynamic system is composed

of a cylinder with two partitions A and B where two different gases A and B are
contained (Fig. 3.14). Plates a and b are free to move inside the cylinder and we
can suppose that plate b is permeable to gas A, allowing this gas to mix with
gas B in partition B. The pressure is maintained constant by the three balance
weights and the atmospheric pressure and the temperature is kept constant with
the help of a thermal bath.
We know that

dG = −SdT + V d P + μd N .
.
118 3 The Equilibrium State

Fig. 3.14 Minimum

principle for the Gibbs T
energy

B
gas B

gas A

Fig. 3.15 Osmotic pressure

Show that the Gibbs energy has a minimum principle when the temperature T
and pressure P are maintained constant.
2. Osmotic pressure. A membrane permeable to water separates a solution of water
and salt in A from pure water in B (see Fig. 3.15).
In equilibrium, the pressure of the brine . PA will be larger than the water pressure
. PB and the difference . PA − PB is called osmotic pressure. Suppose that the

solution of water + salt may be considered an ideal solution with the Gibbs
energy given by
3.7 Minimum Principles 119
∑ ∑
. G(T, P, x1 , . . . , xr ) = Ni gi, pur o (T, P) + N RT xi ln xi . (3.35)
i i

(a) Calculate the chemical potential of the water in the solution and in the pure
water phase and show that

RT
. PA − PB = ln xwater . (3.36)
νwater (T )

(b) In the following show that when .xsalt is very small

RT
. PA − PB = xsalt , (3.37)
νwater (T )

which is the van’t Hoff’s law.15

15 Jacobus Henricus van ’t Hoff Jr. (1852–1911) was a Dutch physical chemist. A highly influential
theoretical chemist of his time, van ’t Hoff was the first winner of the Nobel Prize in Chemistry.
His pioneering work contributed to the fundamentals of the modern theory of chemical affinity,
chemical equilibrium, chemical kinetics, and chemical thermodynamics.
Chapter 4
Phase Equilibrium

Abstract We begin this chapter by following the efforts of Maxwell to show that
the attractive forces between molecules were responsible for vapor condensation,
at temperatures below the critical temperature found by Andrews for .CO2 . In his
1875 paper ‘On the dynamical evidence of the molecular constitution of bodies’,
Maxwell recognizes the impact of the 1873 Doctoral Thesis of Johannes Diderik van
der Waals for clearing the role of the interaction of real molecules on phase transition
and does not omit to mention the important work of Willard Gibbs who, in Maxwell’s
words, “ has given us a remarkably simple and thoroughly satisfactory method of
representing the relations of the different states of matter by means of a model”. Van
der Waals equation of state leads to a two-parameter law of corresponding states in
the sense that only the critical temperature and pressure are needed for describing
the liquid-vapor states of a substance. Most substances require a third parameter for
their description. The acentric factor, introduced by Pitzer in 1955, gave rise to a
sequence of empirically derived equations of state of substantial practice importance
in deriving liquid-vapor phase diagrams. The chapter ends with the presentation of
a method for calculating the thermodynamic functions of liquid-vapor systems.

4.1 Andrews, van der Waals and Maxwell

The molecular theory of the continuity of the liquid and gaseous states form the subject of an
exceedingly ingenious thesis by Mr. Johannes Diderik van der Waals, a graduate of Leyden.
There are certain points in which I think he has fallen into mathematical errors and his final
result is certainly not a complete expression for the interaction of real molecules, but his
attack on this difficult question is so able and so brave, that it cannot fail to give a notable
impulse to molecular science. It has certainly directed the attention of more than one inquirer
to the study of the Low Dutch language in which it is written. (Maxwell 1875)1

1J.C. Maxwell (1875), On the dynamical evidence of the molecular constitution of bodies, Nature
11 357- 359 [117].
© Associação Paranaense de Cultura 2024 121
P. C. Philippi, Thermodynamics, https://doi.org/10.1007/978-3-031-49357-7_4
122 4 Phase Equilibrium

Fig. 4.1 Andrews

experimental results for
carbon dioxide at page 583
of Ref. [5]

Pressure

Density

Andrews’s experiments published in 18692 showed that carbon dioxide has a

critical temperature of about 31.◦ C. Above this temperature, there is one fluid state
with a fixed density for each pressure and temperature. Below the critical temperature,
there are two densities for each pressure and temperature, the higher being the one of
the liquid and the lower corresponding to the vapor in equilibrium with it (Fig. 4.1).
Johannes Diderik van der Waals passed his doctoral examinations in December
1871. Eighteen months later he submitted his thesis ‘On the continuity of the gaseous
and liquid state’ [181]. It carries the date 14 June 1873, which was the day of his
public defense of it. “Like the early 19th century workers in kinetic theory, he was
very much the outsider and brought to the subject a new vision, but unlike them, he
was well versed in mathematics and physics and so was able to handle his subject
in a way that commanded respect even when it attracted criticism” [147].

2T. Andrews (1869), On the continuity of the gaseous and liquid states of matter, Phil. Trans. Roy.
Soc. 159 575–90 [5].
4.1 Andrews, van der Waals and Maxwell 123

P P

M M

Ps L I B V
A

m m
b b

v vls vvs
v’’ v’ v v
(a) (b)

Fig. 4.2 a Van der Waals equation of state has 3 roots for each pressure; b Maxwell’s area rule

He had two problems to solve. First, how to take into account the effect on the
pressure of attractive forces of unknown form but, he believes, of essentially short
range, that is, of a range comparable with the sizes of the molecules. He was one
of the first to emphasize that the cooling of real gases on expansion observed by
Joule and Thomson was a direct evidence for the existence of attractive forces in
gases. The molecules at the surface of a fluid are pulled inwards and the decay on
the pressure, . P, due to this attraction effect should be proportional to the probability
of two molecules being found together i.e., something proportional to the square of
the number density of molecules, .n 2 .
His second problem was to calculate the amount .b by which the observed volume
must be reduced by the space taken up by the molecules so as to give an effective
volume in which they move. He considered his molecules as hard spheres with a
given size.
The equation of state proposed in his thesis had, thus, the form
RT a
. P= − . (4.1)
v̄ − b v̄2

Van der Waals’s equation is a cubic equation in the volume. So, at a fixed pressure
and temperature it has three real roots (Fig. 4.2). The lowest .v'' and highest .v real
roots correspond to vapor and liquid states but the third root .v' at an intermediate
density has no real existence for it is a state for which .(∂ P/∂v)T is positive, and so
is to be considered unstable. Such a state, if formed, would spontaneously break up
into a mixture of vapor and liquid.
Indeed, from Eq. (3.13), the differential equation for the Helmholtz energy is

.d f = −Pdv
124 4 Phase Equilibrium

when the temperature is constant. An equilibrium state at a constant temperature

requires . f to be a minimum, or that
( (
∂2 f
. > 0.
∂v2 T

But
( ( ( ( ( (
∂2 f ∂ ∂f ∂P
. = =− > 0.
∂v2 T ∂v ∂v T ∂v T

Therefore, stable equilibrium states must satisfy

( (
∂P
. ≤ 0,
∂v T

and all the states between .m and . M in Fig. 4.2 must be considered as ‘non-physical’.
But, how to find the pressure where the vapor phase will change to the liquid
state?…
In his 1875 Nature paper [117], Maxwell proposed what is presently known as
the ‘area rule’: transition at a given isotherm will happen when area B under the
isotherm and above a line of constant pressure . Ps in Fig. 4.2b is equal to the area A
above this same isotherm and below line . Ps .
Maxwell’s paper is somewhat hard to follow for readers not familiar with kinetic
theory. It was entitled “On the dynamical evidence of the molecular constitution
of bodies”. In fact, Maxwell was more interested in the molecular theory than in a
practical rule for finding the saturation pressure. He wrote the equation of state as
⎛ ⎞
2⎝ ∑1 (|| ||)
.PV = Ec − Ri j · ||r j − ri || ⎠ (4.2)
3 i, j
2

where . E c is the kinetic energy of the molecules, .Ri j is the attraction force
(|| between ||)
any two molecules separated by.r j − ri . He writes: “The quantity . 21 Ri j · ||r j − ri || ,
or half the product of the attraction into the distance across which the attraction is
exerted, is defined by Clausius as the ‘virial of the attraction’…When a gas is in an
extremely rarefied condition, the number of particles within a given distance of any
one particle will be proportional to the density of the gas. Hence the virial arising
from the action of one particle on the rest will vary as the density, and the whole
virial in unit of volume will vary as the square of the density”.
So by dividing Eq. (4.2) by the volume Maxwell concluded that

1 2 2 2
. P= ρξ − Aρ
3 3
4.1 Andrews, van der Waals and Maxwell 125

where .ξ is the mean molecular speed and . A “is a quantity which is nearly constant
for small densities” . The above equation is very similar to the van der Waals equation
of state.
At page 359, he finally states his area rule (see Fig. 4.2b): “Now let us suppose the
medium to pass from V to L along the hypothetical curve VMImL in a state always
homogeneous, and to return along the straight line LV in the form of a mixture of
liquid and vapor. Since the temperature has been constant throughout, no heat can
have been transformed into work. Now the heat transformed into work is represented
by the excess of the area LmI over VMI. Hence the condition that determines the
maximum pressure of the vapor at a given temperature is that the line LV cuts off
equal areas from the curve above and below”.
At the end of his paper, he mentions the 1873 paper of Gibbs [61] recognizing that
“The purely thermodynamic relations of the different states of matter do not belong
to our subject, as they are independent of particular theories about molecules. I must
not, however, omit to mention a most important American contribution to this part
of thermodynamics by Prof. Willard Gibbs, of Yale College, U.S., who has given a
remarkably simple and thoroughly satisfactory method of representing the relations
of the different states of matter by means of a model. By means of this model, problems
that had long resisted the efforts of myself and others may be solved at once”.

4.1.1 Exercise

1. We know that
.d ū = 0,

for an adiabatic free expansion of a gas and that

( ( (
∂P
.d ū = c̄v dT + T − P d v̄
∂ T v̄

Consider 1 mole of carbon dioxide at 320.K occupying a volume of 1.l and

that this gas suffers a free adiabatic expansion doubling its initial volume. For
CO.2 suppose the van der Waals parameters are known: .a = 0.359 .Pa mol2 /m6
and .b = 4.27 × 10−5 .m3 /mol and consider .cv = 29 .J/ (mol K). Find the CO.2
temperature after the expansion. What would be this temperature if the inter-
molecular attractive forces between the CO.2 molecules were null?
126 4 Phase Equilibrium

4.2 Gibbs Potential to Find the Transition Pressure

Table 4.1 shows a sample worksheet for calculating the thermodynamic functions
based on the van der Waals equation of state, starting from a reference state .(v 0 , T )
and following an isothermal by successive volume increments .Δv of the molar vol-
ume.
Pressure . P is calculated using Eq. (4.1).
The entropy calculation in the third column is based on Eq. (3.32). In isothermal
conditions

∫v ( (
∂P
s (T, v) − s (T, v 0 ) =
. dv.
∂T v
v0

For a van der Waals equation of state

v−b
.s (T, v) − s (T, v 0 ) = ln .
v0 − b

The integration of Eq. (3.33) enables to find

∫v ( ( ( (
∂P
.u (T, v) − u (T, v 0 ) = T − P dv.
∂T v
v0

in the fourth column.

For a van der Waals equation of state
( (
∂P a
.T − P = 2,
∂T v v

and
( (
1 1
. u (T, v) − u (T, v 0 ) = a − .
v0 v

Table 4.1 Sample worksheet for the calculation of thermodynamic functions

.v .P .s − s 0 .u − u 0 .h − h 0 .g − g0
.v 0 . P0 .0 .0 .0 .0

.v 0 + Δv … … … … …
.v 0 + 2Δv … … … … …
… … … … … …
4.2 Gibbs Potential to Find the Transition Pressure 127

Fig. 4.3 a Calculation of the transition pressure using a worksheet. a. P − v diagram corresponding
to a single isotherm. b Gibbs energy for different pressures

Calculation of the enthalpy follows

. h (T, v) − h (T, v 0 )
= [u (T, v) − u (T, v 0 )] + P
,,,, × v
,,,, − P0 v 0
, ,, ,
fourth column second column first column

Finally, the Gibbs energy is given by

. g (T, v) − g (T, v 0 )
[ ]
= h (T, v) − h (T, v 0 ) − T (s − s 0 )
, ,, , , ,, ,
fifth column third column

Figure 4.3a shows the . P − v̄ diagram corresponding to a single isotherm and

Fig. 4.3b shows the Gibbs energy .ḡ (T, v̄) − ḡ (T, v̄0 ) for different pressures.
Let us consider that we start from the low-pressure state O where the substance
is in its vapor state. By increasing the pressure we will find state 1. From O to 1 the
Gibbs energy increases due to the compression work that was furnished to the vapor.
State 1’ is unstable whereas state 1” belongs to the condensed phase, with a very
small value of the isothermal compressibility
128 4 Phase Equilibrium
( (
1 ∂v
k =−
. T .
v ∂P T

By further increasing the pressure, we arrive at state .t and, now, both .t in the
vapor phase and .t '' in the condensed phase have the same Gibbs energy. So, if the
substance we are dealing remains in its vapor phase it will increase its Gibbs energy,
while if it changes to the liquid state .t '' in Fig. 4.3a, it will maintain its Gibbs energy.
The Gibbs energy has a minimum when the temperature and pressure are main-
tained constants.
So, if .G is the Gibbs energy of the system

. G = Gv + Gl,

then, at constant temperature and pressure

( )
.dG = μv − μl d Nv = 0,

meaning that the Gibbs potentials—or molar Gibbs energies .ḡ—must be identical
when phases .v and .l are in equilibrium. When the pressure is increased from O to
M, this state of minimal Gibbs energy is found, as the one represented by point .t,
when .μ̄v (or .ḡv ) and .μ̄l (or .ḡl ) are identical as shown in Fig. 4.3b.

4.2.1 Maxwell Area Rule

The Maxwell area rule can be derived from the condition

μv − μl = 0,
.

at the transition point. In fact, from the Gibbs-Duhem equation, Eq. (3.8)

dμ = vd P,
.

at a constant temperature.
Figure 4.4 shows the van der Waals equation of state considering the volume .v in
terms of the pressure . P for a given temperature, By integrating the above equation
between points V and L

∫L
0=
. vd P,
V
4.2 Gibbs Potential to Find the Transition Pressure 129

Fig. 4.4 Maxwell area rule

derived from the equality of
the Gibbs potentials .μl and
v
.μv

A M
I

m B
L

or
∫M ∫I ∫m ∫L
. vd P + vd P + vd P + vd P = 0,
V M I m

or

∫M ∫M ∫I ∫L
. vd P − vd P = vd P − vd P .
V I m m
, ,, , , ,, ,
area A area B

The conclusion is that the shaded areas A and B in Fig. 4.4 must be identical at
the transition pressure, leading to

∫vvs
. P (v) dv = Ps (v vs − vls ) . (4.3)
vls

This equation, together with the equation of state is what is needed for finding
the transition pressure. Since the resulting system of equations is non-linear, this is
usually performed with the help of numerical methods.
130 4 Phase Equilibrium

4.2.2 Exercise

Suppose that carbon dioxide follows a van der Waals equation of state. For .CO2 ,
a = 0.359Pa moles2 /m6 and .b = 4.27 × 10−5 m3 /moles.
.

(a) Using a worksheet such as the one of Table 4.1, find the transition pressure . Ps
for carbon dioxide at .T = 280 .K.
(b) Use an iteration numerical method for solving the Maxwell area rule, Eq. (4.3)
and find the transition pressure . Ps for carbon dioxide at this temperature.

4.3 The Transition State

When all the states between V and L in Fig. 4.5 are equilibrium states the pressure
will be the transition pressure . Ps (T ). Therefore if we displace the piston in Fig. 4.5
to the left in a quasi-static process, starting from point V, a fraction of the vapor
phase will change to the liquid phase, but the pressure will remain constant.
In reality, this piston displacement will produce non-equilibrium states in the
liquid-vapor system. In fact, a mass .Δm of vapor needs to release heat to be con-
densed. If the pressure is constant this released heat will correspond to the enthalpy
variation between the liquid and vapor states (the latent heat),

.δ Q = Δm (h vs − h ls ) .

Nevertheless, if the time .Δt corresponding to the piston displacement is smaller

than the time this latent heat needs to be transferred to the heat bath, the vapor
pressure close to the piston head will increase above . Ps giving rise to a pressure
wave that propagates to the left. Pressure . Ps will be restored, accompanied by the
condensation of the mass fraction .Δm of vapor, only after .δ Q is entirely transferred
to the thermal bath. This is the reason why condensation happens adjacent to the wall
that separates the vapor from the liquid bath and not as it is drawn in Fig. 4.5, which
is only a sketch for showing the amount of liquid and vapor when the piston is in an
intermediate position between L and V.
Consider now that the volume of the system liquid + vapor is

. V = Vv + Vl .

The volume that is occupied by the vapor is the product of its specific volume by
the mass of vapor in the system

. Vv = m v v (T, P) .
4.3 The Transition State 131

V
Ps L

b
T
vls v vvs
v

Fig. 4.5 During the phase transition at pressure . Ps and temperature .T , the molar volume of vapor
and liquid do not change and are .vvs and v.ls respectively

For the equilibrium states along the isobaric . Ps , the specific volume of vapor is
always the same .vvs (T ) between points V and L because neither .T nor . P vary along
the transition line V-L. The same comment applies to the liquid phase.
So

. V = m v vvs + m l vls ,

and if we divide the above equation by the total mass .m = m v + m l we get

v = xvvs + (1 − x) vls ,
.

where the mass fraction .x = m v /m is named the ‘vapor quality’ in the system.3

3The name ‘vapor quality’ is derived from the steam engine. Low-quality steam would contain a
high liquid fraction and, therefore, damage the turbine blades more easily.
132 4 Phase Equilibrium

4.3.1 Exercise

Show that for any state between L and V in Fig. 4.5

f = xfvs (T ) + (1 − x) fls (T ) ,
.

where f is any extensive property per unit mass or mole, e.g., .s, .u, .h, .g, . f .

4.4 The Liquid-Vapor Phase Diagram

As the temperature increases, the specific volume for the vapor phase, .vvs , becomes
closer to .vls and the latent heat .h lv = h vs − h ls approaches zero. We define the
critical temperature .Tc as the temperature above which an isothermal compression
does not produce a phase transition.
The critical point corresponds to the point where . P (v, Tc ) has an inflection point,
or, when (Fig. 4.6)

∂ P (v̄, Tc ) ∂ 2 P (v̄, Tc )
. = = 0. (4.4)
∂ v̄ ∂ v̄2
The existence of a critical point was first discovered by Charles Cagniard de la
Tour in 18224 and named by Dmitri Mendeleev in 18605 and Thomas Andrews in
1869 [5].

4.5 Clausius-Clapeyron Equation

The Clausius-Clapeyron equation was initially derived in 1834 by Clapeyron [33]

from the analysis of the Carnot cycle for the condensation of steam in thermal equi-
librium with the liquid. Clausius refined the equation in 1850 [34] and extended it
to other phase transitions (Fig. 4.7).
Let us consider a thermodynamic system where the liquid and vapor phases are in
equilibrium at temperature .T . In this condition the Gibbs potential will be the same
in both phases

.μv = μl .

4 Charles Cagniard de la Tour (1822) “Exposé de quelques résultats obtenu par l’action combinée
de la chaleur et de la compression sur certains liquides, tels que l’eau, l’alcool, l’éther sulfurique
et l’essence de pétrole rectifiée”, Annales de Chimie et de Physique, 21: 127–132.
5 In accordance with Landau, L.D. and Lifshitz, E.M. (1980) Course on Theoretical Physics Vol V,

Statistical Physics, p. 83. Third Edition of Elsevier.

4.5 Clausius-Clapeyron Equation 133

Fig. 4.6 Phase diagram for NH.3 showing the isotherms of 360 K (blue), 390 K (black) and 400 K
(green) modeled with the Peng-Robinson equation of state. The acentric factor was taken as .ω =
0.25. Pressure transition was found using the Maxwell area rule for each one of these isotherms. It
is seen that the molar volume variation .vlv = vvs − vls between the saturated liquid and saturated
vapor decreases with the temperature and reaches the zero value at the critical point. Above the
critical temperature, there is no phase transition. For NH.3 the critical temperature is .Tc = 405.55
.K and the critical pressure is . Pc = 112.78 .bar

Now, suppose that the temperature has an infinitesimal variation .dT . Conse-
quently, both potentials will vary, and, since they stay in equilibrium,

.dμv = dμl .

Nevertheless, from the Gibbs-Duhem equation, Eq. (3.8)

. − svs dT + vvs d Ps (T ) = −sls dT + vls d Ps (T ) ,

or, in other words, the variation .dT will produce a variation .d Ps in the saturation
pressure because . Ps is dependent on the temperature.
From the above equation
d Ps svs (T ) − sls (T )
. = .
dT vvs (T ) − vls (T )
134 4 Phase Equilibrium

Fig. 4.7 The

Clausius-Clapeyron equation T Ps(T)

vapor

liquid

Now, remark that from Eq. (3.17)

.dh = T ds + vd P,

and, since when the vapor state is attained from the liquid state, the pressure and the
temperature remain constant

.dh = T ds,

or

. lvh = h vs − h ls = T (svs − sls ) ,

leading to
d Ps h lv (T )
. = , (4.5)
dT T (vvs − vls )

expressing the slope .d Ps /dT in terms of the latent heat .h lv (T ).

4.5.1 Exercise

Using Eq. (4.4) show that for the van der Waals equation of state
4.6 The Law of Corresponding States 135

RTc 27R 2 Tc2

b=
. , v̄c = 3b, a = .
8Pc 64Pc

4.6 The Law of Corresponding States

The principle of Corresponding States was stated by van der Waals and reads: “Sub-
stances behave alike at the same reduced states. Substances at the same reduced
states are at corresponding states.”
Consider a gas that follows the van der Waals equation of state

RT a
. P= − 2,
v̄ − b v̄

and let . P ∗ = PPc , .T ∗ = T

Tc
, .v∗ = v̄
v̄c
. With these new variables, the van der Waals
equation becomes

RT ∗ Tc a
. P ∗ Pc = ∗
− ∗ 2.
v v̄c − b (v v̄c )

From Eq. (4.4) it can be found that van der Waals parameters satisfy

RTc 27R 2 Tc2

. b= , v̄ = 3b, a = ,
8Pc 64Pc

and so
( (2
∗3T ∗ RTc 27 1 RTc
.P = − ,
3v∗ − 1 Pc v̄c 64 v∗2 Pc v̄c

or, since . R Tc
Pc v̄c
= 8
3

8T ∗ 3
. P∗ = ∗
− ∗2 . (4.6)
3v − 1 v

Therefore, the above reduced equation is a universal relation . P ∗ = P ∗ (T ∗ , v∗ ),

independent of the substance we are dealing with. In other words, “the states of all
real gases are corresponding for a given reduced temperature and reduced volume”
136 4 Phase Equilibrium

1.1

0.9

0.8

0.7
Z

0.6
T*=1.0
0.5 T*=1.2
0.4 T*=1.5
T*=2.0
0.3

0.2
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9 3

P*

Fig. 4.8 Compressibility factor in terms of the reduced pressure and temperature for the van der
Waals equation of state

Now, let us define the compressibility factor,

P v̄ 3 P ∗ v∗
. Z= = .
RT 8 T∗
Since, for an ideal gas

RT
v̄
. ideal = ,
P
it follows that
v̄ 3 P ∗ v∗
. Z= = ,
v̄ideal 8 T∗

or that
8Z T ∗
v∗ =
. .
3P ∗
Replacing the above equation into Eq. (4.6)
( ) 27 ( ∗ ∗ ) 27 ∗2
8T ∗ Z 3 − T ∗ P ∗ + 8T ∗ Z 2 +
3 2
. P T Z− P = 0. (4.7)
8 64
The above equation can be solved for finding . Z = Z (P ∗ ) for a given isotherm
∗
. T . This relationship is pictured in Fig. 4.8.
4.7 Acentric Factor 137

4.7 Acentric Factor

The principle of corresponding states (PCS) that we have just discussed was origi-
nally outlined by van der Waals. In reality, it is the simplest version of the principle of
corresponding states, and it is referred to as the two-parameter PCS. This is because
it relies on two parameters (critical pressure and temperature) for defining a “corre-
sponding state.”
With the passing of time, more accurate principles of corresponding state formula-
tions have made use of more than two parameters. In any case, the general statement
of PCS remains untouched: “Substances at corresponding states behave alike”. What
makes the difference is the definition of “what a corresponding state is.”
The acentric factor “.ω” is a concept that was introduced by Pitzer in 1955 [134],
[135], and has proven to be very useful in the characterization of substances. It has
become a standard for the proper characterization of any single pure component,
along with other common properties, such as molecular weight, critical temperature,
critical pressure, and critical volume.
Pitzer came up with this factor by analyzing the vapor pressure curves of vari-
ous pure substances. From thermodynamic considerations, the vapor pressure curve
for pure components can be mathematically described by the Clausius-Clapeyron
equation, here written on a molar basis,

d Ps h̄ lv (T )
. = ,
dT T v̄lv (T )

where .v̄lv = v̄vs − v̄ls . Now, the compressibility factor variation between the liquid
and vapor state, at a temperature .T is given by
Ps (T ) v̄lv (T )
ΔZ = Z v − Z l =
. .
RT
Therefore

d Ps h̄ lv (T ) dT
. = ,
Ps RΔZ (T ) T 2
or
d Ps∗ dT ∗
. = a (T ) , (4.8)
Ps∗ T ∗2

where
h lv (T )
a (T ) =
. .
RTc ΔZ (T )
138 4 Phase Equilibrium

-0.5
C H4
log10Ps*

sim
ple
-1.0 flui
d
C(CH3)4

-1.5

1.0 1.1 1.2 1.3 1.4 1.5

1/T*

Fig. 4.9 Estimation of the acentric factor based on experimental data for the saturation pressure

Now, suppose that both .h lv (T ) and .ΔZ (T ) decrease at the same rate when the
temperature is increased. In this case, parameter .a is a constant and Eq. (4.8) can be
integrated to give
a
. ln Ps∗ = − + b.
T∗
Pitzer wrote the above equation as
A
. log10 Ps∗ = − + B,
T∗
where . A = a log10 e and . B = b log10 e. If the two-parameter corresponding state
principle was to hold true for all substances, the parameters . A and . B should be the
same for all substances. In other words, if the plot is of the form “.log10 Ps∗ ” versus
“1/.T ∗ ”, all lines should show the same slope . A and intercept . B. Noble gases (such as
Ar, Kr, and Xe) happen to follow the two-parameter corresponding states theory very
closely. Hence, they yield themselves amenable to acting as a reference to evaluate
“compliance” with the two-parameter equation of state (Fig. 4.9).
Pitzer decided to use noble gases as the base for comparison. Analyzing vapor
pressure data for noble gases, Pitzer showed that a value of .log10 Ps∗ = −1 was
achieved at approximately .T ∗ = 0.7. So he defined the acentric factor as
( )
ω = − log10 Ps∗ T ∗ =0.7 − 1
.

Noble gases, being the reference themselves, have an acentric factor value of zero
(.ω = 0). Substances with an acentric factor of zero are called “simple” substances.
4.7 Acentric Factor 139

Table 4.2 Acentric factors found by Pitzer

Substance .ω

Argaom .− 0.002

Krypton .− 0.002

Simple fluid .0

Xenon .0.002

Methane .0.13

Nitrogen .0.040

Ethane .0.0944

Propane .0.152

n-Butane .0.201

Benzene .0.215

Carbon dioxide .0.225

n-Pentane .0.252

Ammonia .0.25

Water .0.348

Table 4.3 Saturation pressure for Ethane

. T (K ) 100 150 200 250 300 305.33
. Ps (MPa) 0.0000111 0.0096780 0.2174 1.3008 4.3559 4.8714

The acentric factor is said to be a measure of the non-sphericity of the molecules.

Therefore, the three-parameter corresponding state theory of Pitzer reads: “Fluids
that have the same value of .ω will behave alike at the same conditions of reduced
pressure and temperature.”
Table 4.2 shows the acentric factors of some substances, found by Pitzer [135].

4.7.1 Exercises

1. Figure 4.2 shows the Pitzer [134] experimental results for some hydrocarbons,
where the base 10 logarithm of the reduced saturation pressure is plotted against
the inverse of the reduced temperature. Using the figure as a reference estimate
the acentric factor of Methane and Neopentane6
2. Table 4.3 shows the experimental values of the saturation pressure for ethane.
Estimate the acentric factor .ω for ethane based on this data and compare it with

6 Neopentane, also called 2,2-dimethylpropane, is a double-branched-chain alkane with five carbon

atoms. Neopentane is an extremely flammable gas at room temperature and pressure that can
condense into a highly volatile liquid on a cold day, in an ice bath, or when compressed to a
higher pressure.
140 4 Phase Equilibrium

Satura on data for water

1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2 2.1 2.2 2.3 2.4 2.5
0.1
-0.1
-0.3
-0.5
-0.7
-0.9
-1.1
-1.3
-1.5
-1.7 sim
log10P*s

-1.9 ple
-2.1 flui
-2.3
d
-2.5
-2.7
-2.9
-3.1
-3.3
-3.5
-3.7
-3.9
-4.1
-4.3
-4.5
1/T*

Fig. 4.10 .log10 Ps∗ in terms of the inverse of the reduced temperature .T ∗ for water

the value that was obtained by Pitzer. The critical temperature for ethane is
305.33 .K.
3. Figure 4.10 shows the saturation data for water where .log10 Ps∗ is plotted against
∗
.1/T . Find the acentric factor for water and compare with the value found by

Pitzer.

4.8 Some Popular Cubic Equations of State

4.8.1 Redlich-Kwong

The van der Waals equation of state does not take into account the temperature
dependence of the force parameter.a, considered as a constant. When the temperature
increases, the cohesion forces become less effective with respect to the dispersion
forces and a correction on the van der Waals equation was proposed by Redlich-
Kwong in 1949 to take this effect into account [145]
√
RT a/ T
.P = − ,
v̄ − b v̄ (v̄ + b)
4.8 Some Popular Cubic Equations of State 141

with 5
R 2 Tc2 RTc
.a = 0.42747 , b = 0.08664 .
Pc Pc

4.8.2 Soave
√
In 1972 G. Soave [159] replaced the 1/. T term of the Redlich-Kwong equation with
a function .α (T, ω) involving the temperature and the acentric factor (the resulting
equation is also known as the Soave-Redlich-Kwong equation). The .α function was
devised to fit the vapor pressure data of hydrocarbons and the equation does fairly
well for these substances.
It is written as

RT a (Tc ) α (T, ω)
. P= − ,
v̄ − b v̄ (v̄ + b)

with

R 2 Tc2 RTc
a (Tc ) = 0.427
. , b = 0.08664
Pc Pc

and

( ( √ ))2
.α (T, ω) = 1 + κ 1 − T ∗ ,
κ = 0.48508 + 1.55171ω − 0.15613ω2 .

4.8.3 Peng-Robinson

The Peng- Robinson equation was developed in 1976 at The University of Alberta
by Ding-Yu Peng and Donald Robinson [128] in order to satisfy the following goals:
(i) the parameters should be expressible in terms of the critical properties and the
acentric factor; (ii) the model should provide reasonable accuracy near the critical
point, particularly for calculations of the compressibility factor and liquid density.
It is written as

RT a (Tc ) α (T, ω)
. P= − , (4.9)
v̄ − b v̄ (v̄ + b) + b (v̄ − b)
142 4 Phase Equilibrium

where

R 2 Tc2 RTc
a (Tc ) = 0.45724
. b = 0.0778 ,
Pc Pc
( ( √ ))2
α (T, ω) = 1 + κ 1 − T ∗ ,
κ = 0.37464 + 1.54226ω − 0.26992ω2 .

For the most part, the Peng- Robinson equation exhibits performance similar to
the Soave equation, although it is generally superior in predicting the liquid densities
of many materials, especially non-polar ones.

4.8.4 Exercises

Carbon dioxide has a measured saturation pressure . Ps = 35.642 .bar at .T = 274 K.

1. Suppose that carbon dioxide follows a Redlich-Kwong equation of state. For
.CO2 , . Tc = 304.13 .K, . Pc = 73.77 bar.

(a) Using a worksheet such as the one of Table 4.1, find the transition pressure
. Ps for carbon dioxide at . T = 280 .K.

(b) Using the Maxwell area rule, Eq. (4.3) find the transition pressure . Ps for
carbon dioxide at this temperature.
2. Repeat Exercise 1, now supposing carbon dioxide follows a Soave-Redlich-
Kwong equation of state. For .CO2 , the acentric factor is .ω = 0.225 (Table 4.2).

4.9 The Virial Equation of State

The virial equation of state gives the compressibility factor . Z as a power series of
the reciprocal of the molar volume
P v̄ B C
. Z= = 1 + + 2 + ··· , (4.10)
RT v̄ v̄
where . B is the second virial coefficient and .C is the third virial coefficient. All virial
coefficients are independent of pressure and density and are only dependent on the
temperature for pure components. The virial equation can also be written in terms
of the pressure
P v̄
. Z= = 1 + BP P + C P P 2 + · · · , (4.11)
RT
4.9 The Virial Equation of State 143

Fig. 4.11 PVT experimental data allows the determination of the virial coefficients B and C for
each isotherm. Virial parameter B (T) is found from the interception of the isotherm T with the
vertical axis. Parameter C(T) is given by the slope of the isotherm T when .n = 1/v̄ → 0

where [138],

B C − B2
. BP = , CP = .
RT (RT )2

The virial coefficients can be determined from experimental PVT data, using
( ( ( (
∂Z 1 ∂2 Z
. B = lim , C = lim
n→0 ∂n T n→0 2! ∂n 2 T

by rearranging Eq. (4.10) as

( (
P v̄ C
v̄
. −1 = B+ , (4.12)
RT v̄

The symbol is used here to denote the inverse of the molar volume .n = 1/v̄.
Figure 4.11 gives a sketch for this determination.

4.9.1 Second Virial Coefficient

Tsonopoulos [177–179] proposed the following correlation for the second virial
coefficient
B Pc ( ) ( )
. B∗ = = F0 T ∗ + ωF1 T ∗ , (4.13)
RTc
144 4 Phase Equilibrium

where

( ) 0.330 0.1385 0.021 0.000607

. F0 T ∗ = 0.1445 − − − ∗3 − ,
T∗ T ∗2 T T ∗8
( ) 0.331 0.423 0.008
F1 T ∗ = 0.0637 + ∗2 − ∗3 − ∗8 .
T T T
Tsonopoulos proposition provides a good correlation for the second virial coef-
ficients of normal fluids. A normal fluid is one whose molecules are of moderate
size, non-polar or else slightly polar, and do not associate strongly (e.g., by hydrogen
bonding).
For polar fluids, Tsonopoulos [180] suggested that Eq. (4.13) be amended to read
( ) ( ) ( )
. B ∗ = F0 T ∗ + ωF1 T ∗ + F2 T ∗ , (4.14)

where
( ) a b
. F2 T ∗ = ∗6 − ∗8 .
T T
Parameter .a is also dependent on the temperature. Tsonopoulos suggests separate
equations, one for each group of different chemical species. For ketones, aldehydes,7
alkyl nitriles,8 ethers,9 and carboxylic-acid esters,10

7 Ketones and aldehydes are simple compounds that contain a carbonyl group (a carbon-oxygen
double bond). They are considered “simple” because they do not have reactive groups like -OH or
-Cl attached directly to the carbon atom in the carbonyl group, as in carboxylic acids containing
-COOH. Many ketones are known and many are of great importance in industry and in biology.
Examples include many sugars (ketoses) and the industrial solvent acetone, which is the smallest
ketone.
8 A nitrile is any organic compound that has a -C=N functional group. Nitriles are found in many

useful compounds, including methyl cyanoacrylate, used in super glue, and nitrile rubber, a nitrile-
containing polymer used in latex-free laboratories and medical gloves. Nitrile rubber is also widely
used in automotive and other seals since it is resistant to fuels and oils. Organic compounds con-
taining multiple nitrile groups are known as cyanocarbons.
9 Ethers are a class of organic compounds that contain an ether group—an oxygen atom connected

to two alkyl or aryl groups—of general formula R–O–R.' . These ethers can again be classified into
two varieties: if the alkyl groups are the same on both sides of the oxygen atom, then it is a simple
or symmetrical ether, whereas if they are different the ethers are called mixed or unsymmetrical
ethers. A typical example of the first group is the solvent and anesthetic diethyl ether, commonly
referred to simply as “ether” (CH.3 –CH.2 –O–CH.2 –CH.3 ). Ethers are common in organic chemistry
and pervasive in biochemistry, as they are common linkages in carbohydrates and lignin.
10 Esters are chemical compounds derived from an acid (organic or inorganic) in which at least

one –OH (hydroxyl) group is replaced by an –O–alkyl (alkoxy) group. Usually, esters are derived
from carboxylic acid and alcohol. Glycerides, which are fatty acid esters of glycerol, are important
esters in biology, being one of the main classes of lipids, and making up the bulk of animal fats and
vegetable oils. Esters with low molecular weight are commonly used as fragrances and found in
essential oils and pheromones. Phosphoesters form the backbone of DNA molecules. Nitrate esters,
4.9 The Virial Equation of State 145

a = −2.14 × 10−4 μ∗ − 4.308 × 10−21 μ∗ .

8
.

In the above equation

μ2 Pc
μ∗ = 0.9869 × 105
.
Tc2

where .μ is the molecular dipole moment in Debye, . Pc is the critical pressure in bar
and .Tc is the critical temperature in Kelvin.
For water Tsonopoulos recommends .a .= − 0.0109 and .b = 0.

4.9.2 Third Virial Coefficient

Several corresponding states correlations are available for the third virial coefficients
of gases and gas mixtures. Typically these correlations give the reduced third virial
coefficient through the generalized correlation,

C Pc2 ( ) ( )
C∗ =
. = F0 T ∗ + ωF1 T ∗ , (4.15)
R 2 Tc2

where . F0 (T ∗ ) and . F1 (T ∗ ) are regressed from experimental data.

Orbey and Vera [124] found the following correlations for non-polar gases

( ) 0.02432 0.00313
. F0 T ∗ = 0.01407 + − ,
T ∗2.8 T ∗10.5
( ) 0.01770 0.040 0.003 0.00228
F1 T ∗ = −0.02676 + + ∗3 − ∗6 − .
T ∗2.8 T T T ∗10.5

4.9.3 Exercises

1. The Soave-Redlich-Kwong equation

RT a (T )
. P= − ,
v − b v (v + b)

can be rewritten as
RT (b − B) RT
. P= − ,
v−b v (v + b)

such as nitroglycerin, are known for their explosive properties, while polyesters are important
plastics.
146 4 Phase Equilibrium

where
a (T )
. B =b−
RT
Find the virial coefficient . B from the Soave-Redlich-Kwong parameters and
compare your results with Tsonopoulos correlation, Eq. (4.13).

4.10 Calculation of State Variables

State variables such as entropy, internal energy, and enthalpy can be calculated for
any given state with respect to a reference state given by its temperature .T0 and molar
or specific volume .v0 .
In Fig. 4.12, consider, for instance, that we want to calculate the entropy .s (T, v)
with respect to the entropy at the state .T0 , v0 , i.e., .s (T, v) − s (T0 , v0 ). We know that
when the entropy is expressed in terms of .v and .T ,
( (
∂P
.ds = cv dT + dv, (4.16)
∂T v

Fig. 4.12 Calculation of state variables

4.10 Calculation of State Variables 147

Therefore, all we need for calculating .s (T, v) − s (T0 , v0 ) is an equation of state

and the thermodynamic data for .cv . Nevertheless:
( )
• in general . ∂∂ TP v is temperature dependent;
• the relationship .cv (T, v) is generally not known in the range of states we desire
to calculate the entropy.

Both problems can be solved: i) by calculating the entropy variation separately along
the isotherms and following an isochoric line and ii) by extending the integration
limit to .v∞ beyond which the vapor behaves as an ideal gas, or, beyond which the
compressibility factor . Z is very close to 1,

. |Z − 1| ≤ tol,

where .tol means the allowable error involved in the calculation, below which we
consider the results to be satisfactory.
The reason for extending the integration limit to .v∞ is that beyond this limit the
molar or specific heat is only dependent on the temperature

c = cv0 (T ) ,
. v

where the index ‘0’ means that .cv was determined at low pressures . P → 0 (or
high volumes .v → ∞). In fact, .cv0 (T ) is known for most substances and given in
accordance with a polynomial regression in terms of the temperature, e.g.,

c0 = A + BT + C T 2 + DT 3 .
. v

Constants . A, B, C and . D can be found in many handbooks and textbooks.

Using this reasoning, Eq. (4.16) is first integrated along the isotherm .T0 between
.v0 and .v∞ giving

⎞
∫v∞ ( (
∂P
.s (T0 , v∞ ) − s (T0 , v0 ) = dv⎠ .
∂T v
v0 T0

Then Eq. (4.16) is integrated along the isovolumetric .v∞ between .T0 and .T giving

∫T
cv0 (T )
.s (T, v∞ ) − s (T0 , v∞ ) = dT.
T
T0

Finally, Eq. (4.16) is integrated along the isotherm .T between .v∞ and .v giving
⎞
∫v ( (
∂P
.s (T, v) − s (T, v∞ ) = dv⎠ .
∂T v
v∞ T
148 4 Phase Equilibrium

When the three last equations are added, we get

. s (T, v) − s (T0 , v0 )
⎞ ⎞
∫v∞ ( ( ∫T 0 ∫v ( (
∂P c (T ) ∂ P
= dv⎠ + v
dT + dv⎠ . (4.17)
∂T v T ∂T v
v0 T0 T0 v∞ T

4.10.1 Exercises

1. The critical data for ethane is .Tc = 305.33 K, . Pc = 48.714 bar, and its acentric
factor was measured as.ω = 0.0944. The transition pressure for ethane calculated
with the Soave-Redlich-Kwong equation gives. Ps = 3.37217 bar for.T = 210 K,
and . Ps = 28.50510 bar for .T = 280 K. The ideal gas molar heat of ethane at
constant pressure is given by

c̄0 = 6.9 + 17.27 × 10−2 T − 6.406 × 10−5 T 2 + 7.285 × 10−9 T 3 .

. p

Consider now that you have saturated vapor of ethane at 210 K and that this
vapor is compressed reaching a final temperature of 280 K.

(a) Using the Soave-Redlich-Kwong equation of state calculate the molar vol-
ume of the saturated vapor of ethane at 210 K.
(b) Consider the compression to be isentropic and, based on Eq. (4.17), find the
pressure and molar volume of ethane in its final superheated state.
2. The following experimental data is available for the refrigerant R134a.

. p c̄0 = 19.4 + 0.258531T − 1.29665 × 10−4 T 2

T = 374.23K ,
. c Pc = 40.48bar, ω = 0.403

Consider that the steady-state expansion of the refrigerant R134a through the
expansion valve of a refrigerating cycle is adiabatic an, so, isenthalpic. The
refrigerant is in its saturated liquid state at .T = 313 .K at the entrance of the
expansion valve and leaves it at a temperature of 253 .K.
The transition pressures for R134a that were calculated with the Peng-Robinson
equation of state gave . Ps = 1.08323 bar for .T = 253 K, and . Ps = 9.44018 bar
for .T = 313 K

(a) Find the molar volume of R134a for the liquid and vapor phases in the
liquid-vapor transition region at 313 .K and 253 .K
4.10 Calculation of State Variables 149

(b) By supposing that the R134a is well described by the Peng-Robinson equa-
tion of state and using the data given above, find the vapor content of R134a
at the outlet of the expansion valve. Base your calculations of the enthalpy
on the same method of Sect. 4.10 (you may consider the saturated liquid at
253K as the reference state).
Chapter 5
Non-ideal Mixtures

Abstract Consider a multicomponent system in the gaseous phase. For low pres-
sures, the gaseous mixture can be considered as ideal with a Clapeyron equation
of state . P V = N RT . In this state, the mean free path is very high, and the inter-
molecular electrostatic potentials do not intercept, each component in the mixture
behaving as an ideal gas. When the pressure increases or the volume decreases the
molecules become closer and intermolecular attraction forces begin to be apprecia-
ble, deviating the whole system from the ideal gas behavior. In these conditions,
the several components in the mixture may stand as a homogeneous system or seg-
regate into separate phases. In this chapter, we deal with homogeneous mixtures,
when the attraction forces between different molecules predominate over the attrac-
tion forces between identical molecules and there is no segregation. After exposing
the theoretical fundamentals of multicomponent systems, the concepts of fugacity
and activity are introduced and used for calculating the bubble and dew points of
homogeneous liquid solutions in equilibrium with vapor. The functional group semi-
empirical method of Prausnitz and collaborators is also exposed and used in the
estimation of the activity coefficients.

5.1 Partial Volumes and the Amagat’s Law

A mixture in equilibrium with r-components each one of them labeled as .i =

1, . . . , r can be described by .r + 2 independent variables, which can be chosen
as .T, P, N1 , . . . , Nr .
So, the volume can be written as a function of these .r + 2 variables

. V = V (T, P, N1 , . . . , Nr ) .

In an equilibrium state, the homogeneity condition means

. V (T, P, λN1 , . . . , λNr ) = λV,

© Associação Paranaense de Cultura 2024 151

P. C. Philippi, Thermodynamics, https://doi.org/10.1007/978-3-031-49357-7_5
152 5 Non-ideal Mixtures

for any arbitrary partition parameter .λ. So, if we consider this partition parameter as
the only variable in a homogeneous mixture,

∂V ∑ ∂ V ∂ (λNi )
. = = V.
∂λ i
∂ (λNi ) ∂λ

Nevertheless,
∂ (λNi )
. = Ni ,
∂λ
and, so, for .λ = 1
∑ ( )
∂V
. V = Ni . (5.1)
∂ Ni T,P,N j/=i

( )
The term . ∂∂NVi is defined as the partial molar volume of component .i,
T,P,N j/=i
( )
∂V
.v̄i = , (5.2)
∂ Ni T,P,N j/=i

and Eq. (5.1) is, here, called Amagat’s law, meaning that the total volume of a mixture
is the sum of the partial volumes .Vi = Ni v̄i of each component in the mixture.1
By partial volume .Vi we mean the volume that the . Ni moles of component .i
would occupy in the mixture at the same temperature .T and pressure . P of the
mixture. Amagat’s law, as defined by Eq. (5.1) is a general law for homogeneous
mixtures and can be used both for ideal and non-ideal systems. In general the partial
molar volume .v̄i of each component .i is dependent on the amounts . N1 , . . . , Nr of
the components that are present in the mixture as well as on the temperature and
pressure

. i v̄ = v̄i (T, P, N1 , . . . , Nr ) ,

and this means that a variation in the number of moles of any component . j in the
mixture will affect the molar volume .vi of each component. This is the case for non-
ideal mixtures because .vi is affected by the cross interactions between molecules .i
and . j.
In fact, since ideal gas mixtures are ruled by the Clapeyron equation
(∑ )
i Ni RT
.V = , (5.3)
P

1 In accordance with Ref. [138], the Amagat’s law was derived based on the partial molar volume
of each component in the pure state .v̄i, pur e (T, P) and not on the molar partial volume .v̄i as defined
by Eq. (5.2). In fact,.v̄i is modified by the addition or removal of components and will be only equal
to .v̄i, pur e in a mixture of ideal gases.
5.1 Partial Volumes and the Amagat’s Law 153

the partial molar volume

RT
v̄ =
. i
P

is only dependent on the temperature and pressure. So, the partial volume.Vi = Ni RPT
does not depend on the molar amounts of the other components that are present in
the mixture and can be interpreted as the volume that component .i would occupy if
it were alone in the mixture at the same temperature and pressure of the mixture.
In the same way, if we consider that each component occupies the whole volume
of the mixture, Eq. (5.3) can be rewritten as
∑
.P = Pi , (5.4)
i

where . Pi = Ni RT /V is the partial pressure of component i in the mixture. Equa-

tion (5.4) is Dalton’s law. In contrast with Amagat’s law, Dalton’s law only applies
to ideal gas mixtures. Figure 5.1 shows a sketch that tries to clarify the meanings of
the Amagat and Dalton’s laws.
Consider now, as another example, a non-ideal mixture whose equation of state
is a virial equation
P v̄ P
. Z= =1+ B , (5.5)
RT RT

P=1 bar
P=1 bar

79 21
N2 (0.79 bar)
+ N2 O2
O2 (0.21 bar)

100 100
Fig. 5.1 If oxygen in the atmospheric air is separated from nitrogen it will occupy a volume of
21 L for every 100 L of air at atmospheric pressure
154 5 Non-ideal Mixtures

where the second virial coefficient is related to the intermolecular attractive forces
by
∑
.B = yi y j Bi j , (5.6)
i, j

with
√
. Bi j = δi j Bi B j . (5.7)

Equation (5.5) can be used for mixtures submitted to low and moderate pressures
[138] and parameter .δi j = 1 for .i = j but different from zero when .i /= j. In fact,
parameter .δi j for .i /= j, is related to the cross intermolecular interactions between
molecules .i and . j. Experimental values for . Bi j and models for their estimation may
be found in [57, 137, 138].
Now, we rewrite Eq. (5.5) as
∑ RT 1 ∑
. V = Ni +∑ N j Nk B jk . (5.8)
i
P i Ni j,k

For simplicity, consider a binary mixture with only two components

RT N 2 B1 + 2N1 N2 B12 + N22 B2

. V = (N1 + N2 ) + 1 . (5.9)
P N1 + N2

The molar partial volume of component 1 is calculated in accordance with Eq. (5.2)
as
( )
∂V RT
v =
. 1 = + B1 + y22 [2B12 − (B1 + B2 )] . (5.10)
∂ N1 T,P,N2 P

In the same way

( )
∂V RT
.v2 = = + B2 + y12 [2B12 − (B1 + B2 )] . (5.11)
∂ N2 T,P,N1 P

Therefore, in addition to the ideal-gas term .RT /P, the partial molar volume
v is dependent on the strength of the molecular attraction forces between the .i-
. i
molecules which is represented by an additive term . Bi in the above equations, but
is also dependent on the amount of component j present in the mixture and on the
magnitude of the cross-interactions.2B12 with respect to the self-interactions between
identical molecules . B1 + B2 .
5.1 Partial Volumes and the Amagat’s Law 155

For diluted mixtures, when . y2 → 0 and . y1 → 1, Eqs. (5.10)–(5.11) become

RT
v̄ →
. 1 + B1 ,
P

RT
v̄ →
. 2 + 2B12 − B1 .
P
So, while .v1 is unaffected by the presence of molecules 2 in the mixture, the
non-ideal part of the molar partial volume .v2 is dependent on the cross interaction
term . B12 and on the intermolecular forces term . B1 between molecules 1, but not on
the self-interaction term . B2 . This was to be expected since molecules 2 are diluted
and do not interact between themselves.
Another important limiting case of Eqs. (5.10)–(5.11) is represented by systems
where the intermolecular forces are similar . B1 ∼ B2 ∼ B12 . In this case

RT
. 1 v̄ ≈ + B1 ,
P

RT
. 2 v̄ ≈ + B2 ,
P
meaning that the molar volume of each component is independent of the presence
of the remaining components in the mixture. Mixtures with such properties are
called ideal solutions, which cannot be confused with mixtures of ideal gases whose
molecules do not interact with attractive forces.

5.1.1 Exercise

The equations of state for mixtures are commonly written in a form analogous to the
equations of state for pure components. Consider, for instance, the van der Waals
equation of state,

RT a
. P= − 2
v̄ − b v̄

For a mixture, parameter .b is the mean molecular volume,

∑
.b = yi bi
i
156 5 Non-ideal Mixtures

and the force parameter .a is given by

∑
a=
. yi y j ai j
i, j

where, similarly to Eq. (5.7) parameter .ai j takes account of the cross interactions by
an interference factor .δi j /= 0 for .i /= j.
√
. ija = δi j ai a j

Find the molar partial volumes for a binary mixture with a van der Waals equation
of state and consider the possible limiting cases, trying to find a physical meaning
for the resulting expressions.

5.2 Gibbs–Duhem Equation for Mixtures

For an open multicomponent system, the internal energy can be modified by the
addition or removal of each one of the components in the mixture and Eq. (3.5) can
be written as
∑
.dU = T d S − Pd V + μ̄i d Ni .
i

In Sect. 3.4 we found that for a thermodynamic system in equilibrium

U = T S − P V + μ̄N .
.

For a multicomponent system in equilibrium the above homogeneity relationship

is written as
∑
.U = T S − P V + μ̄i Ni ,
i

meaning that for every partition parameter .λ, the intensive variables .T, P, μ̄i remain
the same in every partitioned part of the thermodynamic system. Therefore, by dif-
ferentiating the above equation or by supposing that the multicomponent system
changes from an equilibrium state to another
∑ ∑
.dU = SdT − V d P + Ni d μ̄i + T d S − Pd V + μi d Ni .
i i
, ,, ,
=dU
5.3 Ideal Gases 157

So, we get
∑
. Ni d μ̄i = − SdT + V d P (5.12)
i

Nevertheless, as we have just seen in Sect. 5.3, volume .V may be written in terms
of the molar partial volumes
∑
.V = Ni v̄i . (5.13)
i

In the same way, considering the entropy . S as a function of .T, P, N1 , . . . , Nr and

using the homogeneity condition for equilibrium states
∑
.S = Ni s̄i , (5.14)
i

where
( )
∂S
s̄ =
. i , (5.15)
∂ Ni T,P,N j/=i

is the molar partial entropy of component .i.

By inserting Eqs. (5.13) and (5.14) into Eq. (5.12), we find

d μ̄i = − s̄i dT + v̄i d P.

. (5.16)

Equation (5.16) is the Gibbs–Duhem equation for multicomponent systems in

equilibrium. On a mass basis, it can also be written as

dμi = − si dT + vi d P.
.

5.3 Ideal Gases

In this section, we investigate the thermodynamic properties of mixtures whose

components behave as ideal gases. We start by considering a theoretical process
where a mixture of ideal gases is formed by adding components to component .i
without changing the temperature and total pressure (Fig. 5.2).
The effect of this addition is to reduce the pressure of the component .i from . P to
. Pi and so, in accordance with Eq. (5.16) its chemical potential will change by

.d μ̄i = v̄i d P. (5.17)

158 5 Non-ideal Mixtures

T P T P

i 1,2,....,r

Fig. 5.2 A theoretical process where a mixture of ideal gases is formed by adding.r − 1 components
to component .i without changing the temperature and total pressure

The molar volume .v̄i is given by

RT
v̄ =
. i ,
P
and so by integrating Eq. (5.17)

μ̄i (T, P, N1 , . . . , Nr ) = μ̄i, pur e (T, P) + RT ln yi ,

. (5.18)

where . yi = Pi /P is the molar fraction of component i in the mixture.

Since . yi ≤ 1, the above equation means that the chemical potential of component
i in the mixture will be reduced with respect to its value when in a pure state at the
same temperature and pressure. Exercise 1 of Sect. 5.3.1 is recommended for people
who want to get a deeper understanding of the meaning of the chemical potential.
Now, from the Gibbs–Duhem equation for mixtures, Eq. (5.16), the molar partial
entropy of component .i can be found by deriving Eq. (5.18) with respect to the
temperature,
( )
∂ μ̄i ∂ [ ]
.s̄i = − =− μ̄i, pur e (T, P) P − R ln yi .
∂ T P,N1 ,...,Nr ∂T

Nevertheless,
∂ [ ]
. − μ̄i, pur e (T, P) P = s̄i, pur e (T, P) ,
∂T
5.3 Ideal Gases 159

and so
( )
∂ μ̄i
s̄ =
. i = s̄i, pur e (T, P) − R ln yi .
∂T P,N1 ,...,Nr

Therefore, in accordance with Eq. (5.14),

∑ ∑
.S = Si, pur e (T, P) − R N yi ln yi .
i i

Therefore, when .r components are mixed at a constant temperature and pressure,

the entropy of the resulting system is not an additive property and differs from the
sum of the entropy of the pure components by a positive quantity, which is called
entropy of mixing
∑
. Smi xing = − R N yi ln yi .
i

The molar partial volume can be also retrieved from Eq. (5.16) as
( )
∂ μ̄i ∂ [ ]
.v̄i = = μ̄i, pur e (T, P) T = v̄i, pur e (T, P) .
∂ P T,N1 ,...,Nr ∂P

Since .v̄i, pur e (T, P) = RT /P in an ideal gas mixture, the total volume of the
multicomponent system will be
∑ RT RT
. V = Ni =N ,
i
P P

as expected.
Consider now the internal energy
∑
.U= Ni u i ,
i

or
∑
u=
. yi u i ,
i

where
( )
∂U
. iu = ,
∂ Ni T,P,N j/=i

is the molar partial energy of component .i.

160 5 Non-ideal Mixtures

From the principle of energy conservation, internal energy is, necessarily, an

additive property, meaning that the internal energy of a multicomponent system
must be equal to the sum of the internal energies of the individual components, when
these components are mixed at constant temperature and pressure
∑
.u = yi u i, pur e ,
i

and since .u i, pur e = u i, pur e (T ) for ideal gases,

∑
.u (T, y1 , . . . , yr ) = yi u i, pur e (T ) .
i

Another thermodynamic property of interest is the Helmholtz energy

. f = u − T s.

For a multicomponent system

∑ ( ) ∑
.f = yi u i, pur e − T s i, pur e + RT yi ln yi ,
i

or
∑ ∑
. f (T, P, y1 , . . . , yr ) = yi f i, pur e (T, P) + RT yi ln yi .
i

Finally, the Gibbs energy

∑
. g (T, P, N1 , . . . , Nr ) = yi g i .
i

Since .g i = μi , we can use Eq. (5.18) to find

∑ ∑
. g (T, P, y1 , . . . , yr ) = yi μi, pur e (T, P) + RT yi ln yi .
i i

5.3.1 Exercises

1. Two spherical chambers each one with a volume of 1 L are separated by a

semipermeable membrane (Fig. 5.3). Chamber A contains, initially, 21% of
Oxygen and 79% of Nitrogen, and chamber B is filled with Oxygen. The mem-
5.3 Ideal Gases 161

T=300K T=300K
P=1bar P=1bar
A B

O2 (21%)+N2(79%) O2

1 liter 1 liter
Membrane
permeable
to O2

Fig. 5.3 Oxygen will flow from chamber B to chamber A because its chemical potential in chamber
B is greater than its chemical potential in chamber A

brane is permeable to .O2 . The temperature is the same for the two chambers and
equal to 300 K and at time .t = 0, the pressure is 1 bar in both chambers. Since
the chemical potential of the Oxygen in chamber B is greater when compared to
the Oxygen in chamber A, there will be a transfer of Oxygen from chamber B
to chamber A. Find the pressures . PA and . PB and the molar fraction of Oxygen
in chamber A after equilibrium is reached.
Hint:

Use the Gibbs–Duhem equation for pure components

.d μ̄ = − s̄dT + v̄d P,

to find

. μ̄ pur e (T, PA ) − μ̄ pur e (T, PB ) ,

after the equilibrium state is reached.

2. In Exercise 1 calculate the entropy variation between the initial and the equilib-
rium states, (a) in chamber A; (b) in chamber B; (c) in the whole system A +
B.
3. The enthalpy of a multicomponent system is given by
∑
.H = Ni h i ,
i

where
162 5 Non-ideal Mixtures
( ) ( ) ( )
∂H ∂U ∂V
h =
. i = +P
∂ Ni T,P,N j/=i ∂ Ni T,P,N j/=i ∂ Ni T,P,N j/=i

and so

h = u i + Pvi
. i

Find an expression for the enthalpy of a mixture of ideal gases and show that it
is also an additive property.

5.4 The Lewis Concept of Fugacity

The concept of fugacity was introduced by Gilbert Newton Lewis in 1901 [105] as a
‘escaping tendency’ of a given molecular species when it migrates from one phase
to another till equilibrium is reached as regards the distribution of species.2
For introducing the concept of fugacity, consider that a mixture of .r -components
is subjected to a change of pressure in accordance with the sketch given in Fig. 5.4.

T P0 T P

1,2,....,r 1,2,....,r

Fig. 5.4 The concept of fugacity

2 In accordance with Lewis [105] “If any phase containing a given molecular species is brought
into contact with any other phase not containing that species, a certain quantity will pass from
the first phase to the second. Every molecular species may be considered, therefore, to have a
tendency to escape from the phase in which it is. In order to express this tendency quantitatively
for any particular state, an infinite number of quantities could be used, such, for example, as the
thermodynamic potential of the species, its vapor pressure, its solubility in water, etc. The quantity
which we shall choose is one which seems at first sight more abstruse than any of these but is in fact
simpler, more general, and easier to manipulate. It will be called the fugacity, represented by the
symbol .ψ and defined by the following conditions: (1) The fugacity of a molecular species is the
same in two phases when these phases are in equilibrium as regards the distribution of that species.
(2) The fugacity of gas approaches the gas pressure as a limiting value if the gas is indefinitely
rarefied. In other words, the escaping tendency of a perfect gas is equal to its gas pressure”.
5.4 The Lewis Concept of Fugacity 163

Since the temperature remains the same, the change in the chemical potential of
the .i-component will be given by

∫P
μi (T, P, y1 , . . . , yr ) − μi (T, P0 , y1 , . . . , yr ) =
. vi d P.
P0

If the system is a mixture of ideal gases

RT
v =
. i ,
P
and
P
μi (T, P, y1 , . . . , yr ) − μi (T, P0 , y1 , . . . , yr ) = RT ln
. .
P0

The fugacity concept of Lewis is now introduced by replacing . P and . P0 by .ψi

and .ψi,0 , respectively, in such a way that, for a non-ideal mixture
ψi
μi (T, P, y1 , . . . , yr ) − μi (T, P0 , y1 , . . . , yr ) = RT ln
. .
ψi,0

As pointed out by Lewis, the fugacity of a gas must approach the gas pressure as
a limiting value, if the gas is indefinitely rarefied. In the case of mixtures,

ψi → yi P when P → 0.
.

Now, when the temperature is constant

.dμi = vi d P.

So by integrating the above equation between . P0 and . P

∫P
ψi 1
. ln = vi d P. (5.19)
ψi,0 RT
P0

Therefore, fugacity replaces the concept of chemical potential for measuring the
escaping tendency of a given molecular species, in accordance with
ψi
. ln = μi (T, P, y1 , . . . , yr ) − μi (T, P0 , y1 , . . . , yr ) .
ψi,0

Now, consider rewriting . ψψi,0i in Eq. (5.19) as

164 5 Non-ideal Mixtures

ψi ψi P yi P0
. = ,
ψi,0 yi P P0 ψi,0

in such a manner that Eq. (5.19) becomes

∫P ( )
ψi yi P0 vi 1
. ln + ln = − d P.
yi P ψi,0 RT P
P0

When . P0 → 0

ψi,0 → yi P0 ,
.

and so

∫P ( )
vi 1
. ln φi = − d P, (5.20)
RT P
0

where
ψi
φ =
. i , (5.21)
yi P

is called the fugacity coefficient and is a measure of the non-ideality of component

.i in the multicomponent system (for mixtures of ideal gases the fugacity coefficient
.φi = 1).

5.4.1 Calculation of the Fugacity Coefficient for a Virial

Equation of State

Equation (5.5) can be used for estimating the compressibility . Z of a multicomponent

system under low and moderate pressures,
P v̄ P
. Z= =1+ B , (5.22)
RT RT
5.4 The Lewis Concept of Fugacity 165

where ∑
. B= yi y j Bi j .
i, j

In this case, the molar partial volume of component i is given by

RT ∑
v̄ =
. i +2 y j Bi j − B,
P j

and, from Eq. (5.20)

⎛ ⎞ ⎛ ⎞
∫P ∑ ∑
1 ⎝2 P
. ln φi = y j Bi j − B ⎠ d P = ⎝2 y j Bi j − B ⎠ .
RT j
RT j
0

As an example, consider a binary mixture. In this case

P ( )
. ln φ1 = B1 + y22 [2B12 − (B1 + B2 )] , (5.23)
RT
and
P ( )
. ln φ2 = B2 + y12 [2B12 − (B1 + B2 )] .
RT
But, in accordance with Eqs. (5.10)–(5.11) the terms between the parenthesis in
the above equations are related to the molar partial volumes of components 1 and 2
by

RT
v̄ −
. 1 = B1 + y22 [2B12 − (B1 + B2 )] ,
P

RT
v̄ −
. 2 = B2 + y12 [2B12 − (B1 + B2 )] .
P
Therefore, for .i = 1, 2,
( )
P RT Pvi
. ln φi = v̄i − = − 1,
RT P RT

and this is exactly what is wanted with the concept of the fugacity coefficient: a
measure of the deviation from the ideal gas behavior for the components in a non-
ideal multicomponent system. When the molar volume .vi is the one an ideal gas

φ = 1.
. i
166 5 Non-ideal Mixtures

Another interesting particular case is one of the ideal solutions when .2B12 ∼
(B1 + B2 ). In this case
P Bi
. ln φi = .
RT

5.4.2 Calculation of the Fugacity Coefficient for a Cubic

Equation of State

For most equations of state, the pressure . P is the dependent variable and its depen-
dence with respect to the independent variables is given as

. P = P (T, V, N1 , . . . , Nr ) ,

and the calculation of the molar partial volume of each species .i requires to invert
the derivative in accordance with
( )
( ) ∂P
∂V ∂ Ni T,V,N
j/ =i
.v̄i = = −(∂P ) .
∂ Ni T,P,N j/=i ∂ V T,N ,...,N 1 r

Therefore, Eq. (5.20) becomes

⎡ ( ) ⎤
∫V ∂P ( )
⎢ 1 ∂ N i T,V,N j/=i 1 ⎥ ∂P
. ln φi = ⎣− ( ) − ⎦ d V,
RT ∂∂ VP T,N ,...,N P ∂ V T,N1 ,...,Nr
∞ 1 r

or
∫∞ [ ( ) ( ) ]
1 ∂P 1 ∂P
. ln φi = + d V. (5.24)
RT ∂ Ni T,V,N j/=i P ∂V T,N1 ,...,Nr
V

Introducing the compressibility factor

Z N RT
. P= ,
V
we will have
( ) ( )
1 ∂P 1 1 ∂Z
. =− +
P ∂V T,N1 ,...,Nr V Z ∂V T,N1 ,...,Nr
5.4 The Lewis Concept of Fugacity 167

So, Eq. (5.24) becomes

∫∞ [ ( ) ] ∫∞ ( )
1 ∂P 1 ∂ ln Z
. ln φi = − dV + d V,
RT ∂ Ni T,V,N j/=i V ∂V T,N1 ,...,Nr
V V

or
∫∞ [ ( ) ]
1 ∂P 1
. ln φi = − d V − ln Z , (5.25)
RT ∂ Ni T,V,N j/=i V
V

because when .V → ∞, . Z → 1.
As an example consider the van der Waals equation with the form,
∑ ∑
RT i Ni i, j Ni N j
.P = ∑ − . (5.26)
V − i Ni bi V2

The fugacity coefficient in Eq. (5.25) can be calculated for this equation of state
as (see Exercise 1 of Sect. 5.4.3),
∑
v̄ bi j y j ai j
. ln φi = ln + − − ln Z . (5.27)
v̄ − b v̄ − b v̄RT

Considering the definition of the fugacity coefficient

ψi ψi v̄
φ =
. i = ,
yi P yi Z RT

we have

yi RT
. ln φi = ln ψi − ln − ln Z ,
v̄
and so, Eq. (5.27) may be rewritten in terms of the fugacity .ψi ,
∑
yi RT bi j y j ai j
. ln ψi = ln + − . (5.28)
v̄ − b v̄ − b v̄RT

5.4.3 Exercises

1. Starting from the van der Waals equation of state, Eq. (5.26), derive Eq. (5.27)
for the fugacity coefficient .φi .
168 5 Non-ideal Mixtures

2. Show that the van der Waals equation of state can be written for moderate pres-
sures as
P v̄ ( a ) P ( )
. Z= =1+ b− + O P2
RT RT RT
and relate the fugacity coefficient calculated with the above virial equation with
the one calculated with Eq. (5.27) for a multicomponent system.

5.5 Ideal Solutions

For a mixture of ideal gases the chemical potential of each component is given by
Eq. (5.18)

μ̄i (T, P, N1 , . . . , Nr ) = μ̄i, pur e (T, P) + RT ln yi ,

. (5.29)

meaning that, in addition to being dependent on the temperature and pressure, .μ̄i
is only dependent on the molar fraction of the .i-component (and not on the molar
fraction of the remaining components).
An ideal solution is defined as a solution whose chemical potential for each com-
ponent .i is given by

μ̄i (T, P, N1 , . . . , Nr ) = μ̄i, pur e (T, P) + RT ln xi ,

. (5.30)

where .xi is the molar fraction of component .i in the solution.

Therefore, when components . j are totally or partially added or removed from the
solution without affecting the temperature, pressure and the total number of moles,
the chemical potential .μ̄i of component .i is not affected. This, absolutely, does not
mean that the molecules do not interact or that the system is ruled by a Clapeyron
equation of state. In fact, each i-component in the solution is non-ideal and they only
form an ideal solution due to the particular nature of the electrostatic potential of
their molecules.
Ideal solutions are seldom found in nature but from a theoretical point of view they
are important as giving a reference framework, enabling us to estimate the deviations
from the ideality of real solutions.
From the definition of fugacity,
ψi (T, P, xi )
μ̄i (T, P, xi ) − μ̄i, pur e (T, P) = RT ln
. ,
ψi, pur e (T, P)

and so

ψi (T, P, xi ) = xi ψi, pur e (T, P)

.
5.5 Ideal Solutions 169

A common strategy for calculating the fugacity of pure component .i is to write

it in terms of the fugacity of this component in its state of saturated liquid, based on
Eq. (5.19)

∫P
ψi, pur e (T, P) 1
. ln ( )= vi d P, (5.31)
ψi, pur e T, Ps,i RT
Ps,i

where . Ps,i (T ) is the saturation pressure of component .i for the temperature .T of the
system.
Since the compressibility factor of liquids is very small, the molar volume .vi can
be approximated as a constant and equal to the molar volume of the .i-component
saturated liquid, .vls,i (T ), in the range between . Ps,i and . P.
Therefore, Eq. (5.31) can be written as
( ) vls,i
.ψi, pur e (T, P) = ψi, pur e T, Ps,i e RT ( P−Ps,i ) . (5.32)

Equation (5.32) is often called Poynting correction.

5.5.1 Raoult’s Law

J’ai entrepris de rechercher comment varie la tension de vapeur des liquides volatils, quand
on y dissout des substances fixes, de différentes natures. Les quantités qu’il faut déterminer
dans ce genre de recherches sont la tension de vapeur d’un dissolvant volatil pur et la tension
de 1a vapeur du même dissolvant, tenant en dissolution un poids connu de substance fixe, la
température restant la même. Je les ai mesurées par la méthode statique ou par la méthode
dynamique suivant les cas. F. M. Raoult (1889)3

Consider the equilibrium distribution of a component in a multicomponent system

between a liquid and a vapor phase (Fig. 5.5a). We seek a simple relation describing
the distribution of the components between the phases, i.e., an equation relating x,
the mole fraction in the liquid phase, to y, the mole fraction in the vapor phase.
For this purpose, the liquid phase will be considered as an ideal solution and the
vapor phase will be considered a mixture of ideal gases.
For each .i-component, the equilibrium condition is the equality of the fugacities
in the liquid . L and vapor .V phases,

ψiL = ψiV .
. (5.33)

3F. M. Raoult. Recherches expérimentales sur les tensions de vapeur des dissolutions. J. Phys.
Theor. Appl., 1889, 8 (1), pp. 5–20 [142].
170 5 Non-ideal Mixtures

Fig. 5.5 Raoult’s law: a the

liquid phase is considered as
T P T Ps,i
an ideal solution and, the gas
phase, a mixture of ideal
gases; b in its saturated
condition the fugacity of a Vapor phase Vapor phase
liquid is its saturation
pressure when the vapor is 1,2,...,r components Pure component i
considered as an ideal gas

1,2,...,r components

Liquid phase Liquid phase

(a) (b)

Since the vapor phase is considered as a mixture of ideal gases

ψiV = yi P,
.

and if we neglect the Poynting correction in Eq. (5.32), supposing . P ∼ Ps,i for all
the .r components in the liquid solution,
( )
ψiL = xi ψi,Lpur e T, Ps,i .
. (5.34)
( )
For finding .ψi,Lpur e T, Ps,i consider Fig. 5.5b, showing pure component .i in its
saturated condition. In equilibrium, the fugacities of the liquid and vapor phase are
identical

ψi,Lpur e = ψi,Vpur e ,
.

and since the vapor of component i was considered as an ideal gas, .ψi,Vpur e = Ps,i
and

.ψi,Lpur e = Ps,i . (5.35)

Therefore, by inserting Eq. (5.35) into Eq. (5.34), the equilibrium condition for
component .i, Eq. (5.33) becomes

. iy P = xi Ps,i . (5.36)
5.5 Ideal Solutions 171

Equation (5.36) was firstly empirically discovered by François-Marie Raoult

(1830–1901) in 1887.4 It states that the total pressure over an ideal solution is the
sum of the saturation pressures of its components weighted by their molar fractions
[142].
Let us apply Raoult’s law for a binary mixture A + B, supposing that component
A is more volatile than component B.
The total vapor pressure over the solution will be given by

( )
. P = x A Ps,A (T ) + (1 − x A ) Ps,B (T ) = Ps,B + x A Ps,A − Ps,B , (5.37)

meaning that . P has a linear variation with the molar fraction .x A .

On the other hand,
PA x A Ps,A
y =
. A = ,
P P
and since
P − Ps,B
x =
. A ,
Ps,A − Ps,B

the total vapor pressure can be also expressed in terms of . y A

Ps,B
. P=( ( )) . (5.38)
Ps,B
1 − yA 1 − Ps,A

So for a given pressure . P the amount of component A in the liquid solution

will be given by its molar fraction, .x A , solution of Eq. (5.37) and the amount of this
component in the vapor phase by . y A , solution of Eq. (5.38). This is shown in Fig. 5.6.

5.5.2 Exercise

1. Consider that you have a binary mixture of vapors A and B at a constant tem-
perature T and that the amount of A in this mixture is given by . y A,1 (Fig. 5.7).
Component A is more volatile than component B at temperature T. When the
pressure is increased, it will reach the value . P1 and the first drops of liquid will
appear (this is the reason the continuous black line is called dew point line). The
amount of A in these liquid drops is given by .x A,1 . By further increasing the
pressure both the liquid and the vapor phases become richer in component A,
till the pressure attains the value . P2 when the mixture finally condensates and

4F. M. Raoult. Comptes rendus des séances de l’Académie des Sciences (6 décembre 1886 et 4
avril 1888.
172 5 Non-ideal Mixtures

T
140
120
liquid Ps,A
100
80
P

60
40 vapor
Ps,B 20

0 xA 0.4 yA
0 0.2 0.6 0.8 1.0
xA,yA
Fig. 5.6 Raoult’s law: for a given pressure . P the amount of component A in the liquid solution
will be given by its molar fraction, .x A , solution of Eq. (5.37) and the amount of this component in
the vapor phase by . y A , solution of Eq. (5.38)

T
140
120
liquid Ps,A
100 int line
le po e
80 bubb nt
lin
P

oi
de wp
60 2
40 1 vapor
Ps,B 20

0 xA,1 xA,2=yA,1 yA,2 1.0

0 0.2 0.4 0.8
xA,yA
Fig. 5.7 Raoult’s law

the molar fraction of A is . y A,2 in the remaining vapor bubbles. Try to explain
the following questions: (a) why the pressure must increase from . P1 to . P2 for
the complete condensation of the vapor mixture? … (b) why does component
A condensate at pressures smaller than its saturation pressure . Ps,A ? … (c) why
component B remains in the vapor state at pressures greater than its saturation
pressure . Ps,B ? … (d) why the first liquid drops that appear in the system when
. P = P1 are poorer in component A and richer in component B? … (e) why we

say that for pressures greater than . P2 all the vapor mixture is in the liquid state?
…
5.6 Non-ideal Solutions 173

5.6 Non-ideal Solutions

For an ideal solution, we saw in Sect. 5.5 that the fugacity may be written as

ψi (T, P, xi ) = xi ψi, pur e (T, P) .

.

For non-ideal solutions, we define the activity coefficient .γi by

ψi (T, P, x1 , . . . , xr )
γ (T, P, x1 , . . . , xr ) =
. i . (5.39)
xi ψi, pur e (T, P)

In this way, .γi is a measure of the non-ideality of the solution: when .γi ∼ 1 for
all .i, the solution is close to an ideal solution and when .γi is far from 1, appreciable
deviations from ideality must be expected.
As a very simple example consider a non-ideal binary solution with a virial equa-
tion of state as in Sect. 5.4.1.
The fugacity coefficient of the first component is given by Eq. (5.23)
P ( )
. ln ψ1 = ln (x1 P) + B1 + x22 [2B12 − (B1 + B2 )] .
RT
So,
P B1
. ln ψ1, pur e = ln P + ,
RT
and
P ( 2 )
. ln γ1 = x2 [2B12 − (B1 + B2 )] .
RT
Therefore, the activity coefficient will be 1 in the following cases: (a) when.2B12 ≈
(B1 + B2 ) meaning that the solution is ideal and (b) when component 1 is in its pure
state (.x2 = 0). So, the deviations of the activity coefficient .γi from 1 may be regarded
as due to the cross interactions between molecules .i and the remaining molecules in
the solution.

5.6.1 Gibbs Energy in Excess

The Gibbs energy of a non-ideal solution is given by

∑
. G (T, P, N1 , . . . , Nr ) = Ni μi (T, P, N1 , . . . , Nr ) .
i
174 5 Non-ideal Mixtures

We define the Gibbs energy in excess by

. G E (T, P, N1 , . . . , Nr ) = G (T, P, N1 , . . . , Nr ) − G ideal (T, P, N1 , . . . , Nr ) .

Therefore,
∑
. G E (T, P, N1 , . . . , Nr ) = Ni μiE , (5.40)
i

where

.μiE = μi − μideal
i .

Now, consider writing .μiE in terms of the fugacity. From the definition of fugacity
( )
ψi ψi
.μi − μi = RT ln ideal = RT ln .
ideal
ψi xi ψi, pur e

But
ψi
. = γi ,
xi ψi, pur e

and so the chemical potential in excess is related to the activity coefficient by

μiE = RT ln γi .
.

Equation (5.40) can thus be written as

∑
. G E (T, P, N1 , . . . , Nr ) = Ni RT ln γi . (5.41)
i

Equation (5.41) provides a method for calculating the activity coefficients based
on heuristic models for .G E
( )
∂G E (T, P, N1 , . . . , Nr )
.RT ln γi = (5.42)
∂ Ni T,P,N j/=i
5.6 Non-ideal Solutions 175

5.6.2 Margules Model

The simplest heuristic model for the Gibbs energy in excess was introduced by Max
Margules in 1895.5 For a binary mixture with molecules 1 and 2 of the same molecular
size but different polarity,

GE N1 N2
. =A = N Ax1 x2 , (5.43)
RT N1 + N2

where . A = A (T, P) can be considered as an experimental parameter independent

of the composition. The model satisfies the condition

. G E = 0,

when. N1 or. N2 is null. In fact, in this case, the mixture is reduced to a pure component
without cross-attraction forces between molecules belonging to different species,
which are the sources of non-ideality.
The Margules model can be extended to binary mixtures of components with
different molecular sizes in accordance with

GE
. = N x1 x2 (A21 x1 + A12 x2 ) .
RT
From the above equation
( )
GE N1 N2 N1 N2
. = A21 + A12 .
RT N1 + N2 N1 + N2 N1 + N2

So, from Eq. ()

. ln γ1 = [A12 + 2 (A21 − A12 ) x1 ] x22 , (5.44)

and

. ln γ2 = [A21 + 2 (A12 − A21 ) x2 ] x12 . (5.45)

Equation (5.44) enables to find parameter . A12 from the limiting value of the
activity coefficient .γ1 in accordance with6

5 Margules, Max (1895). Über die Zusammensetzung der gesättigten Dämpfe von
Misschungen. Sitzungsberichte der Kaiserliche Akadamie der Wissenschaften Wien
Mathematisch-Naturwissenschaftliche Klasse II. 104: 1243–1278. https://archive.org/details/
sitzungsbericht10wiengoog.
6 Remember that when component 1 is pure .γ = 1. In the same way, .γ = 1 for pure component
1 2
2.
176 5 Non-ideal Mixtures

. lim ln γ1 = A12 .
x1 →0

In the same way

. lim ln γ2 = A21 .
x2 →0

For similar molecules . A12 = A21 = A and the activity coefficients become

. ln γ1 = Ax22 and ln γ2 = Ax12 .

5.6.3 Exercises

1. Table 5.1 shows the values of the activity coefficients for a water–methanol
solution at .T = 50 ◦ C obtained by using the UNIFAC method of Prausnitz and
co-workers [57, 137, 138]. For the calculations, the vapor phase was considered
a mixture of ideal gases. Using the data presented in Table 5.1 calculate Margules
parameters . A12 and . A21 . In the following, calculate the activity coefficients .γ
for the several molar fractions of methanol presented in Table 5.1 and compare
these obtained values with the Margules model with the ones in Table 5.1.
2. Henry’s law is a gas law that states that the amount of a dissolved gas labeled i
is proportional to its partial pressure in the gas phase,7

x = H Pi .
. i

The proportionality factor . H is called Henry’s law constant. Henry’s law was
formulated by the English chemist William Henry, who studied the topic in the
early nineteenth century. In his publication about the quantity of gases absorbed
by water, he described the results of his experiments: “water takes up, of gas
condensed by one, two, or more additional atmospheres, a quantity which, ordi-
narily compressed, would be equal to twice, thrice, etc. the volume absorbed
under the common pressure of the atmosphere.”
Considering the mixture of vapors in equilibrium with a liquid solution as a mix-
ture of ideal gases and starting from the Margules model for similar molecules
given by Eq. (5.43) show that when the dissolved gas is diluted in the liquid
phase (.xi → 0), Henry’s constant is given by
( )
A
. H = exp ψi, pur e (T, P) .
RT

7Henry, William (1 January 1803). Experiments on the Quantity of Gases Absorbed by Water, at
Different Temperatures, and under Different Pressures. Philosophical Transactions of the Royal
Society. London. 93: 29–274. https://doi.org/10.1098/rstl.1803.0004.
5.7 Activity Coefficients from Molecular Groups 177

Table 5.1 Activity coefficients .γ for a water–methanol solution at .T = 50 ◦ C

Molar fraction of methanol Water Methanol
0.005 1.0000 2.2415
0.010 1.0001 2.2101
0.050 1.0033 1.9912
0.100 1.0124 1.7809
0.200 1.0436 1.4957
0.500 1.2078 1.1232
0.800 1.4476 1.01479
0.900 1.5423 1.0034
0.990 1.6637 1.0000

5.7 Activity Coefficients from Molecular Groups

It is often convenient to consider a molecule as an aggregate of functional groups; as

a result, some thermodynamic properties of pure fluids can be calculated by summing
group contributions. Extension of this concept to mixtures was suggested long ago by
Langmuir [97], and several attempts have been made to establish group-contribution
methods for activity coefficients. Here we will be restricted to the UNIFAC method for
activity coefficients .γ , which is particularly useful for making reasonable estimates
for those non-ideal mixtures for which data are sparse or totally absent.
The UNIFAC method (UNIquac Functional-group Activity Coefficients) is a
semi-empirical method for the prediction of non-electrolyte activity in non-ideal
mixtures that uses the functional groups present on the molecules that make up the
liquid mixture to calculate activity coefficients. By using interactions for each of
the functional groups present on the molecules, as well as some binary interaction
coefficients, the activity of each of the solutions can be calculated. This informa-
tion can be used to obtain information on liquid equilibria, which is useful in many
thermodynamic calculations.
The UNIFAC model was first published in 1975 by Fredenslund, Jones and Praus-
nitz [57], a group of chemical engineering researchers from the University of Cali-
fornia.
The molecular activity coefficient is separated into two parts: one part.γic provides
the contribution due to differences in molecular size and shape, and the other .γiR
provides the contribution due to molecular interactions.
In a multicomponent system, the activity coefficient of (molecular) component .i
is considered as

. ln γi = ln γic + ln γiR . (5.46)

,,,, , ,, ,
combinatorial residual
178 5 Non-ideal Mixtures

5.7.1 Combinatorial Activity Coefficient

In accordance with the UNIFAC method the combinatorial activity coefficient of

each molecule takes into account the geometric parameters of its individual groups
through the following expression
φi z θi φi ∑
. ln γic = ln + qi ln + li − xjlj, (5.47)
xi 2 φi xi j

where .φi is the volume contribution of component .i in the solution

ri x i
φ =∑
. i ,
j rjxj

and .θi is the area contribution of component .i in the solution

qi x i
θ =∑
. i .
j qjxj

Parameter .li is given by

z
l =
. i (ri − qi ) − (ri − 1)
2
with .z = 10.
The volume parameter .ri considers the volume contribution of all functional
groups in the molecule
∑
.ri = νki Rk
k

where .νki is the number of groups .k in molecule .i and . Rk is the volume contribution
of group .k (Fig. 5.8).
The area parameter .qi considers the area contribution of all functional groups in
the molecule
∑
.qi = νki Q k
k

where . Q k is the area contribution of group .k (Fig. 5.8).

5.7.1.1 Examples

The butane, Fig. 5.9 is an organic compound with the formula C.4 H.10 that is an alkane
with four carbon atoms. Butane is a gas at room temperature and atmospheric pres-
5.7 Activity Coefficients from Molecular Groups 179

Fig. 5.8 Volume and area

Group Numbers
group contributions for the
combinatorial activity Main Secondary Name R Q
coefficient. Reproduced from 1 CH3 0.9011 0.8480
Table 8.23 of Ref. [137] 2 CH2 0.6744 0.5400
1
3 CH 0.4469 0.2280
4 C 0.2195 0.0000
5 CH2=CH 1.3454 1.1760
6 CH=CH 1.1167 0.8670
2 7 CH2=C 1.1173 0.9880
8 CH=C 0.8886 0.6760
9 C=C 0.6605 0.4850
10 ACH 0.5313 0.4000
3
11 AC 0.3652 0.1200
12 ACCH3 1.2663 0.9680
4 13 ACCH2 1.0396 0.6600
14 ACCH 0.8121 0.3480
5 15 OH 1.0000 1.2000
6 16 CH3OH 1.4311 1.4320
7 17 H2O 0.9200 1.4000
8 18 ACOH 0.8952 0.6800
19 CH3CO 1.6724 1.4880
9
20 CH2CO 1.4457 1.1800

Fig. 5.9 The molecule of

n-butane
CH3

CH2
180 5 Non-ideal Mixtures

sure. The term may refer to either of two structural isomers, n-butane or isobutane,
or to a mixture of these isomers. In the IUPAC8 nomenclature, however, “butane”
refers only to the n-butane isomer (which is the isomer with the unbranched struc-
ture). Butanes are highly flammable, colorless, easily liquefied gases.
We can identify the following molecular groups: 2 groups CH.2 and 2 groups CH.3 .
The volumetric . R and surface . Q parameters for the functional groups CH.2 and CH.3
are given in Fig. 5.8.
So
∑
.r butane = νkbutane Rk = 2 × 0.6744 + 2 × 0.9011 = 3.151
k

∑
q
. butane = νkbutane Q k = 2 × 0.540 + 2 × 0.848 = 2.776
k
z
l
. butane = (rbutane − qbutane ) − (rbutane − 1)
2
10
= (3.151 − 2.776) − (3.151 − 1) = − 0.276
2
Consider now a solution of n-pentane with acetone (Fig. 5.10). Pentanes are
relatively inexpensive and are the most volatile liquid alkanes at room temperature,
so they are often used in the laboratory as solvents that can be conveniently and rapidly
evaporated. However, because of their non-polarity and lack of functionality, they
dissolve only non-polar and alkyl-rich compounds. Pentanes are miscible with most
common nonpolar solvents such as chlorocarbons, aromatics, and ethers. Acetone
is the organic compound with the formula (CH.3 ).2 CO. It is a colorless, volatile,
flammable liquid, and is the simplest ketone. Acetone is miscible with water and
serves as an important solvent in its own right, typically for cleaning purposes. About
6.7 million tonnes were produced worldwide in 2010, mainly for use as a solvent and
production of methyl methacrylate and bisphenol A. It is a common building block
in organic chemistry. Familiar household uses of acetone are as the active ingredient
in nail polish remover and as paint thinner.
Tables 5.2 and 5.3 shows the calculation of the combinatorial activity coefficients
in a solution of acetone and n-pentane with 25% of acetone using Eq. (5.47).

5.7.2 Residual Activity Coefficient

The residual activity coefficient takes the molecular interaction between the molecu-
lar groups into account. The molecular interaction parameters are shown in Fig. 5.11,
for some groups. Table 8.24 of Poling et al. [137] gives a complete set of values of

8 International Union of Pure and Applied Chemistry.

5.7 Activity Coefficients from Molecular Groups 181

CH3CO

n-pentane acetone

Fig. 5.10 Normal-pentane and acetone molecules

Table 5.2 Calculation of the geometric parameters r, q, and .l for the molecules of n-pentane and
acetone
Groups R Q r q .l

n-pentane 2CH.3 + 3.8254 3.316 .− 0.2784

3CH.2
Acetone 1CH.3 + 2.5735 2.336 .− 0.386
1CH.3 CO
CH.3 0.9011 0.8480
CH.2 0.6744 0.5400
CH.3 CO 1.6724 1.4880

Table 5.3 Combinatorial activity coefficient for n-pentane and acetone, based on data given in
Table 5.2, when the molar fraction of acetone is x.acetone = 0.25
Molar fraction .φ .θ .ln γc .γc

n-pentane 0.75 0.8168 0.8098 .− 0.0031 0.9969

Acetone 0.25 0.1832 0.1902 .− 0.0356 0.9650

the molecular interaction parameters for 50 molecular groups and is suggested for
interested readers.
Parameter .amn is considered as a reference when .m = n, i.e., .amn = 0 when .m =
n. Considering, for instance, the molecular groups .CH3 and .CH2 , both groups belong
to group 1 in accordance with Fig. 5.8 and their interaction parameter is null. In the
same way, .a19 = 476.4 K9 means that the attractive potential between groups 1
and 9 is 476.4 K larger than the attractive potential between groups 1 and 1 and
.a91 = 26.76 K, means that the attractive potential between groups 9 and 1 is 26.76 K

larger than the attractive potential between groups 9 and 9. A negative value for
.a56 = − 137 K means that the attractive potential between molecules 5 and 6 is
137 K smaller than the attractive potential between molecules 5 and 5.

9 Parameters are normalized by the Boltzmann constant, so their unit is the Kelvin (K).
182 5 Non-ideal Mixtures

Main Group 1 2 3 4 5 6 7 8 9
1 0 86.02 61.13 76.5 986.5 697.2 1318 1333 476.4
2 -35.36 0 38.81 74.15 524.1 787.6 270.6 526.1 182.6
3 -11.12 3.446 0 167 636.1 637.4 903.8 1329 25.77
4 -69.7 -113.6 -146.8 0 803.2 603.3 5695 884.9 -52.1
5 156.4 457 89.6 25.82 0 -137.1 353.5 -259.7 84
6 16.51 -12.52 -50 -44.5 249.1 0 -181 -101.7 23.39
7 300 496.1 362.3 377.6 -229.1 289.6 0 324.5 -195.4
8 275.8 217.5 25.34 244.2 -451.6 -265.2 -601.8 0 -356.1
9 26.76 42.92 140.1 365.8 164.5 108.7 472.5 -133.1 0

Fig. 5.11 Interaction parameters .amn in Kelvin. Reproduced from Table 8.24 of Poling et al. [137]

Table 5.4 Functional groups in the acetone and n-pentane solution

.CH3 .CH2 .CH3 CO

Identifying number 1 2 3
Main group 1 1 9

The effect of temperature on the molecular interaction is taken into account by

introducing the interaction energy in a dimensionless form
( a )
mn
.Ψmn = exp − (5.48)
T
and the residual coefficient of activity is given by
∑ ( )
. ln γi = νki ln [k − ln [ki
r
(5.49)
k

where the symbol .k is used for the molecular groups, the gamma factor .[k is related
to the .k-group and is calculated in accordance with
( )
∑ ∑ θm Ψkm
. ln [k = Q k 1 − ln θm Ψmk − ∑
m m n θn Ψnm

The term .[ki in Eq. (5.49) means that the gamma factor must be calculated in a
reference solution of molecules.
For concreteness, consider an acetone-pentane solution. From Fig. 5.8, the func-
tional groups in the solution acetone (labeled as 1) and n-pentane (labeled as 2) are
identified in Table 5.4.
The interaction parameters are given below

.a11 .a12 .a21 .a22 .a13 .a31 .a33 .a23 .a32

0 0 0 0 .476.40 .26.76 0 .476.40 .26.76
5.7 Activity Coefficients from Molecular Groups 183

Fig. 5.12 Method for calculating the gamma factors in the reference solutions [137]

5.7.2.1 Calculation of the Gamma Factor in the Reference Solutions

The .[ factors in the reference solutions are calculated in accordance with the method
presented in Fig. 5.12 [137].
In our example, there are two molecules present in the solution: the acetone
molecule referred to as 1 and the n-pentane molecule referred as 2. So the first ref-
erence solution is a solution of acetone, where the two groups CH.3 and CH.3 CO are
supposed to be the solution components.

Group Fractions

In the acetone solution (labeled as 1) there are two functional groups CH.3 (1) and
CH.3 CO (3) in equal proportions. So

ν11 1
. X 11 = = ,
ν11 + ν31 2

ν31 1
. X 31 = = .
ν11 + ν31 2

In the n-pentane solution (labeled as 2) there are 2 functional groups CH.3 (1) and
3 functional groups CH.2 (2) in proportions

ν12 2
. X 12 = = ,
ν1 + ν2
2 2 5

ν22 3
. X 22 = = .
ν1 + ν2
2 2 5
184 5 Non-ideal Mixtures

Area Fractions

The area fractions in the acetone solution (1) are given for the functional group CH.3
(1) by
X 11 × Q 1
θ
. 11 = = 0.36301,
X 11 × Q 1 + X 31 × Q 3

and for the functional group CH.3 CO (3), by

X 31 × Q 3
θ
. 31 = = 0.63699.
X 11 × Q 1 + X 31 × Q 3

The area fractions in the n-pentane solution (2) are given for the functional group
CH.3 (1) by
X 12 × Q 1
θ
. 12 = = 0.51146,
X 12 × Q 1 + X 22 × Q 2

and for the functional group CH.2 (2), by

X 22 × Q 2
θ
. 22 = = 0.48854.
X 12 × Q 1 + X 22 × Q 2

Let the temperature be .T = 307 K. The interaction energies .Ψmn between the
functional groups at this temperature are given below

.Ψ11 .Ψ12 .Ψ21 .Ψ22 .Ψ13 .Ψ31 .Ψ33 .Ψ23 .Ψ32

1 .1 1 .1 .0.21187 .0.91652 .1 .0.21187 .0.91652

Gamma Factors

Following the method presented in Fig. 5.12, in the acetone reference solution, the
gamma factors for the group CH.3 (1) and CH.3 CO (3) are given by

( )
(1 − ln (θ11 × Ψ11 + θ31 × Ψ31 ) )
. ln [1 = Q1 = 0.40891,
1
− θ11 ×Ψθ1111×Ψ11
+θ31 ×Ψ31
+ θ11 ×Ψθ3113×Ψ13
+θ31 ×Ψ33

( )
(1 − ln (θ11 × Ψ13 + θ31 × Ψ33 ) )
. ln [3 = Q3 = 0.13891.
1
− θ11 ×Ψθ1111×Ψ31
+θ31 ×Ψ31
+ θ11 ×Ψθ3113×Ψ33
+θ31 ×Ψ33
5.7 Activity Coefficients from Molecular Groups 185

In the n-pentane solution, all the gamma factors are unitary because the interaction
energies are the reference ones

( )
(1 − ln (θ12 × Ψ11 + θ22 × Ψ21 ) )
. ln [1 = Q1 = 0,
2
− θ12 ×Ψθ1211×Ψ11
+θ22 ×Ψ21
+ θ12 ×Ψθ2212×Ψ12
+θ22 ×Ψ22

( )
(1 − ln (θ12 × Ψ12 + θ22 × Ψ22 ) )
ln [22 = Q 2 = 0.
− θ12 ×Ψθ1211×Ψ + θ12 ×Ψθ2212×Ψ
. 21 22
+θ22 ×Ψ21 +θ22 ×Ψ22

Gamma Factors for the Acetone-Pentane Solution

Consider a solution with a very small fraction of acetone, x.acetone = x = 0.047. Let
us calculate the fractions of the 3 molecular groups.
For the CH.3 group present in the two molecules

x × ν11 + (1 − x) × ν12
. X1 = = 0.40193.
x × ν11 + x × ν31 + (1 − x) × ν12 + (1 − x) × ν22

For the CH.3 CO present in the acetone (1) molecule

x × ν31
. X3 = = 9.6728 × 10−3 .
x× ν11 +x× ν31 + (1 − x) × ν12 + (1 − x) × ν22

For the CH.2 present in the n-pentane (2) molecule

(1 − x) × ν22
. X2 = = 0.58839.
x × ν11 + x × ν31 + (1 − x) × ν12 + (1 − x) × ν22

So

CH.3 (1) CH.2 (2) CH.3 CO (3)

.X1 = 0.40193 . X 2 = 0.58839 . X 3 = 9.6728 × 10−3

Therefore, the area fractions will be given by,

X 1 × Q1
θ =
. 1 = 0.50647,
X 1 × Q1 + X 3 × Q3 + X 2 × Q2

X 3 × Q3
θ =
. 3 = 2.1388 × 10−2 ,
X 1 × Q1 + X 3 × Q3 + X 2 × Q2
186 5 Non-ideal Mixtures

X 2 × Q2
θ =
. 2 = 0.47214.
X 1 × Q1 + X 3 × Q3 + X 2 × Q2

So

CH.3 (1) CH.2 (2) CH.3 CO (3)

.θ1 = 0.5067 .θ2 = 0.47214 .θ3 = 2.1388 × 10−2

The gamma factors are calculated in accordance with

( )
∑ ∑ θm Ψkm
. ln [k = Q k 1 − ln θm Ψmk − ∑ .
m m n θn Ψnm

So,

. ln [1
⎛ ⎞
1
⎜ − ln (θ1 × Ψ11 + θ3 × Ψ31 + θ2 × Ψ21 ) ⎟
⎜ ⎛ ⎞ ⎟
⎜ θ1 ×Ψ11 ⎟
= Q1 ⎜ ⎟ = 1.37 × 10−3 ,
⎜ ⎜ θ1 ×Ψ11 +θ3θ×Ψ 31 +θ2 ×Ψ21
2 ×Ψ12 ⎟ ⎟
⎝ − ⎝ + θ1 ×Ψ12 +θ3 ×Ψ32 +θ2 ×Ψ22 ⎠ ⎠
θ3 ×Ψ13
+ θ1 ×Ψ13 +θ3 ×Ψ33 +θ2 ×Ψ23

. ln [2
⎛ ⎞
1
⎜ − ln (θ1 × Ψ12 + θ3 × Ψ32 + θ2 × Ψ22 ) ⎟
⎜ ⎛ ⎞ ⎟
⎜ θ1 ×Ψ21 ⎟
= Q2 ⎜ θ1 ×Ψ11 +θ3 ×Ψ31 +θ2 ×Ψ21 ⎟ = 8.72 × 10−4 ,
⎜ ⎜ θ2 ×Ψ22 ⎟ ⎟
⎝ − ⎝ + θ1 ×Ψ12 +θ3 ×Ψ32 +θ2 ×Ψ22 ⎠ ⎠
+ θ1 ×Ψ13 +θθ33 ×Ψ
×Ψ23
33 +θ2 ×Ψ23

. ln [3
⎛ ⎞
1
⎜ − ln (θ1 × Ψ13 + θ3 × Ψ33 + θ2 × Ψ23 ) ⎟
⎜ ⎛ ⎞ ⎟
⎜ θ1 ×Ψ31 ⎟
= Q3 ⎜ ⎟ = 2.207.
⎜ ⎜ θ1 ×Ψ11 +θ3θ×Ψ 31 +θ2 ×Ψ21
2 ×Ψ32 ⎟ ⎟
⎝ − +
⎝ θ1 ×Ψ12 +θ3 ×Ψ32 +θ2 ×Ψ22 ⎠ ⎠
+ θ1 ×Ψ13 +θθ33 ×Ψ
×Ψ33
33 +θ2 ×Ψ23
5.7 Activity Coefficients from Molecular Groups 187

Residual Activity Coefficients

Finally, the residual activity coefficients are calculated for acetone (1) and n-pentane
(2), in accordance with
∑ ( )
. ln γi = νki ln [k − ln [ki ,
r

where the index .i refers to the solution component (acetone or n-pentane) and the
index .k refers to the functional groups that are present in the molecule of the .i-
component.
For the acetone,

( ) ( )
. ln γ1r = ν11 ln [1 − ln [11 + ν31 ln [3 − ln [31
( )
= ν11 1.3694 × 10−3 − 0.40891 + ν31 (2.207 − 0.13891)
= 1.6605,

and so

γ r = exp (1.6605) = 5.2619.

. 1

For the n-pentane,

( ) ( )
. ln γ2r = ν12 ln [1 − ln [12 + ν22 ln [2 − ln [22
( ) ( )
= ν12 1.3694 × 10−3 − 0 + ν22 8.7205 × 10−4 − 0
= 5.3550 × 10−3 ,

and so
( )
γ r = exp 5.3550 × 10−3 = 1.0054.
. 2

Activity Coefficients

The activity coefficients of acetone and n-pentane are calculated in accordance with
Eq. (5.46)

. ln γi = ln γic + ln γiR .
,,,, , ,, ,
combinatorial residual
188 5 Non-ideal Mixtures

For acetone, present with a molar fraction .xacetone = 0.047 the combinatorial
activity coefficient was calculated as .γ1c = 0.9486 and so

. ln γ1 = − 0.0527 + 1.6605 = 1.6078.

Therefore

γ = 4.9918.
. 1

For n-pentane, the combinatorial activity coefficient was calculated as .γ2c =

0.9999 for this molar fraction of acetone and so

. ln γ2 = − 1.01769 × 10−4 + 5.3550 × 10−3 = 5.2532 × 10−3 .

Therefore

γ = 1.0053.
. 2

5.7.3 Exercises

1. Use the functional groups’ method presented in Sect. 5.7 for finding the activity
coefficients of a solution of acetone and n-pentane at 307 K when the molar
fraction of n-pentane is .x pentane = 0.047.
2. Liquefied petroleum gas (LPG), also referred to as simply propane or butane,
are flammable mixtures of hydrocarbon gases used as fuel in heating appliances,
cooking equipment, and vehicles. Repeat Exercise 1 for a solution of propane
and n-butane at 300 K when the molar fraction of propane is .x pr opane = 0.1, 0.2,
0.5, 0.8 and 0.9. What is your conclusion about the deviations from the ideality
of such mixtures?

5.8 Phase Equilibrium

Consider a multicomponent system .1, 2, . . . , r at temperature .T and pressure . P. We

seek to find how the . N1 moles of component 1, . N2 moles of component 2, …, . Nr
moles of component .r , will be redistributed among the liquid .l and vapor .v phases.
In equilibrium condition the fugacities of each component .i must be the same in
the liquid and vapor phases

.ψil = ψiv , for i = 1, 2, . . . , r, (5.50)

5.8 Phase Equilibrium 189

and we can write .ψil in terms of the activity coefficient

ψil = γil xi ψi,l pur e (T, P) .

.

On the other hand, using the Poynting correction, the fugacity of pure component
i can be found in terms of the
.

( )
l l
( ) vls,i ( )
.ψi, pur e (T, P) = ψi, pur e T, Ps,i exp P − Ps,i , (5.51)
RT
( )
in accordance with Eq. (5.40). In the above equation, .ψi,l pur e T, Ps,i is the fugacity
of pure component .i in its saturated liquid state at .T (when the vapor pressure is
. Ps,i ) and .vls,i is the molar volume of component .i in its saturated liquid state at

temperature .T . ( )
The term .ψi,l pur e T, Ps,i can be written in accordance with Eq. (5.21) as
( ) ( )
ψi,l pur e T, Ps,i = φi,l pur e T, Ps,i Ps,i ,
.

and the fugacity coefficient .φi,l pur e can be found from the equation of state of com-
ponent i

∫Ps,i ( )
l
( ) vi,l pur e
1
. ln φi, pur e T, Ps,i = − d P.
RT P
0

For the vapor phase

.ψiv = yi Pφiv (T, P, y1 , . . . , yr ) , (5.52)

where

∫P ( )
viv 1
. ln φi (T, P, y1 , . . . , yr ) = −
v
d P. (5.53)
RT P
0

Finally, Eq. (5.50) takes the form

( )
( ) vls,i ( )
γ l xi φi,l pur e T, Ps,i Ps,i exp
. i P − Ps,i
RT
= yi Pφi (T, P, y1 , . . . , yr )
v
(5.54)

So, if we start considering that we want to find the vapor composition . y1 , . . . , yr

and pressure . P from a given composition .x1 , . . . , xr in the liquid state and tem-
190 5 Non-ideal Mixtures

perature .T , our task will be to solve the .r equations given by Eq. (5.50) for the .r
unknowns10

. 1 y , . . . , yr −1 , P.

This is a difficult task and requires solving .r non-linear equations, since, in

Eq. (5.53), the molar volume of each component in the vapor phase .viv is depen-
dent on the composition.
To ease this task we rewrite Eq. (5.54) as

γ l xi Ps,i = yi P K i ,
. i (5.55)

where
φiv (T, P, y1 , . . . , yr )
Ki = ( )) ).
vls,i ( (
.
exp R T
P − Ps,i φi,l pur e T, Ps,i

When the vapor phase can be considered a mixture of ideal gases,

.ψiv = yi Pφiv = yi P,

and so, .φiv = 1. On the other hand,

.ψi,l pur e = Ps.i φi,l pur e = Ps.i .

Therefore, .φi,l pur e is also 1 and if we neglect the Poynting correction

. K i = 1.

5.8.1 Phase Equilibrium at Low Pressures

A reasonable simplifying assumption for multicomponent systems at low pressures

O (1 bar) is to suppose that the vapor phase behaves as a mixture of ideal gases. In
.
this case .φi = 1 and

.ψiv = yi P.

10 Remember that
∑
. yi = 1.
i
5.8 Phase Equilibrium 191

Fig. 5.13 Water–methanol Activity coefficients for

activity coefficients at 25 .◦ C
in terms of the molar fraction T=298.15 K
of methanol Molecule
x
water methanol
1.0126 1.7484
0.2 1.0435 1.4702
0.3 1.0869 1.2997
0.4 1.1398 1.1893
0.5 1.2006 1.1157
0.6 1.2685 1.0664
0.7 1.3431 1.0339
0.8 1.4240 1.0139
0.9 1.5111 1.0032

So, Eq. (5.50) becomes

γ l xi ψi,l pur e (T, P) = yi P, for i = 1, 2, . . . , r,

. i

and, by neglecting the Poynting correction,

γ l xi Ps.i (T ) = yi P, for i = 1, 2, . . . , r.
. i (5.56)

For concreteness, consider a binary water–methanol system at .T = 25 ◦ C. By

summing Eq. (5.56) over the components

γl x
. meth meth Ps,meth (T ) + γwl (1 − xmeth ) Ps,w (T ) = P. (5.57)

Therefore, for each molar fraction of methanol in the liquid solution, the activity
coefficients of methanol and water can be determined following the method presented
in Sect. 5.7 and the pressure can be calculated using the above equation. After the
pressure is calculated the molar fraction of methanol in the vapor phase is calculated
from Eq. (5.56)
l
γmeth xmeth Ps.meth (T )
y
. meth = (5.58)
P
At .T = 25 ◦ C the saturation pressures of water and methanol are respectively, 24
and 127 mm Hg. Their activity coefficients are presented in Fig. 5.13.
192 5 Non-ideal Mixtures

140.00

120.00

100.00
Pressure (mm Hg)

80.00

60.00

40.00

Bubble line functional groups

Dew line functional groups
20.00 Experimental values
Experimental values
Raoult's bubble line
Raoult's dew line
0.00
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Molar fraction of methanol

Fig. 5.14 Water–methanol bubble and dew lines at .25 ◦ C calculated from functional groups and
compared with experimental values [89] and with Raoult’s law

The bubble and dew lines for .T = 25 ◦ C resulting from the solution of Eqs. (5.57)
and (5.58) are presented in Fig. 5.14 and compared with experimental values.11 In
the same figure, we also plot the bubble and dew line when the mixture is supposed
to be ideal and follows Raoult’s law. Appreciable deviations from ideality can be
appreciated in all the ranges of methanol molar fractions.

5.8.1.1 Exercise

Replace the methanol with ethanol in the mixture for a temperature .T = 25 ◦ C.

Calculate the activity coefficients for molar fractions of ethanol that vary between
0 and 1 and, considering the vapor phase as a mixture of ideal gases, draw a phase
diagram such as the one shown in Fig. 5.14 based on the functional groups theory.
Compare your results with Raoult’s law and with available experimental data.

11Kontogeorgis GM, V. Yakoumis I, Meijer H, et al. (1999) Multicomponent phase equilibrium

calculations for water–methanol–alkane mixtures. Fluid Phase Equilib 158–160:201–209. https://
doi.org/10.1016/S0378-3812(99)00060-6 [89].
5.8 Phase Equilibrium 193

5.8.2 Phase-Equilibrium Using the Virial Equation of State

for the Vapor Phase

When the vapor phase cannot be considered as a mixture of ideal gases, the fugacity
coefficient .φiv given in Eq. (5.52) must be calculated for each component .i. In other
words, we need to know a suitable equation of state for describing the vapor mixture.
The mixture of vapors can be described by the truncated form of the virial equation
of state, Eq. (5.22) and the fugacity coefficient can be calculated as
⎛ ⎞
P ∑
. ln φi =
v ⎝2 y j Bi j − B ⎠ .
RT j

For a binary mixture of components labeled as 1 and 2

P ( ( ) )
. ln φ1v = B1 + 1 − y12 [2B12 − (B1 + B2 )]
RT
and
P ( )
. ln φ2v = B2 + y12 [2B12 − (B1 + B2 )] .
RT
Therefore, the equilibrium condition is given by the following set of non-linear
equations

γ l x1 Ps,1 = y1 P K 1 ,
. 1 (5.59)

γ l (1 − x1 ) Ps,2 = (1 − y1 ) P K 2 ,
. 2 (5.60)

where

P − Ps,1 ( ) P ( )
. ln K 1 = B1 − vls,1 + 1 − y12 [2B12 − (B1 + B2 )] ,
RT RT

P − Ps,2 ( ) P 2
. ln K 2 = B2 − vls,2 + y [2B12 − (B1 + B2 )] ,
RT RT 1
for the pressure . P and the molar fraction . y1 of component 1 in the vapor phase, when
the temperature and the molar fraction .x1 of this component in the liquid phase are
known.
The l.h.s. of Eqs. (5.59) and (5.60) are known, when experimental values are
disposable or it is possible to estimate the activity coefficients .γ1l and .γ2l in terms of
the temperature and the molar fraction .x1 of component 1 in the liquid phase. The
194 5 Non-ideal Mixtures

r.h.s. of these equations are non-linear functions of the pressure . P and the molar
fraction . y1 .
Due to the high non-linearity, the solution procedure is iterative and based on the
following steps:
1. Consider . K 1 = K 2 = 1 and calculate the pressure . P and the molar fraction . y1
by solving Eqs. (5.59) and (5.60);
2. With the calculated values of . P and . y1 , calculate . K 1 and . K 2 ;
3. Solve again Eqs. (5.59) and (5.60) with the new values of . K 1 and . K 2 ;
4. Compare the new with the old values of . P and . y1 till the desired convergence is
achieved.
Ideal Solutions

An interesting particular case is represented by ideal solutions. If the solution is ideal,

meaning that .2B12 − (B1 + B2 ) and .γ1l = 1, then, for .i = 1, 2,
P − Ps,i ( )
. ln K i = Bi − vls,i , (5.61)
RT
and we obtain
( P−Ps,i )( Bi −vls,i )
y Pe
. i RT = xi Ps,i ,

which can be considered an extension of Raoult’s law, when the vapor mixture cannot
be considered a mixture of ideal gases or, in other words, when the attraction forces
between the molecules cannot be neglected but are similar.

5.8.2.1 Exercise

Consider a mixture of ethanol and water at .25 ◦ C. Calculate the activity coefficients
l
.γi using the functional groups theory of Sect. 5.7, and find the virial coefficients
. Bwater , . Bethanol and . Bwater −ethanol , and the vapor pressures for water and ethanol at
this temperature that are available in the literature. With these data find the bubble
and dew lines of water-ethanol mixtures at .25 ◦ C using the method presented in
Sect. 5.8.2. Compare your results with the ones previously obtained when the vapor
phase was considered as a mixture of ideal gases.
5.8 Phase Equilibrium 195

5.8.3 Phase Equilibrium Using a Cubic Equation of State for

the Vapor Phase

Although the methodology is the same for all equations of state with the form . P =
P(T, v, y1 , . . . , yr ), we restrict our analysis to the Peng–Robinson equation of state
and its modification by Stryjek and Vera, in the present section.
The Peng–Robinson equation for mixtures has the same form as in Eq. (4.9)
RT a
. P= − , (5.62)
v̄ − b v̄ (v̄ + b) + b(v̄ − b)

where the force and volume parameters are dependent on the composition by mixing
rules. The simplest ones are12
∑ √ ∑
.a = yi y j ai j , ai j = δi j ai a j , b = yi bi ,
i, j i

and the pure component force parameters are calculated in accordance with
( )
a = ai Tc,i αi (T, ωi ) ,
. i

with
( )2
( ) RTc,i RTc,i
a Tc,i
. i = 0.45724 , bi = 0.07780 ,
Pc,i Pc,i

and
[ ( √ )]2
T
α = 1 + κi 1 −
. i .
Tc,i

12Other mixing rules based on experimental data have been proposed. See, for instance, Wilczek-
Vera G, Vera JH (1987) A comparative study of mixing rules for cubic equations of state. Fluid
Phase Equilib 37:241–253. https://doi.org/10.1016/0378-3812(87)80054-7, Poling BE, Prausnitz
JM, O’Connell JP (2001) The properties of gases and liquids. McGraw-Hill, New York and Vera
JH, Wilczek-Vera G (2016) Classical thermodynamics of fluid systems: principles and applications.
CRC-Press, New York.
196 5 Non-ideal Mixtures

Parameter .κi is related to the acentric factor of component .i by13

κ = 0.37464 + 1.54226ωi + 0.26992ωi2 .

. i

By defining
aP bP
. p= , q= .
(RT ) 2 RT

Equation (5.62) can be rewritten as

. Z 3 + AZ 2 + B Z + C = 0, (5.63)

where
( )
. A = (q − 1) , B = p − q (3q + 2) , C = q q + q 2 − p

This is a very convenient form for a cubic equation of state since it enables us to
find the compressibility factor in terms of temperature, pressure and composition,
avoiding dependence on the molar volume .v.
In the same way, the fugacity coefficient for the Peng–Robinson equation of state
can be found in Eq. (5.27) as

qi
. ln φi = (Z − 1) − ln (Z − q)
q
⎛ ⎞ ⎛ ( √ ) ⎞
p qi 2 ∑ Z + 1 + 2 q
+ √ ⎝ − y j pi j ⎠ ln ⎝ ( √ ) ⎠ (5.64)
2 2q q p j Z + 1− 2 q

13 A modification of the Peng–Robinson equation of state proposed by Stryjek and Vera [163–
165] reproduces well the vapor pressure data of pure compounds. This cubic equation of state,
the PRSV equation has one adjustable parameter per pure compound. When coupled with suitable
two-parameter mixing rules, it may be used to reproduce vapor–liquid equilibrium data of mixtures
with accuracy comparable to that of some common dual methods using activity coefficients for the
liquid phase and equations of state for the vapor phase.
In PRSV EOS, parameter .κ is written as
( √ )( )
T T
.κi = κi,0 + κi,1 1 + 0.7 − ,
Tc,i Tc,i
where

.κi,0 = 0.378893 + 1.489715ωi − 0.171384ωi2 + 0.0196544ωi3 ,

and .κi,1 is the single pure compound adjustable parameter.
5.8 Phase Equilibrium 197

5.8.3.1 Example

For concreteness, let us consider a methanol–water binary solution and that we want
to find the pressure and the molar fraction . y1 of methanol in the vapor phase when
this solution is in equilibrium with its vapor at .T = 50 ◦ C and the molar fraction of
methanol in the liquid phase is .x1 = 0.3.
The equilibrium condition is given by

γ l x1 Ps,1 = y1 P K 1
. 1 (5.65)

γ l (1 − x1 ) Ps,2 = (1 − y1 ) P K 2
. 2 (5.66)

In these equations, the only unknowns are the pressure . P and the molar frac-
tion . y1 .( Let us) first, calculate
( the )fugacity coefficients for the pure components
l l
.φ1, pur e T, Ps,1 and .φ2, pur e T, Ps,2 at saturated conditions. For .i = 1, 2

. ln φi,l pur e = (Z i − 1) − ln (Z i − qi )
⎛ ( √ ) ⎞
pi Z i + 1 + 2 qi
− √ ln ⎝ ( √ ) ⎠ (5.67)
2 2qi Z i + 1 − 2 qi

where
ai Ps,i bi Ps,i
. pi = , qi = ,
(RT )2 RT

and . Z i is the solution of

. Z i3 + Ai Z i2 + Bi Z i + Ci = 0,

where
( )
. Ai = (qi − 1) , Bi = pi − qi (3qi + 2) , Ci = qi qi + qi2 − pi .

For each pure compound (methanol or water), we have three roots for the com-
pressibility factor. Since
Ps,i v̄i
. Zi = ,
RT
the one that corresponds to the pure component in the liquid phase is the lowest one.
198 5 Non-ideal Mixtures

Table 5.5 Peng–Robinson parameters, compressibility factor, and fugacity coefficient for pure
components methanol and water in the saturated liquid state at 50 .◦ C
Methanol Water
( 3)
Jm
.ai 2 1.6621 0.97206
( mol ) −5 −5
.bi m /mol .4.1016 × 10 .1.8919 × 10
3

. Ps,i (mm Hg) 417.4 92.5

. pi .1.2646 × 10
−2 .1.639 × 10
−3

.qi .8.3845 × 10
−4 .8.5706 × 10
−5

. Ai .− 0.99916 .− 0.99991

. Bi .1.0967 × 10
−2 .1.4676 × 10
−3

.C i .− 9.8995 × 10
−6 .− 1.3313 × 10
−7

.Zi .9.9228 × 10
−4 .9.7141 × 10
−5
( 3 ) −5 −5
.vls,i m /mol .4.8541 × 10 .2.1443 × 10
l
.φi, pur e (T, Ps.i ) .0.64058 .0.66289

Table 5.5 shows the Peng–Robinson parameters, compressibility factor, and fugac-
ity coefficient for pure components methanol and water in the saturated liquid state
at 50 .◦ C.
When we neglect the Poynting correction, parameters . K 1 and . K 2 in Eqs. (5.65)
and (5.66) are given by
φ1v (T, P, y1 )
. K1 = l
,
φ1, pur e (T, Ps.1 )

φ2v (T, P, y1 )
. K2 = l
.
φ2, pur e (T, Ps.2 )

In the above equations

q1
. ln φ1v = (Z − 1) − ln (Z − q)
q
⎛ ( √ ) ⎞
( ) Z + 1+ 2 q
p q1 y1 p1 + (1 − y1 ) p12
+ √ −2 ln ⎝ ( √ ) ⎠
2 2q q p Z + 1− 2 q
5.8 Phase Equilibrium 199

q2
. ln φ2v = (Z − 1) − ln (Z − q)
q
⎛ ( √ ) ⎞
( ) Z + 1 + 2 q
p q2 (1 − y1 ) p2 + y1 p12
+ √ −2 ln ⎝ ( √ ) ⎠
2 2q q p Z + 1− 2 q

where . Z is the root of Eq. (5.63). Parameters .q1 and .q2 are dimensionless molecular
volumes
b1 P b2 P
q =
. 1 , q2 = ,
RT RT
parameter .q is a weighted average considering the molecular volumes of molecules
1 and 2,

q = y1 q1 + (1 − y1 ) q2 ,
.

parameters . p11 = p1 and . p22 = p2 are dimensionless force parameters for the inter-
actions between identical molecules and . p12 for the interactions between different
molecules
a1 P a2 P a12 P
. p11 = p1 = , p22 = p2 = p12 = ,
(RT )2 (RT )2 (RT )2

and parameter . p is the weighted average of the above force parameters

. p = y12 p11 + 2y1 (1 − y1 ) p12 + (1 − y1 )2 p22 .

The solution procedure is the same as the one shown in Sect. 5.8.2 for the virial
equation of state. We first consider parameters . K 1 = K 2 = 1 and calculate . P and . y1
by solving Eqs. (5.65) and (5.66). Then parameters . K 1 and . K 2 are recalculated with
these first estimations for . P and . y1 and Eqs. (5.65) and (5.66) are solved again. This
is repeated till convergence.

5.8.3.2 Exercises

1. Starting from Eq. (5.62), find the form of this equation of state given by Eq. (5.63).
2. Consider a methanol–water mixture in the vapor phase at .T = 50 ◦ C and . P =
150 mm Hg when the molar fraction of methanol is . ymeth = 0.8. Supposing that
this vapor mixture is well described by the Peng–Robinson equation of state,
find the compressibility factor . Z{ and the molar volume of the mixture. }
Answer: The roots are . Z = 3.8037 × 10−4 , 3.9453 × 10−3 , 0.99535 , but
only the larger root applies because at this temperature and pressure, the mixture
200 5 Non-ideal Mixtures

of methanol–water is in the vapor phase (see Fig. 5.14). The molar volume is
v = 0.11291 .m3 /mol.
.
3. Solve Eq. (5.62) and find the molar volume for the methanol–water mixture in
the same conditions of the previous exercise.
4. Consider a mixture of acetone and n-hexane at .T = 318 K. Calculate the activity
coefficients for molar fractions of acetone varying between 0 and 1 and, consider-
ing the vapor phase as a mixture of ideal gases, draw a phase diagram such as the
one shown in Fig. 5.14 based on the functional groups theory. In the following,
consider that the mixture( of vapors ) follows a Peng–Robinson equation of state,
find the fugacity .φi,l pur e T, Ps,i of these compounds in their saturated states at
318 K and the fugacity of each compound in the vapor phase .φiv (T, P, yi ). Use
these results to redraw the phase diagram and compare these last results with the
previous ones. Finally, try to compare your results with available experimental
results.
5. Consider a mixture of methylcyclohexane and ethylene at.50 ◦ C. Find the activity
coefficients based on functional
( ) groups and using the PRSV equation of state,
find the fugacity .φi,l pur e T, Ps,i of these compounds in their saturated states at
◦
.50 C and the fugacity of each compound in the vapor phase .φi (T, P, yi ). Use
v

these results to find . K 1 and . K 2 , the pressure and the molar fraction of ethylene
in the vapor phase when the molar fraction of ethylene in the liquid phase is
. x ethy varies from 0 to 1. Plot these values in a phase diagram such as the one

shown in Fig. 5.14 and compare your results with Raoult’s law. Functional group
parameters are given in Tables 8.23 and 8.24 of [137] and parameters .ω and .κ1
are given in [141].

5.9 Summary

In contrast with single-component systems, when the pressure is increased over a

mixture of vapors, the vapor–liquid transition pressure does not remain constant
during the vapor condensation process at a constant temperature. Therefore, for a
given vapor composition, the liquid and vapor will coexist in equilibrium in a range
of pressures between the dew and bubble states defined, respectively, as the states
where the first drop of liquid appears and the last bubble of vapor condensates.
The calculation of the dew and bubble states is dependent on the knowledge of:

• The activity coefficients.γil of the several components that form the liquid solution;
• The fugacity coefficient .φiv of these components in the vapor state.

The activity coefficient .γil is a measure of the non-ideality of the liquid solution
and the fugacity coefficient .φiv is a measure of the non-ideality of the mixture of
vapors.
For ideal solutions .γil = 1 and for mixtures of ideal gases .φiv = 1.
5.9 Summary 201

In this simplest situation, the distribution of components among the liquid and
vapor phases can be estimated by Raoult’s law.
Unfortunately, Raoult’s law cannot be used for most liquid solutions which are
non-ideal and require experimental values or a method for estimating the activity
coefficient.
In this chapter, we considered the molecules as composed of functional groups,
and the activity coefficients were estimated by taking into account their geometrical
and molecular potential contributions.
On the other hand, the estimation of the vapor fugacity coefficients is dependent
on the hypothesis it is possible to make about the thermodynamic state of the vapor
phase.
When the pressure is low enough, the vapor phase may be considered a mixture
of ideal gases. Otherwise, we should use an equation of state we judge appropriate
for the mixture of vapors.
Two alternatives were presented in this chapter for the estimation of the fugacity
coefficients based, respectively: (i) on a virial equation of state and (ii) on a cubic
equation of state.
Chapter 6
Surface Physics

Abstract Scientific interest in surface physics began very early with the XVIII
century efforts trying to explain surface tension and capillarity. Molecules on the
surface of a liquid are pulled inwards by the remaining molecules in the liquid phase
and this produces a state of tension on the liquid surface. On the other hand, when
interacting with a solid surface liquid molecules are subjected to attractive forces from
the solid molecules, and, when these forces prevail over the forces between liquid
molecules, this will give rise to a contact angle smaller than 90.◦ and to capillary
ascension inside small bore tubes. Surface tension forces liquid surfaces to contract
to the minimal area and Laplace’s law relates the surface tension with the internal
pressure that is produced and to the curvature of the liquid drops. Nevertheless,
surface tension has a molecular origin and is independent of surface curvature. In this
chapter, we try to give a deeper insight into the molecular scale for explaining surface
tension under the light of van der Waals theory considering the surface as a region
where the fluid density has a strong decay. It is then possible to relate the underlying
molecular parameters with the Helmholtz energy in excess with respect to the liquid
and vapor bulk phases. In liquid–solid and vapor–solid interaction, interfacial excess
energies in the triple line region are also the controlling macroscopic mechanisms
for the contact angle when a fluid interacts with a hydrophilic or hydrophobic surface
and are important macroscopic parameters in dynamical problems where the contact
angle gets away from its equilibrium value. A section is dedicated to exposing Cahn’s
theory for finding the liquid–solid and vapor–solid interfacial energies. The chapter
ends with a brief overview of emulsions.

6.1 Introduction

The cohesive forces among liquid molecules are responsible for the phenomenon
of surface tension. In the bulk of the liquid, each molecule is pulled equally in
every direction by neighboring liquid molecules, resulting in a null net force. The
molecules at the surface do not have the same molecules on all sides of them and
therefore are pulled inwards. This creates some internal pressure and, also, forces
liquid surfaces to contract to the minimal area. Surface tension is responsible for

© Associação Paranaense de Cultura 2024 203

P. C. Philippi, Thermodynamics, https://doi.org/10.1007/978-3-031-49357-7_6
204 6 Surface Physics

the shape of liquid droplets. Although easily deformed, droplets of water tend to be
pulled into a spherical shape by the imbalance in cohesive forces of the surface layer.
In the absence of other forces, including gravity, drops of virtually all liquids would
be spherical.
Another way to view surface tension is in terms of energy. A molecule in contact
with a neighbor is in a lower state of energy than if it were alone (not in contact with a
neighbor). The interior molecules have as many neighbors as they can possibly have,
but the boundary molecules are missing neighbors (compared to interior molecules)
and therefore have higher energy. For the liquid to minimize its energy state, the
number of higher energy boundary molecules must be minimized. The minimized
quantity of boundary molecules results in a minimal surface area. As a result of
surface area minimization, a surface will assume the smoothest shape it can. Since
any local curvature in the surface shape results in a greater area, a higher energy will
also result. Consequently, the surface will push back against any curvature singularity.
Research in surface physics was first undertaken to explain capillarity. In fact,
capillarity seemed to contradict the laws of fluid equilibrium. In consequence, many
worthless theories have been proposed with a view to explain apparent anomalies.
In accordance with Bashfort [8], “after long groping in the dark, it was found
to be desirable to discover by experiment what were the actual phenomena which
required explanation”. Hauksbee1 found that the height to which a fluid would rise
in a capillary tube of a given radius was the same for all thicknesses of the tube wall.
From this, it was apparent that the attracting force of the tube was situated at or near
the inner surface of the tube. But Hauksbee does not appear to have taken account
of the mutual attractions of the fluid particles.
James Jurin,2 a physician educated at Cambridge and Leiden made an important
experiment that added a new fact to those discovered by Hauksbee; the height to
which water rose in a tube depended only on the diameter at the position of the
meniscus. A tube that was wide at the bottom but narrow at the top could therefore
hold in suspension a greater volume of water than one of a uniform bore. This fact
undermined Hauksbee’s not very coherent explanation that the rise was due to a
diminution of the “gravitating force” by a horizontal attraction of the whole of the
glass wall. Jurin claimed to have found “the real cause of that phenomenon, which is
the attraction of the periphery, or section of the surface of the tube, to which the upper
surface of the water is contiguous and coheres”. He expounded six propositions,
such as, for example, that water particles attract water but not as strongly as they
are attracted to glass, whereas mercury (quicksilver) attracts mercury more strongly
than mercury is attracted to glass. He established that the depression of mercury in a
capillary tube, like the rise of water, is the reciprocal of the bore.

1 Francis Hauksbee (1710) ‘An Account of an Experiment Touching the Ascent of Water Between
Two Glass Planes, in a Hyperbolic Figure’. Phil. Trans. 27, pp. 539–540 [64].
2 James Jurin (1717) ‘An Account of Some New Experiments, relating to the Action of Glass Tubes

upon Water and Quicksilver’. James Jurin, M. D. Reg. Soc. and Coll. Med. Lond. Soc. Phil. Trans.
30, pp.1083–1096 [82].
6.1 Introduction 205

Clairaut3 was the first to attempt to explain capillary phenomena on right princi-
ples, by referring them to the mutual attraction of the particles of the fluid, and to
the attraction of the particles of the solid on the particles of the fluid and supposing
these attractions to depend upon the same function of the distance. After this precise
and promising start his analysis quickly goes astray. He assumes that all the liquid in
a tube of, say, one-twentieth of an inch in diameter is within the attractive range of
the glass walls, an assumption tentatively made by Hauksbee in spite of the evidence
of his own experiments.
Shortly afterwards Segner4 introduced the supposition that forces of attraction of
both the particles of the solid and of the fluid “were not sensible at sensible distances”.
He concluded that these forces gave a constant tension to the free capillary surfaces,
and tried to calculate the forms of sessile drops of fluid with a view to compare them
with their measured forms. But in his calculations, he took into account only the
curvature of the vertical sections made by a plane passing through the axis of the
drop. In addition, his measurements of the actual forms appear not to have been very
precise.
In 1805 Thomas Young wrote5 : “It has already been asserted, by Mr. Monge and
others, that the phenomena of capillary tubes are referable to the cohesive attraction
of the superficial particles only of the fluids employed” … “Segner, who appears to
have been the first that maintained a similar opinion, has shown in what manner the
principle may be deduced from the doctrine of attraction, but his demonstration is
complicated, and not perfectly satisfactory; in applying the law to the forms of drops,
he has neglected to consider the very material effects of the double curvature, which
is evidently the cause of the want of a perfect coincidence of some of his experiments
with his theory”.
Young criticizes Segner’s notion of a tension only on surfaces of variable cur-
vature and couples this idea with the assertion that there is a fixed angle of contact
between any given pair of liquid and solid, an assertion which he describes as “one
observation, which appears to be new, and which is equally consistent with theory
and with experiment”. He uses these two facts to produce the first satisfactory phe-
nomenological treatment of capillary rise. He writes: “It is well known, and it results
immediately from the composition of forces, that where a line is equably distended,
the force that it exerts, in a direction perpendicular to its own, is directly as its cur-
vature; and the same is true of a surface of simple curvature; but where the curvature
is double, each curvature has its appropriate effect, and the joint force must be the
sum of the curvatures in any two perpendicular directions”.
Young’s work was and would have been seen by his predecessors as a triumphant
success. The next advance came at once; it required Laplace’s combination of phys-
ical insight and a mathematical grasp which grew from the resurgence of French
mathematics at the turn of the XVIII century.

3 M. Clairaut (1743), ‘Theorie de la figure de la Terre’, David Fils, Paris.

4 I. A. Segner (1751), ‘De figuris superficierum fluidarum’, Comment. Soc. Reg. Sci. Gottingensis
1, pp. 301–372.
5 T. Young (1805), ‘An essay on the cohesion of fluids’, Phil. Trans. Roy. Soc. 95, pp. 65–87.
206 6 Surface Physics

6.2 Laplace

In the field of capillarity it is usual to consider together the work of Young and Laplace,
and it is true that they both obtained some of the same important results within a year of
each other. Their aims and methods were, however, quite different. In reading Young we are
reading 18th-century natural philosophy; in reading Laplace we are reading 19th-century
theoretical physics. This ‘sea-change’ in the early years of the new century is as dramatic as
that of the scientific revolution of the 17th century, and was due to the efforts of the great
French school of mathematical physics of that time. This is not the place to discuss the origin
of this second revolution but to concentrate only on how it led to a revival of the subject of
cohesion and to a second period of advance. The man responsible was Laplace. Rowlinson
[147]

Laplace, in his ‘Traité de Mécanique Céleste’ published in 18056 is considered

to be the first who formally derived the law relating the pressure with the surface
tension that is presently known as Young–Laplace’s law.
In his 156 pages dedicated to the study of the surface tension phenomena, he gave
a very detailed treatment of the forces responsible for capillarity: “Jurin analyse
avec exactitude, toutes les forces qui peuvent concourir á élever l’eau dans un tube
de verre. Mais sa théorie, exposée avec l’élégance qui caractérise son bel Ouvrage,
laisse á desirer l’explication de la loi de cette ascension qui, d’après l’ expérience,
est en raison inverse du diamètre du tube”.
He, also, attributed the hydrophilic or hydrophobic behavior of solid surfaces to the
short range interaction between the fluid molecules and the solid surface, responsible
for the contact angle “Ainsi l’attraction des tubes capillaires n’a d’influence sur
l’élévation ou sur l’abaissement des fluides qu’ils renferment qu’ en déterminant l’
inclinaison des premiers plans de la surface du fluide intérieur, extrêmement voisins
des parois du tube, inclinaison dont dépend la concavité ou la convexité de cette
surface, et la grandeur de son rayon”.
Figure 6.1 shows Figure 6 of Laplace’s supplement ‘Sur l’action capillaire’ [99].
For deriving Laplace’s law consider a liquid droplet (phase .α) in mechanical
equilibrium with a phase .β (a gas, another liquid or its own vapor), in absence
of gravity (Fig. 6.2). In these conditions, the droplet surface will be spherical. In
fact, as commented above, whereas, in the bulk phases, the intermolecular forces are
balanced, on the surface, the liquid molecules are subjected to an unbalanced field
of intermolecular forces. These molecules will attract themselves with electrostatic
forces, being as closest as possible one from another and the sphere offers the minimal
superficial area for this purpose.
Suppose now that, by any means, we try to detach the surface element shown in
Fig. 6.2 from the liquid droplet. The remaining liquid molecules will react to this
detachment with a force .γ 2π (R sin θ ) directed outward the surface element and
acting along the perimeter of this element shown in Fig. 6.2. Thus .γ is considered

6P. S. Laplace (1805), Sur l’ action capillaire, Supplement au Dixième Livre du Traité de Mécanique
Céleste, Courcier, Paris, pp. 398–553 [99].
6.2 Laplace 207

Fig. 6.1 Figure 6 of

Laplace’s supplement
Sur l’action capillaire

Fig. 6.2 Derivation of

Laplace’s law

to be related to the—short-range—attraction forces between contiguous molecules

along the perimeter of the detached surface element.
While the horizontal component of this force is null, its vertical component
.γ 2π (R sin θ) sin θ will be balanced by the pressure difference . Pα − Pβ on the two
faces of the curved surface, in such a way that
( )
. Pα − Pβ π (R sin θ )2 = γ 2π (R sin θ) sin θ,
208 6 Surface Physics

Fig. 6.3 Capillary rise inside a cylindrical tube

or
2
. Pα − Pβ = γ . (6.1)
R
The above equation can be generalized for any curved interface by taking into
account its curvature radius .R, given by . R1 and . R2 in accordance with

1 1 1
. = + .
R R1 R2

Equation (6.1) is the famous Laplace’s equation, proposed in 1805 [99], which
relates the pressure imbalance in curved surfaces of liquids with the interface tension.

6.2.1 Exercises

1. Consider water rising inside a capillary cylindrical tube with radius . R (Fig. 6.3).
By making a balance of forces for the volume indicated by the dashed lines show
that, in equilibrium, the height .h of the liquid column is given by Jurin’s law
[82],
2γ cos θ
.h = ,
ρg R
6.3 The Form of a Sessile Drop 209

where .θ is the contact angle between the liquid and the solid surface, .g is the
local acceleration due to gravity and .ρ is the density of water. What is the liquid
pressure on the top of the liquid column?
2. Starting from the Laplace’s equation, Eq. (6.1), and supposing the meniscus
surface to be spherical with a radius of curvature .R, retrieve the Jurin’s law of
Exercise 1.
3. Consider now that during the capillary rising the height .h = h (t) and that the
liquid rises into the tube due to the imbalance of capillary . Fcap , gravitational . Fg
and viscous . Fvisc forces in such a way that
( )
d 2 dh
. πR h = Fcap − Fg − Fvisc ,
dt dt

in accordance with Newton’s second law. Find a differential equation for .h (t)
and try to solve this (nonlinear) equation7 using, e.g. a Runge–Kutta method,
plotting .h in terms of the time .t. What are your conclusions? By considering a
Hagen–Poiseuille flow, the viscous force is given by
dh
. Fvisc = 8π ηh
dt
where .η is the dynamic viscosity of the liquid.

6.3 The Form of a Sessile Drop

Many years have elapsed since this work was commenced, and it is even now only partially
completed. My object was to test the received theories of Capillary Action, and through
them the assumed laws of molecular attraction, on which they are founded. To this end it
was proposed to compare the actual forms of drops of fluid resting on horizontal planes
they do not wet, with their theoretical forms. After some trials, a satisfactory micrometrical
instrument was constructed for the measurement of the forms of drops of fluid, but my
attempts to calculate their forms as surfaces of double curvature failed entirely, and my
undertaking must have ended here if I had depended upon my own resources. But at this
point Professor J. C. Adams furnished me with a perfectly satisfactory method of calculating
by quadratures the exact theoretical forms of drops of fluids from the Differential Equation
of Laplace, an account of which he has now had the kindness to prepare for publication.
Francis Bashforth (1883), “An attempt to test the theories of capillary action by comparing
the theoretical and measured forms of drops of fluids”, Cambridge University Press, p. 158
[8]
As it was commented at the beginning of this chapter, a liquid drop deposed on a
horizontal surface will have a spherical form when it not subjected to gravity.
Consider, now, a liquid drop that is deposed on a horizontal surface and under the
action of gravity .g (Fig. 6.4). Considering radial symmetry the problem is to find the

7 The Bosanquet equation [18].

210 6 Surface Physics

Fig. 6.4 A sessile drop and its meridional and transverse plans
6.3 The Form of a Sessile Drop 211

equation of the drop surface, i.e., the relationship between the vertical coordinate .z
and the horizontal coordinate .x under equilibrium conditions in the meridional plan
shown in the figure.
For each point in the surface, we have two curvature radii. . R1 is the curvature
radius on the meridional plan and . R2 is the curvature radius on the transverse plan.
The projection of . R2 on the horizontal direction of the meridional plan is the ordinate
. x. So

. x = R2 sin φ,

where .φ is the angle between the normal to the surface and the vertical direction.
Nevertheless,
dz
. tan φ = ,
dx
and so
sin φ
cos φ tan φ
. sin φ = =( )1 .
sec φ 1 + tan2 φ 2

It results
dz
1
= (
dx
.
( dz )2 ) 21
.
R2
x 1+ dx

On the other hand, the curvature radius . R1 is given by

1 dφ dφ d x
. = = .
R1 ds d x ds

But
dx 1 1
= cos φ = =( ,
( dz )2 ) 21
.
ds sec φ
1+ dx

and
( ) d2z
dφ d dz dx2
. = arctan = ( )2 .
dx dx dx 1 + ddzx

It results that
d2z dz
dx2 ∆P
( + dx
= . (6.2)
( dz )2 ) 23 ( ( dz )2 ) 21
.
γ
1+ dx
x 1+ dx
212 6 Surface Physics

The pressure difference can be written as

( )
∆P = ρ − ρg gz + C,
.

where .C is a constant that can be determined considering that when .z → 0 the drop
surface becomes spherical with . R1 = R2 = b. So
2γ
C=
. ,
b
and Eq. (6.2) becomes
d2z
( )
dx2
dz
ρ − ρg gz 2
( + dx
1 = + . (6.3)
( dz )2 ) 23 ( )
.
( dz )2 2 γ b
1+ dx
x 1+ dx

Numerical solutions of the sessile drop problem are known for more than a century
[8, 126] and results of this theory are found in many applications. For example, a
technique to measure surface tension experimentally is based on comparing measured
and calculated profiles of a drop of suitable size [23].
In addition to numerical solutions, analytical solutions were provided by some
authors in particular situations. See, e.g., [32].
The solution of Eq. (6.3) is specially important for measuring the surface tension
by the pendant drop method. Song and Springer [160] developed a computer-aided
method based on the comparison between the theoretical profile given by Eq. (6.3)
and the profile that is extracted from a pendant liquid drop (Fig. 6.5) using image
analysis techniques. In this case, the drop apex is chosen at the lower point of the
drop and the ordinate .z is directed upward.

Pendant drop Segmented image Extracted profile

Fig. 6.5 Measurement of the interface tension by the pendant drop method
6.4 Surface Tension and the Helmholtz Energy 213

6.4 Surface Tension and the Helmholtz Energy

A fluid interface separating two phases.a and.b is a region where some thermodynamic
properties have a strong variation. This is the case of the density when phase .a is a
liquid and phase .b its vapor. From a molecular standpoint, the interface is a region
where molecules of phase.a are simultaneously subjected to intermolecular attraction
forces from its own phase .a and from the nearby molecules of phase .b. When the
kinetic energy of a molecule in phase.a is large enough to compensate for its potential
energy—which is always negative—this molecule will change to phase .b. The same
comment for molecules in phase .b.
In this section, we are not concerned with phase transition, but with the mechanical
properties of fluid interfaces. So we will suppose that there is no mass transfer
between the two phases or that the two phases are already in equilibrium with respect
to mass transfer, i.e., that the Gibbs potentials .μia and .μib are the same in the two
phases for .i = a, b. In this aspect, we can consider that the interface has an energy
excess with respect to the bulk phases. In fact, the potential energy related to the
intermolecular attraction between molecules in phase .a and a single molecule .a in
the interface can be quantified as

(1−2)
φ (x) =
. aa φaa (||r − x||) n a (r) dr ≤ 0, (6.4)
||r−x||≥σa

(1−2)
where .φaa (||r − x||) is the potential energy related to the intermolecular attraction
between any two molecules.a, supposed to be only dependent on the distance.||r − x||
that separates a molecule .a in .r from the molecule .a in the position .x of the interface,
.σa is the molecular diameter of molecule .a and .n a is the number of molecules .a per
unit volume.
In the bulk phase .n a is constant, but, in the interface, .n a decreases and .φaa → 0.
This gradient in the potential energy is responsible for a force on molecule .a
∇φaa
g
. aa =− ,
m
directed toward the bulk phase .a. Using the same reasoning we can conclude that
phase .b will attract the molecule .a with a force
∇φab
g
. ab =− ,
m
and this force is responsible for the reduction of the interface tension .γab with respect
to the surface tension .γa , i.e., with respect to the surface force that arises, when a
liquid is in equilibrium with its own vapor.
Since the potential energy is always negative, a reduction on its absolute value
means an excess in energy. For the molecules in the interface to minimize this energy
excess, the number of higher energy boundary molecules must be minimized.
214 6 Surface Physics

From a macroscopic point of view, the Helmholtz energy of this two-phase system
can be written as

. F = Fa + Fb + Fab ,

where . Fab is the Helmholtz energy in excess in the interface. In this case, our ther-
modynamic system is considered to be a drop of phase .a immersed on phase .b.
If a small perturbation results on an volume increase .d Va and, consequently, a
volume decrease .d Vb = − d Va at constant temperature,

d Fa = − Pa d Va ,
.

d Fb = Pb d Va ,

and

d Fab = f ab d Aab ,
.

where . f ab is the Helmholtz energy per unit area of the interface.

Since in equilibrium, . F must be a minimum when the total volume and tempera-
ture are kept constants

. (Pb − Pa ) d Va + f ab d Aab = 0,

and we retrieve Laplace’s law Eq. (6.1)

γab
. Pa − Pb = ,
R
where .γab = f ab and
1 d Aab
. = .
R d Va

To convince ourselves that the interface tension is the Helmholtz energy in excess
in the interface, let us derive Laplace’s law based, this time, on the principle of
maximum entropy, instead of the principle of minimum Helmholtz energy. In this
case, we must keep the internal energy constant and the entropy variations that follows
a small volume variation of phase .a will be

1 Pa
d Sa =
. dUa + d Va ,
Ta Ta
1 Pb
d Sb = dUb + d Vb ,
Tb Tb
d Sab = sab d Aab .
6.5 A Deeper Insight into the Physics of Interfaces 215

On the other hand

.dU = dUa + dUb + u ab d Aab = 0,

as a condition for the entropy to be maximum in equilibrium.

Therefore
( ) ( ) ( )
1 1 Pa Pb u ab
.d S = − dUa + − d Va − − sab d Aab
Ta Tb Ta Tb Tb

In equilibrium conditions

T = Tb = T,
. a

and

d S = 0.
.

So
⎛ ⎞
⎜ ⎟ d Aab
. Pa − Pb = ⎝u ab − T sab ⎠
d Va
= f ab

as it was to be demonstrated.

6.4.1 Exercise

Show that Laplace’s equation can be also derived using the minimum principle for
the internal energy (at constant entropy).

6.5 A Deeper Insight into the Physics of Interfaces

To get a deeper insight into the physics of interfaces let us go back to Eq. (6.4)
expressing the potential energy of a single molecule .a in the interface, supposing,
now, that phase .a is a liquid and phase .b its vapor. So omitting the index .a and .b let
.n (r) be developed in a Taylor series around .n (x)

1
.n (r) = n (x) + ηα ∂α n + ηα ηβ ∂αβ n + · · · ,
2
216 6 Surface Physics

where .η = r − x. Inserting this expansion into Eq. (6.4), since .φ (1−2) (||η||) is an
even function of .η,

φ (x) = − 2an (x) − κ∇ 2 n + · · ·

. (6.5)

where

1
.a=− φ (1−2) ||x|| dη ≥ 0,
2
||η||≥σa
1
κ=− φ (1−2) (||η||) η2 dη ≥ 0,
6
||η||≥σa

are molecular parameters that depend on the interaction potential between the
molecules. In fact, parameter .a can be shown to be related to the attractive part
of the van der Waals equation of state and parameter .κ to the surface tension.
The force acting on a single molecule .a and directed toward the liquid bulk phase
will, thus, be given by
2a κ
.g= ∇n + ∇∇ 2 n,
m m
and the force per unit volume acting on .n liquid molecules can be written as

ρg = 2an∇n + κn∇∇ 2 n
.
( κ )
= ∇ an 2 + κn∇ 2 n + (∇n)2 − ∇ · (κ∇n∇n) .
2
In general the radius of curvature .R is much larger than the interface thickness
and this last expression can be simplified if we consider that the interface is very
thin and the density variation happens along the normal direction .x, neglecting the
1
curvature . R of the interface,
( ( ) )
d d 2n κ dn 2
.ρg x = an + κn 2 − .
2
dx dx 2 dx

When the interface is in equilibrium and at rest this force creates a gradient
in the density and, consequently, in the ideal-gas pressure . P0 = nkT , where .k =
R/N Avogrado is the Boltzmann constant
d (nkT )
.ρgx = , (6.6)
dx
6.5 A Deeper Insight into the Physics of Interfaces 217

in such a way that

( ( )2 )
d d 2n κ dn d ( )
. κn 2 − = nkT − an 2 . (6.7)
dx dx 2 dx dx
pressure gradient
surface force

The above equation means that the surface force is locally equilibrated, in the
interface, by a pressure gradient.8 We can identify the term between the parenthesis
in the r.h.s. of the above equation as the pressure. P when the molecules are considered
as material points subjected to intermolecular attractive forces.
By integrating Eq. (6.7)
( )2
d 2n κ dn
.κn − = P + C,
dx2 2 dx

where .C is an integration constant that is to be assigned to the saturation pressure in

the bulk phases, since there are no gradients in the bulk phases (the liquid and vapor
phase were assumed to be in equilibrium)

C = − Ps (T ) .
.

Therefore,
( )2
d 2n κ dn
κn
.
2
− = P − Ps (T ) . (6.8)
dx 2 dx

The Helmholtz energy per unit volume can be written as

. ψ = n f,

and if we identify, in accordance with van der Waals, the Helmholtz energy in excess
in the interface as [181]
( )2
κ dn
.ψexc = . (6.9)
2 dx

Equation (6.8) can be written as

dψexc
n
. − ψexc = P − Ps (T ) ,
dn

or, by dividing the above equation by .n 2 ,

8A more rigorous derivation of Eq. (6.7) can be given, considering it as a particular case of the
momentum balance equation for two-phase systems. Interested readers are invited to read Sect. 8.2.
218 6 Surface Physics

1 dψexc 1 1
. − 2 ψexc = 2 (P − Ps (T )) ,
n dn n n
or
( )
d ψexc 1
. = (P − Ps (T )) .
dn n n2

Now, we integrate the above equation between the saturated number density of
molecules in the vapor phase .n vs (T ) and an arbitrary number density .n inside the
interface,
n
1 ( ( ' ) )
.ψexc (n, T ) = n '2 P n , T − Ps (T ) dn ' . (6.10)
n
n vs

Therefore, we get an equation expressing the Helmholtz energy in excess in the

interface in terms of the equation of state of the substance with which we are dealing.
This result is not surprising since is based on the equilibrium between surface forces
and pressure gradient inside the interface, expressed by Eq. (6.7). So the effect of the
surface forces is measured by the pressure gradient they produce across the interface.
In the bulk liquid and vapor phases this pressure gradient is null and, also, the excess
energy.
In fact, when the number density .n becomes .n s in the liquid bulk phase .ψexc = 0
and so
n s
1 ( ( ' ) ) '
. P n , T − Ps (T ) dn = 0,
n'2

n vs

or, by replacing .n by .v = 1/n as the integration variable,

vvs

. Pdv ' = Ps (vvs − v s ) ,

v s

which is the well-known Maxwell area rule, enabling to find the saturation pressure
Ps (T ) corresponding to a given equation of state . P (v, T ).
.
Equation (6.10) can be written using dimensionless variables
n∗
∗ ∗ 1 ( ∗ ( ') ∗
( ∗ )) '
.ψexc =n P n − Ps T dn , (6.11)
(n ' ) 2
n ∗vs

where
6.5 A Deeper Insight into the Physics of Interfaces 219

0.9 T*=0.6

0.8 T*=0.7
T*=0.9
0.7
exc

0.6
*

0.5

0.4

0.3

0.2

0.1

0
0 0.5 1 1.5 2 2.5
n*
Fig. 6.6 Helmholtz excess energy in the liquid–vapor interface, calculated with Eq. (6.10) for
different reduced temperatures. The van der Waals equation of state was used for the calculation

∗ ψexc n P T
.ψexc = , n∗ = , P∗ = , T∗ = .
Pc nc Pc Tc

Although Eq. (6.7) was derived considering a system of material points with added
attractive forces from the surface, more rigorous derivations can show that pressure
. P can be written in terms of .n and . T , using any arbitrary equation of state [156].

Figure 6.6 shows the excess function that is obtained with the van der Waals
equation of state, which can be written in dimensionless form as
8n ∗ T ∗
P∗ = − 3n ∗ .
2
. (6.12)
3 − n∗
∗
Results show that the Helmholtz energy in excess in the interface, .ψexc , is always
non-negative and becomes zero in the bulk phases in equilibrium when .n ∗ = n ∗vs and
∗ ∗ ∗
.n = n s . Furthermore, as would be expected, a temperature increase reduces .ψexc ,
which converges to zero when .T approaches the critical temperature .Tc .
220 6 Surface Physics

6.5.1 Exercises

∗
1. Equation (6.11) giving the Helmholtz energy in excess .ψexc (n ∗ , T ∗ ) in a liquid–
vapor interface is dependent on the equation of state that we find suitable for the
∗
substance we are dealing with. Find the expression for .ψexc when the Redlich–
Kwong equation of state is used.
2. Repeat Exercise 1 for the Soave equation of state.

6.5.2 Density Profile

Equation (6.9) gives the excess function in terms of the surface parameter .κ. In
dimensionless form, it can be written as
dn ∗
.dx∗ = √ ∗ ,
2ψexc

where .x ∗ = √
x
κ .
nc Pc
Therefore, by integrating the above equation between a point .x in the interface
n ∗ +n ∗
where the density .n ∗ = vs 2 s and an arbitrary point in the interface
n∗
dn '
. x∗ − x∗ = √ ,
∗ (n ' )
2ψexc
n∗ ∗
vs +n s
2

it is possible to find the density profile .n ∗ (x ∗ ) along the interface thickness.

This profile is shown in Fig. 6.7 for a van der Waals equation of state. The pressure
difference . P ∗ − Ps∗ is also shown in the same figure. In the vapor phase, the pressure
is the saturation pressure . Ps . Starting from the vapor phase, as we leave this phase the
pressure has a slight increase due to the increase in the density .n. Nevertheless, as we
approach the liquid phase, the attractive term .an 2 , in the van der Waals equation of
state, becomes dominant with respect to the dispersion term .nkT / (1 − bn) and the
pressure decays to a minimum. After this point, the denominator .1 − bn approaches
zero and the dispersion term begins to prevail over the attraction term restoring the
saturation pressure when .x approaches the liquid phase.

6.5.2.1 Exercise

Find the liquid–vapor density profile for water at a reduced temperature .T ∗ = 0.6,
using a Soave equation of state.
6.5 A Deeper Insight into the Physics of Interfaces 221

n*
2
P*- P*s
vapor phase

liquid phase
1
attractive forces

-1.0 -0.5 0.5 1.0 1.5

-1 x*-x*

-2

Fig. 6.7 Figure shows the density profile.n ∗ and the pressure. P ∗ − Ps∗ along the interface thickness

6.5.3 Interface Thickness

The interface thickness is commonly defined as in Fig. 6.8,

( )
dn n s − n vs
. = ,
d x x=x λ

or using Eq. (6.9)

Fig. 6.8 Definition of the

interface thickness vapor

density profile
liquid
x
222 6 Surface Physics

n c n ∗s − n ∗vs
. λ= / √ .
2Pc ∗ (x)
ψexc
κ

6.5.4 Surface Tension

Consider an isothermal one-dimensional liquid–vapor system at rest with a plane

interface orthogonal to the x-axis. The Helmholtz free energy in excess . exc in
the interface per unit area orthogonal to the x-axis must take the van der Waals
contribution into account [181], i.e., the effect of strong density gradients on the
local excess function .ψexc (n, T ),
⎛ ⎞
∞ ⎜ ( )2 ⎟
⎜ ⎟
= ⎜ψexc (n, T ) + κ dn ⎟ d x, (6.13)
. exc ⎜ 2 dx ⎟
−∞
⎝ ⎠
f (n, ddnx ,T )

where

ψexc (n, T ) = ψ (n, T ) − ψ0

. (6.14)

In the above equation .ψ0 = ψ0 (n ∞ , T ) is the Helmholtz free energy of the homo-
geneous phases and .n ∞ is the density of the homogeneous phase when .x → − ∞,
i.e., the vapor phase, where .n ∞ = n vs , or when .x → ∞, the liquid phase, where
.n ∞ = n s .

The problem we want to solve here is to find the excess energy .ψexc (n, T ) in
equilibrium conditions, in such way that the integral in Eq. (6.13) is a minimum.
This problem can be solved using variational calculus (see Appendix A). Since
the integrand . f in Eq. (6.13) does not depend explicitly on .x, the Euler–Lagrange
equations reduces to
∂f
n'
. − f = C,
∂n '

where .n ' = dn
dx
and .C is a constant. In the present case,
( )2 ( )2
dn κ dn
κ
. − ψexc (n, T ) − = C.
dx 2 dx

Since .ψexc .→ 0 and . ddnx → 0 when .x → ± ∞, we can conclude that the solution
we seek is
6.5 A Deeper Insight into the Physics of Interfaces 223
( )2
κ dn
.ψexc (n, T ) = .
2 dx

In equilibrium conditions, the minimal Helmholtz free energy in excess is the

surface tension .γ and so, from Eq. (6.13),
∞ ( )2 n s ( )
dn dn
γ =κ
. dx = κ dn, (6.15)
dx dx
−∞ n vs

or
n
√
s
√
.γ = n c 2κ Pc ∗ dn ∗ .
ψexc
n vs

This last equation can be rewritten as

γ2
κ=
. , (6.16)
I 2 n 2c 2Pc

where
n ∗s
√
. I = ∗ dn ∗
ψexc (6.17)
n ∗vs

∗
Since the excess function .ψexc can be determined for a given equation of state,
Eq. (6.16) relates the surface tension .γ with the molecular parameter .κ and can be
used to find the value .κ of from experimental values of .γ .

6.5.5 Sample Case: Water

Table 6.1 summarizes the main thermodynamic properties of water obtained with a
Peng–Robinson equation of state,
n RT aα (T, ω)
. P= − ,
1 − bn 1 − (bn)2 + 2bn

where .n is the number of moles per unit volume (otherwise, we would use the Boltz-
mann constant .k, instead of . R, in the case.n was assigned to the number of molecules
per unit volume),
224 6 Surface Physics

Table 6.1 Surface properties of water ( )

.T
∗ .γ (N/m) ∗
.n vs
∗
.n s I .κ
N
.λ (Å)
(kg/m3 )2
0.4220 0.07560 0.00002 3.75343 8.26090 .4.86366 × 10−17 2.69
0.4622 0.07183 0.00009 3.69318 7.52090 .5.29792 × 10
−17 2.97
0.5008 0.06773 0.00034 3.62967 6.82980 .5.71060 × 10
−17 3.27
0.5997 0.05568 0.00400 3.43707 5.15170 .6.78435 × 10
−17 4.17
0.7001 0.04183 0.02193 3.18226 3.55090 .8.06069 × 10
−17 5.49
0.8005 0.02752 0.07806 2.83531 2.07030 .1.02591 × 10
−16 7.92
0.9009 0.01250 0.23065 2.32105 0.78810 .1.46081 × 10
−16 13.92
0.9395 0.00700 0.35270 2.03483 0.38959 .1.87617 × 10
−16 20.44
0.9596 0.00370 0.44937 1.84592 0.21757 .1.67950 × 10
−16 30.00
0.9797 0.00201 0.59576 1.60197 0.07998 .3.66696 × 10
−16 50.72

( ( √ ))2 R 2 Tc2 RTc

α = 1 + κ 1 − T∗
. , a = 0.45724 , b = 0.0778 ,
Pc Pc

with

κ = 0.37464 + 1.54226ω − 0.26992ω2 ,

.

considering an acentric factor .ω = 0.344. The critical properties of water are .Tc =
647.3 K, . Pc = 221.2 bar.
The table was constructed using Maxwell’s area rule, to finding the saturation
densities .n ∗vs and .n ∗s and the integral . I in the fifth column was calculated using
∗
Eq. (6.17). The excess energy .ψexc was calculated using Eq. (6.11) and is given, in
this case, by

⎡ ⎤
1−Bn ∗
ln
∗
( ) ⎢ 1+An ∗
⎥
.ψexc = E × α T ∗, ω × n∗ × ⎣ ⎦
1−Bn ∗∞
− ln 1+An ∗∞
( [ )]
∗ ∗ n ∗ 1 − Cn ∗∞
+ D × n × T × ln ∗
n ∞ (1 − Cn ∗ )
⎡ n ∗∞ T ∗ ⎤
( ∗ )
D × 1−Cn
n ⎢ ( ∗
∞ ) ⎥
+ 1− ∗ ×⎣ ⎦ (6.18)
n∞ ∗
α(T ,ω)n ∞ ∗2
− F × 1−C 2 n ∗2 +2Cn ∗
∞ ∞
6.5 A Deeper Insight into the Physics of Interfaces 225

where . A = 0.58443, . B = 0.10027, .C = 0.24208, . D = 3.1116, . E = 6.4655, . F =

4.4269. The density .n ∗∞ corresponds to the dimensionless density of the bulk phases,
indifferently .n ∗vs or .n ∗s .
Parameter .κ was calculated from Eq. (6.16), based on experimental values of .γ .
The expression
−17
.κ = − 5 × 10 ln γ − 9 × 10−17

gives a reasonable approximation for expressing .κ in terms of .γ and this approxi-

mation can be useful in practical calculations.
The interface thickness .λ is presented in the last column of Table 6.1, given in
Angstroms. Interfaces are very thin for temperatures far from the critical point and
their .O(Å) thicknesses are a challenge for a continuous approach like the one that
was used in this section. Molecular dynamics simulation is suggested to interested
readers (see, e.g., Ref. [169]). The interface thickness, .λ has experimentally been
measured via both ellipsometry [85] and X-ray reflectivity [20]. Braslau et al. [20]
found an interface thickness for the water liquid–vapor interface of about 3 Å at room
temperature.
Molecular dynamics simulations appear to confirm the adequacy of the continuum
approach (Fig. 6.9) for estimating the liquid–vapor interface thickness. Nevertheless,
both MD and Peng–Robinson models underestimate the interface thickness, when
compared with the experimental ellipsometry results of Kinosita et al. [85]. Never-
theless, experiments are rather difficult in this kind of measurement and this underes-
timation cannot be considered as conclusive. In fact, this subject is still challenging
and full of open questions.

18.00

16.00
Exp. ellipsometry
14.00 MD Matsumoto
Interface thickness, A

12.00 MD Taylor et al.

10.00 Peng-Robinson

8.00

6.00

4.00

2.00
0.00
250 300 350 400
T (K)
Fig. 6.9 Comparison of the continuum (PR) and molecular dynamics (MD) approach with ellip-
sometry experimental results for estimating the interface thickness of water. MD simulation results
are from Matsumoto and Kataoka [112] and Taylor et al. [169]. Experimental results are from
Kinosita and Yokota [85]
226 6 Surface Physics

6.5.6 Exercise

Table 6.1 for water was built based on the Peng–Robinson equation of state and
experimental values of the surface tension .γ . Consider now carbon dioxide. Surface
tension data for .CO2 can be found in, e.g., the Dortmund Data Bank (DDB). Use the
Soave equation and find the interface thickness in terms of the reduced temperature
for the carbon dioxide.

6.6 Interaction Between Fluids and Solid Surfaces

It is frequently stated that Thomas Young established the relationship between the
contact angle and the surface energies in his ‘An essay on the cohesion of fluids’
published by the Royal Society in 1805 [189]. Undoubtedly, this paper clearly states
the significance of the contact angle, but as commented by Vogle [183] “Close reading
fails to detect clear statements that can be formulated as the familiar. So it is tempting
to conclude that the Young equation, as we currently enjoy it, is more of a digestion
or a composite of Laplace and Young and possibly Dupré”.
In the present section, the henceforth called Young–Laplace–Dupré equation is
derived by using the approach given by Gibbs [59, 60], by considering the interface
as a region where the free energy has an excess with respect to the bulk phases. So,
consider a system as shown in Fig. 6.10, consisting of three phases . , .v and .s and
their interfaces . v, . s and .vs. The interface is supposed to be spherical when at rest
and in the absence of gravity effects.
The Helmholtz free energy of this system is given by

. F = Fs + F + Fv + Fs + Fsv + F v
bulk phases interfaces

Alv+dAlv

V
dAlscos
L

dAls
S O

Fig. 6.10 Contact angle between a fluid and a solid surface

6.6 Interaction Between Fluids and Solid Surfaces 227

So by supposing, as depicted in Fig. 6.10 that the liquid drop suffers a small
expansion at a constant temperature, the system will undergo a variation in its free
energy, given by

. F = d Fs + d F + d Fv + d Fs + d Fsv + d F v .

As it is well known from equilibrium thermodynamics,

.d F = − SdT − Pd V + μdm,

for the bulk phases and for an interface .αβ

d Fαβ = γαβ d A.
.

So, by considering that.dm = 0, meaning that the involved phases do not exchange
mass or that this mass exchange is negligible during the droplet expansion, and taking
into account that

d Vs = 0, d Vv = − d V ,
.

and that each increase in the liquid–solid contact area corresponds to an equal
decrease in the vapor–solid contact area

d Avs = − d A s ,
.

we can conclude that

d F = (Pv − P ) d V + γ v d A v + (γ s − γvs ) d A
. s
+ γ v d A s cos θ.

In equilibrium conditions .d F = 0. In addition, from the Young–Laplace–Dupré

equation the first two terms of the second member cancel themselves, since in
mechanical equilibrium
dA v γv
. (P − Pv ) = γ v = , (6.19)
dV R

where .R is the curvature radius. Therefore,

. (γ s − γvs ) d A s + γ v d A s cos θ = 0.

This last equation leads to the Young–Laplace–Dupré equation that expresses the
equilibrium contact angle in terms of the surface energies .γ s , .γvs , and . y v
228 6 Surface Physics

γvs − γ s
. cos θ = .
γv

The adhesion energy . E adh will be here defined as9

. E adh = γvs − γ s

and the ratio between . E adh and the cohesion energy, . E coh = y v , is related to
the wettability of the solid surface with respect to the fluid in contact with it.
When . E adh > E coh the fluid is considered to completely spread on the solid sur-
face without attaining any equilibrium state at a given equilibrium contact angle.
When .1 ≥ E adh /E coh ≥ 0 the fluid attains an equilibrium state with a contact angle
◦
.0 ≤ θ ≤ 90 . When the adhesion energy is a negative fraction of the cohesion

energy, the cohesion prevails and the liquid attains an equilibrium contact angle
◦ ◦
.90 ≤ θ ≤ 180 .
The theoretical limit .θ = 90◦ corresponds to an adhesion energy . E adh = 0, when
both .γvs and .γ s solid–fluid surface energies are zero. In other words, the fluid
molecules do not interact with the solid surface in this limit.

6.6.1 Exercise

Figure 6.11 shows a balance of forces for an element .e of the surface of a liquid
droplet supposed to be spherical and deposed on a solid surface. The surface forces
act only on the perimeter area of this element .2π R sin α. In fact, when we try to
detach .e from the liquid droplet, the remaining liquid molecules, on their side, will
react trying to prevent this detachment with a force .γ v 2π R sin α in the perimeter
area, directed along the liquid surface and outward from .e. These forces are null
in the lower and upper surfaces of .e, because the potential energy related to the
intermolecular interaction has a null gradient when we move into the liquid phase
or into the vapor phase. Using the same reasoning as above, make a force balance
in the . s and .vs interfaces and try to find the Young–Laplace–Dupré law as given in
Eq. (6.19).

9 There is some controversy in the definitions of the adhesion energy [183]. The adhesion energy
is here defined as the adhesion tension or .γvs − γ s because it represents the net effect of the solid
surface on the fluid molecules and because it appears to be more universal than the usual concept
of adhesion work. When two liquids 1 and 2 interact with a solid surface, liquid 1 will be adherent
to the surface when .γ1s − γ2s is positive. Otherwise, liquid 2 will be adherent to the surface.
6.6 Interaction Between Fluids and Solid Surfaces 229

Rsin
Potential energy

Surface element e
PV

lv
lv PL
L V
V
(b) L R

(a)
Fig. 6.11 a Force balance on a surface element .e of a liquid droplet. b Potential energy related to
the intermolecular interaction

6.6.2 Dynamic Contact Angle

Interest in gaining a deeper understanding of the physical processes that control the
fluid–fluid interface displacement at the pore level is currently increasing due to
(a) the growing possibility of direct simulation of immiscible displacement in com-
plex geometry offered by the significant development of computers and numerical
methods that has occurred over the recent years and (b) the increasing accuracy of
three-dimensional representations of the porous structure promoted by the develop-
ment of three-dimensional reconstruction methods [4, 55, 77, 106, 118] and by the
increasing spatial resolution that it is possible to achieve with microtomography-
based methods [6, 56, 65].
Classically, the immiscible displacement of fluids is approached by considering
that the transition layer has a null thickness and by performing a momentum balance
around this layer. In this way, in addition to the intrinsic physical properties of the
involved fluids, the interface tension is another important macroscopic parameter.
The main physical mechanisms that control the interface displacement are the
fluid–fluid and the fluid–solid intermolecular potentials, so, the correct knowledge
of this process requires downscaling to the molecular scale. Intermolecular potentials
may be of many different types, including polar forces among permanent dipoles,
induction forces, and dispersion forces among symmetrical molecules. In the transi-
tion region between two immiscible fluids, a molecule is predominantly subjected to
attractive fields from its own phase that act as a potential barrier and give rise to the
fluid–fluid interface tension. In addition, molecules that are found in this transition
230 6 Surface Physics

layer are subject to collisions that tend to mix the two fluids and are responsible
for diffusion. Consequently, the thickness of the transition layer is controlled by the
strength and interaction length of intermolecular cohesion forces and by the binary
diffusivity, .D12 .
The theoretical difficulty is strongly increased when these two fluids interact
with a solid surface around a three-phase region, called triple or contact line. Under
equilibrium conditions, the interfacial energies related to the interaction between
two fluids and the solid surface are the main macroscopic parameters governing the
interaction between the fluids and the solid surface. These interfacial energies lead to
an equilibrium contact angle. Nevertheless, when the interface advances or recedes
along a solid surface, dynamic effects will change the contact angle with respect to
its equilibrium value. The changes in the contact angle will depend on the interaction
potentials, on the surface roughness, and on the chemical heterogeneities, in addition
to the imposed flow conditions summarized by the capillary number,
μv
Ca =
.
γ

where .μ is the viscosity of the displaced fluid, .v is the interface velocity and .γ is the
interface tension.
Figure 6.12a and b give a sketch of the variation of the dynamic contact angle
in accordance with the flow direction. When a wetting fluid displaces a non-wetting
fluid, inside a channel, the contact angle increases, but when the displaced fluid is
the wetting one, the contact angle is reduced.
It is usual, in studies of capillary-rise dynamics, to consider that the contact angle
varies according to the expression:

. cos θd = cos θ − aCa b (6.20)

where .θd is the dynamic contact angle and a and b are constants. In Fig. 6.12c it
is shown the dependence of .cos θ on the capillary number obtained from lattice-
Boltzmann simulations. The graph shows that the expression given by Eq. (6.20)
represents adequately the contact angle dependence on .Ca. In fact, Eq. (6.20) fits
very well with simulation results when .a = 18.21 and .b = 0.88.
Finally, Fig. 6.12d shows the streamlines simulated with the lattice-Boltzmann
method, for a capillary rising process. The flow near the interface can be very com-
plex, with the formation of vortices near the triple line as can be seen in the figure.
Reference [43] is suggested for further details (Fig. 6.12).
Therefore, due to the dependence of the dynamic contact angle on both the flow
conditions that are imposed and on the geometry and properties of the solid surface,
the equilibrium, or even the dynamic contact angles cannot be used as a parameter
in dynamic simulations.
Fluid invasion into a tube as a result of unbalanced capillary forces is a classical
problem in the literature devoted to fluid dynamics. It is of great scientific interest
and is important in various applications, such as ink printing, dyeing of textiles,
6.6 Interaction Between Fluids and Solid Surfaces 231

0.95

Flow
direction 0.90

0.85

d
cos
eq
d 0.80
d

eq
0.75 Fit curve (a=18.21 b=0.88)
Lattice-Boltzmann simulation
Flow
0.70
direction
0.000 0.001 0.002 0.003 0.004 0.005 0.006
Ca
(a) (b) (c)

(d)

Fig. 6.12 Sketch of the variation of the dynamic contact angle in accordance with the flow direction
(a, b); dependence of the contact angle with the capillary number (c, d) streamlines showing the
formation of vortices near the triple line [43]

and enhanced oil recovery. Lucas10 and Washburn11 published the first attempts
to describe the dynamic process involved in capillary rise. They verified that the
process can be described using a mechanical balance between the capillary, viscous,
and gravitational forces, leading to the well-known Lucas–Washburn equation. Their
model has been verified experimentally even for very thin capillaries and is gener-
ally accepted as a valid approach under a laminar flow regime [2]. However, the
Lucas–Washburn equation fails in the prediction of the first stages of the meniscus
rising, predicting that the initial velocity is infinite at time zero, a non-physical solu-

10 R. Lucas, Kolloid. Zh. 23 (1918) 15–22.

11 E. W. Washburn, Phys. Rev. 17 (1921) 273–283.
232 6 Surface Physics

tion, resulting from the lack of an inertial term [193]. This problem was solved by
Bosanquet [18] who included an inertial term in the Lucas–Washburn equation but
neglected any influence from the fluid flow in the reservoir. For such a case, it was
verified that the inclusion of the inertial effects avoided the non-physical initial con-
dition, allowing the definition of an initial velocity at time zero, called the Bosanquet
velocity. High-resolution experiments showed that the observed initial velocities are
different from those predicted by the Bosanquet equation [90]. Szekely et al. [168]
carried out an analysis of the capillary rise with inertial and entrance effects by using
arguments based on the balance of the mechanical energy equation. They modified
the inertial term by adding an apparent mass related to the fluid flow from the reser-
voir and a vena contracta effect due to the inlet flow. Considering that the analysis
by Szekely et al. was for high Reynolds number flow, while most cases of capillary
rising are low Reynolds number problems, Levine et al. [104] performed a more
detailed analysis of the momentum and mechanical energy balance equations, devel-
oping one of the most complete models found in the literature. The Levine et al.
model presents different factors for the inertial terms and an additional term, related
to the inlet viscous dissipation. Dreyer et al. [44] and Stange et al. [161] carried out
similar analysis, including some improvements, and obtained governing equations
that are similar to those found by Levine et al. These equations were confirmed by
experiments under microgravity conditions. As mentioned above, the capillary rise
problem has been the subject of several discussions related to the short-time aspects
of meniscus formation and the resulting flow velocity. Its complete understanding
continues to be, nevertheless, an open subject, particularly due to the complex physic-
ochemical interactions between the liquid and the solid surface and the fundamental
role of the triple line in the liquid displacement.
Figure 6.13, extracted from Wolf et al. [187] shows how complex can be the
first steps of the meniscus formation in capillary rise, and Fig. 6.14 shows the time
evolution of the dynamic contact angle during capillary rise between two parallel
plates Simulations were performed with a lattice-Boltzmann model.

Fig. 6.13 First steps of the

meniscus formation in
capillary rise between
parallel plates simulated with
a lattice-Boltzmann model.
∗
.h =
∗ ∗
2R , . x = 2R , .t = t0
h x t

are dimensionless variables

[187]
6.7 Surface Energies: Cahn’s Theory 233

Fig. 6.14 Time evolution of the dynamic contact angle during the capillary rise between two parallel
plates when the equilibrium contact angle is .44◦ [187]

In the next section, we present Cahn’s theory [24] for predicting the surface
energies .γ f s related to the intermolecular interaction between a fluid . f —in its liquid
or vapor physical state—and a solid surface.s. Cahn’s theory was conceived to retrieve
the surface energies .γ s and .γvs in terms of the equilibrium contact angle .θeq for a
system liquid–vapor–solid in equilibrium.
In dynamical problems, these surface energies are to be used instead of the mea-
sured values of the equilibrium contact angle.

6.7 Surface Energies: Cahn’s Theory

When gas molecules are close to solid surfaces, they are subjected to the attraction
fields from the solid molecules and form an adsorbed layer.
De Boer and Zwicker12 explained the adsorption of non-polar molecules on adsor-
bents by assuming that the uppermost layer of the adsorbent induces dipoles in the
first layer of adsorbed molecules, which in turn induce dipoles in the next layer and
so on until several layers are built up (Fig. 6.15).
Physical adsorption is a subject of fundamental interest and received outstanding
contributions since Langmuir in 1918.13 References [21, 38] are suggested for readers
interested in historical papers on this subject. Nevertheless, since our scale of interest
is much larger than the molecular dimensions corresponding to the thickness of the

12De Boer and Zwicker, Z. physik. Chem., B3, 407 (1929) [38].
13Langmuir I (1918) The adsorption of gases on plane surfaces of glass, mica, and platinum. J Am
Chem Soc 40:1361–1403. https://doi.org/10.1021/ja02242a004 [98].
234 6 Surface Physics

Fig. 6.15 The electrostatic field produced by a solid surface induces polarization on non-polar
molecules. These polarized molecules induce dipoles in the next layers until several layers are built
up

adsorbed layer, it can be supposed that the semi-infinite fluid interacts with this layer
and not with the solid surface itself.
When a drop of liquid is deposed on a solid surface, the only available experi-
mental information is the contact angle .θ after the liquid drop reaches an equilibrium
configuration. From the Young–Laplace–Dupré equation,
γvs − γ s
. cos θ = , (6.21)
γv

we know that this contact angle is dependent on the difference between the Helmholtz
energy in excess in the vapor–solid, .γvs , and in the liquid–solid, .γ s , transition layers.
Since both the liquid and vapor phases are, in fact, interacting with an adsorbed layer
of fluid, we can suppose that this interaction is ruled by a single potential which is
only dependent on the amount of adsorbed molecules.
Cahn [24] assumed that the interactions between the surface and the fluid are
sufficiently short-range in such a manner that the contribution to the free energy of a
unit area of this layer is . (ρs ), where .ρs is the limiting density of the fluid at .x = 0.
So the contribution of the adsorbed layer to the free energy is of an additive constant.
Therefore, following the same reasoning of Sect. 6.5.4, the question we seek to
solve is to get the density profile and excess free energy due to the adsorbed layer by
finding the function which minimizes the excess free energy of a unit area of surface,
∞( ( )2 )
κ dρ
. exc = (ρs ) + ψexc (ρ, T ) + d x, (6.22)
2 dx
0

where

ψexc (ρ, T ) = ψ (ρ, T ) − ψ0 (ρ∞ , T ) ,

.

is here interpreted as the free energy required to create a unit volume of uniform fluid
( )2
of composition from a large reservoir with density .ρ∞ , . κ2 ddρx is the van der Waals
6.7 Surface Energies: Cahn’s Theory 235

contribution due to the presence of a gradient and .ρ∞ is the bulk density of the fluid,
far from the solid surface.
As it was seen in Sect. 6.5.4, in equilibrium conditions the solution to the mini-
mization problem is given by,
( )2
κ dρ
ψexc (ρ, T ) =
. . (6.23)
2 dx

Cahn supposed that the free energy at the surface is responsible for the density
gradient at .x = 0,
)
dρ
. (ρs ) = ρs κ .
d x x=0

Therefore, a surface property related to the wettability can be defined as

)
dρ
.φ = − κ ,
d x x=0

considered as a macroscopic property related to the net balance of the intermolecular

forces between the fluid molecules and the solid and between the fluid molecules
close to the solid and the fluid phase.
When the surface is hydrophilic,.φ > 0 and the fluid density decreases as.x → ∞.
When .φ < 0 the surface is hydrophobic and the fluid density increases as .x → ∞
(Fig. 6.16).

(x) (x)

x x

bulk fluid bulk fluid

(a) (b)

Fig. 6.16 Density distribution at the fluid–solid interface: a hydrophilic surface,.φ > 0, b hydropho-
bic surface, .φ < 0
236 6 Surface Physics

From Eq. (6.23)

/
dρ 2ψexc (ρ, T )
. =± . (6.24)
dx κ

So
√
φ = ± 2κψexc (ρ, T ).
. (6.25)

If .φ is assumed
) to be positive by taking the plus sign in the above equation, this
means that . d x dρ
< 0, or that .ρs > ρ > ρ∞ . If it is assumed to be negative by
x=0 )
taking the minus sign in the above equation, this means that . ddρx > 0, or that
x=0
.ρs < ρ < ρ∞ .
The minimum value of the excess free energy .ψexc is the interface surface tension
.γ f s . So, by replacing Eq. (6.23) into Eq. (6.22) we get

∞ ( )2
dρ
γfs =
. (ρs ) + κ dx
dx
0
ρ∞
dρ
= (ρs ) + κ dρ. (6.26)
dx
ρs

The last term on the r.h.s. of the above equation is always positive. So if.φ is positive
(hydrophilic surfaces) and the plus sign is taken in Eq. (6.25), then Eq. (6.26) will
be written as
ρs
√
γfs =
. (ρs ) + 2κψexc (ρ, T )dρ. (6.27)
ρ∞

On the other hand, if .φ is negative (hydrophobic surfaces) and the minus sign is
taken in Eq. (6.25), then Eq. (6.26) will be written as
ρ∞
√
. γfs = (ρs ) + 2κψexc (n, T )dρ. (6.28)
ρs

In dimensionless variables Eq. (6.25) can be written as

φ √
.φ∗ = √ ∗ (ρ, T ).
= ± 2ψexc (6.29)
κ Pc
6.7 Surface Energies: Cahn’s Theory 237

Considering the van der Waals equation of state for the fluid, the Helmholtz’s
excess energy is given by

( ) ( )
∗ ρ∗ 8 ρ ∗ 3 − ρ∞
∗
.ψexc = 3ρ ∗ ρ∞
∗
1 − ∗ + ρ ∗ T ∗ ln ∗ (6.30)
ρ∞ 3 ρ∞ (3 − ρ ∗ )
( )( )
8ρ ∗ T ∗ ρ∗
+ ( ∞ ∗ ) − 3n ∗∞
2
. 1− ∗ .
3 − ρ∞ ρ∞

6.7.1 Hydrophilic Surfaces

Figure 6.17 shows how to use Eq. (6.29) for finding the fluid density .n ∗s at the solid
surface for a hydrophilic surface. Equation (6.29) has 2 solutions for the vapor–solid
interaction and 2 other solutions for the liquid–solid interaction. Since in hydrophilic
surfaces, .ρs∗ > ρ∞
∗ ∗
, where .ρ∞ is the density of the bulk phase (vapor or liquid), the
∗ ∗
two solutions where .ρs < ρ∞ are to be discarded.
The solution of the non-linear equation
√
.φ∗ = ∗ (ρ, T ),
2ψexc (6.31)

can be performed by using a numerical method such as the Newton–Raphson method

and leads to the two solutions for the vapor and liquid densities at the solid surface.
So, when .φ ∗ = 0.4 as in Fig. 6.17, the reduced densities of the fluid at the solid
surface will be .ρs∗ = 0.49337 for the vapor phase and .ρs∗ = 2.14093 for the liquid
phase.
Equation (6.27) can now be used for the calculation of the interface tensions .γvs
and .γ s ,
√
. (ρs ) = − ρs φ = − ρs∗ ρc φ ∗ κ Pc .

The water critical properties are . Pc = 221.2 × 105 Pa, .ρc = 322 .kg/m3 . For the
van der Waals equation of state (see Table 6.2) at .T ∗ = 0.8. .ρvs∗
= 0.23962, ρ ∗s =
−15
1.93279, .κ = 1.37541 × 10 N
. So, for the vapor phase
(kg/m3 )2

N
. (ρs ) = − 6.5038 × 10−3 ,
m
and
238 6 Surface Physics

*s=0.49337 *s=2.14093
*vs=0.23962
*
*ls=1.93279

Fig. 6.17 Dimensionless excess function for water at T* = 0.8 calculated using the van der Waals
EOS plotted against the reduced density. The information on the solid–fluid interaction is given by
the dimensionless specific free energy.φ ∗ considered in this case as.φ ∗ = 0.4 (hydrophilic surface). In
accordance with Cahn’s theory, the water reduced density at the solid surface will be .ρs∗ = 0.49337
and.ρs∗ = 2.14093 corresponding, respectively, to the vapor and liquid in equilibrium with the solid

0.49337
−3
√ √
γ
. vs = − 6.5038 × 10 + ρc 2κ Pc ∗ dρ ∗
ψexc
0.23962
N
= − 0.0046598 .
m
For the liquid phase,
N
. (ρs ) = − 0.028223 ,
m
and
2.14093
√ √
γ s = − 0.028223 + ρc 2κ Pc
. ∗ dρ ∗
ψexc
1.93279
N
= − 0.026859 .
m
6.7 Surface Energies: Cahn’s Theory 239

Table 6.2 Surface properties of water with the van der Waals equation of state
( )
.T
∗ ∗
.ρv∞ .ρ ∞
∗ .κ .
N
(kg/m3 )2
0.4220 0.00725 2.56345 .4.68314 × 10−16
0.4622 0.01336 2.51190 .5.18317 × 10
−16

0.5023 0.02238 2.45787 .5.71457 × 10

−16

0.6027 0.06127 2.30967 .7.25571 × 10

−16

0.7016 0.12958 2.13966 .9.35490 × 10

−16

0.8005 0.23962 1.93279 .1.37541 × 10

−15

0.9025 0.43234 1.65075 .2.13247 × 10

−15

0.9426 0.55217 1.49691 .2.88014 × 10

−15

The water surface tension at this temperature is

N
γ
. v = 0.0264 .
m
Finally, the contact angle can be calculated using the Young–Laplace–Dupré equa-
tion
( )
γvs − γ s
. cos θ = ,
γv

or .θ = 32.77◦ .
In the above calculations, we notice that the surface free energies .γvs and .γ s are
both negative when the surface is hydrophilic and that .γ s is more negative than .γvs .
Indeed, the dominant term in Eq. (6.27) is the surface free energy . (n s ) which is
negative when the surface is hydrophilic, indicating that work is needed for freeing
the adsorbed fluid layer from the solid surface. The reason for .γ s be more negative
than .γvs is that, while the specific free energy at the surface, .φ, was written to be only
dependent on the fluid–solid molecular interaction, the amount of adsorbed water
in the liquid–solid interface is, in the present case, around 6 times the amount of
adsorbed water in the vapor–solid interface.
Figure 6.17 also shows that when .φ ∗ = 0, both .γ s and .γvs are null, meaning that
there is no interaction between the fluid and the solid surface, or that .θ = 90◦ . On
the other hand the maximum value of .φ ∗ for which .θ = 0◦ is the one satisfying
γvs − γ s
. = 1.
γv
240 6 Surface Physics

Fig. 6.18 Hydrophobic surface

6.7.2 Hydrophobic Surfaces

Figure 6.17 shows how to use Eq. (6.29) for finding the fluid density .n ∗s at the solid
surface for a hydrophilic surface. Equation (6.29) has 2 solutions for the vapor–solid
interaction and 2 other solutions for the liquid–solid interaction. Since in hydrophobic
surfaces, .n ∗s < n ∗∞ , where .n ∗∞ is the density of the bulk phase (vapor or liquid), the
two solutions where .n ∗s > n ∗∞ are to be discarded.
So, when .φ ∗ = − 0.4 as in Fig. 6.18, the reduced densities of the fluid at the solid
surface will be .n ∗s = 0.09120 for the vapor phase and .n ∗s = 1.62903 for the liquid
phase.

6.7.3 Contact Angle and Adhesion Energy

∗
Table 6.3 shows the reduced densities at the solid surface for the vapor .ρs−v and
∗ ∗
liquid .ρs− phases of water for a reduced temperature .T = 0.8. The van der Waals
equation of state, Eq. (6.30) was used in the solution of the non-linear equation,
Eq. (6.31), by a Newton–Raphson method. The interface tensions .γsv and .γs are
also shown in this table. They were calculated for each surface parameter .φ ∗ using
Eq. (6.27) for hydrophilic surfaces and Eq. (6.28) for hydrophobic surfaces. Finally,
6.7 Surface Energies: Cahn’s Theory 241

Table 6.3 Reduced densities at the solid surface, interface tensions and equilibrium contact angle
for several solid surface parameters .φ ∗ when .T ∗ = 0.8 and the van der Waals equation of state
.φ
∗ ∗
.ρs−v
∗
.ρs− .γsv × 10
3 .γs × 10
2 .θ

Reduced values N/m N/m .

◦

0.478 0.56798 2.17530 .− 6.0199 .− 3.2404 1.97

0.475 0.56483 2.17401 .− 5.9640 .− 3.2189 6.59
0.470 0.55965 2.17185 .− 5.8714 .− 3.1832 10.47
0.460 0.54947 2.16751 .− 5.6888 .− 3.1117 15.60
0.440 0.52988 2.15876 .− 5.3334 .− 2.9692 22.68
0.400 0.49337 2.14093 .− 4.6598 .− 2.6859 32.77
0.300 0.41412 2.09429 .− 3.1695 .− 1.9883 50.72
0.20 0.35100 2.04500 .− 2.7092 .− 1.3889 64.95
0 0.23976 1.93255 0 0 90.00
.− 0.20 0.15600 1.79600 1.2934 1.2364 114.79
.− 0.30 0.12171 1.72002 1.7501 1.8174 128.47
.− 0.40 0.09120 1.62903 2.0996 2.3703 144.92
.− 0.44 0.08009 1.58795 2.2123 2.5826 153.44
.− 0.47 0.07215 1.55480 2.2875 2.7382 161.90
.− 0.475 0.07086 1.54905 2.2993 2.7638 163.70
.− 0.48 0.06958 1.54323 2.3108 2.7893 165.70
.− 0.49 0.06705 1.53138 2.3333 2.8400 170.89
.− 0.495 0.06579 1.52533 2.3443 2.8652 175.21
∗
Reduced densities .ρs−v ∗
and .ρs− were found from the non-linear equation, Eq. (6.31), using a
Newton–Raphson method

the Young–Laplace–Dupré equation was used for calculating the equilibrium contact
angle .θ .
The adhesion energy is defined as

. E adh = γvs − γ s ,

and is a measure of the interaction energy between the liquid and solid surface.
Figure 6.19 shows the dependence of the adhesion energy on the solid surface
specific energy .φ ∗ . A linear fit appears to be an excellent approximation for the adhe-
sion energy for almost the entire range of .φ ∗ . The figure also shows the dependence
of the equilibrium contact angle on the surface parameter .φ ∗ for hydrophilic and
hydrophobic interactions.
Figure 6.20 shows the reduced fluid densities at the solid surface in terms of the
equilibrium contact angle for vapor–solid and liquid–solid interactions. The knowl-
edge of the liquid and vapor densities at the solid surface in terms of the equilibrium
contact angle is of crucial importance because they are the only parameters that
are to be used in dynamic simulations of the liquid–vapor interface displacement in
microcavities. This is the case of capillary rising. As we have seen in Sect. 6.6.2 it is
not possible to consider the contact angle as a constant when a liquid column rises
242 6 Surface Physics

Fig. 6.19 Adhesion energy and contact angle in terms of parameter .φ ∗

Fig. 6.20 Vapor .ρs−v∗ and liquid .ρs−∗ densities at the solid surface in terms of the equilibrium
contact angle .θ for water at .T ∗ = 0.8, when a van der Waals equation of state is used

inside a capillary as it was assumed in Bosanquet model [18]. In accordance with

Fig. 6.14, the contact angle has a strong variation at the beginning of the rising and
only approaches its equilibrium value at the end of ascension. During this process,
both the meniscus shape and the contact angle are dependent on the capillary, gravity,
viscous, and inertial forces.
6.8 Emulsions 243

6.7.4 Exercise

The equilibrium contact angle between water and glass was measured as .13◦ at
. T = 300 K. Suppose that water follows a van der Waals equation of state and find
parameter .φ ∗ and the surface energies .γ s and .γvs . In the following, calculate param-
eter .φ ∗ corresponding to the contact angles .θ = 0◦ and .θ = 180◦ .

6.8 Emulsions

Emulsions in one form or another, have been used for centuries and early societies
learned to take advantage of this natural phenomenon. The ancient Egyptians used
eggs to emulsify berry extracts with oils to form crude emulsified paints. In medicine,
emulsions are delivered in the form of medical ointments. Cream and butter are liter-
ally the same thing. To make butter, you simply mix cream until the emulsion reverses;
that is, it transforms from an oil-in-water emulsion into a water-in-oil emulsion. But
despite this being the only difference between cream and butter, the effect on taste
and texture are significant. Many natural and processed foods consist either partly
or wholly as emulsions or have been in an emulsified state at some time during
their production; such foods include milk, cream, butter, margarine, fruit beverages,
soups, cake batters, mayonnaise, cream liqueurs, sauces, desserts, salad cream, and
ice cream.
Oil and water are immiscible liquids and do not mix. The reason behind this
behavior must be searched in the molecular scale. The intermolecular attraction forces
between water molecules are much stronger than the ones between oil molecules.
Water is a polar molecule and the distribution of its unpaired electrons enforces this
force in the form of hydrogen bonds. Oil has apolar molecules that attract themselves
with weak London forces. So, each water molecule will push other water molecules in
its neighborhood, segregating the oil molecules that are found in this neighborhood.
Suppose now that you add, e.g., oil to water inside a container. Since oil is less

Fig. 6.21 The molecule of

cetyl alcohol
CH.3 (CH.2 ).15 OH
244 6 Surface Physics

water

oil oil

(a)

oil

water water

(b)

Fig. 6.22 Due to its amphiphilic nature, when surfactants are added to an oil–water system their
molecules migrate to the interface. a When oil is fragmented and dispersed in a continuous water
phase the polar heads of the emulsifier at the interface attract the water molecules in their neighbor-
hoods preventing the oil drops from coalescing. b When water is dispersed in oil, the long tails of
the emulsifier prevent their polar heads to attract the water molecules in the contiguous water drop,
because they are beyond the interaction length of these intermolecular polar forces

dense than water, their segregated molecules will move up in the container forming
a separate oil layer above the water level.
An emulsion of oil droplets dispersed inside water can be formed by vigorously
shaking the liquids in the container. Nevertheless, this emulsion is drastically unsta-
ble, and, very soon, water and oil will segregate again forming two separated layers.
Emulsions can be stabilized by adding emulsifiers to it. Emulsifiers are amphiphilic
substances, whose molecules have a polar hydrophilic head and a long hydropho-
bic tail. An example of such a substance is cetyl-alcohol, whose molecule is shown
in Fig. 6.21. Due to its amphiphilic nature, when these substances are added to an
oil–water system their molecules migrate to the interface. When oil is fragmented
6.8 Emulsions 245

Oil

Additional attraction force

Attraction force between water and surfactant
between a water heads
molecule in the interface
and the oil phase
Attraction force
between a water molecule
in the interface and the water
Water phase

Fig. 6.23 When a surfactant is added to an oil–water system, the polar heads in the interface will
create an additional force from the surface that will reduce the force imbalance responsible for the
interface tension

and dispersed in a continuous water phase as in Fig. 6.22a, the polar heads of the
emulsifier at the interface will attract the water molecules in their neighborhoods
with a force that is much stronger than the one it attracts the oil molecules. So a
water layer will be stabilized by these segregating forces between any two oil drops,
preventing the oil drops from coalescing.
When water is dispersed in oil as in Fig. 6.22b the long tails of the emulsifier
will prevent their polar heads to attract the water molecules in the contiguous water
drop, because they will be beyond the interaction length of these intermolecular polar
forces.
Emulsifiers are found in milk, mayonnaise, butter, ice cream, chocolate, and sev-
eral other food products. Proteins and phospholipids prevent the milk fat dispersed
in water from coalescing. Other emulsifiers are the lecithin present in the egg yolk
and responsible for stabilizing the mayonnaise. Cetyl alcohol is used as an emulsifier
in the cosmetic industry in the production of skin creams and lotions.
Surfactants are responsible for the reduction in the interface tension between two
immiscible fluids. In fact, the interface tension may be thought as the result of the
imbalance between the forces that attract a fluid molecule in the interface to its bulk
phase and the forces from the surface that attract this molecule to its contiguous
phase. So, when a surfactant is added to an oil–water system as in Fig. 6.23, the polar
heads in the interface will create an additional force from the surface that will reduce
this force imbalance.
246 6 Surface Physics

Fig. 6.24 Effect of the

addition of cetyl alcohol on 50
the interface tension of a

Interface Tension mN/m

water–nujol system at 30 .◦ C.
Experimental results from 40
Wolf [186]
30

20

10

0 1 2 3 4 5 6 7 8
5
Cetyl alcohol concentration mol/g x 10

Figure 6.24 shows the effect of the addition of small amounts of cetyl alcohol
on the interface tension of a water–nujol system at 30 .◦ C [186]. Nujol is a mixture
of alkanes .Cn H2n+2 with a very large value of .n and is often used for simulating
water–oil displacements in petroleum rocks. The water–nujol interface tension at
30 .◦ C was measured as 51.6 .mN/m.
 Part III
Introduction to Non-equilibrium
Thermodynamics
Chapter 7
Non-equilibrium States

Abstract Non-equilibrium thermodynamics provides us with a general framework

for the macroscopic description of thermodynamic systems in non-equilibrium states,
where the state variables are treated as field variables, i.e., as continuous functions
of space coordinates and time. In this field, the balance equation for entropy plays a
central role, expressing the fact that the entropy of a volume element changes with
time for two reasons. First, it changes because entropy flows into the volume element,
and second, because there is an entropy source due to irreversible phenomena inside
the volume element. In this chapter, the differential equations for density, momentum,
and energy are presented when these thermodynamic variables are considered field
variables. An application of the theory is given considering the heating and dissipation
processes when a cavity filled with an ideal gas is heated. The chapter ends with the
entropy balance equation in its differential and integral forms.

7.1 Introduction

In accordance with de Groot and Mazur [39], thermodynamic considerations were

first applied to the treatment of irreversible processes by W. Thomson in 1854,1 in
the study of the thermoelectric phenomena. Duhem,2 Natanson,3 Jaumann4 and later
Eckart5 attempted to obtain expressions for the rate of change of the local entropy

1 Thomson, W. (1882) On the Dynamical Theory of Heat, with numerical results deduced from

Mr Joule’s equivalent of a Thermal Unit, and M. Regnault’s Observations on Steam, Mathematical

and Physical Papers, 174–332 [174].
2 Duhem, P. (1911) Traité d’ énergétique ou de thermodynamique générale (2 vols.). Gauthier-

Villars [47].
3 Natanson, L. (1896). Uber die Gezetze nicht umkehrbarer Vorgange. Z. Phys. Chem., 193 [121].
4 Jaumann, G. (1905). Die Grundlagen der Bewegungslehre. Barth Verlag [75]. Jaumann, G. (1911).

Geschlossenes System physikalischer und chemischer Diffrerentialgesetze. K. k. Hof- und Staats-

druckerei [76].
5 Eckart, C. (1940a). The Thermodynamics of Irreversible Processes. I. The Simple Fluid. Physical

Review, 58(3), 267–269 [49]. Eckart, C. (1940b). The Thermodynamics of Irreversible Processes.
II. Fluid Mixtures. Physical Review, 58(3), 269–275 [50].
© Associação Paranaense de Cultura 2024 249
P. C. Philippi, Thermodynamics, https://doi.org/10.1007/978-3-031-49357-7_7
250 7 Non-equilibrium States

in nonuniform systems by combining the second law of thermodynamics with the

macroscopic laws of conservation of mass, momentum, and energy.
The field of non-equilibrium thermodynamics provides us with a general frame-
work for the macroscopic description of irreversible processes. As such it is a branch
of macroscopic physics, which has connections with other macroscopic approaches
such as fluid dynamics and heat transfer, insofar as the latter disciplines are also
concerned with non-equilibrium situations.
Thus non-equilibrium thermodynamics should be set up from the start as a con-
tinuum theory, treating the state variables of the theory as field variables, i.e., as
continuous functions of space coordinates and time. Moreover one would like to
formulate the basic equations of the theory in such a way that they only contain
quantities referring to a single point in space at one time, i.e. in the form of local
equations. In equilibrium thermodynamics such a local formulation is generally not
needed, since the state variables are usually independent of the space coordinates.
In non-equilibrium thermodynamics, the so-called balance equation for entropy
plays a central role. This equation expresses the fact that the entropy of a volume
element changes with time for two reasons. First, it changes because entropy flows
into the volume element, and second because there is an entropy source due to
irreversible phenomena inside the volume element.
As we have seen in Sect. 2.2 the entropy source is always a non-negative quantity.
It only vanishes when the process is reversible. This is the local formulation of the
second law of thermodynamics. The main aim is to relate the entropy source explicitly
to the various irreversible processes that occur in a system.
To this end, one needs the macroscopic conservation laws of mass, momentum,
and energy, in local, i.e. differential form. These conservation laws contain a number
of quantities such as the diffusion flows, the heat flow and the pressure tensor, which
are related to the transport of mass, energy, and momentum. It turns out that the
entropy source has a very simple appearance: it is a sum of terms each one being
a product of a flux characterizing an irreversible process, and a quantity, called
thermodynamic force, which is related to the non-uniformity of the system. e.g., the
gradient of the temperature or to the deviations of some internal state variables from
their equilibrium values. The entropy source strength can thus serve as a basis for
the systematic description of the irreversible processes occurring in a system.
This set of equations contains the irreversible fluxes as unknown parameters and
can therefore not be solved with given initial and boundary conditions for the state
of the system. At this point, we must therefore supplement our equations with an
additional set of equations that relate the irreversible fluxes and the thermodynamic
forces appearing in the entropy source strength. In the first approximation, the fluxes
are linear functions of the thermodynamic forces. Fick’s law of diffusion, Fourier’s
law of heat conduction, and the stress–strain relationship in fluid mechanics, for
instance, belong to this class of linear phenomenological laws.
7.2 The Local Approach 251

In this sense, an outstanding contribution to elucidate the relationships between

fluxes and forces in complex systems was made by Onsager6 establishing the reci-
procity theorem, connecting the coefficients, which occur in the linear phenomeno-
logical laws that describe the irreversible processes.
Finally, Prigogine7 formulated a consistent phenomenological theory of irre-
versible processes, incorporating both Onsager’s reciprocity theorem and the explicit
calculation for a certain number of physical situations of the so-called entropy source
strength. In this way, a new field of thermodynamics of irreversible processes was
born.
The problem of justifying the principles of non-equilibrium thermodynamics can
alternatively be approached from the viewpoint of the kinetic theory. The irreversibil-
ity itself is already contained in the fundamental equation of the kinetic theory, the
Boltzmann integro-differential equation. One may then justify the use of thermody-
namic relations for systems outside equilibrium, and derive the Onsager reciprocal
relations (see, e.g., Philippi and Brun [129]).
In this part, we limit ourselves to presenting the transport equations for non-
equilibrium systems and, in Chaps. 8 and 9 to use the entropy balance equation
as a tool for finding the fundamental balance equations in multiphase, multicom-
ponent systems, by using the Gibbs’s formal relations. The book of de Groot and
Mazur [39] is suggested to interested readers aiming to improve their understanding
of irreversible phenomena, including chemical reactions, cross-effects in multicom-
ponent systems and electromagnetic theory.

7.2 The Local Approach

Consider a thermodynamic system with a single component. When this system is in

equilibrium, the energy of its molecules, or the internal energy .U , is a homogeneous
function of the entropy . S, the volume .V and the mass .m,

.U = U (S, V, m) .

Since the temperature .T can be defined as .T = (∂U/∂ S)V,m , the pressure as

. P = − (∂U/∂ V ) S,m and the chemical potential as .μ = (∂U/∂m) S,V the following
differential equation can be derived

dU = T d S − Pd V + μdm,
.

6 L. Onsager (1931), Reciprocal Relations in Irreversible Processes. I. Phys. Rev. 37, 405–426 [122].
L. Onsager (1931), Reciprocal Relations in Irreversible Processes. II. Phys. Rev. 38, 2265–2279
[123].
7 I. Prigogine, Etude thermodynamique des phenomenes irreversibles (Dunod, Paris and Desoer,

Liege, 1947).
252 7 Non-equilibrium States

or, in terms of intensive variables, as

. du = T ds − Pdv, (7.1)

because in homogeneous systems,

u = T s − Pv + μ,
.

and

. dμ = − sdT + vd P, (7.2)

in accordance with the well-known Gibbs–Duhem equation.

In this section, we will prefer to work with the mass density .ρ = 1/v instead of
the specific volume .v as an independent variable. So Eq. (7.1) will be written as
P
ρdu = ρT ds +
. dρ. (7.3)
ρ

Equation (7.3) is strictly valid for homogeneous systems in thermodynamic equi-

librium. Non-equilibrium systems can, nevertheless, be described with this equation
using the separation of scales hypothesis in a continuum system. With this assump-
tion, each point of the space is considered to have a characteristic length much greater
than the molecular dimensions and, at the same time, much smaller than the charac-
teristic length of the whole system. So, each point can be thought as an elementary
system in local thermodynamic equilibrium, although its local thermodynamic prop-
erties differ from the ones of its neighbor points.
In fluid dynamics, within the framework of continuum mechanics, a fluid element
or fluid particle is a very small amount of fluid, identifiable throughout its dynamic
history while moving with the fluid flow. As it moves, the mass of a fluid element
remains constant, while—in a compressible flow—its volume may change. The shape
of this fluid particle may also change due to the distortion by the flow.
Therefore by assuming the continuum hypothesis and, consequently, by suppos-
ing each point to be an elementary representative volume in local thermodynamic
equilibrium, the governing differential equations for this fluid in non-equilibrium
states will be established.
Let us first rewrite Eq. (7.3) in the form
ds ρ du P dρ
ρ
. = − , (7.4)
dt T dt ρT dt

where .d/dt refers to the material time derivative, .d/dt = ∂t + v · ∇, i.e., to the
changes in the course of time that an elementary material volume will suffer in its
thermodynamic properties when flowing along its trajectory.
7.3 Mass Conservation 253

A trajectory or path line of a fluid particle can be described by a function

x = x (x0 , t) where .x is the position of the particle at time .t and .x0 is the posi-
.
tion this particle had at time .t = 0. The initial coordinates .x0 of a particle is referred
to as the material coordinates of the particle. The term Lagrangian coordinates is
also used. The spatial coordinates .x of the particle can be referred to as its position
or place. Associated with the spatial and Lagrangian descriptions are two derivatives
with respect to time. We denote them by
( )
∂
.∂t = = local time derivative
∂t x
( )
d ∂
. = = material time derivative
dt ∂t x0

Thus .∂t F is the rate of change of F as observed at a fixed point .x, whereas . ddtF
is the rate of change as observed when moving with the particle. In particular, the
material derivative of the position of a particle is its velocity.
This allows us to establish a connection between the two derivatives,

( )
dF ∂ ∂ ∂F ∂x
. = (F (x (x0 , t) , t))x0 = (F)x + ·p = ∂t F + v · ∇ F.
dt ∂t ∂t ∂x ∂t x0

Therefore, the material derivative of the entropy in Eq. (7.4) is dependent on the
material derivatives of the internal energy and the density. These derivatives will be
given in the next two sections in relation to the mass conservation and the internal
energy balance equations.

7.3 Mass Conservation

The law of mass conservation is attributed to Antoine Lavoisier.8

In its modern integral form, the mass conservation equation is related to the
invariance of mass in a material or Lagrangian volume .V (t) as depicted in Fig. 7.1

8 It is generally accepted that Lavoisier’s great accomplishments in chemistry largely stem from
his changing the science from a qualitative to a quantitative one. Lavoisier is most noted for his
discovery of the role oxygen plays in combustion. He recognized and named oxygen (1778) and
hydrogen (1783) and opposed the phlogiston theory. Lavoisier helped construct the metric system,
wrote the first extensive list of elements, and helped to reform chemical nomenclature. He predicted
the existence of silicon (1787) and was also the first to establish that sulfur was an element (1777)
rather than a compound. He discovered that, although matter may change its form or shape, its mass
always remains the same [101].
254 7 Non-equilibrium States

V(t+ t) A(t+ t)

V(t)

A(t)

Fig. 7.1 Material or Lagrangian volume .V (t) moving inside a fluid has its volume and shape
affected by pressure and viscous forces, but the mass .m is always the same

(
dm d
. = ρd V = 0.
dt dt
V (t)

By using the Reynolds transport theorem9 we obtain

( ( (
d
. ρd V = ∂t ρd V + ρv · nd A = 0. (7.5)
dt
V (t) V (t) A(t)

The divergence theorem

( (
. ρv · nd A = ∇ · p (ρv) d V,
A(t) V (t)

can, thus, be used for the surface integral resulting in the mass conservation equation
in its differential form

∂ ρ + ∇ · (ρv) = 0
. t (7.6)

or, in terms of the material derivative of the density

dρ
. = ∂t ρ + v · p∇ρ,
dt

9 The Reynolds transport theorem is derived in fluid mechanics textbooks, see, e.g., [7, 93, 127].
7.3 Mass Conservation 255

as
dρ
. = − ρ∇ · v (7.7)
dt
The local form of the mass conservation equation, Eq. (7.6) can be generalized to
any arbitrary thermodynamic property . F. So, let . f be the value of this property per
unit mass. The time variation of this property in a material volume .V (t), is given by
( ( (
d
. ρ f dV = − j ( f ) · nd A + P ( f ) d V. (7.8)
dt
V (t) A(t) V (t)

The above equation is a general conservation law and means that the time variation
of the thermodynamic property . F in a material volume .V (t) is due to the diffusive
flux .j ( f ) across its surface boundary and to a source term .P ( f ) that generates the
property . F inside this material volume.
By applying the Reynolds transport equation, Eq. (7.5) and the divergence theo-
rem, Eq. (7.8) can be written in its differential form

∂ (ρ f ) + ∇ · (ρ f v + j) = P ( f ) ,
. t (7.9)

where .ρ f v is the advective flux of . f .

7.3.1 Exercises

1. Integrate Eq. (7.6) in an Eulerian volume .V0 , invariant with time and show that
( (
d
. ρd V = − ρv · nd A,
dt
V0 A0

where . A0 denotes the external area of the Eulerian volume. What is the physical
meaning of the r.h.s. of the above equation?
2. Consider now Eq. (7.5). What is the physical meaning of
(
. ρv · nd A,
A(t)

when .V (t) is a Lagrangian volume with a constant mass?

3. Consider a mixture of two non-reacting fluids A and B and .Ji , .i = A, B to be
the total flux of each one of these fluids. The flux .J A has two components: the
advective flux .ρ A v and the diffusive flux .j A = ρ A (v A − v).
256 7 Non-equilibrium States

(a) What would be the form of the differential equation, Eq. (7.9), for .ρ A and
.ρ B ?
(b) Write these equations in accordance with an integral Lagrangian form.
(c) Show that.j A = − j B , i.e., that the two diffusive flux must cancel themselves.

7.4 Momentum Balance Equation

The momentum per unit mass is represented by . f = v. It is a vectorial quantity and

is transported in accordance with Eq. (7.9), where .ρ f v = ρvv is the momentum per
unit volume .ρv that is advected with the velocity .v. In this case a external force field
.g—the gravity field, for instance—is a source of momentum at each point of the
physical domain and Eq. (7.9) is written as

∂ (ρv) + ∇ · (ρvv + P) = ρg.

. t (7.10)

Pressure tensor .P represents the diffusive flux of momentum and has two compo-
nents, an elastic component . Pδ and a viscous component, the viscous stress tensor .τ ,

.P = Pδ + τ ,

where .δ is the Kronecker-delta tensor and . P is the local thermodynamic pressure.

In components notation, Eq. (7.10) is also written as
( )
∂ (ρvα ) + ∂β ρvα vβ + Pδαβ + ταβ = ρgα .
. t (7.11)

By using the mass conservation equation, Eq. (7.7), the above equation can be
written as
dvα
.ρ = ρgα − ∂α P − ∂β ταβ , (7.12)
dt
which is the usual form of the momentum balance equation that is found in most
textbooks on fluid mechanics.
For Newtonian fluids that satisfy the Stokes hypothesis10 the relationship between
the viscous stress and strain rate tensors is given by

10 The Stokes hypothesis [162] supposes that the trace of the viscous stress tensor is null

.τ11 + τ22 + τ33 = 0.

This hypothesis is confirmed by the kinetic theory for systems of particles limited to translational
degrees of freedom. The paper of Buresti [22] is suggested for further reading.
7.4 Momentum Balance Equation 257

2
τ
. αβ = − 2ηSαβ + η∇ · pv, (7.13)
3
where .η is the dynamical viscosity and
1( )
S
. αβ = ∂α vβ + ∂β vα ,
2
is the strain rate tensor.
The Lagrangian integral form of Eq. (7.10) gives, perhaps, a better picture of the
transport of momentum in a fluid,
( ( ( (
d
. ρvd V = − Pnd A − τ · nd A + ρgd V, (7.14)
dt
V (t) A(t) A(t) V (t)

which can be interpreted as giving the forces that affect the momentum of a material
volume .V (t):

• the normal pressure forces . Pn and the viscous forces .τ · n acting on the external
surface . A (t) of the material volume;
• the external forces .ρg acting in all the points of .V (t).

7.4.1 Exercises

1. With the help of Eq. (7.13) show that for incompressible flow with constant
viscosity,
dvα
.ρ = ρgα − ∂α P + η∇ 2 vα . (7.15)
dt
2. Neglecting the gravity, consider the incompressible, two-dimensional, and full-
developed, Poiseuille channel flow as depicted in Fig. 7.2. In this case

.∇ · v = 0,

v = v1 (x2 ) ,
. 1

and

v = 0.
. 2
258 7 Non-equilibrium States

x2
inlet V 2b outlet
x1
velocity profile

Fig. 7.2 A two-dimensional channel flow

(a) Simplify Eq. (7.15) for the components 1 and 2 of the momentum and
show that the pressure P is only dependent on the horizontal coordinate,
. P = P (x 1 ).
(b) Show that .∂ 2 v1 /∂ x22 is a constant and related to .d P/d x1 .
(c) Find the velocity profile .v1 = v1 (x2 ).
(d) In the Eulerian volume V shown in the figure, integrate the resulting equation
for the horizontal velocity component momentum equation and show that
the pressure difference between the inlet and the outlet must counterbalance
the friction forces between the fluid and the channel wall, for maintaining
the flow.

7.5 Energy Conservation

The principle of energy conservation was enunciated by Hermann Ludwig Ferdinand

von Helmholtz (Fig. 7.3). Based on the works of Sadi Carnot, Émile Clapeyron and
James Prescott Joule, he postulated a relationship between mechanics, heat, light,
electricity and magnetism by treating them all as manifestations of a single force
(or energy). He published his theories in his book ‘Über die Erhaltung der Kraft’
(On the conservation of force) in 1847 [68].
In a fluid, the total energy ( ) can be written as the sum of its internal .(U ),
kinetic .(E c ), and potential . E p energies

. E = U + Ec + E p .

In this section, we derive the balance equation for the internal energy from the
balance equations for the kinetic and potential energy and from the conservation
equation for the total energy.
7.5 Energy Conservation 259

Fig. 7.3 Helmholtz in 1848

Balance Equation for the Kinetic Energy

The kinetic energy balance equation can be retrieved by making the inner product of
the momentum balance equation, Eq. (7.11) by the local velocity .v. Therefore, for
the time derivative term
( )
v ∂ (ρvα ) = ∂t ρv2 − ρvα ∂t (vα ) .
. α t

=vα ∂t (ρvα )−v2 ∂t ρ

So
( )
1 2 v2
v ∂ (ρvα ) = ∂t
. α t ρv + ∂t ρ. (7.16)
2 2

On the other hand, for the momentum advection flux

( ) ( )
v ∂
. α β ρvα vβ = ∂β ρv2 vβ − ρvα vβ ∂β vα .
=vα ∂β (ρvα vβ )−v2 ∂β (ρvβ )

So
( )
( ) 1 2 v2 ( )
.vα ∂β ρvα vβ = ∂β ρv vβ + ∂β ρvβ . (7.17)
2 2
260 7 Non-equilibrium States

Summing Eqs. (7.16) and (7.17), we have,

( ) ( )
( ) 1 2 1 2
v ∂ (ρvα ) + vα ∂β ρvα vβ = ∂t
. α t ρv + ∂β ρv vβ
2 2
v2 ( ( ))
+ ∂t ρ + ∂β ρvβ . (7.18)
2
=0

For the remaining terms

( )
v ∂
. α β Pδαβ = vα ∂α P = ∂α (Pvα ) − P∇ · v, (7.19)

and
( )
v ∂ τ
. α β αβ = ∂β ταβ vα − ταβ ∂β vα . (7.20)

From Eqs. (7.18), (7.19), and (7.20) results

( ) ( )
1 2 1 2
∂
. t ρv + ∂β ρv vβ + Pvα + ταβ vα = ρgα vα + P∇ · v + ταβ ∂β vα ,
2 2

or in vector form

( ) ( )
1 2 1 2
∂
. t ρv + ∇ · ρv v + Pv + τ · v = ρg · v + τ : ∇v + P∇ · pv. (7.21)
2 2

In its Lagrangian integral form, the above equation reads as

( ( (
d 1 2
. ρv d V = − Pv · nd A − τ · v · nd A
dt 2
V (t) A(t) A(t)
(
+ (ρg · v − ϕ + P∇ · v) d V.
V (t)

In the above equation

• the first term on the r.h.s. represents the compression work exerted by normal
forces on the boundary surface . A (t);
• the second term represents the change of kinetic energy due to shear forces between
. V (t) and the external fluid.
7.5 Energy Conservation 261

There are three sources of kinetic energy inside the Lagrangian volume:

• the external body force .g;

• the viscous forces in the form of a dissipation term .ϕ == − τ : ∇v ≥ 0 that is
responsible for the irreversible conversion of kinetic energy into heat;
• in compressible flows when.∇ · v /= 0 the pressure also acts as a (reversible) source
of kinetic energy.

Balance Equation for the Potential Energy

The balance equation for the potential energy is given in its Lagrangian form as
( (
d
. ρϒd V = − ρg · vd V, (7.22)
dt
V (t) V (t)

where .ϒ is the potential energy per unit mass. In fact, there is no diffusive flux for
the potential energy, which is only transported by advection of the fluid and can be
transformed into kinetic energy when the inner product between the external force
.g and the local velocity .v is greater than zero. In its differential form, this balance
equation is written as

∂ (ρϒ) + ∇ · (ρϒv) = − ρg · v.
. t (7.23)

Energy Conservation
Consider now the total energy . E = U + E c + E p . Due to the principle of energy
conservation, there are no sources of total energy
( )
P (E) = P (U ) + P (E c ) + P E p = 0,
.

and since
( )
P (E c ) + P E p = τ : ∇v + P∇ · v,
.

this result means that the source of thermodynamic internal energy is to be written
as

P (U ) = ϕ − P∇ · v.
.

Therefore, the balance equation for the internal energy .ρu per unit volume, will
be written as

∂ (ρu) + ∇ · (ρuv + q) = ϕ − P∇ · v,
. t (7.24)

where .q is the Fourier diffusive flux of heat.

262 7 Non-equilibrium States

The Eulerian integral form of this equation is given by

( ( (
d
. ρud V = − (ρuv · n + q · n) d A + (ϕ − P∇ · v) d V. (7.25)
dt
V0 A0 V0

In its differential form, the internal energy balance equation is more usually written
as
du
ρ
. + ∇ · q = ϕ − P∇ · v. (7.26)
dt
The total energy conservation equation can be written by summing Eqs. (7.21),
(7.23) and (7.24),
( ) [( ) ]
1 ρu + 21 ρv2 + ρϒ v
.∂t ρu + ρv2 + ρϒ + ∇ · = 0. (7.27)
2 + q + Pv + τ · v

The meaning of the several terms that contribute to the transport of energy can
be better appreciated when this equation is written in its Eulerian integral form. So,
consider an Eulerian fixed volume .V0 delimited by an external surface with area . A0 .
Integrating Eq. (7.27) in the volume .V0 and using the Reynolds theorem, Eq. (7.8)

( ( )
d 1
. ρu + ρv2 + ρϒ d V
dt 2
V0
( ( ) (
1
=− ρu + ρv2 + ρϒ v · nd A − Pv · nd A (7.28)
2
A0 A0
( (
. − q · nd A − τ · v · nd A.
A0 A0

The first term on the r.h.s. represents the net flux of total energy through the
Eulerian volume .V0 .
The total energy . E of .V0 is also affected by:

• the pressure work .− A0 Pv · nd A that is released to or from .V0 at, respectively,

the inlet and outlet;
• the net heat flux .− A0 q · nd A exchanged with the external environment;
• the friction work.− A0 τ · v · nd A that is transferred through the boundary surface
. A 0 (Fig. 7.4).
7.6 The Heated Cavity Problem 263

Pressure and friction work

Air
intake

To combustion
chamber

Fig. 7.4 Pressure work and friction are responsible for transferring the kinetic energy of the com-
pressor blades to the air

Fig. 7.5 The heated cavity

thermal insulation
problem
x

y
L

Heat flux

7.6 The Heated Cavity Problem

In this section, we apply the differential equations derived in the previous sections
to formulate and find the temperature and velocity fields that are developed when a
fluid inside a two-dimensional square cavity is heated from its right face (Fig. 7.5).
We consider the fluid as an ideal gas submitted to a constant heat flux .q0 during a
certain interval of time .∆t. After this interval, the cavity’s right face is insulated and
we follow what happens with the temperature and velocity fields.
264 7 Non-equilibrium States

Further, we suppose that the heating process only reduces the local density.ρ (x, y)
by an small amount given by

∆ρ = − ρ0 β (T − T0 ) ,
.

where .ρ0 is the unperturbed density at the initial temperature .T0 and
( )
1 ∂ρ
.β = − ,
ρ0 ∂ T P

is the volumetric expansion coefficient at a constant pressure.

We can, thus, use the Boussinesq’s approximation11 considering the flow to be
incompressible, driven by a buoyancy force .− ρ0 gβ (T − T0 ) opposed to the gravity
force .ρ0 g along the coordinate . y. When the temperature is high enough in such
a way that .β (T − T0 ) > 1 the gas inside the cavity will flow upwards by natural
convection whereas in the colder regions, the gas moves downwards, closing the
convection loop.
With these approximations, the potential energy inside the cavity is a constant
and given by

( (L (L
L2 × 1
. − (ρ0 y) d V = − ρ0 dx ydy = − ρ0 .
2
V0 0 0

Therefore, from an integral point of view the equation to be used during the heating
time is
( ( )
d 1
. ρ0 u + ρ0 v d V = q0 L × 1,
2
dt 2
V0

which is a suitable form of Eq. (7.28) for the problem since there is no flow through
the boundary of the cavity.
It is, nevertheless, important to note that this invariance of the total potential
energy is a consequence of the mathematical model used for describing the problem.
A more realistic model would consider the local variations of the density with time
due to heating and the incompressibility assumption would not apply to the problem.
The differential equations for the horizontal.vx and vertical.v y velocity components
are,
∂vx ∂v y
. + = 0,
∂x ∂y

11Boussinesq, Joseph (1897). Théorie de l’écoulement tourbillonnant et tumultueux des liquides

dans les lits rectilignes a grande section. Vol. 1. Gauthier-Villars.
7.6 The Heated Cavity Problem 265
( )
∂vx ∂vx ∂vx 1 ∂P ∂ 2 vx ∂ 2 vx
. + vx + vy =− +ν + , (7.29)
∂t ∂x ∂y ρ0 ∂ x ∂x2 ∂ y2

∂v y ∂v y ∂v y 1 ∂ (P − ρ0 gy)
. + vx + vy = − gβ (T − T0 ) −
∂t ∂x ∂y ρ0 ∂y
( 2 )
∂ vy ∂ 2vy
+ν + .
∂x2 ∂ y2

Since the flow was supposed to be incompressible the pressure P lost its meaning
as a function of .ρ and .T given by the Clapeyron equation of state and became an
additional field variable.
For closing the system given by Eq. (7.29), we use the Fourier law

q = − λ∇T,
.

for heat conduction and the internal energy balance equation, Eq. (7.24) is written as
[ ] ( 2 )
∂T ∂T ∂ T ∂2T
.ρ0 cv ∂t T + ρ0 cv v x + vy =λ + + ϕ, (7.30)
∂x ∂y ∂x2 ∂ y2

where.cv is the specific heat,.λ the thermal conductivity and.η, the dynamical viscosity
of the gas are supposed to be constants.
The last term on the r.h.s. of Eq. (7.30) corresponds to the irreversible transfor-
mation of kinetic energy into heat. For incompressible flows (see Sect. 7.7),
(( ) ( ) ) ( )
∂vx 2 ∂v y 2 ∂vx ∂v y 2
.ϕ = 2ηS : S = 2η + +η + .
∂x ∂y ∂y ∂x

Equations (7.29) and (7.30) can be written using dimensionless variables as

∂ v̂x ∂ v̂ y
. + = 0,
∂ x̂ ∂ ŷ

( )
∂ v̂x ∂ v̂x ∂ v̂x ∂ P̂ ∂ 2 v̂x ∂ 2 v̂x
. + v̂x + v̂ y =− + + ,
∂ tˆ ∂ x̂ ∂ ŷ ∂ x̂ ∂ x̂ 2 ∂ ŷ 2

( )
∂ v̂ y ∂ v̂ y ∂ v̂ y Ra ∂ P̂ ∂ 2 v̂ y ∂ 2 v̂ y
. + v̂x + v̂ y =− θ− + + , (7.31)
∂t ˆ ∂ x̂ ∂ ŷ Pr ∂ ŷ ∂ x̂ 2 ∂ ŷ 2
[ ] ( )
∂θ ∂θ 1 ∂ 2θ ∂ 2θ
.∂tˆ θ + v̂ x + v̂ y = + 2 + Ecϕ̂,
∂ x̂ ∂ ŷ Pr ∂ x̂ 2 ∂ ŷ
266 7 Non-equilibrium States

where .tˆ = νt/L 2 , .x̂ = x/L, . ŷ = y/L, .v̂α = Lvα /ν,

T − T0 P − ρ0 gy
θ=
.
q0 L
, P̂ = ρ0 ν 2
.
λ L2

The problem is driven by three non-dimensional parameters:

• the Rayleigh number related to the ratio between the buoyancy and the viscous
forces . Ra = ρo βgc
4
v q0 L
νλ2
;
• the Prandtl number giving the ratio of momentum diffusivity to thermal diffusivity,
ρo cv ν
. Pr = ;
λ
• the Eckert number related to the dissipation of kinetic energy, . Ec = L 3νcvλq0 .
2

Equation (7.31) are highly non-linear and were solved by using a finite-volumes
numerical method.
Figure 7.6 shows the temperature field and streamlines during the heating time of
the cavity. The heated fluid near the right face moves upwards by natural convection
driven by a buoyancy force and downwards in the colder regions. We can notice that
the velocity and temperature gradients are very large close to the heated right face.
After the heating process is finished and the right face of the cavity is insulated,
the convection loop is dissipated, and the kinetic energy of the flow being converted
into internal energy (Fig. 7.7).
From Eq. (7.25), during the heating process, the internal energy increases with
time in accordance with
( (
d
. ρ0 cv (T − T0 ) d V = q0 L × 1 + ϕd V,
dt
V0 V0

meaning that the internal energy increase is due to the heat flow into the cavity and
the viscous dissipation of kinetic energy. Remark that dissipation can be very strong
in the heating time, due to the strong velocity gradients near the right face of the
cavity.
During the dissipation time, after the heating is ceased and the right face of the
cavity is insulated, the kinetic energy is entirely converted into internal energy by
viscous dissipation,
( (
d
. ρ0 cv (T − T0 ) d V = ϕd V.
dt
V0 V0

The numerical solution of the differential equations, Eq. (7.31) enables to find the
temperature field .T (x, y, t) and, consequently, the total internal energy of the cavity
as a function of time
(
.U (t) = ρ0 cv (T (x, y, t) − T0 ) d V.
V0
7.6 The Heated Cavity Problem 267

Fig. 7.6 a Temperature field 8.80x10-2

and b streamlines during the
heating time of the cavity 7.33x10-2
(.tˆ = 0.008). Dimensionless
parameters are Pr = 1,
Ra = .106 , Ec = .10−4 [73] 5.87x10-2

4.40x10-2

2.93x10-2

1.47x10-2

(a)

(b)

Figure 7.8 shows the integrated result for the total internal energy .U (t) for the
cavity problem. When the gas is heated both the total internal and total kinetic energy
increase with time producing a convection loop that is further extinguished by viscous
dissipation when heating is ceased.
Viscous dissipation was also measured using the simulation results for the velocity
field. The results for
(
. D̂ = Ec ϕ̂d V̂ ,
V̂

as a function of the dimensionless time .tˆ are presented in Fig. 7.9.

Viscous dissipation grows very fast during the heating period due to the strong
velocity gradients close to the heated face. During the dissipation time, it is responsi-
268 7 Non-equilibrium States

Fig. 7.7 a Temperature field 1.56x10

-2

and b streamlines at the

dissipation period .tˆ = 0.026. 1.30x10
-2

Dimensionless parameters
are Pr = 1, Ra = .106 , Ec = -2
−4 [73] 1.04x10
.10

-3
7.80x10

-3
5.20x10

-3
2.60x10

(a)

(b)

ble for converting the kinetic energy of the flow into internal energy and for reducing
these velocity gradients.
Oscillations in both Figs. 7.8 and 7.9 do not appear to have any physical meaning
and are possibly due to numerical errors.
Equilibrium .× Non-equilibrium Thermodynamics
In spite of the simplifications that were made for numerically solving the problem, the
heated cavity is an important sample problem for distinguishing the main features of
non-equilibrium thermodynamics when compared to equilibrium thermodynamics.
The models and methods of equilibrium thermodynamics that were seen in Parts I
and II of this book only apply to thermodynamic systems in equilibrium. The first
law of thermodynamics

δ Q + δW = dU,
.
7.6 The Heated Cavity Problem 269

Heating Dissipation
1.00

Kinetic energy

0.75
Total energy

0.50

Internal Energy U(t) as a fraction of the

total added energy q0A t
0.25

0.00
0.00 0.01 0.02 0.03 0.04
time ^t

Fig. 7.8 A constant heat flow .q0 into a cavity filled with an ideal gas increases its temperature and
produces natural convection currents. After heating is ceased these currents are dissipated, and their
kinetic energy is transformed into internal energy

3 Heating Dissipation
^
Viscous dissipation D

0
0.01 0.02 0.03 0.04
time ^t

Fig. 7.9 Dimensionless viscous dissipation .ϕ̂ as a function of time [73]

for instance, can only be used when we are considering a thermodynamic process
between any two equilibrium states. It cannot be used, for, e.g., finding the variation
of the internal energy, when one or both states are non-equilibrium states.
Therefore, for the cavity problem, the variation of the internal energy between the
initial 1 and final 2 equilibrium states is the total heat that was added to the cavity
during the heating time .∆t,

. Q 1−2 = q0 L × 1 × ∆t = U2 − U1 ,
270 7 Non-equilibrium States

in accordance with the first law of thermodynamics.

In any intermediary non-equilibrium state, the kinetic energy is different from
zero, the influx of heat being used for creating the convection loop and increasing
the temperature. The source of kinetic energy (and momentum) is, in this Boussinesq
approximation, the buoyancy force.

7.7 The Law of Increasing Entropy

By inserting Eqs. (7.7) and (7.26) into Eq. (7.4), the following differential relationship
can be derived,
ds 1
ρ
. = − (∇ · q − ϕ) .
dt T
The first term on the r.h.s. of the above equation can be rewritten in the form
1 (q) 1
. ∇ ·q=∇ · + 2 q · ∇T, (7.32)
T T T
so
ds
ρ
. = − ∇ · qs + P (s) , (7.33)
dt
where
q
q =
. s ,
T
is the flux of entropy and
( )
1 ϕ
.P (s) = − 2 q · ∇T + ≥ 0,
T T

is the irreversible source of entropy.

The Lagrangian integral of this equation is easily found as,

( ( ( ( )
d q·n 1 ϕ
. ρsd V = − dA + − 2 q · ∇T + dV . (7.34)
dt T T T
V (t) A(t) V (t)

dS de S di S
dt dt dt

This equation, to be attributed to Ilya Prigogine [140], states that the material
derivative of the entropy . ddtS is due to the net influx of heat through the external
7.7 The Law of Increasing Entropy 271

surface of the material volume and to the existence of internal non-equilibrium in the
temperature and velocity fields. By the second law of thermodynamics, the internal
production of entropy . ddti S must be non-negative
di S
. ≥ 0.
dt
Let us consider a Newtonian fluid that satisfies the Stokes hypothesis (for which the )
relationship between the viscous stress.ταβ and the rate of strain. Sαβ = 21 ∂α vβ + ∂β vα
tensors is given by
2
τ
. αβ = − 2ηSαβ + η∇ · vδαβ .
3
Using the symmetry with respect to a change of index, the viscous dissipation
term can be written as
1( )
ϕ = − ταβ : ∂α vβ = − ταβ
. ∂α vβ + ∂β vα = − ταβ Sαβ ,
2
or
( )
2
ϕ = 2ηSαβ Sαβ − η (∇ · v)2 ,
.
3

because
1( )
S δ
. αβ αβ = 2∂x vx + 2∂ y v y + 2∂z vz = ∇ · v.
2
Therefore
( )2
1
ϕ = 2η Sαβ − ∇ · v ,
.
3

because

( )2
1 2 1
. Sαβ − ∇ ·v = Sαβ Sαβ − Sαβ δαβ ∇ · v + (∇ · v)2 δαβ δαβ
3 3 9
2 1
= Sαβ Sαβ − (∇ · v)2 + (∇ · v)2 ,
3 3
since, in three dimensions

δ δ
. αβ αβ = 3.
272 7 Non-equilibrium States

On the other hand, by restricting ourselves to isotropic fluids that satisfy the
Fourier’s law for heat conduction

q = − λ∇T,
.

the heat conduction contribution to the source of entropy can be written as

( )
q · ∇T ∇T 2
. − = λ ,
T2 T

in such a way that

( [ ( ) ( )2 ]
di S ∇T 2 2η 1
. = λ + Sαβ − ∇ · v d V ≥ 0, (7.35)
dt T T 3
V (t)

and we are led to the conclusion that for a Newtonian and isotropic fluid both the
dynamical viscosity .η and the thermal conductivity .λ must be positive, otherwise the
second law of thermodynamics would be broken. This appears to be an obvious con-
clusion since we always expect that heat will flow from higher to lower temperatures
and that momentum will flow from higher to smaller velocities, but the important
meaning of Eq. (7.35) is that dissipation will always happen every time we have a
gradient in our thermodynamic system. In other words, gradients are the source of
irreversibility and will always increase the entropy.
When a thermodynamic system is subjected to a process, starting from an equi-
librium state 1 at .t = 0 to another equilibrium state 2 at .t = t∞ , where .t∞ is the
time required by the system for reaching the equilibrium state 2, Eq. (7.34) can be
integrated between .t = 0 and .t = t∞ giving,

S − S1 = (S2 − S1 )e + (S2 − S1 )i
. 2

For concreteness let us reconsider the abrupt isothermal expansion sample prob-
lem of Sect. 2.5.1 and shown in Fig. 7.10. Numerical simulation of the velocity and
temperature field during the abrupt gas expansion is a difficult task for this problem
since it involves a compressible flow with moving boundaries and subjected to fric-
tion forces between the piston and the cylinder walls. As commented in Sect. 2.1.2
the friction work is absorbed by the thermal pond and will not contribute to defining
the final equilibrium state. Nevertheless, these forces have an important contribution
to the dynamics of gas expansion, or to the time the system will take for reaching
the final equilibrium state.
7.7 The Law of Increasing Entropy 273

50

50 50

weightless

300 K
P
1

30
0K
Patm

Thermal insulator irreversible

process

Ideal gas

Thermal pond Q12=778 J 2

V
10 cm2

Fig. 7.10 Isothermal expansion of an ideal gas

We can easily find the total variation of entropy as

V2 J
S − S1 = R ln
. 2 = 3.10 .
V1 K

On the other hand, since the boundary surface is maintained at a constant temper-
ature during the whole process, the entropy variation due to the heat transfer from
the thermal pond is

(t∞ (
q·n 1
. (S2 − S1 )e = − d A = Q 1−2 ,
T T
0 A(t)

giving
J
. (S2 − S1 )e = 2.5933 .
K
Therefore the amount
J
. (S2 − S1 )i = 0.5067 ,
K
274 7 Non-equilibrium States

corresponds to the entropy increase by the internal dissipation processes due to the
temperature and velocity gradients inside the system that are produced when the gas
has an abrupt expansion

(t∞ ( ( )
1 ϕ
. (S2 − S1 )i = − 2 q · ∇T + d V.
T T
0 V (t)

7.8 Exercises

1. Consider the momentum balance equation in its differential form

∂ (ρv) + ∇ · (ρvv + P) = ρg.

. t (7.36)

(a) Write the Eulerian and Lagrangian integral forms of this equation.
(b) Consider the 2D channel drawn in Fig. 7.11. In this channel a fluid flows
along a horizontal direction, driven by a pressure gradient .(P0 − PL ) /L.
Use the integral Eulerian form of the momentum balance equation supposing
the flow to be incompressible and considering the control volume inside the
dashed line shown in the figure. What is the physical meaning of the several
terms that appear in the formulation? Simplify the equations for a stationary
flow.

2. An ideal gas in chamber A at 20 .◦ C is put in thermal contact with a liquid bath

at 90 .◦ C (Fig. 7.12). In consequence, a Bénard cellular convection process will
take place in the gas phase (consider that the liquid bath remains with .T = 90 ◦ C
during the whole process). Using the Boussinesq approximation shows that the
momentum balance equation for the gas can be written as12
dv
ρ
. 0 = − ∇ · τ − ∇ P + ρ0 (1 − β (T − T0 )) g (7.37)
dt
source of momentum

where .g is the gravity, .ρ0 is the fluid density at .t = 0 and .β is the fluid volumetric
expansion coefficient
( )
1 ∂ρ
.β = −
ρ ∂T P

12 In the Boussinesq approximation the flow is supposed to be incompressible with .∇ · v = 0. So

the density variation produced by heating is supposed to be small.
7.8 Exercises 275

Fig. 7.11 Channel flow

Fig. 7.12 Bénard cellular

convection
A

gas
200C
Thermal insulation

liquid bath
900C

Therefore, the buoyancy force .ρ0 (1 − β (T − T0 )) g is a source of momentum

contrary to gravity, and responsible for the convection process.

(a) Multiply Eq. (7.37) by .v and find the balance equation for the advection
kinetic energy of the fluid.
(b) Write the balance equation for the potential energy.
(c) Find the balance equation for the total energy in its differential and integral
form.
(d) Find the balance equation for the internal energy in its differential and
integral form.
(e) Find the balance equation for the entropy in its differential and integral form.
Finally, try to identify and explore the possible symmetries of the prob-
lem and to write/identify a closed system of differential equations and
276 7 Non-equilibrium States

their boundary conditions, for the velocity, temperature, and entropy whose
(numerical) solution will enable to find these fields in the course of time till
the equilibrium is reached. Suppose a plane geometry described by coordi-
nates .x and .z.
Also, try to explain the role of each term in the internal energy balance and
entropy equations in their integral forms.
Chapter 8
Multiphase Systems

Abstract Two-phase liquid–vapor systems in non-equilibrium conditions are often

found in nature and devices such as cooling systems of power plants, as sprays in the
combustion chamber of internal combustion engines, and in boiling and industrial
devices. The liquid–vapor interface has a thickness of a few nanometers and the fluid
density has strong variation along the interface thickness. In numerical simulations,
the way the interface position is determined plays a key role in modeling two-phase
systems and we can classify the several methods that were developed for two-phase
systems in two families. The first one is based on the singular interface approach when
the interface is considered as a boundary separating the two phases. The second
one is based on the diffuse interface approach when the interface is considered a
diffused zone. These two families of methods are presented in this chapter. For
illustrating purposes, the lattice-Boltzmann simulation results for the coalescence
process between two liquid droplets and vapor condensation are presented at the end
of the chapter.

8.1 Introduction

Many processes in nature and industry are dependent on the correct understanding of
multiphase systems. These include omnipresent phenomena and applications, such
as rain clouds or fuel injection into combustion chambers, rocket engines, or spray-
based production processes and petroleum exploitation.
Droplets have been the focus of scientific interest for a long time: the generation
of droplets from liquid jets under the influence of surface tension was studied by
Plateau1 and Lord Rayleigh [143] in the nineteenth century. The Plateau–Rayleigh
instability describes how a liquid jet breaks up into droplets. In Fig. 8.1, we can
notice the waviness in the jet before the breakdown. These undulations are nothing
but small disturbances, which grow with the distance from the jet exit. In accordance
with Laplace’s law, the tiniest curvature in the jet causes an imbalance in the liquid
pressure due to surface tension. Liquid pressure is higher in the jet regions with

1 Plateau JAF (1873) Statique expérimentale et théorique des liquides soumis aux seules forces

moléculaires, vol 2. Gauthier-Villars.

© Associação Paranaense de Cultura 2024 277
P. C. Philippi, Thermodynamics, https://doi.org/10.1007/978-3-031-49357-7_8
278 8 Multiphase Systems

Fig. 8.1 The

Plateau–Rayleigh instability
High pressure

Low pressure

High pressure

smallest curvature radius and the liquid flows from the high to the small pressures.
Considering the inset in Fig. 8.1, the downward movement of the liquid is assisted
by the pressure difference between the superior and central regions. In contrast, it
is resisted by the pressure difference between the inferior and central regions. This
promotes the growth of an undulation, which eventually transforms the central region
into a droplet. Because the system is unstable, any small changes in the pressure due
to the undulations will become larger, ultimately resulting in the jet breaking into
droplets.
As pointed out by Lefebvre and McDonnel in their seminal book [102], in the
process of atomization, a liquid jet is disintegrated by the kinetic energy of the
liquid, by the exposure to high-velocity air or gas, or by mechanical energy applied
externally through a rotating or vibrating device. Combustion of liquid fuels in diesel
engines, spark-ignition engines, gas turbines, rocket engines, and industrial furnaces
are dependent on effective atomization to increase the specific surface area of the
fuel and thereby achieve high rates of mixing and evaporation.
Another subject of great scientific interest is the liquid–vapor spinodal decompo-
sition (Fig. 8.2). As was pointed out in Chap. 4, at a given temperature, a thermody-
namic system that is suddenly put into a state where the density is between the vapor
and liquid saturated densities, cannot subsist in this state as a single-phase system
and any small perturbation will break up this system into a two-phase liquid–vapor
system. Phase separation is very fast in the beginning, however, it becomes slower
due to the large number of liquid droplets competing for those molecules that are in
the vapor phase. Smaller drops evaporate faster than larger drops. This is due to the
effects of surface tension on the interface with the greatest curvature, which leads to
a greater pressure difference in smaller droplets, leading them to evaporate earlier,
while larger drops tend to grow, by condensation or coalescence between neighbor
droplets.
Spinodal decomposition is not restricted to liquid–vapor systems, but to liquid–
liquid multicomponent systems and metallic alloys.
8.2 The Singular Interface Approach 279

Fig. 8.2 The first stages of spinodal decomposition in liquid–vapor systems

Typically, a material interface separating two different phases is on the order of

few nanometers, and, in an engineering context, can be considered as a sharp discon-
tinuity. However, the way the interface is treated and how its position is determined
plays a key part in multi-species, multiphase flow modeling, and differentiates the
numerous methods introduced over the last three decades. Following a physical rigor
principle, one can consider two families of methods, one that rigorously does not
allow interface mixing, namely, the singular interface methods, and the other one
that considers the interface as a diffused zone, namely, the diffuse interface methods.
In this chapter, these two approaches are exposed, leading to the macroscopic
equations for the density, momentum, and internal energy.

8.2 The Singular Interface Approach

In the singular surface approach, the transport equations are solved in each phase
and the interface is considered as a boundary, where the field variables from the two
phases are linked. In the following, these boundary conditions are derived.

8.2.1 Mass Conservation

Consider a material volume.Ω = Ωa ∪ Ωb as shown in Fig. 8.3. This material volume

involves the interface .∑ between phases .a and .b and the interface is supposed to be
moving with a velocity .W.
280 8 Multiphase Systems

Fig. 8.3 The singular

interface approach b
Phase b

t n
rface W
I nt e Interf
ace
ve
loc
ity

Phase a
a

Since the mass .m inside .Ω is constant

⎛ ⎞
d ⎝
. ρa d V + ρb d V ⎠ = 0. (8.1)
dt
Ωa Ωb

But from the Reynolds transport theorem

d ∂ρa
. ρa d V = dV + ρa va · nd A + ρa W · nd A
dt ∂t
Ωa Ωa ┌α ∑
∂ρa
= dV + ρa va · nd A + ρa va · nd A
∂t
Ωa ┌a ∑

− ρa va · nd A + ρa W · nd A,
∑ ∑

and so
( )
d ∂ρa
. ρa d V = + ∇ · (ρa va ) d V − ρa (va − W) · nd A.
dt ∂t
Ωa Ωa ∑
8.2 The Singular Interface Approach 281

In the same way,

d ∂ρb
. ρb d V = dV + ρb vb · nd A + ρb W · (− n) d A
dt ∂t
Ωb Ωb ┌b ∑
∂ρb
= dV + ρb vb · nd A − ρb vb · (− n) d A
∂t
Ωb ┌β ∑

+ ρb vb · (− n) d A − ρb W · (− n) d A,
∑ ∑

because the normal .n is oriented from phase .a to phase b in .∑. So,

( )
d ∂ρb
. ρb d V = + ∇ · (ρb vb ) d V + ρb (vb − W) · nd A.
dt ∂t
Ωb Ωb ∑

Replacing the above equations in Eq. (8.1),

( )
∂ρa ∂ρb
. + ∇ · (ρa va ) + + ∇ · (ρb vb ) d V
∂t ∂t
Ω

=0

+ ρb (vb − W) · nd A − ρa (va − W) · nd A
∑ ∑
= 0,

or

. ρa (va − W) · nd A = ρb (vb − W) · nd A.
∑ ∑

Since the interface .∑ is arbitrary and can be made as small as possible, this
equation can be written in its differential form as
·
ρ (va − W) · n = ρb (vb − W) · n = m.
. a (8.2)
·
The above equation means that the mass .m (per unit area) that exits phase .a, per
unit time, is the same as the one that enters into phase .b.
·
When phases .a and .b are immiscible, .m = 0 and

v · n = vb · n = W · n.
. a (8.3)
282 8 Multiphase Systems

If, in addition, it is possible to suppose that there is no slip between the two phases,
i.e., the tangential velocities of the two phases are the same in the interface

v = vb , in ∑.
. a (8.4)

8.2.2 Momentum Balance Equation

Neglecting the body forces in the volume element .Ω

⎛ ⎞
d ⎝
. ρa va d V + ρb vb d V ⎠
dt
Ωa Ωb

=− Pa · nd A − Pb · nd A
┌a ┌b

+ (∇γab · t + (γab ∇ · n) n) d A. (8.5)

∑

surface forces

In the above equation .∇ · n is the curvature of the surface or the inverse of the
curvature radius.R and.(γab ∇ · n) n is the normal or Laplace component of the surface
forces whereas .∇γab · t corresponds to the tangential or Marangoni component of
these forces,2

.Ps = Ps δ + τ s

2 Marangoni forces were first identified in the so-called tears of wine by James Thomson in 1855.
(James Thomson. On certain curious Motions observable at the Surfaces of Wine and other Alcoholic
Liquors. The London, Edinburgh, and Dublin Philosophical Magazine and Journal of Science. 1855.
pp. 330–333.) The general effect is named after Italian physicist Carlo Marangoni, who studied it
for his doctoral dissertation at the University of Pavia and published his results in 1865. (Carlo
Marangoni. Sull’espansione delle goccie d’un liquido galleggianti sulla superficie di altro liquid,
Pavia, Italy: Fratelli Fusi. 1869.) A complete theoretical treatment of the subject was given by Gibbs
[59, 60]. Since a liquid with a high surface tension pulls more strongly on the surrounding liquid
than one with a low surface tension, the presence of a gradient in surface tension will naturally
cause the liquid to flow away from regions of low surface tension. The surface tension gradient can
be caused by a concentration gradient or by a temperature gradient. The tears of wine effect is a
consequence of two factors: (i) alcohol has a lower surface tension and (ii) alcohol is more volatile
than water. The water/alcohol solution rises up the surface of the glass as a film. Alcohol evaporates
from the film leaving behind liquid with higher surface tension (more water, less alcohol). This
region with a lower concentration of alcohol (greater surface tension) pulls on the surrounding fluid
more strongly than the regions with a higher alcohol concentration (lower in the glass). The result
is the liquid is pulled up until its own weight exceeds the force of the effect, and the liquid drips
back down the vessel’s walls.
8.2 The Singular Interface Approach 283

for .s = a, b, is the pressure tensor, . Ps is the thermodynamic pressure and .τ s is the

viscous stress tensor in phase .s.
But,

d
. ρa va d V
dt
Ωa
∂
= (ρa va ) d V + (ρa va va ) · nd A
∂t
Ωa ┌a

+ (ρa va W) · nd A
∑

+ (ρa va va ) · nd A − (ρa va va ) · nd A,
∑ ∑

where .ρa va is the momentum of phase .a in the interface .∑ and .W is the interface
velocity, which transports this momentum in .∑.
It results

d
. ρa va d V
dt
Ωa
( )
∂
= (ρa va ) d V + ∇ · (ρa va va ) d V
∂t
Ωa

− (ρa va (va − W)) · nd A,

∑

or

d
. ρa va d V
dt
Ωa

= (− ∇ · Pa ) d V − (ρa va (va − W)) · nd A.

Ωa ∑
284 8 Multiphase Systems

In the same way, for phase .β,

d
. ρb vb d V
dt
Ωb
( )
∂
= (ρb vb ) d V + ∇ · (ρb vb vb ) d V
∂t
Ωb

+ (ρb vb (vb − W)) · nd A,

∑

or

d
. ρb vb d V
dt
Ωb

= (− ∇ · Pb ) d V + (ρb vb (vb − W)) · nd A.

Ωb ∑

So by replacing the above equations into Eq. (8.5)

. (− ∇ · Pa ) d V − (ρa va (va − W)) · nd A

Ωa ∑

+ (− ∇ · Pb ) d V + (ρb vb (vb − W)) · nd A

Ωb ∑

=− Pa · nd A − Pb · nd A (8.6)
┌a ┌b

. + (∇γab · t + (γab ∇ · n) n) d A.
∑

By using the divergence theorem

. (− ∇ · Pa ) d V = − Pa · nd A − Pa · nd A,
Ωa ┌a ∑

(− ∇ · Pb ) d V = − Pb · nd A + Pb · nd A.
Ωb ┌b ∑
8.2 The Singular Interface Approach 285

So, in Eq. (8.6), the integrals along the surfaces .┌α and .┌β cancel themselves and
we get

. (Pb − Pa ) · nd A
∑

+ (vb ρb (vb − W)) · nd A − (va ρa (va − W)) · nd A

∑ ∑

= (∇γab · t + (γab ∇ · n) n) d A,
∑

or

. (Pb − Pa ) · n + (vb ρb (vb − W)) · n − (va ρa (va − W)) · n

= (∇γab · t + (γab ∇ · n) n) · n

·
Now, from Eq. (8.2), we know that .ρb (vb − W) = ρa (va − W) · n = m. So,
·
. (Pb − Pa ) · n + (vb − va ) m = (∇γab · t + (γab ∇ · n) n) .

8.2.3 Particular Cases

·
When the phases are immiscible .m = 0, so

. (Pb − Pa ) · n = (∇γab · t + (γab ∇ · n) n) .

When the phases are immiscible and at rest, .τ a = τ b = 0 and so, .Pb = Pb δ and
Pa = Pa δ, resulting
.

. (Pb − Pa ) n = (∇γab · t + (γab ∇ · n) n) .

Making the inner product by .n

γab
. Pb − Pa = γab ∇ · n = ,
R
which is the Laplace’s law.
286 8 Multiphase Systems

8.2.4 Exercises

1. (a) Show that .∇ · n = 2/R on the surface of a sphere of radius . R (Hint: use
spherical coordinates).
(b) Using the same method that was presented in this section, derive the total
and internal energy balance equations for the interface.
(c) Marangoni force. Using the same reasoning as in Exercise 1 explain how a
gradient in the interface tension along the interface will produce a tangential
unbalanced force.
2. A typical Rayleigh–Taylor instability (RTI) occurs when a heavier fluid sits on
top of a lighter fluid under the effect of gravity, characterizing an unstable inter-
face. As the heavy fluid goes down, the RTI develops spikes with a characteristic
mushroom shape along with the appearance of plumes flowing upwards. At the
same time, the light fluid forms bubbles as it goes up to occupy the place of the
heavy fluid (Fig. 8.4). It receives its name after Rayleigh [144], who was the first
to study this phenomenon in the nineteenth century, and later Sir G. I. Taylor in
the 1950s [170].
By considering the heavier (h) and lighter (l) fluid as immiscible and neglecting
the Marangoni forces what would be the flow equations for fluids h and l and the
boundary conditions at the interface? You can suppose the problem to be two-
dimensional and the lateral boundary conditions to be periodic. Try to explain
the physical reasons for the mushroom shape of the interface.

Heavier fluid

Lighter fluid

Fig. 8.4 The Rayleigh–Taylor instability

8.3 The Diffuse Interface Approach 287

8.2.5 Numerical Methods for the Singular Interface

Approach

Numerical methods based on Riemann solvers were developed by Fechter and col-
laborators [51–53], Rohde and Zeiler [146] and Schleper [151].
The level set method (LSM) developed by Osher and Sethian [125] and Sussman
[166], uses a distance function from the interface, where the zero level marks the
interface location, and the positive values correspond to one of the fluids and the
negative values to the other one. In the original version, this method was affected by
spurious oscillations that were corrected in the work of Fedkiw et al. [54]. In this
technique, named ghost fluid method, a ghost cell storing the properties of the fluid b
is defined in correspondence with each computational cell storing the properties of the
fluid a. After each time step, the distance function, calculated with a level set equation,
determines which set of properties has to be used to compute the solution, whether
the one stored in the ghost or the computational cell [11]. This set of methods has
the advantage of allowing a sharp definition of the interface for fluids with different
equations of states. However, general drawbacks they share are the difficulty of
enforcing mass conservation at the discontinuities.
Interested readers are suggested to read these papers.

8.3 The Diffuse Interface Approach

The singular interface approach is physically well-grounded as long as the interfaces

are separated. However, it comes with a severe disadvantage when the interfaces
meet, which is the case with the coalescence of droplets and droplet splitting. This
renders the singular interface approach to fail. For such reasons, diffuse-interface
models have been suggested to describe the dynamics of two-phase flow.

8.3.1 Surface Forces

The study of two-phase liquid–vapor systems was pioneered by Johannes Diderik

van der Waals who in 1873 proposed a theoretical model3 for condensable gases in
his doctoral thesis.4

3 J. D. van der Waals (1873), Over de Continuiteit van den Gas en Vloeistoftoestand, Ph.D. thesis,
Leiden University [181].
4 The main interest of van der Waals was in the field of thermodynamics. He was influenced by

Rudolf Clausius’ 1857 treatise entitled Über die Art der Bewegung, welche wir Wärme nennen (On
the Kind of Motion which we Call Heat). Van der Waals was later greatly influenced by the writings
of James Clerk Maxwell, Ludwig Boltzmann, and Willard Gibbs. Clausius’ work led him to look
for an explanation of Thomas Andrews’ experiments that had revealed, in 1869, the existence of
288 8 Multiphase Systems

Consider a system composed of a single fluid where the liquid .( ) and the vapor
(v) phases coexist separated by an interface. Let this interface be a diffuse one, i.e., a
.
region where the thermodynamic properties vary continuously from the liquid to the
vapor phase and where it is possible to suppose local thermodynamic equilibrium at
each point.
With these assumptions, the fundamental relationships of equilibrium thermody-
namics can be used for each point in the space separating the two phases. So, consider
the Helmholtz free energy per unit mass . f in its differential form,
P
d f = − sdT +
. dρ.
ρ2

The Helmholtz energy can be rewritten in a volumetric basis as .ψ = ρ f , in such

a way that the above equation becomes
ψ+P
.dψ = − ρsdT + dρ.
ρ

Since the chemical potential .μ (or the Gibbs energy per unit mass, .g) satisfies
ψ+P
μ=g=
. , (8.7)
ρ

we can also write

dψ = − ρsdT + μdρ,
.

and, so, the chemical potential can be calculated from the Helmholtz energy as
( )
∂ψ
.μ = .
∂ρ T

Therefore, from Eq. (8.7), the thermodynamic pressure can be expressed in terms
of the Helmholtz energy per unit volume.

critical temperatures in fluids. He managed to give a semi-quantitative description of the phenomena

of condensation and critical temperatures in his 1873 thesis, entitled Over de Continuïteit van den
Gasen Vloeistoftoestand (On the continuity of the gas and liquid state). This dissertation represented
a hallmark in physics and was immediately recognized as such, e.g. by James Clerk Maxwell who
reviewed it in Nature in a laudatory manner. In this thesis he derived the equation of state bearing
his name. This work gave a model in which the liquid and the gas phases of a substance merge into
each other in a continuous manner. It shows that the two phases are of the same nature. In deriving
his equation of state van der Waals assumed not only the existence of molecules (the existence of
atoms was disputed at the time), but also that they are of finite size and attract each other. Since he
was one of the first to postulate an intermolecular force, however rudimentary, such a force is now
sometimes called a van der Waals force.
8.3 The Diffuse Interface Approach 289

Fig. 8.5 Excerpt of van der

Waals doctoral thesis [181]

( )
∂ψ
. P=ρ − ψ. (8.8)
∂ρ T

In homogeneous systems the free energy .ψ can be considered as a function of

the density and temperature. Nevertheless, in the interface, the density gradients can
be too strong in such a way that .ψ can be supposed to be also dependent on the
local density gradient, .ψ = ψ (ρ, T, ∇ρ). Therefore, in the interface the Helmholtz
potential has an excess due to the density gradients,

.ψ = ψ0 (ρ, T ) + ψexc , (8.9)

where .ψ0 (ρ, T ) is the free energy of an homogeneous system. The simplest form of
this excess energy is given by,
κ
ψexc =
. |∇ρ|2 ,
2
firstly proposed by van der Waals (see Rowlinson and Widom [148]) and where .κ
can be shown to be a molecular parameter related to the attractive electrostatic forces
between the fluid molecules [66, 156] (Fig. 8.5).
Therefore, in liquid–vapor interfaces, supposing that.ρ, T and.∇ρ are independent
variables
290 8 Multiphase Systems

dψ = − ρsdT + μdρ + κ∇ρ · d (∇ρ) .

.

The internal energy .u = f + T s can be written in terms of the volumetric

Helmholtz energy as

.ρu = ψ + ρT s,

and so,

ρdu = (μ + T s − u) dρ + ρT ds + κ∇ρ · d (∇ρ) .

. (8.10)

Since
ρμ − ψ
μ + Ts − u =
. ,
ρ

the difference .ρμ − ψ can be considered as a scalar pressure . p, in accordance with

Eq. (8.7). Nevertheless, this pressure differs from the thermodynamic pressure in the
interface by terms that depend on the surface force parameter .κ.
So, Eq. (8.10) can be rewritten as
p
. ρdu = dρ + ρT ds + κ∇ρ · d (∇ρ) ,
ρ

or, in terms of material derivatives,

du p dρ ds d (∇ρ)
ρ
. = + ρT + κ∇ρ · , (8.11)
dt ρ dt dt dt

giving for the material derivative of the entropy the following expression
ds ρ du p dρ κ d (∇ρ)
ρ
. = − − ∇ρ · . (8.12)
dt T dt ρT dt T dt

On the other hand, the entropy balance equation satisfies the general balance
equation
ds
ρ
. = − ∇ · qs + P (s) ,
dt

where .qs is the flux of entropy and .P (s) is the irreversible source of entropy due to
non-equilibrium.
So, from Eq. (8.12), the production of entropy in liquid–vapor phase-transition
systems can be written as,
ρ du p dρ κ d (∇ρ)
P (s) =
. + ∇ · qs − − ∇ρ · . (8.13)
T dt ρT dt T dt
8.3 The Diffuse Interface Approach 291

For the last term in the above equation,

( )
d (∇ρ) dρ dρ ( )( )
.∇ρ · = ∂β ∂β ρ − ∂β ∂β ρ − (∂α ρ) ∂β ρ ∂β vα ,
dt dt dt
or
( )
d (∇ρ) dρ dρ
.∇ρ · = ∇ · ∇ρ − ∇2ρ − ∇ρ∇ρ : ∇v. (8.14)
dt dt dt

From the mass conservation equation,

dρ
. = − ρ∇ · v (8.15)
dt
On the other hand, the internal energy balance equation can be written in terms
of the divergence of a heat flux vector and a source term given by the contraction
of the pressure tensor .P—the diffusive flux of momentum—and the gradient of the
flow velocity,
du
ρ
. = − ∇ · q − P : ∇v. (8.16)
dt
By replacing Eqs. (8.14), (8.15) and (8.16) into Eq. (8.13)

1 [ ( ) ]
P (s) = −
. P − κ∇ρ∇ρ − p − κρ∇ 2 ρ δ : ∇v + ∇ · qs
T
1 1
− ∇ · q + κ∇ · [ρ (∇ρ) (∇ · v)] . (8.17)
T T
Now,
1 (q) 1
. ∇ ·q=∇ · + 2 q · ∇T, (8.18)
T T T
and the last term on the r.h.s. of Eq. (8.17) may be written as

1 [κ ] 1
. κ∇ · [ρ (∇ρ) (∇ · v)] = ∇ · ρ (∇ρ) (∇ · v) − 2 κ [ρ (∇ρ) (∇ · v)] · ∇T.
T T T
Replacing the above equalities into Eq. (8.17),
292 8 Multiphase Systems

1 [ ( ) ]
.P (s) = − P − κ∇ρ∇ρ − p − κρ∇ 2 ρ δ : ∇v
T [ ]
1
+ ∇ · qs − (q − κρ (∇ρ) (∇ · v))
T
1
− 2 (q − κρ (∇ρ) (∇ · v)) · ∇T. (8.19)
T
The entropy production is always non-negative. In the above expression, this is
assured by making

q = − λ∇T + κρ (∇ρ) (∇ · v) ,
. (8.20)

reducing the flux of entropy to

λ∇T
q =−
. s ,
T
and eliminating the divergence second term on the r.h.s. of Eq. (8.19).
Also, since .τ : ∇v ≤ 0, the entropy production is assured to be non-negative if,
in addition to the above condition, we impose that
( )
P = τ + κ∇ρ∇ρ + p − κρ∇ 2 ρ δ.
. (8.21)

Therefore, entropy production becomes

1 1
.P (s) = − τ : ∇v + 2 λ (∇T )2 ≥ 0 (8.22)
T T

Equations (8.20) and (8.21) are sufficient conditions for .P (s) ≥ 0, but not nec-
essary. These conditions are, nevertheless, confirmed by kinetic theory.
In liquid–vapor interfaces the heat flux, Eq. (8.20), appears to be modified by an
extra term related to the density gradient in the interface. From a molecular point of
view, this term does not correspond to a flux of thermal energy, but is related to a
work produced by intermolecular forces [48].
The term

.S = κ∇ρ∇ρ,

in Eq. (8.21) was postulated by Korteweg [91] and received his name, but it is also
known as capillary tensor. This term is directly related to surface tension.
8.3 The Diffuse Interface Approach 293

8.3.2 Momentum Balance Equation

Considering Eq. (8.21) for the pressure tensor, the momentum balance equation reads
as

∂ (ρv) + ∇ · (ρvv + Ps δ) = ρg − ∇ · τ − ∇ · S,
. t (8.23)

where . Ps is a scalar pressure given by

. Ps = p − κρ∇ 2 ρ. (8.24)

Pressure . p can be found from Eq. (8.8)

( )
∂ψ
.p = ρ − ψ, (8.25)
∂ρ T

where .ψ is the Helmholtz free energy in the interface, given by

κ
ψ = ψ0 +
. (∇ρ)2 .
2
So,
( )
∂ψ0 κ
. p=ρ − ψ0 − (∇ρ)2 ,
∂ρ T 2
=P

where the thermodynamic pressure is defined as in Eq. (8.8), but with the calculations
performed with the homogeneous part of the Helmholtz free energy .ψ0 ,
( )
∂ψ0
.P = ρ − ψ0 .
∂ρ T

Therefore the scalar pressure . Ps in Eq. (8.24) becomes

( )
1
. Ps = P − κ (∇ρ) + ρ∇ ρ ,
2 2
2

in the liquid–vapor interface.

8.3.3 Summary

Table 8.1 summarizes the different forms of the momentum balance equation
294 8 Multiphase Systems

Table 8.1 Single and two-phase flows

Kind of flow Single phase Two phases
( )
Pressure . Ps=P . Ps= P − κ 21 (∇ρ)2 + ρ∇ 2 ρ
Surface tensor .S = 0 .S = κ∇ρ∇ρ

∂ (ρv) + ∇ · (ρvv + Ps δ) = ρg − ∇ · τ − ∇ · S,
. t

when written for single and two-phase flows.

Therefore, under the assumption of diffuse interfaces, the momentum balance
equation must take the liquid–vapor interface forces into account. The relationship
between the surface parameter .κ and the surface tension .γ was deduced in Sect. 6.5.4
as

γ2
. κ= , (8.26)
2I 2 n 2c Pc

where .γ is the experimental value of the surface tension evaluated at the local tem-
perature, .n c is the critical value of the molecular density [number of molecules/unit
volume], . Pc is the critical pressure, and the dimensionless parameter . I is given in
terms of the excess Helmholtz energy in the interface as
n ∗s
√
. I = ∗ dn ∗ ,
ψexc
n ∗vs

where .n ∗ = n/n c and .ψexc

∗
= ψexc /Pc .

8.3.4 Exercise

Consider a plane interface normal do the direction .x. Suppose that this interface is
in at rest (.v = 0). (a) Simplify Eq. (8.23) for this case and show that
( )2
d 2n κ dn
. P − κn + = C,
dx2 2 dx

where . P = P (x) is the thermodynamic pressure in the liquid–vapor interface. (b)

Considering .x → ± ∞ show that the constant .C is the equilibrium pressure in the
bulk phases . P∞ , which is only dependent on the temperature, and must be the same
in both phases since the interface curvature is null. (c) Define
8.3 The Diffuse Interface Approach 295
( )2
1 dn
.H = κ ,
2 dx

and show that this function satisfies

dH
.n − H = P − P∞ .
dn
(d) Show that
( )
d H 1
. = (P − P∞ (T )) .
dn n n2

(e) Integrate the above equation between a point in the vapor phase and a point in
the interface and show that the function . H can be determined from the equation of
state of the system
n
1 ( ( ' ) )
. H (n, T ) = n '2 P n , T − P∞ (T ) dn ' .
n
n v∞

(f) Using the last result and remembering that . H is null in the bulk phases, show
that
vv∞

. Pdv ' = P∞ (vv∞ − v ∞) , (8.27)

v ∞

where .v = n1 is the molecular volume and Eq. (8.27) is the Maxwell area rule.
By going back to Sect. 6.5.4 it can be concluded that function . H is, in fact, the
Helmholtz energy in excess .ψexc and that the surface tension is related to the excess
energy we have in the liquid–vapor interface, due to the unbalanced intermolecular
forces and following Eq. (8.26).

8.3.5 Internal Energy Balance Equation

The internal energy balance equation is given by Eq. (8.16), here reproduced,
du
.ρ = − ∇ · q − P : ∇v. (8.28)
dt
In the present case,
κ
ρu = ψ + ρT s = ψ0 + ρT s +
. |∇ρ|2 , (8.29)
2
=ρu 0
296 8 Multiphase Systems

showing that the internal energy in the interface has two contributions: the internal
energy per unit volume .ρu 0 from an homogeneous system and a contribution from
the surface forces . κ2 |∇ρ|2 .
Therefore,
du du 0 dρ d (∇ρ)
ρ
. =ρ + (u 0 − u) + κ∇ρ . (8.30)
dt dt dt dt
From Eq. (8.29)
κ
. (u 0 − u) = − |∇ρ|2 ,
2ρ

and so, Eq. (8.30) becomes

du du 0 κ dρ d (∇ρ)
ρ
. =ρ − |∇ρ|2 + κ∇ρ . (8.31)
dt dt 2ρ dt dt

The last term on the r.h.s. of the above equation is given by Eq. (8.14). Taking the
mass conservation equation into account, it results

du du 0 κ
.ρ =ρ + (∇ρ)2 ∇ · v − κ∇ · (ρ∇ρ∇ · v)
dt dt 2
+ κρ∇ 2 ρ∇ · v − κ∇ρ∇ρ : ∇v.

On the other hand the heat flux vector in Eq. (8.28) was deduced in accordance
with Eq. (8.20) in such a way that

. − ∇ · q = ∇ · (λ∇T ) − κ∇ · (ρ∇ρ∇ · v) . (8.32)

In the same way, using the deduced expression for the pressure tensor .P in
Eq. (8.21)
( )
. − P : ∇v = − τ : ∇v − κ∇ρ∇ρ : ∇v − p − κρ∇ 2 ρ ∇ · v.

So, in its differential form the internal energy balance equation becomes
du 0
ρ
. = ∇ · (λ∇T ) − (τ + Pδ) ∇ · v, (8.33)
dt
in the interface.
The internal energy balance equation for non-homogeneous systems takes, thus,
the same form as its related form for the bulk phases, where .∇ρ = 0. Indeed, the
surface force terms cancel themselves and do not have any direct effect on the tem-
perature distributions. These effects are only indirect, since surface forces affect the
density distributions in the interface and, consequently, the transport properties.
8.3 The Diffuse Interface Approach 297

Nevertheless, the independence of the temperature distribution on the surface

forces does not mean that the internal energy does not depend on the surface forces
in the interface. This dependence can be easily seen in Eq. (8.29).
Explicit expressions for the internal energy equation, Eq. (8.33) in terms of .ρ and
. T can be derived by using

( ) )
du 0 dT 1 ∂P dρ
. = cv + 2 P−T ,
dt dt ρ ∂ T ρ dt

where .cv is the isovolumetric specific heat capacity.

8.3.6 Numerical Methods for the Diffuse Interface Approach

Two subclasses of numerical methods for the diffuse interface approach have emerged
in literature, the first of which, also named phase-field method, considers a Cahn–
Hilliard [25] visco-capillary structure for the mixture cells. The second subclass is to
some extent an evolution of the volume of fluid method, where the classical system of
conservation laws is extended with additional phase mass conservation and volume
fraction advection equations. A very recent review of these methods is provided by
Maltsev et al. [110].
The continuous development of more rigorous molecular or atomistic models
of fluid systems, in conjunction with improvements in computational resources has
resulted in a more in-depth understanding of the underlying physics of these sys-
tems. Nevertheless, molecular simulation models have excessively large computa-
tional requirements and are limited to extremely small time and/or space scales and
simple geometries. In this context, mesoscopic models provide an alternative way for
producing numerical tools that can improve the understanding of complex physical
phenomena that are very difficult to describe at the hydrodynamic scale.
In the past two decades several numerical methods for liquid–vapor flows with
phase transitions were proposed using mesoscopic modeling, especially in the frame-
work of the Lattice Boltzmann Method (LBM), first introduced by Shan and Chen
[152, 153], in the early 1990s. Although Shan and Chen (SC) isothermal model was
not concerned with the first law of thermodynamics, there is a lack of thermodynamic
consistency related to the failure of the model in predicting the liquid–vapor coex-
istence curves in agreement with the ones derived from the thermodynamic theory.
Moreover, it was shown that the simulated coexistence curves depend on the relax-
ation time [72]. The use of this model is also limited to liquid–vapor density ratios of
about 50 [190]. Another negative aspect of the SC model is that the interface tension
and the attraction term of the equation of state are dependent on a single parameter
. G. Sbragaglia et al. [150] proposed a later modification of the Shan–Chen model by
using two independent parameters on the pseudo-potential enabling to fix this draw-
back. Other SC model drawbacks are thoroughly discussed in Philippi et al. [133].
298 8 Multiphase Systems

With the purpose of improving the Shan–Chen model, Sankaranarayanan et al.

[149] and Yan and Schaeffer [190] proposed to write the effective mass of the SC
model in terms of a given equation of state, incorporated into the model. In conse-
quence, the interface tension becomes controlled, solely, by the temperature in the
equation of state (EOS). It is observed that the maximum liquid to vapor density ratio
is strongly dependent on the equation of state that is chosen: between 7 for the van der
Walls EOS and superior to 1000 for the Carnahan Starling EOS. Nevertheless, as it
was shown by Kupershtokh et al. [94], the agreement with the Maxwell liquid–vapor
coexistence curves is only achieved by tuning up the relaxation time.
Swift et al. [167] developed a LBM model for multiphase flows based on the
free energy excess related to a strong number density gradients inside the interface.
Nevertheless, the model is not Galilean invariant [108]. As it was shown by He and
Doolen [66], the imposed condition in Swift et al.’s model, equating the equilibrium
distribution second order moments to the pressure tensor( ) was the main factor leading
to a break of Galilean invariance, with a spurious . O u 2 term in the momentum bal-
ance equation. Ad-hoc corrections to the Swift and co-workers model were proposed
and presented in Holdych et al. [70], Inamuro et al. [74] and Wagner [184].
Both pseudo-potential and free-energy models were based on ad-hoc assumptions.
The successful establishment of the theoretical foundation of the LB method in the
framework of kinetic theory has led to more rigorous ways of incorporating molecu-
lar interactions into the lattice-Boltzmann equation (LBE). The original Boltzmann
equation leads to an ideal equation of state and is only suitable for the description of
the non-equilibrium states of rarefied gases. The description of real fluids requires
the attractive forces between the molecules to be considered. He and Doolen [66]
considered that the integral term in the Boltzmann kinetic equation can be split into
a short-range repulsion term and a long-range attraction term.
A major breakthrough in LB theory was the direct derivation of the LB equation
from the continuous Boltzmann equation by He and Luo [67]. Shan et al. [154] and
Philippi et al. [132] reopened the prospect by establishing a systematic link between
the kinetic theory and the lattice Boltzmann method, determining the necessary
conditions for the discretization of the velocity space. The lattices obtained through
the method proposed by these authors, a prescribed abscissas quadrature, proved to
be stable in flows over a wide range of parameters [158], by the use of high-order
lattice Boltzmann schemes, leading to velocity sets which, when used in a discrete
velocity kinetic scheme ensures accurate recovery of the high-order hydrodynamic
moments and increasingly higher-order of isotropy of the lattice tensors.
The LB thermodynamic consistency for liquid vapor phase transitions was care-
fully investigated by Wagner [184] and Siebert et al. [156] leading to the conclusion
that high-order approximations to the stream term are necessary for assuring the
retrieval of the liquid–vapor coexistence curves. The essence is to derive LBEs from
the discrete-velocity Boltzmann equation so that the resulting leading-order error
terms are of higher-order with respect to the physical interface interaction terms.
Such high-order thermodynamic consistent LBEs were applied for simulating two-
phase liquid–vapor systems by Siebert et al. [156].
8.3 The Diffuse Interface Approach 299

Fig. 8.6 Coalescence between two liquid droplets in a vapor atmosphere in different time steps .δ
[157]

Coalescence Between Two Liquid Droplets

Figure 8.6 shows the coalescence process between two liquid droplets in a vapor
atmosphere. The numerical simulation was performed using a kinetic theory-based
LB model developed by Siebert et al. [156]. When two liquid droplets are close
enough, the molecules at the surface of each droplet are attracted by the intermolec-
ular forces from the other droplet and form a liquid neck between them. The diameter
of this neck increases with time, leading to coalescence. In accordance with Aarts et
al. [1], the increase of the neck diameter . D0 with time is dependent on the Reynolds
number . Re given, in this case, by
ργ D0
. Re = ,
2η2

where the capillary velocity is given by .u cap = γ /η, .γ is the surface tension and
η the liquid dynamical viscosity. Experimental data indicate that for . Re < 1 the
.

capillary velocity is governed by viscous forces, and the neck diameter has a linear
dependence with time, whereas for . Re > 1 the inertial forces are dominant and the
neck diameter has a .t 1/2 dependence.
Lattice-Boltzmann simulations of the inertial regime showed to be in very good
agreement with experimental data. Further details may be found in [157].
Vapor Condensation
Figure 8.7 shows what happens when a certain amount of vapor is suddenly cooled
to a temperature in such a way that the volumetric average density .<ρ> is between
the saturated vapor density .ρvs (T ) and the saturated liquid density .ρ s (T ).
The simulation was performed in two dimensions with the lattice-Boltzmann
method in the absence of gravity and considering periodic conditions at the bound-
aries. These simplifications avoid dealing with fluid–solid interactions and give a
better picture of the dynamics of droplet formation by vapor condensation.
Being a mesoscopic method, LBM is based on the discretization of the Boltzmann
equation, the attractive forces between molecules being modeled with molecular
parameters .a and .κ of Sect. 6.5. Therefore, the molecular velocity .ξ is discretized,
in such a way that, at each node, the amount of molecules in this node is distributed
300 8 Multiphase Systems

Fig. 8.7 Vapor condensation in the absence of gravity simulated with the lattice-Boltzmann method

among different populations with velocity .ξi . The simulation starts by randomly
distributing the fluid populations among the nodes (Fig. 8.7a). This gives rise to
vapor condensation in the form of liquid droplets. Some liquid droplets grow by
coalescence with their neighbors while others are unstable and reevaporate. Pressure
waves are visible during simulation (Fig. 8.7b). Figure 8.7c shows a later stage of
the simulation.
At the end of the simulation, a single liquid drop is visible with density .ρ s (T )
inside a vapor bath with density .ρvs (T ).
Chapter 9
Multicomponent Systems

Abstract Besides their technological importance in the chemical and petroleum

industry, there is a great scientific interest in studying multicomponent systems due
to the challenges to overcome when dealing with their dynamics and patterns in
non-equilibrium states. Homogeneous non-ideal mixtures may segregate into two or
even more phases when the temperature is decreased. The segregation mechanisms
have a molecular origin and are due to the attractive forces between the molecules
of the same species when these forces are dominant in comparison with diffusion
and cross-interactions. The correct understanding of these complex systems requires,
thus, a multiphysics approach and this complexity is a major difficulty for numerical
simulations based on discrete forms of the macroscopic equations. On the other
hand, the knowledge of the macroscopic equations that govern the highly non-linear
physical processes when a non-ideal mixture changes from an equilibrium state to
another is important for establishing a reference framework that every multiphysics
model must satisfy. In this chapter, these macroscopic equations are derived based
on the second law of thermodynamics. Particular attention is given to the molecular
origin of segregation. Some lattice-Boltzmann simulation results are presented at the
end of the chapter.

9.1 Introduction

From a molecular standpoint, the interface between two phases is a transition layer,
where the molecules of each fluid are subjected to two opposite processes: the random
thermal motion of the molecules, responsible for mixing, and the intermolecular
attractive forces on each molecule from its own phase, promoting segregation.
These forces are of an electrical nature and are due to the permanent and temporar-
ily asymmetric distribution of the electronic clouds around each molecule. Electrical
neutral molecules have been addressed by several workers [138], mainly based on the
ideas of Keesom, Debye, and London. The relative importance of these two compet-
ing effects is dependent on the thermodynamic state and on the molecular properties
of the involved fluids. Indeed, miscibility is not an intrinsic property of a pair of
liquids but is dependent on the thermodynamic state of the mixture. Two liquids that

© Associação Paranaense de Cultura 2024 301

P. C. Philippi, Thermodynamics, https://doi.org/10.1007/978-3-031-49357-7_9
302 9 Multicomponent Systems

Fig. 9.1 Dodecane-ethanol

phase diagram (from
Dortmund Data Bank)
critical temperature
Ethanol
10
rich
solution
Dodecane
rich solution

T ( C)
5

0
Two-phase system

0 0.2 0.4 0.6 0.8 1

Ethanol mole fraction

can mix in all proportions, forming a homogeneous solution at a given temperature

can segregate when the temperature is decreased. The amount of solute that remains
in the solution at this lower temperature is a measure of solubility and this property
is dependent on the temperature and pressure.
Figure 9.1 shows an isobaric phase diagram of a thermodynamic system com-
posed of dodecane and ethanol. When the temperature is above .13 ◦ C the two fluids
form a homogeneous solution in all proportions. Below this critical temperature, the
dodecane-ethanol mixture is unstable for intermediate molar fractions of ethanol and
suffers a spinodal decomposition breaking up into a two-phase system. This unstable
region is shown as a bell in the phase diagram. On the left side of the bell, we have
a single-phase system as a dodecane-rich solution and for thermodynamic states on
the right side of the bell, we have an ethanol-rich solution (Fig. 9.1).
Considering their great technological importance, the equilibrium properties of
real, non-ideal liquids and their mixtures were exhaustively studied in the last cen-
tury, after van der Waals first attempt to understand the thermodynamic properties
of vapors and phase transitions [117]. Equations of state and mixing rules have been
heuristically proposed and successfully validated against pressure–temperature–
volume (PVT) experiments. In the same manner, solubility and its temperature and
pressure dependence have been measured for a large number of solutions.
In non-equilibrium conditions, the macroscopic equations for non-ideal fluids and
their mixtures, in the miscibility range of these fluids, are usually written in the same
form as in the ideal case, with an appropriate equation of state for the thermodynamic
pressure. In a completely distinct framework, the flow of immiscible fluids has been
numerically approached in classical computational fluid dynamics (CFD) by the
introduction of an extra source term in the momentum balance equations, related to
the surface forces and giving the Korteweg stress tensor. These macroscopic equations
have not, however, been able to realistically describe the segregation processes that
9.2 Mixtures Without Segregation Effects 303

Fig. 9.2 Lattice Boltzmann simulation of the spinodal decomposition. The mixture forms a homo-
geneous solution at the beginning of the simulation (a). Intermolecular attraction forces between
the molecules of the same species segregate the two components forming isolated islands of one
fluid inside the other one (b), till the equilibrium is reached in (c)

occur when two partially immiscible fluids are mixed. This is a class of problems
that cannot be fully understood in the Navier–Stokes (NS) scale, which is restricted
to two extreme conditions: homogeneous non-ideal mixtures and immiscible flows.
In two-phase problems, the flow of individual drops and bubbles can be mon-
itored using a Navier Stokes Euler–Lagrange approach, but heuristic assumptions
must be used if it is necessary to describe a coalescence problem between two drops
or to decide the time at which a liquid drop will snap off when it flows through a
constriction. This does not reduce the value of continuum approaches in the study
of two-phase flows, but knowledge of the underlying physical phenomena that con-
trol the complex two-phase macroscopic phenomena require downscaling toward
molecular scales.
In this chapter, we establish the macroscopic transport equations for non-ideal
mixtures, trying to put into evidence the role of the molecular attraction forces on
segregation. Figure 9.2 shows the results of a lattice-Boltzmann simulation of a spin-
odal decomposition when two fluids . p and .q are mixed in equal proportions forming
a homogeneous solution at the beginning of the simulation (Fig. 9.2a). Intermolec-
ular attraction forces between the molecules of the same species segregate the two
components forming isolated islands of one fluid inside the other one (Fig. 9.2b) till
the equilibrium is reached in (Fig. 9.2c).

9.2 Mixtures Without Segregation Effects

In this section, we deal with mixtures where diffusion mechanisms, leading to mixing,
are dominant with respect to the attractive forces between the molecules. So, these
attractive forces do not lead to phase segregation and, consequently, to the formation
of interfaces, corresponding to the regions outside the bell in Fig. 9.1.
304 9 Multicomponent Systems

For.r -component systems in equilibrium, the Helmholtz free energy. F = U − T S

satisfies
E
.d F = − SdT − Pd V + μi dm i .
i

On the other hand, the Euler equation for homogeneous systems reads, in this
case, as
E
.F = − P V + μi m i .
i

By dividing the last equation by the volume .V

E
.ψ = − P + μi ρi , (9.1)
i

where .ψ is the Helmholtz energy per unit volume.

Multicomponent systems in equilibrium form a homogeneous mixture and satisfy
the homogeneity condition
E E
.ψ = ψi , P = Pi ,
i i

and so

ψi = − Pi + μi ρi .
.

By differentiating the last equation

.dψi = − d Pi + μi dρi + ρi dμi . (9.2)

An expression for .dμi can be found from the Gibbs–Duhem equation

E
. m i dμi = − SdT + V d P. (9.3)
i

Replacing
E E
. S= m i si , V = m i vi ,
i i

in Eq. (9.3), we get

dμi = − si dT + vi d P.
.
9.2 Mixtures Without Segregation Effects 305

Now, .vi is the partial specific volume of component .i in the mixture, or

( )
∂V 1 1
.vi = = ( ∂m i ) = .
∂m i T,P,m j/=i ∂V
ρi
T,P,m j/=i

So
1
.dμi = − si dT + d P.
ρi

Replacing the last equation into Eq. (9.2),

dψi = d (P − Pi ) + μi dρi − ρi si dT.

.

Summing the above equation over the components (remember that . P is an inten-
sive variable and, so, independent of the position in homogeneous systems)
E
.dψ = μi dρi − ρsdT. (9.4)
i

Now, we know that .U = F + T S, and so,

ρu = ψ + ρT s,
.

and so

ρdu = dψ + ρT ds + ρsdT + T sdρ − udρ.

.

By replacing .dψ given by Eq. (9.4) in the above equation, we get

E ψ
ρdu =
. μi dρi + ρT ds − dρ.
i
ρ

The Helmholtz energy is given by Eq. (9.1) and so,

E P E
.ρdu = μi dρi + ρT ds + dρ − μi ωi dρ.
i
ρ i

On the other hand,

dρi = d (ρωi ) = ωi dρ + ρdωi .

.
306 9 Multicomponent Systems

So
E P
. ρdu = μi ρdωi + ρT ds + dρ.
i
ρ

The above equation can be written in terms of Lagrangian derivatives

du E dωi ds P dρ
. ρ = μi ρ + ρT + , (9.5)
dt i
dt dt ρ dt

or
ds ρ du E μi dωi P dρ
ρ
. = − ρ − . (9.6)
dt T dt i
T dt ρT dt

Now, the entropy balance equation reads as

ds
ρ
. = − ∇ · qs + P (s)
dt

where .qs is the diffusive flux of entropy and .P (s) is the entropy production due to
non-equilibrium. Replacing the entropy balance equation into Eq. (9.6),
ρ du E μi dωi P dρ
P (s) = ∇ · qs +
. − ρ − . (9.7)
T dt i
T dt ρT dt

In the above equation, the material derivatives .ρ du

dt
, .ρ dω
dt
i
, . dρ
dt
can be replaced
by using their respective balance equations. The advection–diffusion equation for
component .i reads as
dωi
. ρ = − ∇ · ji , (9.8)
dt
where .ji is the diffusive flux of component .i. The mass conservation equation is
written as
dρ
. = − ρ∇ · v, (9.9)
dt
and the internal energy balance equation, as
du
. ρ = − ∇ · q − P : ∇v (9.10)
dt
where .q is the diffusive flux of heat and .P is the pressure tensor.
9.2 Mixtures Without Segregation Effects 307

Replacing Eqs. (9.8)–(9.10) into Eq. (9.7),

1 1 E μi
P (s) = ∇ · qs −
. ∇ · q − (P − Pδ) : ∇v + (∇ · ji ) . (9.11)
T T i
T

Now,

1 q 1
. ∇ · q = ∇ · + 2 q · ∇T
T T T
μi μi ji ji 1
(∇ · ji ) = ∇ · + μi 2 · ∇T − ji · ∇μi
T T T T
Replacing the above equations into Eq. (9.11), it results

( ( ))
1 E
.P (s) = ∇ · qs − q− μi ji
T i
( )
1 1 E E 1
− (P − Pδ) : ∇v − 2 q − μi ji · ∇T − ji · ∇μi (9.12)
T T i i
T

For the entropy production to be non-negative the divergence term must be null
and so
( )
1 E
.qs − q− μi ji = 0,
T i

and this is satisfied when

E
q = qf +
. μi ji , (9.13)
i
qf
q =
. s . (9.14)
T
The term.μi ji is related to the transfer of energy due the diffusive flux of component
i in addition to the Fourier diffusive flux of heat .q f .
.
Also, to ensure that the entropy production is non-negative, the pressure tensor
must satisfy

P = Pδ + τ ,
.

where .τ is the viscous stress tensor.

308 9 Multicomponent Systems

The entropy production will be thus given by

1 1 E 1
.P (s) = − τ : ∇v − 2 q f · ∇T − ji · ∇μi ≥ 0, (9.15)
T T i
T

as the sum of products between fluxes and the gradients that are the sources of these
fluxes.

9.2.1 Transport Equations

We rewrite the macroscopic equations for locally homogeneous mixtures without seg-
regation effects. The advection–diffusion equation for species .i is given by Eq. (9.8),
dωi
. ρ = − ∇ · ji , (9.16)
dt
where .ji is the diffusive flux of component .i.
Since
d ∂
. = + v · ∇,
dt ∂t
where .v is the local velocity at point .x, Eq. (9.16) can be rewritten in a more usual
form as
∂ρi
. + ∇ · (ρi v + ji ) = 0,
∂t
expressing the total flux of species.i as the sum of an advective flux.ρi v and a diffusive
flux .ji .
On the other hand, the momentum balance equation is given by
∂
. (ρv) + ∇ · (ρvv + P) = ρg,
∂t

where .P = Pδ + τ is the pressure tensor, . P is the thermodynamic pressure and .τ is

the viscous stress tensor.
The balance equation for the internal energy is given by Eq. (9.10), here rewritten
as
du
ρ
. = − ∇ · q − P∇ · v − τ : ∇v.
dt
9.3 Segregation Effects in Non-ideal Mixtures 309

9.2.2 Exercises

1. In a single-phase, locally homogeneous mixture with r-components, in addition to

its temperature dependence, the thermodynamic pressure is also dependent on the
composition . P = P (ρ1 , . . . , ρr , T ). In the same way the transport coefficients
are dependent on the composition and on the temperature. Simplify the transport
equations of Sect. 9.2.1 for a mixture of ideal gases.
2. Consider now a diluted solution of a component . p in a solvent .s. In this case
.ω p → 0 and the viscosity .η and thermal conductivity .λ will be only dependent
on the temperature. Simplify the transport equations of Sect. 9.2.1 for this case.

9.3 Segregation Effects in Non-ideal Mixtures

Similarly to single component systems, if .x is a point belonging to an interface, the

free energy at this point becomes also dependent on the local mass density gradients.
Indeed, following van der Waals [181], the free energy has an excess in the interface
given, in its simplest form, by
E κi j
.ψexc = ∇ρi · ∇ρ j (9.17)
i, j
2

where .κi j is related to the attraction forces between molecules of species i and j, and
are at the origin of the interface tension [133]. In the above expression .κii represents
the intermolecular attraction forces between identical molecules. These forces are
responsible for segregating species i from the other species that are present in the
multicomponent system. When .i /= j, .κi j represent the attraction forces between
different molecules. These forces have a surfactant effect, reducing the interfacial
tension.
When segregating forces are to be taken into account, Eq. (9.4) becomes
E E ( )
.dψ = μi dρi − ρsdT + κi j ∇ρi · d ∇ρ j , (9.18)
i i, j

for all the points .x where the mixture presents concentration gradients.
The last term in the above equation is symmetric with respect to the index .i and
. j because .κi j = κ ji , in such a way that changing the index .i and . j has no effect on

the surface term.

On the other hand, the internal energy .ρu = ψ + ρT s can be differentiated to
give

.ρdu = dψ + ρT ds + ρsdT + T sdρ − udρ. (9.19)

310 9 Multicomponent Systems

By replacing .dψ given by Eq. (9.18) in the above equation, we get

E ψ E ( )
.ρdu = μi dρi + ρT ds − dρ + κi j ∇ρi · d ∇ρ j . (9.20)
i
ρ i, j

The Helmholtz energy is given by Eq. (9.1)

E
.ψ = − p + μi ρi ,
i

where the pressure . p differs from the thermodynamic pressure . P by interfacial

segregating forces.
In addition,

dρi = d (ρωi ) = ωi dρ + ρdωi .

.

Therefore, writing Eq. (9.20) in terms of material derivatives

( )
du p dρ ds E d ∇ρ j E dωi
.ρ = + ρT + κi j ∇ρi · + μi ρ . (9.21)
dt ρ dt dt i, j
dt i
dt

This expression can be compared with Eq. (8.11) for single-component systems
du p dρ ds d (∇ρ)
ρ
. = + ρT + κ∇ρ · .
dt ρ dt dt dt

The last term in the r.h.s. of Eq. (9.21) is related to the effect of the diffusion of
species on the balance of energy in the interface. This term is also present in Eq. (9.5).
As in the previous sections, we rewrite Eq. (9.21) in the following form,

( )
ds ρ du p dρ 1 E dωi 1 E d ∇ρ j
.ρ = − − μi ρ − κi j ∇ρi · . (9.22)
dt T dt Tρ dt T i dt T i, j dt

A general expression for the entropy balance equation is given by

ds
ρ
. = − ∇ · qs + P (s) ,
dt
where .qs is the diffusive flux of entropy and . S (s) is the entropy production due to
non-equilibrium. Replacing the entropy balance equation into Eq. (9.22), we find the
entropy production as given by
9.3 Segregation Effects in Non-ideal Mixtures 311

ρ du P dρ 1 E dωi
P (s) = ∇ · qs +
. − − μi ρ
T dt ρT dt T i dt
( )
1 E d ∇ρ j
− ∇ρi · (9.23)
T i, j dt

For the last term in Eq. (9.23)

( ) ( )
d ∇ρ j dρ j ( )( )( )
.∇ρi · = ∂ρi ∂β − ∂α ρ j ∂β ρi ∂β vα .
dt dt

But
( ) ( ( )) ( )
( ) dρ j dρ j ( ) dρ j
. ∂β ρi ∂β = ∂β ∂β ρi − ∂β ∂β ρi .
dt dt dt

So

( ) ( ( )) ( )
d ∇ρ j dρ j ( ) dρ j ( )( )( )
.∇ρi · = ∂β ∂β ρi − ∂β ∂β ρi − ∂α ρ j ∂β ρi ∂β vα .
dt dt dt

In vector notation

( ) ( ( )) ( )
d ∇ρ j dρ j dρ j
.∇ρi · = ∇ · ∇ρi − ∇ 2 ρi − ∇ρi ∇ρ j : ∇v. (9.24)
dt dt dt

Since .ρ j = ω j ρ and
dρ j dω j dρ
. =ρ + ωj ,
dt dt dt
Equation (9.24) can thus be rewritten as

( ) ( ( )) ( )
d ∇ρ j dω j dρ dω j dρ
.∇ρi · = ∇ · ∇ρi ρ + ωj − ∇ 2 ρi ρ + ωj
dt dt dt dt dt
− ∇ρi ∇ρ j : ∇v. (9.25)

dω
The material derivatives .ρ du
dt
, .ρ dt j , . dρ
dt
can be replaced by their respective balance
equations. The advection–diffusion equation for component .i reads as
dω j
ρ
. = − ∇ · jj, (9.26)
dt
312 9 Multicomponent Systems

where .ji is the diffusive flux of component .i. The mass conservation equation is
written as
dρ
. = − ρ∇ · v, (9.27)
dt
and the internal energy balance equation, as
du
.ρ = − ∇ · q − P : ∇v, (9.28)
dt
where .q is the diffusive flux of heat and .P is the pressure tensor.
But,

1 (q) 1
. ∇ ·q=∇ · + 2 q · ∇T, (9.29)
T ( T ) T
μi μi ji ji 1
. (∇ · ji ) = ∇ · + μi 2 · ∇T − ji · ∇μi . (9.30)
T T T T

Replacing the above relationships into Eq. (9.23) we finally arrive at the following
expression for the entropy production

⎛ ⎛ E ⎞⎞
q− ~
μi ji
⎜ 1 ⎜ i ⎟⎟
. P (s) = ∇ · ⎜ ⎜
⎝qs − T ⎝ E
⎟⎟
⎠⎠
− κi j ∇ρ j (∇ · ji + ρi ∇ · v)
i, j
⎛ ⎛ ⎞ ⎞
1 ⎝ E E
− P− κi j ∇ρi ∇ρ j − ⎝ p − κi j ρi ∇ 2 ρ j ⎠ δ ⎠ : ∇v
T i, j i, j
⎛ ⎞
1 ⎝ E E
− 2 q− ~
μi ji − κi j ∇ρ j (∇ · ji + ρi ∇ · v)⎠ · ∇T
T i i, j
E 1
− ji · ∇~
μi , (9.31)
i
T

where
E
~
μi = μi −
. κi j ∇ 2 ρ j ,
j

is a modified potential that takes the surface forces into account.

9.3 Segregation Effects in Non-ideal Mixtures 313

The entropy production is always non-negative. In the above expression, this is

ensured by making
E E
.q = − λ∇T + ~
μi ji + κi j ∇ρ j (∇ · ji + ρi ∇ · v) , (9.32)
i i, j

reducing the flux of entropy to

λ∇T
q =−
. s ,
T
and eliminating the divergence term on the r.h.s. of Eq. (9.31).
The diffusive heat flux in a non-ideal mixture as expressed in Eq. (9.32) is con-
sidered to be due to the following contributions:
(i) The Fourier contribution .q f = − λ∇T due to temperature gradients;
(ii) The flux of energyE due to the diffusion of the several species that are present
in the interface . i ~
μi ji ; E
(iii) The work performed by the surface forces . i, j κi j ∇ρ j (∇ · ji + ρi ∇ · v).
Also, since .τ : ∇v ≤ 0, the entropy production is ensured to be non-negative if,
in addition to the condition given by Eq. (9.32), we impose that
⎛ ⎞
E E
.P = τ : ∇v + κi j ∇ρi ∇ρ j + ⎝ p − κi j ρi ∇ 2 ρ j ⎠ δ. (9.33)
i, j i, j

The pressure . p is given by

( )
∂ψ
. p=ρ − ψ, (9.34)
∂ρ T

where .ψ is the Helmholtz free energy in the interface,

E κi j
.ψ = ψ0 + ∇ρi · ∇ρ j .
i, j
2

So,
( ) E κi j
∂ψ0
. p=ρ − ψ0 − ∇ρi · ∇ρ j ,
∂ρ T 2
- ~- - i, j
=P

and the pressure tensor in Eq. (9.33) becomes

.P = τ : ∇v + S + Ps δ, (9.35)
314 9 Multicomponent Systems

where the scalar pressure . Ps is now given by

E κi j E
. Ps = P − ∇ρi · ∇ρ j − κi j ρi ∇ 2 ρ j . (9.36)
i, j
2 i, j

This pressure reduces to the thermodynamic pressure . P in the bulk phases.

On the other hand, the Korteweg surface tensor is given by
E
.S = κi j ∇ρi ∇ρ j . (9.37)
i, j

Considering the molecular parameters .κi j as constants, the last term on the r.h.s.
of Eq. (9.36) can be rewritten in a more symmetric form as
E 1 2( 2 ) E
. κi j ρi ∇ 2 ρ j = ∇ ρ κ − κi j ∇ρi ∇ρ j ,
i, j
2 i, j

where
E
κ=
. ωi ω j κi j ,
i, j

is a force molecular parameter averaged by the mass fractions .ωi = ρi /ρ.

Therefore, Eq. (9.36) can also be rewritten as
E κi j 1 ( )
. Ps = P + ∇ρi · ∇ρ j − ∇ 2 ρ 2 κ . (9.38)
i, j
2 2

Replacing Eqs. (9.32) and (9.33) into Eq. (9.31), the entropy production becomes
1 1 E 1
P (s) = −
. (τ : ∇v) − 2 q f · ∇T − ji · ∇~
μi . (9.39)
T T i
T

Equation (9.39) reduces to Eq. (9.15) in the absence of segregating forces.

Exercises
1. Consider Eq. (9.32) expressing the heat flux in a non-ideal mixture when the
segregation forces are taken into account. (a) Show that Eq. (9.32) reduces to
Eq. (9.13) expressing the heat flux in mixtures when the segregation forces can
be discarded; (b) Show that Eq. (9.32) reduces to Eq. (8.20) for single component
systems.
2. Show that the expression for the pressure tensor given by Eq. (9.35) reduces to
Eq. (8.21) for single component systems.
9.3 Segregation Effects in Non-ideal Mixtures 315

9.3.1 Pressure Tensor and the Equation of State

In this section, we consider a non-ideal binary mixture, for simplicity. A molecule of

species p in the interface will be subjected to an attractive potential energy .φ( ps)
related to the intermolecular interaction between this p-molecule and all the .s-
particles around it, .s = p, q. This potential energy can be calculated at a point .x
as [131, 133],
{
( ps)
.φ (x) = φ (1−2)
ps (||r − x||) n s (r) dr, (9.40)
||r−x||≥σ p

where.φ (1−2)
ps corresponds to the potential energy due to the binary interaction between
a single molecule p and a single molecules s, and .σ p is related to the molecular diam-
eter of the p-molecule. Since .φ (1−2)
ps corresponds to an attractive potential, .φ ( ps) ≤ 0.
Let .n s (r) be developed in a Taylor series around .n s (x),
1
n (r) = n s (x) + ηα ∂α n s + ηα ηβ ∂αβ n s + · · · ,
. s
2

where .η = r − x. Inserting this expansion into Eq. (9.40), given that .φ (1−2)
ps (||η||) is
an even function of .η,

φ ( ps) (x) = − 2a ps n s (x) − κ̄ ps ∇ 2 n s + · · ·

. (9.41)

where

{
1
a
. ps =− φ (1−2)
ps ||x|| dη ≥ 0,
2
||η||≥σ p
{
1
κ̄ ps = − φ (1−2)
ps (||η||) η2 dη ≥ 0,
6
||η||≥σ p

are molecular parameters that are only dependent on the interaction potential between
the molecules p and s. Parameter .κ̄ ps is the same force parameter as in the previous
section, but it is here written on a molecular basis.
The force per unit mass acting on a single molecule . p from all the molecules s
around it, will be, thus, given by
2a ps κ̄ ps
g( ps) =
. ∇n s + ∇∇ 2 n s , (9.42)
mp mp
316 9 Multicomponent Systems

and the force per unit volume acting on.n p molecules in the interface can be written as

ρ p g( ps) = 2a ps n p ∇n s + κ̄ ps n p ∇∇ 2 n s .
.

We make use of the following relationship

( )
2a ps n p ∇n s = 2a ps ∇ n p n s − 2a ps n s ∇n p ,
.

in such a way that

( )
.2a pp n p ∇n p = a pp ∇ n 2p , (9.43)

( )
. 2aqq n q ∇n q = aqq ∇ n q2 , (9.44)

( )
.2a pq n p ∇n q + 2a pq n q ∇n p = 2a pq ∇ n p n q . (9.45)

On the other hand,

( ( ( ) ) )
n p ∇∇ 2 n s + n s ∇∇ 2 n p = ∇ · ∇ 2 n p n s − ∇n p · ∇n s δ
.
( )
− 2∇ · ∇n p ∇n s .

So

1 ( ( 2( ) ) )
κ̄ n p ∇∇ 2 n p =
. pp ∇ · ∇ κ̄ pp n p n p − κ̄ pp ∇n p · ∇n p δ
2 ( )
− ∇ · κ̄ pp ∇n p ∇n p , (9.46)

1 ( ( 2( ) ) )
. qqκ̄ n q ∇∇ 2 n q = ∇ · ∇ κ̄qq n q n q − κ̄qq ∇n q · ∇n q δ
2 ( )
− ∇ · κ̄qq ∇n q ∇n q , (9.47)

( ( ( ) ) )
κ̄ n p ∇∇ 2 n q + κ̄ pq n q ∇∇ 2 n p = ∇ · ∇ 2 κ̄ pq n p n q − κ̄ pq ∇n p · ∇n q δ
. pq
( )
− ∇ · 2κ̄ pq ∇n p ∇n q . (9.48)

The total attractive force per unit volume at a given point .x of the binary mixture
is
9.3 Segregation Effects in Non-ideal Mixtures 317

E
. ρ p g( pq) = ρ p g ( pp) + ρ p g (q p) + ρq g ( pq) + ρq g (qq) , (9.49)
p,q

where we should remember that .g ( pp) is the force per unit mass acting on a single
molecule . p from all the molecules p around it, and that .g ( pq) is the force per unit
mass acting on a single molecule . p from all the molecules q around it. The same
reasoning for molecules q. Replacing Eqs. (9.43)–(9.45) and Eqs. (9.46)–(9.48) into
Eq. (9.49) we get the following relationship,

⎛⎛ ⎞ ⎞
an 2
E ⎜⎜ ⎟ ⎟
ρ p g( pq) = ∇ · ⎝⎝ + 2 ∇ κ̄n 2
1 2
. ( ( )2 ( )2 ) ⎠ δ ⎠
p,q − 1
2
κ̄ pp ∇n p + 2κ̄ pq ∇n p · ∇n q + κ̄qq ∇n q
− ∇ · S, (9.50)

where

.a = a pp x 2p + 2a pq x p xq + aqq xq2 , (9.51)

. κ̄ = κ̄ pp x 2p + 2κ̄ pq x p xq + κ̄qq xq2 , (9.52)

and

.S = κ̄ pp ∇n p ∇n p + κ̄qq ∇n q ∇n q + 2κ̄ pq ∇n p ∇n q . (9.53)

In the above equations, we recognize parameter a and .κ̄ as the van der Waals
force parameters for non-ideal mixtures. Both parameters are weighted by the molar
fractions of species p and q in the mixture. We also recognize tensor.S as the Korteweg
surface tensor written on a molecular or molar basis, following the way we consider
symbol n, as the number of molecules or the number of moles per unit volume.
Equation (9.50) can be easily generalized for an arbitrary multicomponent system,
giving

⎛⎛ ⎞ ⎞
an 2
E ⎜⎜ ⎟ ⎟
ρi g(i j) = ∇ · ⎝⎝ 1 E+ 21 ∇ 2 κ̄n 2
. ⎠ δ⎠
i, j
−2 κ̄i j ∇n i · ∇n j
i, j

− ∇ · S. (9.54)
318 9 Multicomponent Systems

It is now time to write the momentum balance equation for a mixture when it is
subjected to segregation effects. We consider the thermodynamic pressure as com-
posed of two parts,

. P = PT + Patt ,

where . PT is the pressure due to the random thermal motion of molecules corrected
by the Enskog’s volume exclusion term [30] and written on a molecular basis as
nkT
. PT = ,
1 − bn

whereas . Patt takes the attractive intermolecular forces into account. Therefore, we
can write the momentum balance equation as
∂ E
. (ρv) + ∇ · (ρvv + τ ) = ρg − ∇ · (PT δ) + ρi g(i j) .
∂t i, j

Replacing the last term of the above equation by Eq. (9.54),

∂ (( ) )
. (ρv) + ∇ · (ρvv + τ ) · = ρg − ∇ · PT − an2 δ
∂t ( (E )
1 2)
− ∇· κ̄i j ∇n i · ∇n j − ∇ 2
κ̄n δ
2 i, j
− ∇ · S,

which can be rewritten in a more compact form as

∂
. (ρv) + ∇ · (ρvv + τ ) ·
∂t
= ρg − ∇ · (Ps δ) − ∇ · S, (9.55)

where the scalar pressure . Ps is given by

1 ( E κ̄i j ∇n i · ∇n j − ∇ 2 κ̄n 2 )
. Ps = P + , (9.56)
2 i, j
and the thermodynamic pressure P is written as

nkT
. P= − an 2 .
1 − bn

Equation (9.56) has exactly the same form as Eq. (9.38). The first approach was
based on pure thermodynamic concepts, or the second law of thermodynamics, while
9.3 Segregation Effects in Non-ideal Mixtures 319

the second approach is based on the mechanical concept of force. Following this
second approach, the resulting equation of state is the one first conceived by van der
Waals.

9.3.2 Interface Physics

To get a better insight into the physics underlying a non-ideal mixture, we consider
a binary mixture with components labeled p and q in isothermal conditions. When
segregation forces are dominant over the random thermal motion of the molecules,
fluids p and q will form separated phases and an interface will be formed between
these two phases.
But,
(i) what are the physical mechanisms that sustain this interface, preventing fluids
p and q to mix?;
(ii) how to model these physical mechanisms and quantify their effects on the
interface formation?;
(iii) what are their contribution to the interface tension, or any other macroscopic
mensurable property?
These are important questions we must answer to get an in-depth understanding of
the macroscopic behavior of immiscible fluids. Nevertheless, their answers require
downscaling to the molecular scale. Using the same approach of Sect. 6.5 we can,
nevertheless, choose to work with a simplified model, considering the molecules as
material points, and, in the case of neutral molecules, these material points as sources
of electrostatic fields produced by permanent or temporary dipoles.
For a p, q binary mixture, in isothermal conditions, the diffusive flux of component
p in the interface was found by Philippi et al. [133], using a kinetic theory approach
as

| |
τ pq − nkT ∇x p + ρω p ωq g( pp) − ρω p ωq g(qq)
m −m ( ) ,
( p)
=
− ρω ω ( p q ) m g( pq) − kT ∇ [n(1 + bnχ )]
j
. (9.57)
χ p q m p mq p n

where .τ pq is a relaxation time for the collisions between molecules p and q related
to the binary diffusivity .D pq , .χ is a dimensionless parameter that takes the volume
.b of the molecules into account

1
χ=
. ,
1 − bn

m s is the mass of molecule .s, s = p, q, xs is the molecular or molar fraction of

.
component s and .ωs is the mass fraction of this component.
320 9 Multicomponent Systems

E
Equation (9.57) satisfies . s j(s) = 0.
In the right-hand side of Eq. (9.57):
(i) the first term is the usual Fickean diffusion term, responsible for mixing the
two fluids and proportional to the temperature;
(ii) the second and third terms are responsible for segregating these two fluids and
are due to the intermolecular electrostatic forces among the molecules of the
same species;
(iii) the fourth term is dependent on the molecular masses of the two involved
species, being null when .m p = m q . Since .g( pq) is related to the cross interac-
tions between . p and .q molecules, the net effect of these cross-interactions in
the interface is to be null, except when the . p and .q molecular masses are dif-
ferent. In this case, the force .g( pq) has a surfactant effect, reducing the interface
tension. The second part of this last term is proportional to the density gradient
and describes the well-known tendency of lighter molecules to be thrown away
from the regions of higher number-density.
For simplicity, consider .m p ≈ m q in such a manner that it is possible to neglect
the last term on the r.h.s. of Eq. (9.57). Consider also that component q is apolar and
that the force .g(qq) ≈ 0. When these conditions are satisfied
τ pq | |
j( p) =
. − nkT ∇x p + ρω p ωq g( pp) , (9.58)
χ

τ pq | |
j(q) =
. − nkT ∇xq − ρω p ωq g( pp) . (9.59)
χ

Therefore, component q will be subjected to a diffusive flux promoted by the force

g( pp) between molecules p, even when velocity .vq is null. This result may appear
.

somewhat strange, but we must remember that the local velocity at a given point in
the interface is given by

v = ω p v p + ωq vq = ω p v p ,
.

and that
( )
. j(q) = ρq vq − v = − ρq v,

and, so, for an observer traveling with the local velocity .v the component q at rest
will appear as moving upstream.

9.3.2.1 Interface Tension

Consider a plane interface orthogonal to the x-axis, separating two liquids . p and
q, and that this interface is at rest. In this condition, the interface tension can be
.

obtained, using
9.3 Segregation Effects in Non-ideal Mixtures 321

{∞
( )
γ ( pq) =
. Px x − Pyy d x. (9.60)
−∞

The pressure tensor is given by Eq. (9.35) and since the spherical tensor . Ps δ is
independent on the direction,
{∞
( pq)
( )
. γ = Sx x − S yy d x. (9.61)
−∞

Using Eq. (9.37), the interface tension can be written as

{∞ ( )2 {∞ ( )2
( pq) ( pp) dρ ( p) (qq) dρ (q)
γ
. =κ dx + κ dx
dx dx
−∞ −∞
{∞ ( ( p) ) ( (q) )
dρ dρ
+ κ ( pq) d x. (9.62)
dx dx
−∞

In the above equation, the molecular parameters .κ (ss) , s = p, q, are related to the
surface tensions .γ (ss) for .s = p, q, through

γ (ss)
κ (ss) = ( ) for s = p, q, (9.63)
{ ∞ ( dρ (s) )2
.

−∞ dx
dx

and the cross-parameter .κ ( pq) can then be related to the interface tension .γ ( pq) by
finding the profiles .ρ ( p) (x) and .ρ (q) (x) inside the . p − q interface.
The result given by Eq. (9.62) can be further compared with the heuristic models
for the interface tension that are widely used in the physical chemistry of surfaces
[88],

γ ( pq) = γ ( p) + γ (q) − ϒ ( pq) ,

. (9.64)

where the .ϒ ( pq) term is always positive, depends on the cross-interaction between
the molecules from species . p and .q, and has different forms in accordance with the
particular model that is being used.
322 9 Multicomponent Systems

9.3.3 Spinodal Decomposition in Multicomponent Systems

In this section, we present the numerical simulation results of a spinodal decomposi-

tion. Simulations are based on a lattice-Boltzmann model proposed by Philippi et al.
[131, 133] for multicomponent systems. For simplifying reasons, a binary mixture
of fluids p and q with identical molecular masses and critical temperatures is consid-
ered. In this case, the surfactant forces are null, as above explained. Therefore, from
Eq. (9.57), the diffusion fluxes become
τ pq | |
.j( p) = − nkT ∇x p + ρω p ωq g( pp) − ρω p ωq g(qq) , (9.65)
χ

and
τ pq | |
j(q) =
. − nkT ∇xq + ρω p ωq g(qq) − ρω p ωq g( pp) . (9.66)
χ

The binary mixture is initially homogeneous and segregation is ruled by the inter-
molecular forces .g( pp) and .g(qq) between identical molecules p and q, respectively.
In accordance with Eq. (9.42) these force are dependent on the molecular parameters
( pp)
.ass and .κ̄ss , .s = p, q, and on the gradient of concentration. Force .g try to rescue
(qq)
molecules p from the mixture to phase p, whereas force .g have the same effect
on molecules q. In the opposite sense, the Fickean diffuse forces try to mix the two
components. These forces are proportional to the temperature in such a way that,
above a certain temperature, they are dominant and there is no segregation.
Figure 9.3 shows the evolution over time of the interface length between the two
components during the segregation process for four values of the reduced temperature
. Tr . The interface length is, in theory, infinite at time .t = 0 and has an exponential

Fig. 9.3 Time evolution of

the interface length [133]
Inverse of the interface lenght (in integers of h)

-4
10

0 .80
Tr=
74
= 0.
Tr

17
Solubility (%)

15
13
11
09
0.74 0.76 0.78 0.80
Tr
-5
10
103 104
Time step (in integers of )
9.3 Segregation Effects in Non-ideal Mixtures 323

Fig. 9.4 When the temperature is below a critical value for a homogeneous mixture of two compo-
nents, the segregating forces will lead to a two-phase system by spinodal decomposition. Snapshots
are related to 500 (a), 1000 (b), 5000 (c), and 10,000 (d) time steps for a reduced temperature
. Tr = 0.74 [133]

decay with time. The exponent of decay in the first 600–10,000 time steps was
found to be 0:51 for .Tr = 0:74, and 0:43 for .Tr = 0:80. After this initial period of
time, the exponential decay is reduced and the segregation proceeds at a slower rate
toward the equilibrium configuration. For reduced temperatures above .Tr = 0:80 no
segregation was noticed in the course of the simulation time and, in some cases, the
simulations were destroyed by increasing high-frequency oscillations. For reduced
temperatures below .Tr = 0.72 the simulations were not stable for the parameters
used in the simulations. The sequence of snapshots in Fig. 9.4 visually confirms
that, in the absence of gravity, the segregation indeed proceeds through spinodal
decomposition [25] with the formation of a very complex fine geometrical structure
including isolated islands of one component inside the other. For better visualization,
Fig. 9.4 presents black and white snapshots after the true gray-scale values were
binarized.
This choice should not confuse the fact that a certain amount of component p will
always be dissolved in phase q (and vice versa). Indeed, the inset in Fig. 9.3 shows
the amount of component p (q), in terms of the temperature, that remains dissolved
in phase q (p) after the equilibrium was reached. It is clearly seen that the simulation
results are physically consistent, retrieving the tendency of solubility to increase with
the temperature.
Chapter 10
Non-equilibrium Thermodynamics
from a Kinetic Standpoint

Abstract The behavior of fluid, especially non-ideal fluids and their mixtures is
largely conditioned by physical processes that require to downscale to the molecular
level for a correct understanding of the underlying physics. Nevertheless, in numerical
simulations of molecular dynamics, the round-off errors prevent the simulations
from being performed beyond very limited spatial and time ranges. This difficulty
increases for very large systems of particles because there is no means to ascertain
the exact position and velocity of each particle at the initial time .t = 0. The best it
is possible to do is to try to prescribe the probability that this particle will have a
certain velocity when it is in a position .x at time t. This probabilistic approach led
to the Boltzmann equation and the finding that a system of particles ruled by the
deterministic Newton’s laws of motion satisfies the second law of thermodynamics.
In this chapter, we derive the Boltzmann equation starting from the Liouville equation
for large systems of particles. The Boltzmann equation is restricted to material points
considered sources of repulsion fields and collisions are binary involving, solely, a
pair of particles. It is thus limited to rarefied gases. Considering their molecular
nature and the great difficulty of the treatment of multiple collisions, the building
up of kinetic equations considering only the information of the molecular scale is
still a great scientific challenge. A kinetic model may be understood as a model that
makes use of the information from the macroscopic scale to improve the simplifying
assumptions that were required when deriving a model considering solely what is
known from the molecular scale. Kinetic models are presented in this chapter for the
collision term and non-ideal fluids and their mixtures.

10.1 Scales of Investigation

The behavior of fluids, especially non-ideal fluids and their mixtures is largely con-
ditioned by the interaction processes between the molecules of the different involved
fluids.
Restricting ourselves to neutral molecules, these interaction processes involve
intermolecular forces of an electrical nature that result from the interaction between
permanent dipoles (Keesom forces and hydrogen bonds), induced dipoles (Debye

© Associação Paranaense de Cultura 2024 325

P. C. Philippi, Thermodynamics, https://doi.org/10.1007/978-3-031-49357-7_10
326 10 Non-equilibrium Thermodynamics from a Kinetic Standpoint

Fig. 10.1 The kinetic scale Macroscopic equations for , v,T and s
as a conceptual bridge
between molecular physics Macroscopic scale
and the macroscopic domain

Statistical averages

Kinetic scale
Boltzmann equation

Probability

Molecular scale
Molecular dynamics (deterministic approach)

forces), and temporary high-frequency fluctuations of the geometric center of the

electronic clouds surrounding the molecules (London dispersive forces).
When these physical processes are important, the correct understanding of the
behavior of thermodynamic systems in non-equilibrium conditions is not possible
based only on the classical macroscopic equations, since the process is affected by
physical phenomena that develop on the molecular scale. This is, e.g., the case of the
displacement of the fluid-solid triple line, the segregation processes in multicompo-
nent systems, and the formation of emulsions.
On the other hand, the description provided by molecular physics methods is
restricted to a very small scale, measured in nanometers.
The kinetic scale may be understood as a conceptual bridge between molecular
physics and the macroscopic domain, by means of certain simplifications resulting
from the up-scaling processes of the detailed phenomena that develop at the molec-
ular level. In this chapter, we deal with the kinetic approach to non-equilibrium
thermodynamics (Fig. 10.1).
On a molecular scale, a fluid is a system of molecules whose thermodynamic prop-
erties result from molecular interactions. This is the molecular dynamics standpoint.
In its simplest classical version, these molecules are considered as material points
with an attraction-repulsion kernel ruled by Newton’s laws of classical mechanics
(Fig. 10.2).
Molecular dynamics in this classical context is, thus, a deterministic approach.
The initial state of a system of . N particles is given by the velocity .ξ i (t = 0) and
position .xi (t = 0) of each particle and the next states are found by solving the
Newton equations
10.2 The Boltzmann Equation 327

Fig. 10.2 Atraction-

repulsion field around a
single molecule
Potential
energy

Repulsion field

Attraction field

dξ i E
m
. = Fi− j , i = 1, . . . , N .
dt j/=i

In numerical simulations of molecular dynamics, the round-off errors prevent the

simulations from being performed beyond a very limited spatial and time ranges.
This difficulty increases for very large systems of particles, when . N → ∞,
because there are no means to ascertain the exact position and velocity of each
particle at the initial time .t = 0. The best it is possible to do is to try to prescribe the
probability that this particle will have a certain velocity when it is in a position .x at
time t.
As it was advanced in Sect. 2.6, this probabilistic standpoint was pursued by Clau-
sius, Maxwell, and Boltzmann in their attempts to explain the full content of the sec-
ond law of thermodynamics and the meaning of entropy, launching the foundations
of the The Mechanical Theory of Heat.
This approach led to the Boltzmann equation [15, 16] and the outstanding finding
that a system of particles ruled by the deterministic Newton’s laws of motion satisfies
the second law of thermodynamics.

10.2 The Boltzmann Equation

In this section, we show the main assumptions when the Boltzmann equation is
derived, starting from the Liouville equation [107]. The books of Chapman and
Cowling [30], Cercignani [29] and Kremer [92] are suggested for further reading
and alternative derivations.
328 10 Non-equilibrium Thermodynamics from a Kinetic Standpoint

10.2.1 The Liouville Equation

Consider a mechanical system of N particles. Let

( )
. f (N ) x1 , ξ 1 , . . . x N , ξ N , t ,

to be the joint probability of finding, at time .t, .dt the particle 1 at the position .x1 , .dx1
with velocity .ξ 1 , .dξ 1 , the particle 2 at the position .x2 , .dx2 with velocity .ξ 2 , .dξ 2 and
so on, until particle . N at the position .x N , .dx N with velocity .ξ N , .dξ N . The Liouville
equation describing the dynamical evolution of this system is given by
E E
∂ f (N ) +
. t ξ i .∂xi f (N ) + gi .∂ξi f (N ) = 0, (10.1)
i i

where .gi is the acceleration due to the force acting on particle .i,

E
N
g = gi(e) +
. i gi j .
j=1
j/=i

Force .gi(e) is related to the force on particle .i due to an external field and .gi j is the
force on particle .i due to its interaction with particle . j,
( )
1 ∂0 xi j
.gi j = − ,
m ∂xi
| |
where .xi j = |xi − x j | and .0 is the potential energy which will be supposed to be
only dependent on the distance between particles .i and . j.
The joint probability . f (N ) gives a too much detailed description of the system,
which is unnecessarily complex, since the dynamical evolution of an arbitrary, but,
single particle, can be a reliable description of the whole mechanical system of
particles, when these particles cannot be individually labeled. Therefore, probability
(N )
.f can be integrated in the phase space .x2 , ξ 2 , . . . x N , ξ N to give the marginal
probability . f (1) of finding, at time .t, .dt the particle .1 at the position .x1 , .dx1 with
velocity .ξ 1 , .dξ 1
{ {
( )
. f (1) x1 , ξ 1 , t = ··· f (N ) dx2 . . . dx N dξ 2 . . . dξ N .

After integrating the Liouville equation, considering the particles to be indistin-

guishable, Eq. (10.1) becomes
10.2 The Boltzmann Equation 329

. ∂t f (1) + ξ .∂x f (1) + g(e) .∂ξ f (1)

= − (N − 1)
{ {
( )
×∂ξ1 g12 f (2) x1 , ξ 1 , x2 , ξ 2 , t dx2 dξ 2 . (10.2)

This result imposes a solution hierarchy since the marginal probability . f (1) is
dependent on . f (2) , . f (2) is dependent on . f (3) , and so on. This hierarchy is called the
BBGKY hierarchy after the works of Born, Bogoliubov, Green, Kirkwood and Yvon
( [13, 17, 86, 87, 191]).
Considering. f = N f (1) and.(N − 1) ≈ N and changing the notation for the target
and the incident particles, the Liouville equation becomes, for large . N ,

. ∂t f + ξ .∂x f + g(e) .∂ξ f

{ {
( )
= −∂ξ · · · g12 N 2 f (2) x, ξ , x1 , ξ 1 , t dx1 dξ 1
{ {
1 ∂0 (|x1 − x|) ( )
= ∂ξ · · · × N 2 f (2) x, ξ , x1 , ξ 1 , t dx1 dξ 1 , (10.3)
m ∂x

which is a Boltzmann equation for the distribution function . f , with a collision term
[.
.
This collision term will now be split in two collision terms1 .[ = [(sd) + [(ld) ,
where .[(sd) is referred to short distance interactions, .|x1 − x| < σ and .[(ld) to
long-range interactions .|x1 − x| > σ , .σ being the distance related to the molecu-
lar diameter below which the attraction forces change to a strong repulsion among
the molecules (Fig. 10.2).
Therefore
(e)
.∂t f + ξ .∂x f + g .∂ξ f = [(sd) + [(ld) , (10.4)

where
{ {
1 ∂0 (|x1 − x|) 2 (2) ( )
.[(sd) = ∂ξ N f x, ξ , x1 , ξ 1 , t dx1 dξ 1 , (10.5)
m ∂x
|x1 −x|<σ

and
{ {
1 ∂0 (|x1 − x|) 2 (2) ( )
[(ld) = ∂ξ
. N f x, ξ , x1 , ξ 1 , t dx1 dξ 1 . (10.6)
m ∂x
|x1 −x|>σ

1 See X. He and G. D. Doolen, [66].

330 10 Non-equilibrium Thermodynamics from a Kinetic Standpoint

10.2.2 The Long-Range Term. Mean Field Theory

Consider first the long-range collision term, Eq. (10.6)

{ {
∂ 1 ∂0 (|x1 − x|) 2 (2) ( )
[(ld) =
. . N f x, ξ , x1 , ξ 1 , t dx1 dξ 1 .
∂ξ m ∂x
|x1 −x|>σ

By making the assumption that, for .|x1 − x| > σ , the molecular chaos prevails
( ) ( )
. N 2 f (2) x, ξ , x1 , ξ 1 = f (x, ξ , t) f x1 , ξ 1 , t ≡ f f 1 ,

one obtains
{ {
∂ f (x, ξ , t) 1 ∂0 (|x1 − x|) ( )
[(ld) =
. · f x1 , ξ 1 , t dx1 dξ 1
∂ξ m ∂x
|x1 −x|>σ
{
1 ∂ f (x, ξ , t) ∂
= · 0 (|x1 − x|) n (x1 , t) dx1 . (10.7)
m ∂ξ ∂x
|x1 −x|>σ

The integrand in the above equation is the mean field, i.e., the potential energy
related to the integrated action of each one of the .n molecules placed in the neigh-
borhood, on a single molecule at the position .x,
{
0m (x) =
. 0 (|x1 − x|) n (x1 , t) dx1 . (10.8)
|x1 −x|>σ

Since the acceleration .g(ld) is the gradient of the potential energy divided by its
mass,
(ld) 1 ∂0m
.g =− , (10.9)
m ∂x
we get
∂ f (x, ξ , t)
[(ld) = − g(ld) ·
. . (10.10)
∂ξ

Therefore the Boltzmann equation can be written as

( )
. t ∂ f + ξ .∂x f + g(e) + g(ld) · ∂ξ f = [(sd)
10.2 The Boltzmann Equation 331

10.2.3 The Short-Range Collision Term

Starting from Eqs. (10.2) and (10.10), the integration of the Liouville equation gives,
after dropping out the subscript index for the target particles and using the subscript
.1 for the integration variables,

( )
. ∂t f (1) + ξ · ∂x f (1) + g(e) + g(ld) · ∂ξ f (1)
= − (N − 1)
{ {
( )
×∂ξ · gx x1 · f (2) x, ξ , x1 , ξ 1 , t dx1 dξ 1 . (10.11)
|x1 −x| <σ

Similarly

∂ f (2) + ξ · ∂x f (2) + ξ 1 · ∂x f (2)

. t
( ) ( )
+ g(e) + g(ld) + gx x1 · ∂ξ f (2) + g1e + gx1 x · ∂ξ1 f (2)
{ {
( )
= −(N − 2) × ∂ξ · gx x2 f (3) x, ξ , x1 , ξ 1 , x2 , ξ 2 , t dx2 dξ 2
{ {
( )
− (N − 2) × ∂ξ1 · gx1 x2 f (3) x, ξ , x1 , ξ 1 , x2 , ξ 2 , t dx2 dξ 2 . (10.12)

The above equations and the related equations for . f (3) , f (4) ,…form the above
referred BBGKY hierarchy where the solution of each distribution . f (i) depends
on the knowledge of . f (i+1) .
For evaluating the short-distance collision term, the following assumptions are
considered [92].

• Assumption 1: Only binary collisions are considered

With this assumption the r.h.s of Eq. (10.12) is negligible and Eq. (10.12) becomes:

∂ f (2) + ξ · ∂x f (2) + ξ 1 · ∂x f (2)

. t
( )
+ g(e) + g(ld) + gx x1 · ∂ξ f (2)
( )
+ g1e + gx x1 · ∂ξ1 f (2) = 0.

• Assumption 2: The external and attractive forces are very small when compared
with the intermolecular forces during a collision

( )
. ∂t f (2) + ξ · ∂x f (2) + ξ 1 · ∂x f (2) + gx x1 · ∂ξ f (2)
( )
+ gx1 x · ∂ξ1 f (2) = 0, (10.13)
332 10 Non-equilibrium Thermodynamics from a Kinetic Standpoint

or, equivalently, considering the time.t to be the instant immediately after the collision

( )
. f (2) x, ξ , x1 , ξ 1 , t
( )
= f (2) x − ξ /t, ξ − gx /t, x1 − ξ 1 /t, ξ 1 − gx1 /t, t − /t , (10.14)

where ./t is a time interval that is supposed to be small with respect to the time
interval between any two collisions, .gx is the intermolecular force evaluated at point
.x and .gx1 is the intermolecular force evaluated at point .x1 .

• Assumption 3: Molecular chaos

. f (2) (x, ξ, x1 , ξ1 , t)
( )
= f (2) x − ξ /t, ξ − gx /t, x1 − ξ1 /t, ξ1 − gx1 /t, t − /t
= f (1) (x − ξ /t, ξ − gx /t, t − /t)
( )
× f (1) x1 − ξ1 /t, ξ1 − gx1 /t, t + /t . (10.15)

• Assumption 4: It is supposed that the distribution function does not vary between
points .x and .x1 at the beginning of the collision process.

Considering now that the time average of . f (1) in Eq. (10.11) is given by

{/t
(1) 1
. f = f (1) ds (10.16)
/t
0

we get
(1) (1) ( ) (1)
. ∂t f + ξ .∂x f + g(e) + g(ld) · ∂ξ f
(N − 1)
=
/t ⎧ | |⎫
{ { ⎨ f (1)((x − ξ /t, ξ − g/t, t − /t) ) ⎬
× × f (1)| x − ξ 1 /t, ξ 1 − g1(/t, t − )| /t dx1 dξ 1 .
⎩ ⎭
− f (1) (x, ξ , t) × f (1) x, ξ 1 , t

Since ./t is very small

x − ξ /t ≈ x − ξ 1 /t ≈ x,
. (10.17)

. t − /t ≈ t, (10.18)
10.2 The Boltzmann Equation 333

Fig. 10.3 Bullet particle

z
trajectory inside the .σ -sphere
of the target particle
b
collision plane
x-z plane

b
’ y

the collision process being considered to only modify the velocities of the target
and bullet particles, respectively, .ξ − g/t = .ξ ' ,before the collision to .ξ , after the
collision and .ξ 1 − g1 /t = ξ '1 before the collision to .ξ 1 , after the collision.
Therefore, we can rewrite the Boltzmann equation as
(1) (1) ( ) (1)
. ∂t f+ ξ .∂x f + g(e) + g(ld) .∂ξ f
{ { | (1) ( ) ( )|
(N − 1) f x, ξ ' , t f (1) (x, ξ '1 , t )
= dx1 dξ 1 , (10.19)
/t − f (1) (x, ξ , t) f (1) x, ξ 1 , t

where .ξ ' , ξ '1 are the velocities of the target and bullet particles before the collision
(at time .t − /t) that will give the velocities .ξ , .ξ 1 after the collision (Fig. 10.3).
When evaluating the right-hand side of the above equation, the position of the
target particle .x is constant. Therefore when evaluating the integral we can replace
.x1 by .x1 − x, by placing the system of coordinates on the center of the target particle.
Writing the element of volume .dη = d (x1 − x) in cylindrical coordinates
(Fig. 10.4),
1
. dη = γ bdbdε, (10.20)
/t
where .γ is the velocity of the bullet particle with respect to the target particle and
γ = ||γ ||.
.

When . N is very large . N − 1 ≈ N and, considering

(1)
. Nf (x, ξ , t) = f (x, ξ , t) , (10.21)
334 10 Non-equilibrium Thermodynamics from a Kinetic Standpoint

Fig. 10.4 Elementary

volume in cylindrical
coordinates
z

t
db
bd
b

in such a way that . f (x, ξ , t) dxdξ as the number of particles that, during the times
between .t and .t + dt, are expected to be in the position between .x and .x+ .dx with
velocity between .ξ and .ξ + .dξ ,

( )
. ∂t f + ξ .∂x f + g(e) + g(ld) .∂ξ f = [, (10.22)

where

{ { { | ( ) ( )|
(sd) f x, ξ ' , t f (x, ξ '1 , t )
[=[
. = γ bdbdεdξ 1
− f (x, ξ , t) f x, ξ 1 , t

The above equation is the Boltzmann equation as originally derived.

10.2.3.1 The Equilibrium Solution

The Boltzmann equation is an integro-differential equation and the collision term

gives the net balance between the molecules that acquire the velocity .ξ and the ones
that lose this velocity. In equilibrium conditions, this balance must be null,

[ = 0,
.

otherwise, the distribution . f would vary in the course of the time and in the space.
This equilibrium solution was found by Boltzmann and Maxwell in terms of the
equilibrium moments as [115], [15, 16]
( ) D/2
1 1 − c02
. f eq = n e 2, (10.23)
2π ξ̄ D
10.2 The Boltzmann Equation 335

√
where .ξ̄ = kT /m is related to a mean molecular speed, .k is the Boltzmann con-
stant, .c0 = ξ 0 − v0 is the dimensionless particle velocity with respect to an observer
moving with .v0 , or the peculiar velocity, .v0 = v/ξ̄ and .ξ 0 = ξ /ξ̄ are, respectively
the dimensionless local and molecular velocity, or the peculiar velocity. Symbol .n
is used for the number density of the particles with mass .m, symbol .v is used for
the local velocity, T for the thermodynamic temperature, and . D for the dimension
of the physical space, . D = 1, 2, 3 for, respectively, one, two, and three-dimensional
physical spaces.

10.2.4 Elementary Properties of the Collision Term

When integrated over all the velocity space domain, the Boltzmann collision term
multiplied by any function of .ξ gives, for each point .x and time t,
{ ({ )
| ' ' |
.I = f f 1 − f f 1 γ bdbdεdξ 1 ϕ (ξ ) dξ . (10.24)
- -- -
F(ξ )

By interchanging .ξ ↔ ξ 1 , the velocities .ξ ' and .ξ '1 will be also interchanged

because

.ξ ' = ξ + cos θ α, (10.25)

becomes (see Fig. 10.5)

ξ ' = ξ 1 − cos θ α,
. 1 (10.26)

and

ξ ' = ξ 1 − cos θ α,
. 1 (10.27)

becomes

. ξ + cos θ α = ξ ' . (10.28)

The only effect of this change in Eq. (10.24) will be

{ ({ )
| ' ' | ( )
. I1 = f f 1 − f f 1 γ bdbdεdξ ϕ ξ 1 dξ 1 (10.29)
- -- -
F (ξ 1 )
336 10 Non-equilibrium Thermodynamics from a Kinetic Standpoint

z z

b b

´
a
´
´ ´

b´ b´
´ ´

Fig. 10.5 Interchanging .ξ and .ξ 1 invert .γ and changes the sign of .α but .θ remains unchanged

( )
since the functional relation . F will be the same in both cases, although . F ξ 1 /=
F (ξ ), unless .ξ 1 = ξ .
Then, since both . I and . I1 are integrals over the whole range of the velocities
.ξ 1 or .ξ ,

I = I.
. 1 (10.30)

Consider now that we interchange .ξ ↔ ξ ' and .ξ1 ↔ ξ1' . With this transformation

.ξ ' = ξ + cos θ α, (10.31)

becomes

. ξ = ξ ' − cos θ α, (10.32)

---- ----
be f or e a f ter

and

. ξ 1 = ξ '1 + cos θ α. (10.33)

---- ----
be f or e a f ter
10.2 The Boltzmann Equation 337

In addition

. γ ' = ξ 1 − ξ = ξ '1 − ξ ' + 2 cos θ α, (10.34)

---- ---- ---- ----
be f or e be f or e a f ter a f ter
- -- -
γ

and since the only effect of this transformation is that .α changes its sign, the same
functional relation(. F is kept as)before but with a change in sign, because the amount
' '
. f f1 − f f1 = − f ' f 1' − f f 1
{ ({ )
| | ( )
. I2 = f f 1 − f ' f 1' gbdbdεdξ '1 ϕ ξ ' dξ ' = − I. (10.35)
- -- -
−F (ξ ' )

For the same reasons above

{ ({ )
| | ( )
. I3 = f f 1 − f ' f 1' gbdbdεdξ ' ϕ ξ '1 dξ '1 = I2 . (10.36)
- -- -
−F (ξ '1 )

Consequently,

{ ({ )
| '
|
. f f 1' − f f 1 gbdbdεdξ 1 ϕ (ξ ) dξ
{
1 | || |
= f ' f 1' − f f 1 ϕ + ϕ1 − ϕ ' − ϕ1' gbdbdεdξ dξ 1 . (10.37)
4

10.2.5 The H-Theorem

Consider

.ϕ = ln f,

in Eq. (10.37) . When . f ' f 1' /= f f 1 , Eq. (10.37) results in an inequality,

{ ( )
1 | | f f1
. I = f ' f 1' − f f 1 ln gbdbdεdξ dξ 1 ≤ 0,
4 f ' f 1'
338 10 Non-equilibrium Thermodynamics from a Kinetic Standpoint

since for .x > 0 and . y > 0

y
. (x − y) ln ≤ 0. (10.38)
x
We consider, now, the Boltzmann equation

. ∂t f + ξ · ∂x f + g · ∂ξ f
{ { {
| ' ' |
= f f 1 − f f 1 γ bdbdεdξ 1 . (10.39)

By multiplying the Boltzmann equation by .(1 + ln f ) and integrating it in the

velocity space we obtain:
(a) For the time derivative term

{ { {
. (1 + ln f ) ∂t f dξ = ∂t f dξ + (∂t f ) ln f dξ
{ { {
= ∂t f dξ + ∂t ( f ln f ) dξ − f ∂t (ln f ) dξ
- -- -
{ ∂t ( f )
= f f dξ

(10.40)
{
= ∂t ( f ln f ) dξ .

We define

{
.H= f ln f dξ .

(b) For the spatial gradient term

{ { {
. (1 + ln f ) ξα ∂α f dξ = ∂α ξα f dξ + ξα (∂α f ) ln f dξ
{ { {
= ∂α ξα f dξ + ξα ∂α ( f ln f ) dξ − ξα f ∂α (ln f ) dξ
- -- -
{ ∂α ( f )
= ξα f f dξ
{
= ∂α ξα ( f ln f ) dξ .
10.2 The Boltzmann Equation 339

Now we decompose .ξα into an advection velocity .vα and a peculiar velocity
ξ − vα
. α

ξ = vα + (ξα − vα ) .
. α

So

{
. ∂α ξα ( f ln f ) dξ
⎛ ⎞ ⎛ ⎞
{ ⎜{ ⎟
⎜ ⎟ ⎜ ⎟
= ∂α ⎜
⎝vα ( f ln f ) dξ ⎟ +∂
⎠ α⎝ ⎜ (ξ α − vα ) ( f ln f ) dξ ⎟ . (10.41)
⎠
- -- - - -- -
=H = jα (H )

(c) For the velocity gradient term, since the acceleration .gα is only dependent on
the spatial coordinate

{ { {
. (1 + ln f ) gα ∂ξα f dξ = gα ∂ξα f dξ + gα ln f ∂ξα f dξ .

Now, remark that

∂ ( f ln f ) = ln f ∂ξα f + ∂ξα f,
. ξα

and so

{
. (1 + ln f ) gα ∂ξα f dξ
{
= gα ∂ξα ( f ln f ) dξ (10.42)
. = 0.

Therefore, from Eqs. (10.40), (10.41) and (10.42)

∂ H + ∇ · (uH) = − ∇ · (j (H)) + S (H) ,

. t (10.43)
340 10 Non-equilibrium Thermodynamics from a Kinetic Standpoint

where {
1 | | ff
.S(H)= f ' f 1' − f f 1 ln ' 1' gbdbdεdξ 1 dξ ≤ 0. (10.44)
4 f f1

is a source of H.
Consider now that Eq. (10.43) is integrated inside a material volume .V (t). From
the Reynolds transport equation

{ { {
d
. ∂t Hd V = Hd V − Hv · nd A.
dt
V (t) V (t) A(t)

Using the divergence theorem

{ {
. ∇ · (vH) d V = Hv · nd A,
V (t) A(t)

and

{ {
. ∇ · (j (H)) d V = j (H ) · nd A,
V (t) A(t)

where . A (t) is the surface that encloses the material volume .V (t).
So, when integrated Eq. (10.43) becomes
{ { {
d
. Hd V = − j (H) · nd A + S (H) d V . (10.45)
dt
V (t) A(t) V (t)
- -- -
≤0

Equation (10.45) is the celebrated H-theorem of Boltzmann which states that for
an isolated system, when there is no flux of H through . A (t), .j (H) = 0 for .x ∈ A,
the total

{
.H= Hd V,
V (t)

decreases with time in non-equilibrium states and has a minimum when . f ' f 1' = f f 1 ,
in equilibrium states.
10.2 The Boltzmann Equation 341

This finding gave a statistical meaning to the entropy of a mechanical system of

particles. On a molecular basis, the local entropy can be written as
{
s = − k H̄ = − k
. f¯ ln f¯dξ 0 ,

where .ξ 0 = ξ /ξ̄ and . f¯ = f ξ̄ D /n are dimensionless variables.

10.2.6 Macroscopic Transport Equations

When the Boltzmann equation, Eq. (10.22), is multiplied by the mass .m of each
molecule and integrated in the velocity space we retrieve the mass conservation
equation
.∂t ρ + ∇ · (ρv) = 0, (10.46)

The momentum balance equations are retrieved as first-order moments of Eq.

(10.22)
(e)
.∂t (ρv) + ∇ · (ρvv + Pδ + τ ) = ρg , (10.47)

where . P = nkT is the thermodynamic pressure and

{
( )
τ
. αβ = m (ξα − vα ) ξβ − vβ f neq dξ ,

is the viscous stress tensor or the flux of the peculiar momentum .m (ξα − vα ) that is
transported along the direction .β by velocity fluctuations, and

. f neq = f − f eq ,

is the non-equilibrium part of the distribution . f .

The thermodynamic internal energy is the statistical average of the peculiar kinetic
energy
{ 1
< > m (ξ − v)2 f dξ
. ec, f =
2 { ,
f dξ

and the internal energy balance equation

( < >) ( < > )
∂ ρ ec, f + ∇ · ρ ec, f v + q = − τ : ∇v − P∇ · v,
. t (10.48)

may be retrieved by subtracting the balance equation for the advection energy
( ) ( )
1 2 1 2
∂
. t ρv + ∇ · ρv v + Pv + τ · v = ρg(e) · v + τ : ∇v + P∇ · v,
2 2
(10.49)
342 10 Non-equilibrium Thermodynamics from a Kinetic Standpoint

from the kinetic energy balance equation

∂ (ρ <ec >) + ∇ · (ρ <ec > v + Pv + τ · v + q) = ρg(e) · v.

. t (10.50)

In the above equations,

{ 1
2{
mξ 2f dξ
. <ec > = ,
f dξ

is the statistical average of the kinetic energy and

{
1
q=
. m (ξ − v)2 (ξ − v) f dξ,
2

is the flux of the peculiar kinetic energy transported by velocity fluctuations, corre-
sponding to the Fourier heat flow vector.

10.2.7 Exercises

1. Starting from the Liouville Eq. (10.1), try to derive Eq. (10.2), considering the
particles to be indistinguishable.
2. Maxwell [115] made a fully statistical model of gases, in which, in equilibrium,
the velocities were “distributed according to the same formula as the errors are
distributed in observations (the normal distribution)”

ξ 2f
. f eq = Ae− B ,

where .ξ f = ξ − v is the peculiar molecular velocity and . A and . B can be deter-

mined by considering that the number density of molecules is given by

{ {∞ {∞ {∞ ξ 2f,x +ξ 2f,y +ξ 2f,z

n=
. f dξ = A
eq
e− B dξx dξ y dξz ,
−∞ −∞ −∞

and that the kinetic energy related to the ‘molecular agitation’ is given by
10.2 The Boltzmann Equation 343
{
1
e
. c, f = f eq mξ 2f dξ
2
{∞ {∞ {∞ ξ 2f,x +ξ 2f,y +ξ 2f,z
1 ( 2 )
= Ae− B m ξ f,x + ξ 2f,y + ξ 2f,z dξx dξ y dξz .
2
−∞ −∞ −∞

Find . A and . B in terms of the number density .n and the mean kinetic energy of
fluctuations .ec, f . What are the independent variables of .n and .ec, f ?
ξ ξ ξ
Hint. Define new integration variables such as . √f,xB = x, √f,yB = y, √f,zB = z and
remember that

{∞ {∞
−x 2
√ 1√
e−x x 2 d x =
2
. e dx = π, π.
2
−∞ −∞

3. Boltzmann derived the equilibrium distribution in the form

f (eq) = ea e−b(ξ −c) = de−b(ξ −c) .

2 2
.

Find .d, .b and .c using

{
.n= f (eq) dξ = number density of molecules,
{
1
v= f (eq) ξ dξ = mean molecular velocity,
n
{
1 1
ec, f = f (eq) m (ξ − v)2 dξ = mean peculiar kinetic energy.
n 2

4. Consider a material volume V.(t), i.e., a volume that is composed of the same
molecules. The total number of molecules in this volume is

{
. N= n (x, t) d V,
V (t)

and the time derivative of . N is null

{
dN d
. = nd V = 0,
dt dt
V (t)
344 10 Non-equilibrium Thermodynamics from a Kinetic Standpoint

Use the Reynolds transport theorem

{ { {
d ∂n
. nd V = dV + nv · en d A
dt ∂t
V (t) V (t) A(t)

where . A (t) is the area of the surface around V.(t) and .en is the unitary vector
pointing outside . A (t), to show that
∂ρ
. + ∇ · (ρv) = 0 (10.51)
∂t
where .ρ = nm and .m is the mass of each molecule.
5. Equation (10.51) can be also written as

∂ρ
. + v · ∇ρ = − ρ∇ · v
∂t
Consider now a Taylor expansion of .ρ (x + /x, t + /t) around .ρ (x, t)

∂ρ
ρ (x + /x, t + /t) = ρ (x, t) +
. /t + ∇ρ · /x + · · ·
∂t
In the limit ./t → 0 we get

ρ (x + /x, t + /t) − ρ (x, t) ∂ρ

. = + v · ∇ρ
/t ∂t
Show that when .∇ · v = 0, the density does not vary along a fluid trajectory in
the physical space (we say that the flow is incompressible).
6. A (system of particles)is described by giving the expected value . f (x, ξ, t) =
f x, y, z, ξx , ξ y , ξz , t of the number of molecules inside an elementary volume
.dx =d xd ydz that have velocities between .ξ x and .ξ x + dξ x , .ξ y and .ξ y + dξ y , .ξz
and.ξz + dξz . Based on Exercise 4 and using words, try to give a physical meaning
for
f (x + /x, ξ + /ξ, t + /t) − f (x, ξ, t)
. lim ,
/t→0 /t

when this system is in equilibrium.

7. The pressure tensor . P can be written as the flux of the momentum .(m (ξα − v) α )
that is transported along a direction .β with the velocity fluctuation . ξβ − vβ
10.3 Kinetic Models for the Collision Term 345
{
( )
. Pαβ = f m (ξα − vα ) ξβ − vβ dξ. (10.52)

(a) Show that in equilibrium, this tensor is isotropic

. Pαβ = Pδαβ

and reduces to the thermodynamic pressure . P = nkT (the Clapeyron

equation).
(b) Under the light of Eq. (10.52) try to describe the meaning of an isotropic
tensor, using your own words.
(c) Consider a 2D case and the integral

{∞ {∞
. f eq mξ f x ξ y dξx dξ y
ξ y =0 ξ f x =− ∞

giving the momentum .mξ f x that is transported into the positive direction .ξ y
and the integral

{0 {∞
. f eq mξ f x ξ y dξx dξ y
ξ y =− ∞ ξ f x =− ∞

giving the momentum .mξ f x that is transported into the negative direction
ξ and answer the question:
. y
In equilibrium, the momentum.mξ f x is transferred along direction. y?…Justify
your answer.

10.3 Kinetic Models for the Collision Term

As originally derived, the Boltzmann equation is restricted to material points con-

sidered sources of repulsion fields and collisions are binary involving, solely, a pair
of particles. It is thus limited to rarefied gases.
Molecules are not material points and, in a liquid, the mean free path has the same
order of magnitude as the molecular diameter, multiple collisions are frequent, and
long-range interactions are important.
346 10 Non-equilibrium Thermodynamics from a Kinetic Standpoint

Considering their molecular nature and the great difficulty of the treatment of
multiple collisions, the numerical simulation of two-phase systems and solutions is
still a scientific challenge.
A kinetic model may be understood as a model that, in the reverse sense uses
the information of the macroscopic scale to improve the simplifying assumptions
that were required when deriving a model considering solely the information of the
molecular scale.

10.3.1 The BGK Collision Model

The Boltzmann equation, Eq. (10.22) is an integrodifferential equation and the eval-
uation of the collision term .[ for each point and time is a very difficult task and
never solved except for very simple problems. This gave rise to kinetic models that
replace the integral form for .[ in Eq. (10.22), by a simpler model. The more widely
used model for .[ is the Bhatnagar, Gross and Krook, BGK model [10]
f eq − f
.[= , (10.53)
τ
meaning that the Boltzmann collision term is replaced by a relaxation term with
a single relaxation time .τ . Although much simpler, this model preserves the main
effect of collisions leading the distribution of particles towards equilibrium, when a
fluid is in a non-equilibrium state.
Kinetic models with the BGK relaxation term have a single independent parame-
ter, the relaxation time .τ . In the numerical simulation of isothermal, incompressible
flow, this parameter can be related to the kinematic viscosity .υ of the fluid, but the
BGK model is unable to be used in the simulation of non-isothermal flows, except
in the case of a unitary Prandtl number.
If we consider numerical simulations performed with discrete forms of the Boltz-
mann equation (the lattice-Boltzmann method-LBM), the strongest weakness of the
BGK model is, nevertheless, that, in the course of the simulation, it very soon leads
to instability issues.
In this field, several alternatives have been proposed in the last two decades for
avoiding or reducing these instability issues.
As a first alternative, we can replace the BGK single relaxation-time collision
term with a sum of relaxation terms for the non-equilibrium moments. This method
is known as the multiple relaxation times method (MRT). This method includes
physical relaxation times related to the relaxation of the viscous stress moment and
supplementary high-order relaxation times without any physical meaning but intro-
duced into the model for increasing stability, the corresponding relaxation times
being tuned with the help of a linear stability analysis [40–42, 96, 188].
10.3 Kinetic Models for the Collision Term 347

The entropic LB scheme [84] appears as a second solution and was conceived
based on the maximization of entropy by locally tuning the single relaxation time of
the Bhatnagar–Gross– Krook (BGK) collision model. Recently, a new extension of
LB schemes was proposed, namely, the entropic stabilizer [83]. Unlike the entropic
LB scheme, the entropic stabilizer does not locally alter the viscosity but rather relies
on modifying the relaxation time for the higher-order moments (i.e., the moments
beyond the stress tensor) which do not contribute to the viscosity. In this respect,
this extension is akin to the already mentioned relaxation parameter tuning for MRT
schemes.
The stability of lattice-Boltzmann schemes can also be improved by adopting
larger discrete velocity sets. This option was investigated by Siebert et al. [158].
Another alternative is to add high-order Hermite polynomial tensors to the equilib-
rium distribution, trying to reduce the effect of their related moments on stability. This
approach is commonly used in MRT and entropic models and was utilized by Siebert
et al. [158] for improving the stability of second-order lattice-Boltzmann equations
(LBE). One can also resort to regularized LBEs, i.e., rewrite a LBE in such a manner
as to filter the undesirable ghost moments from the numerical scheme (see, e.g., [3]
for a further discussion on the ghost moments). This choice was investigated by Ladd
[95], Chen et al. [31], Zhang et al. [192] and by Latt and Chopard [100] for second-
order models; very recently this method received contributions from Mattila et al.
[113], Malaspinas [109], Mattila et al. [114], Hegele et al. [81] and Bazarin et al. [9].
In the next section, we present a method for building collision models with two
relaxation times. The main idea is the same as the MRT method, i.e., to relax each
non-equilibrium moment using independent relaxation parameters, but the purpose
is here to use this model for non-isothermal problems in such a way as to work with
independent parameters for describing the viscous and thermal diffusivities.

10.3.2 Beyond BGK

In its dimensionless form, the equilibrium distribution can be written as

f eq ξ̄ D
. f¯eq = ,
n0

where .n 0 is a number density of reference.

Writing the distribution . f in, also, a dimensionless form

f ξ̄ D
. f¯ = ,
n0
348 10 Non-equilibrium Thermodynamics from a Kinetic Standpoint

the Boltzmann equation, Eq. (10.22). becomes,

( )
∂ f¯ + ξ · ∇ f¯ + g(e) + g(ld) .∂ξ f¯ = [.
. t ¯ (10.54)

When . f¯ is near . f¯eq , we write the collision term as

( )
¯ = L f¯ ,
[
.

where .L is a linear operator.

We now write the distribution

. f¯ = f¯eq + f¯neq ,

as the sum of( an equilibrium

) and a non-equilibrium distribution.
Since .L f¯eq = 0, the collision operator becomes
( )
¯ = L f¯neq ,
[
. (10.55)

¯ = 0.
as would be expected, since the equilibrium distribution is the solution of .[
¯neq
The distribution . f can be written in terms of an orthogonal basis of the Hilbert
space .H that maps the velocity space into the positive real numbers set .R + . Due to
its cartesian symmetry, we choose the Hermite polynomials set

E aθ,(rθ )
. f¯neq = ω0 ( )2 Hθ,(rθ ) (c0 ) ,
θ Hθ,(rθ )

c02
e−
where.ω0 = 2
D is a weight function, and the Hermite polynomial tensor. Hθ,(rθ ) (c0 )
(2π) 2
is written in terms of the peculiar velocity. The symbol .(rθ ) designate a set of indices
r , r2 , . . . , rθ . For instance, . H2,x x , . H2,x y ,. H2,x z ,. H2,yy , . H2,yz , . H2,zz are all designated
. 1
by . H2,(r2 ) and, for each .θ , we use Einstein notation for repeated symbols .(rθ ). There-
fore, for .θ = 2

aθ,(rθ ) a2,αβ
. ( )2 Hθ,(rθ ) = ( )2 H2,αβ .
Hθ,(rθ ) H2,αβ

In two-dimensions

a2,αβ a2,x x a2,yy a2,x y

. ( )2 H2,αβ = H2,x x + H2,yy + H2,x y
H2,αβ 2 2 1
10.3 Kinetic Models for the Collision Term 349

The norm of . Hθ,(rθ ) is given by

{
||Hθ,(rθ ) || =
.
2
ω0 Hθ,(r
2
θ)
d c0 .

So, the Hermitian central moments .aθ,(rθ ) can be found as

{
.aθ,(rθ ) = f¯neq Hθ,(rθ ) d c0 . (10.56)

¯ can be written
On the other hand, following Eq. (10.55) the collision operator .[
as
E aθ,(rθ ) ( )
¯ =
[
. ( )2 L ω0 Hθ,(rθ ) . (10.57)
θ Hθ,(rθ )

Since .L is a linear
( operator,) it maps the Hilbert space .H onto the same Hilbert
(θ)
space .H and .L ω0 Hn x ,n y ,n z can be written in terms of the Hermitian basis of .H
( )
.L ω0 Hθ,(rθ ) = ω0 γ(rθ ),(sθ ) Hθ,(sθ ) , (10.58)

where .γ(rθ ),(sθ ) designates the .(rθ ) , (sθ ) components of a .2θ -rank relaxation tensor.
For instance,.γx x x x ,.γx x x y ,.γx yx x ,.γx x yy , and.γ yyyy for.θ = 2 in two-dimensional spaces.
These tensors can be found as

{ ( )
Hθ,(sθ ) L ω0 Hθ,(rθ ) d c0
γ
. (rθ ),(sθ ) = ,
||Hθ,(sθ ) ||2

due to the orthogonality of the Hermitian polynomials .

Replacing Eq. (10.58) into Eq. (10.57)
E aθ,(rθ )
¯ = ω0
.[ γ(rθ ),(sθ ) H . (10.59)
||H 2 θ,(sθ )
θ θ,(rθ ) ||

This sum can be rewritten as

( ∞
)
E
N
aθ,(rθ ) E aθ,(rθ )
¯ = ω0
.[ γ(rθ ),(sθ ) Hθ,(sθ ) + γ(rθ ),(sθ ) Hθ,(sθ ) .
θ=0
||Hθ,(rθ ) ||2 θ=N +1
||Hθ,(rθ ) ||2
(10.60)
Now, following Gross-Jackson [62], we diagonalize this expansion after a certain
order . N by replacing the tensor .γ(rθ ),(sθ ) by .γ N δ(rθ ),(sθ )
350 10 Non-equilibrium Thermodynamics from a Kinetic Standpoint
( N )
E( ) aθ,(rθ )
¯ = ω0
.[ γ(rθ ),(sθ ) − γ N δ(rθ ),(sθ ) Hθ,(sθ )
θ=0
||Hθ,(rθ ) ||2
⎛ ⎞
EN E∞
aθ,(sθ ) aθ,(s)
+ γ N ω0 ⎝ Hθ,(sθ ) + H
2 θ,(sθ )
⎠
θ=0
||Hθ,(rθ ) ||2 θ=N +1,s
||H θ,(r θ ) ||
θ

or

( N )
E aθ,(rθ )
¯ = −ω0
[
. λ(rθ ),(sθ ) H
2 θ,(sθ )
− λ N f¯neq (10.61)
θ=0
||Hθ,(r θ ) ||

where

( )
λ
. (rθ ),(sθ ) = − γ(rθ ),(sθ ) − γ N δ(rθ ),(sθ ) , λ N = − γ N .

The eigenvalues of .L are null for . H0 , H1,α and non-positive for .θ ≥ 2 as a

consequence of the Boltzmann H-theorem [62]. In addition, .λ(rθ ),(sθ ) = − γ(rθ ),(sθ )
for all off-diagonal components and .γ(rθ ),(rθ ) is greater in absolute value than .γ N for
the diagonal components when .θ ≤ N . Therefore,

λ
. (rθ ),(sθ ) > 0, λ N > 0,

for all .θ ≤ N .
Equation (10.61) can be considered as an Nth-order kinetic model for the collision
term, with an absorption term .λ N f¯neq resulting from the diagonalization of the
relaxation tensors after a given order N, all the moments of order higher than N
being collapsed into a single non-equilibrium term.
When N = 0 or N = 1, Eq. (10.61) gives the well-known BGK model, when all
the collision operator spectra are replaced by a single relaxation term.
Each term in the sum, Eq. (10.61), gives the relaxation to the equilibrium of second
or higher-order kinetic moments that are not preserved in collisions.

10.3.2.1 Hermitian Non-equilibrium Moments

{ }
The tensors .aθ,(rθ ) = a0 , a1,α , a2,αβ , a3,αβγ , . . . given by Eq. (10.56) are the Her-
mitian non-equilibrium moments. Both .a0 and .a1,α are null because .ρ ∗ and .v0 are
equilibrium moments,
10.3 Kinetic Models for the Collision Term 351

On the other hand, the viscous stress tensor is given by

{ {
τ
. αβ = f neq
mcα cβ d c = ρ0 ξ̄ 2
f¯neq c0,α c0,β d c0 ,

and so, the dimensionless viscous stress tensor can be written as

{
ταβ
τ
. 0,αβ = = f¯neq c0,α c0,β d c0 .
ρ0 ξ̄ 2

Now, from Eq. (10.56)

{
a
. 2,αβ = f¯neq H2,αβ d c0 = τ0,αβ . (10.62)

The heat flow vector gives the flux with which the peculiar energy is transported
by diffusion along a given direction. Its dimensionless form is given by

{ {
1 neq 2 1 neq 1
q
. 0,α = f c0 c0,α dc0 = f H3,αββ (c0 ) dc0 = a3,αββ ,
2 2 2

as a sum of non-equilibrium third-order Hermitian moments.

10.3.2.2 Collision Models with Two Relaxation Times

We focus our analysis on second-order collision models (N = 2). In Eq. (10.61)

λ a
. (r2 ),(s2 ) θ,(r2 ) Hθ,(s2 ) = λαβγ δ a2,αβ H2,γ δ .

For Newtonian fluids, we can require isotropy of the 4th-rank tensor .λαβγ δ writing
these tensors as

λ
. αβγ δ = λ1 δαβ δγ δ + λ2 δαγ δβδ + λ3 δαδ δβγ .

Therefore,

λ a
. αβγ δ 2,αβ H2,γ δ = λ1 a2,αα H2,γ γ + (λ2 + λ3 ) a2,αβ H2,αβ .
352 10 Non-equilibrium Thermodynamics from a Kinetic Standpoint

From Eq. (10.62), it is seen that the term .a2,αα is related to the trace of the viscous
stress tensor and so

a
. 2,αα = 0,

because collisions must preserve the kinetic energy.

Therefore,

1 ( )
¯ = − ω0 λμ τ0,αβ c0,α c0,β − δαβ − λ N f¯neq ,
[
. (10.63)
2
where .λμ = λ2 + λ3 .
Equation (10.63) can also be written as

1
.¯ = − ω0 λμ τ0,αβ c0,α c0,β − λ N f¯neq ,
[ (10.64)
2
because .τ0,αβ δαβ = τ0,αα or the trace of .τ0,αβ , which is null for models that are
required to preserve energy.
In two dimensions the collision model would be written as

|τ τ0,yy 2 |
¯ = −ω0 λμ c0,y + τ0,x y c0,x c0,y − λ N f¯neq .
0,x x 2
.[ c0,x + (10.65)
2 2
The present kinetic model with two independent relaxation frequencies is ther-
modynamically consistent and a Chapman-Enskog analysis showed that it is able to
analyze non-isothermal and fully compressible flows [130].

10.3.3 The Stokes Hypothesis

For Newtonian fluids satisfying the conditions of isotropy and symmetry, the rela-
tionship between the viscous stress tensor .ταβ and the strain rate tensor is given
by

( )
τ
. αβ = − μ1 ∂β vα + ∂α vβ + μ2 ∇ · vδαβ

whose trace results in

τ
. αα = − (2μ1 − μ2 D) (∂η vη ) = −χ D(∂η vη ),

where .χ is the bulk viscosity.

10.3 Kinetic Models for the Collision Term 353

For systems of particles with solely translational degrees of freedom, the trace
of the viscous stress tensor is related to the kinetic energy of fluctuations, which
is a conserved quantity. Therefore, the viscous stress tensor is traceless. This is the
Stokes hypothesis meaning that the bulk viscosity is null. In other words,

2
μ2 =
. μ1 .
D
So the viscosity coefficients .μ1 and .μ2 cannot be considered as independent
properties unless the kinetic model takes the internal degrees of freedom of the
molecules into account.
Stokes’s assumption is reasonably accurate for Newtonian fluids. So, Stokes’s
assumption is commonly taken as just another characteristic of Newtonian fluids.
If compression or expansion of a fluid is very rapid, such as a shock or sound wave,
and the molecules have internal degrees of freedom, such as vibration or rotation,
the kinetic model should take account of the several additional relaxation processes
related to the internal modes.
Hanson and Morse [63] built a kinetic model for polyatomic gases by the intro-
duction of the Wang-Chang and Uhlenbeck [185] polynomials in the Gross-Jackson
expansion. These kinetic models were extended for polyatomic mixtures by Philippi
and Brun [129].
At equilibrium, the internal energy should be distributed equally to all internal
modes of motion. However, after a rapid change of state, the energy appears first in
the translation mode, and only after several additional molecular collisions is energy
distributed to the rotation and later on, after more collisions, to the vibration modes.
In certain situations (when the relaxation time is long compared to the flow time)
the bulk viscosity coefficient can be used to model these nonequilibrium effects.
The absorption of sound waves is such a process. Sound absorption in noble gases
follows Stokes’s assumption in agreement with kinetic theory. Sound absorption in
the air has a nonzero bulk viscosity, but its value depends strongly on the water vapor
content (which greatly modifies the relaxation times).

10.3.4 Exercises

(1) Starting with the Boltzmann equation with the BGK collision term, find the mass
conservation equation and the balance equations for the momentum and internal
energy for a thermodynamic system with a single component in non-equilibrium
conditions.
(2) Repeat Exercise 1 when the collision term is replaced by the kinetic model given
by Eq. (10.65).
354 10 Non-equilibrium Thermodynamics from a Kinetic Standpoint

10.4 Non-ideal Fluids

When molecules are close, the intermolecular attraction forces .g(ld) are important
and deviate the fluid from the ideality. In this section, we derive a kinetic model for
non-ideal fluids taking into account two other additional parameters when compared
to single-component thermodynamic systems: the van der Waals bulk-phase force
parameter .a and the surface parameter .κ in the liquid-vapor transition layer. While
the first is related to the deviations of the equation of state from the ideal gas behavior,
the second is the molecular parameter for the surface tension.
Since the distribution . f is close to the equilibrium distribution . f eq , we can write
the force term in the Boltzmann equation as

( ) ( )
. g(e) + g(ld) · ∂ξ f = g(e) + g(ld) · ∂ξ f eq ,

without any effects on the macroscopic equations.

Therefore, we write the kinetic model in the form

(ξ − u) (ξ − u)
∂ f + ξ · ∂x f = [ + g(ld) ·
. t
kT
f eq + g(e) · kT
f eq , (10.66)
m m

where .g(e) corresponds to the force per unit mass due to external body forces, .g(ld)
is related to the intermolecular attraction between the molecules—which we will
suppose to be only dependent on the distance between the centers of the molecules—
and, .[, to the collision term related to the short-range repulsion forces.
The term
(ξ − v)
g(ld) ·
.
kT
f eq , (10.67)
m

related to the long-range interaction, deserves some discussion.

Consider a molecule labelled as 1 located in point .x1 and all the molecules labeled
as 2 around it (Fig. 10.6 ). The attractive potential energy related to the interaction
of all molecules 2 with the single molecule 1 is given by
{
.01 = 0(12) n (x2 ) dx2 , (10.68)
|x2 −x1 |>σ

where .0(12) is the potential energy related to the electrostatic interaction between
particle 1 and a single particle 2.
Defining .η = x2 − x1 , considering that the interaction length has molecular
dimensions and the number of particles per unit volume .n does not have strong vari-
ations, it is possible to consider .n (x2 ) = n (x1 ) inside the whole integration domain
and we obtain,
10.4 Non-ideal Fluids 355

Fig. 10.6 Attractive

potential related to the
dx2
interaction of all molecules 2
with the single molecule 1

x2

x1

01 = − 2an (x1 ) ,
. (10.69)

where
{
1
.a = − 0(12) dη, (10.70)
2
|η|>σ

is a positive constant, because .0(12) < 0. This constant depends solely on the elec-
trostatic field generated by the molecules being a molecular property of the substance
that is being analyzed.
The force .g(ld) on molecule 1 will be thus
∇01
g(ld) = −
. . (10.71)
m
Therefore, the Boltzmann equation takes the form

(ld) (ξ − v)
∂ f + ξ · ∂x f = ----
. t [ + -[-- - + g(e) · kT
f eq , (10.72)
repulsion intermolecular attraction - --
m
-
external body forces

where
2a (ξ − v) eq
[(ld) =
. ∇n · kT f , (10.73)
m m

represents the net gain of particles that acquire the velocity.ξ due to the intermolecular
forces from the other particles around it.
356 10 Non-equilibrium Thermodynamics from a Kinetic Standpoint

10.4.1 Two Phase Liquid-Vapor Systems

When a liquid has a phase transition, the interface liquid-vapor is too much thin to
enable the density .n to be considered constant throughout it, as we have done above.
Due to to the interaction between a molecule in this point and all the molecules
around it, when calculating the potential energy at a point .x1 the expression to be
used is the same as before, still given by Eq. (10.68)
{
01 =
. 0(12) n (x2 ) dx2 .
|x2 −x1 |>σ

but .n (x2 ) is, now, dependent on .|x2 − x1 |.

Nevertheless, using .η = x2 − x1 we can consider a Taylor series for .n,

∂n 1 ∂n ∂n ( )
n (x2 ) = n (x1 ) +
. (x2α − x1α ) + (x2α − x1α ) x2β − x1β + · · · ,
∂ xα - -- - 2 ∂ xα ∂ xβ - -- -- -- -
=ηα =ηα =ηβ

giving,

{
1
.01 = 2n (x1 ) 0(12) dη
2
|η|>σ
- -- -
=−a
{
∂n
+ 0(12) ηα dη
∂ xα
|η|>σ
- -- -
=0
{
1 ∂n ∂n
+ 0(12) ηα ηβ dη.
2 ∂ xα ∂ xβ
|η|>σ

The potential .0(12) depends solely on the norm of .η being an even function of
.η x , η y and .ηz . Therefore

{ {
η2
. 0(12) ηα ηβ dη = δαβ 0(12) dη.
3
|η|>σ |η|>σ
10.4 Non-ideal Fluids 357

On the other hand, in the same way as for the force parameter .a, the term
{ (12) η
2
.
|η|>σ 0 3
dη depends only on the electrical properties of the molecules, being a
constant
{ {
(12) η η2
2
. 0 dη = 2 0(12) dη = − 2κ. (10.74)
3 6
|η|>σ |η|>σ

So

01 = − 2an (x1 ) − κ∇ 2 n (x1 ) + · · · ,

.

and the intermolecular attraction force will be given by

∇01
g(ld) = −
. ,
m
or

ρg(ld) = − n∇01 .
.

So
( )
. ρg(ld) = 2an∇n + nκ∇ ∇ 2 n . (10.75)

In this case, the kinetic equation will be written as

(ld) (ξ − v)
∂ f + ξ · ∂x f = ----
. t [ + -[-- - + g(e) · kT
f eq , (10.76)
repulsion intermolecular attraction - --
m
-
external body forces

with

(ξ − v)
.[(ld) = g(ld) · kT
f (eq)
( m
)
2a κ ( 2 ) (ξ − v) (eq)
= ∇n + ∇ ∇ n · kT f . (10.77)
m m m

The momentum balance equation will be also modified. Considering that,

358 10 Non-equilibrium Thermodynamics from a Kinetic Standpoint
( ) ( ( )) | ( ( ))| ( )
. n∇ ∇ 2 n = n∂α ∂β ∂β n = ∂α n ∂β ∂β n − ∂β ∂β n ∂α n
| ( ( ))| | | ( )
= ∂α n ∂β ∂β n − ∂β ∂β n∂α n + ∂β n ∂β ∂α n, (10.78)

and
( ) 1 | |
. ∂β n ∂α ∂β n = ∂α ∂β n∂β n , (10.79)
2
we obtain

( ) 1
. ρg(ld) = 2an∇n + κ∇ n∇ 2 n + κ∇ (∇n · ∇n) − κ∇ · (∇n∇n)
|( 2 ) |
1
= ∇ · an + κn∇ n + κ (∇n · ∇n) δ − κ (∇n∇n) . (10.80)
2 2
2

The momentum balance equation will thus have the form

. ∂t (ρv) + ∇. (ρvv)
= ρg(e) − ∇P − ∇
- --· τ- − - --· S- , (10.81)
∇
- -- - ----s
external forces scalar pressure viscous forces surface forces

where the scalar pressure is given by

1
. Ps = nkT − ----
---- an 2 − κn∇ 2 n − κ∇n · ∇n , (10.82)
- 2
-- -
repulsion attraction
- -- - surface pressure
P=thermodynamic pressure

and the surface tensor by

S = κ∇n∇n.
. (10.83)

The surface forces have a role solely in the interface liquid-vapor where the
gradients are strong.
The kinetic equation

(ld) (ξ − v)
∂ f + ξ · ∂x f = ----
. t [ + -[-- - + g(e) · kT
f eq , (10.84)
repulsion intermolecular attraction - --
m
-
external body forces

does not consider that the molecules have a volume .b. In consequence, the equation
of state is given by
10.4 Non-ideal Fluids 359

. P= nkT
---- − an 2
---- , (10.85)
repulsion part attraction part

without any volume contribution.

In numerical simulation of kinetic models, it is important to consider this contri-
bution to avoid the colapse of mass2 at certain points in the course of the simulation.
We can, nevertheless, correct the kinetic equation in such a way as to retrieve the
equation of state we want. David Enskog3 was the first who considered a correction in
the Boltzmann equation by taking the molecular volume.b into account. His treatment
is given in the Chapman and Cowling book [30]. For isothermal problems, Enskog’s
correction can be written as

1
[=
. [b=0 − ∇ (nkT (bnχ )) · (ξ − v) f eq . (10.86)
- -- - nkT
repulsion field contribution
- -- -
volume contribution

When volume is attributed to the molecules, the momentum balance equation will
thus have the form

∂ (ρv) + ∇. (ρvv) = ρg(e) − ∇ Ps − ∇ · τ − ∇ · S,

. t (10.87)

where the scalar pressure is given by

1
. Ps = P − κn∇ 2 n − κ∇n · ∇n, (10.88)
2
and the thermodynamic pressure . P is here written in terms of the temperature .T and
number density .n, in accordance with a van der Waals equation of state

nkT
. P= − an 2 .
1 − bn

This suggests to write the kinetic model in a generic form as

. ∂t f + ξ · ∇ f
( ) (ξ − v) eq
= [b=0 + ρg(e) − ∇ (P − nkT ) + nκ∇∇ 2 n · f . (10.89)
nkT

2 The collapse of mass happens in a given point when the repulsion forces are weak and do not
equilibrate the attraction forces. When this happens, the numerical scheme loses stability.
3 David Enskog (22 April 1884, Västra Ämtervik, Sunne—1 June 1947, Stockholm) was a Swedish

mathematical physicist. Enskog helped to develop the kinetic theory of gases by extending the
Maxwell–Boltzmann equations.
360 10 Non-equilibrium Thermodynamics from a Kinetic Standpoint

where the relationship. P (n, T ) can be considered as related to any arbitrary equation
of state judged to be suitable for describing liquid-vapor two-phase systems.

10.4.2 Exercises

1. Using the long-range force term written as in Eq. (10.73), derive the momentum
balance equation as a first-order moment of the kinetic model given by Eq.
(10.72). What is the equation of state for this model?
2. Repeat Exercise 1 using the model given by Eq. (10.77) instead of the one given
by Eq. (10.73). Describe what are the main modifications when the momentum
balance equations are compared.
3. The momentum balance equation can be written in a generic form as

∂ (ρv) + ∇. (ρvv + P) = ρg(e) ,

. t

where the pressure tensor is given by

{
. Pαβ = f mξα ξβ dξ .

Starting from Eq. (10.89) find the momentum balance equation as the first order
moment of the kinetic model and the pressure tensor .P for a two-phase liquid-
vapor system.

10.5 Multicomponent Systems

The Boltzmann equation for a non-ideal .r -component system can be written in con-
tinuous variables as

( )
E
( p) ( p) (e) ( ps)
∂ f
. t + ξ · ∂x f + g + g · ∂ξ f ( p) = [( p) , p = 1, . . . , r,
s
(10.90)
where . f ( p) (x, ξ, t) is the probable amount of . p-particles, which at time .t, dt are
found inside the elementary volume between .x and .x + dx with velocities between
(e)
.ξ and .ξ + dξ and .g is an external body force, usually the gravity force. The short-
range collision term .[( p) means the net amount of . p-particles that acquire velocities
between .ξ and .ξ + dξ in the time interval .dt, due to the collisions they have with the
10.5 Multicomponent Systems 361

other .s-particles, .s = 1, . . . , r that are found inside .x and .x + dx, during this time
interval. The term .[( p) can be written as a sum of . p − s collision terms

E
r
( p)
[
. = [( ps) , (10.91)
s=1

In this section, a heuristic approach is adopted by writing kinetic models for .[( ps)
in such way as to retrieve the macroscopic equations and the equations of state for
the mixture and pure compounds. In fact, the kinetic level of description, leading to
discrete LB equations, may be thought of as a two-way bridge linking the molecular
and the macroscopic levels of description. Therefore, although derived from the
molecular domain and, in consequence, featuring some properties of fluid interfaces
that are inaccessible to the macroscopic level of description, the kinetic model may
be written in such a manner as to recover the correct macroscopic equations. Along
this line of approach, the repulsion term is written considering the Enskog’s volume
correction [30].
| 2 |
∇ bn kT
1−bn
( ps) ( ps) (s) ( p)
.[ = [b=0 − x · (ξ − v) f eq (v) , (10.92)
nkT
where
E
b=
. x (s) b(s) , (10.93)
s

the parameter .b(s) is related to the volume of .s-particles, .x (s) is the number fraction of
(s) (s) ( p)
. s-particles, . x = nn , . f eq (v) is the Maxwell-Boltzmann equilibrium distribution,
written around the center of mass velocity,
E
.v = ω(s) v(s) , (10.94)
s

(s) ( ps)
ω(s) is the mass fraction of species .s, .ω(s) = ρρ . The term .[b=0 represents the short-
.

range repulsion term when the particles are considered without volume, .b = 0, and
must satisfy the following conservation restrictions,
{
( ps)
. m ( p) [b=0 dξ = 0, (10.95)

meaning that the mass of . p−species is not affected in their interaction with the other
species,
E{ ( ps)
. m ( p) ξ [b=0 dξ = 0, (10.96)
s, p
362 10 Non-equilibrium Thermodynamics from a Kinetic Standpoint

and
E{ 1 ( ps)
. m ( p) ξ 2 [b=0 dξ = 0, (10.97)
s, p
2

related, respectively, to the conservation of momentum and kinetic energy in short-

range interactions of material points.
The force term
( ps)
∇0m
g( ps) = −
. (10.98)
m ( p)
is the force per unit mass on the . p-particles resulting from their intermolecular
interaction with the .s-particles. This force is related to the mean potential energy
( ps)
.0m ,

{
( ps)
.0m = n (s) (x + η) 0( ps) (|η|) dη
|η|>σ ps
| |
= − 2a ( ps) n (s) (x) − κ ( ps) ∇ 2 n (s) (x) − · · · , (10.99)

with .σ ps being the distance where the two-particles . p-.s interaction potential .0( ps)
changes from repulsion to attraction,
{
( ps) 1
.a =− 0( ps) (|η|) dη. (10.100)
2
|η|>σ ps

and
{
( ps) 1
κ
. =− η2 0( ps) (|η|) dη. (10.101)
6
|η|>σ ps

Therefore
| |
.ρ ( p) g( ps) = 2a ( ps) n ( p) ∇n (s) (x) + κ ( ps) n ( p) ∇∇ 2 n (s) (x) (10.102)

( p)
Finally, using . f ( p) = f eq (v) for the external force term in Eq. (10.90), and Eq.
(10.92), the following kinetic model is obtained
10.5 Multicomponent Systems 363

∂ f ( p) + ξ.∂x f ( p)
. t
E ( ps)
= [b=0
s
( ( ) )
bn 2 kT E (ξ − v) ( p)
( p) (e) ( p) ( p) ( ps)
+ ρ g −x ∇ + ρ g · f eq (v) . (10.103)
1 − bn s
n ( p) kT

It is important to stress that the purpose of a kinetic model is not to solve the full
Boltzmann equation itself, which is, in fact, unknown, but to consistently retrieve the
macroscopic equations describing, in the present case, the non-ideal behavior of a
mixture. Therefore, numerically solving a kinetic equation, must, firstly, be thought
of as a method for solving a physical problem. In the reverse sense, kinetic models
enable to reveal, or put in evidence, the influence of a number of molecular processes
on the macroscopic behavior of a physical system.

10.5.1 Macroscopic Equations

The first moment of Eq. (10.103) can be found as

( )
. t∂ ρ ( p) + ∇ · ρ ( p) v( p) = 0, p = 1, . . . , r, (10.104)

which is the advection-diffusion equation for species . p.

After summing over all the components, the second moment of Eq. (10.103) is
found to be

∂ (ρv) + ∇ · (ρvv) = ρg(e) − ∇ · P,

. t (10.105)

the pressure tensor .P satisfies

| E |
nkT
.∇ · P = ∇ · τ + ∇ − ρ ( p) g( ps) , (10.106)
1 − bn s, p

and the viscous stress tensor is given by,

E{
.τ = m ( p) (ξ − v) (ξ − v) f neq
( p)
dξ, (10.107)
p

( p)
with the non-equilibrium distribution . f neq given by
( p)
f ( p) = f ( p) − f eq
. neq (v) . (10.108)
364 10 Non-equilibrium Thermodynamics from a Kinetic Standpoint

The pressure tensor is found as

P = τ + Ps δ + S,
. (10.109)

where . Ps is a scalar pressure, which, in the present multicomponent case, is to be

written as
| |
1 ( ) E
. Ps = P − ∇ 2 κn 2 − κ ( ps) ∇n ( p) · ∇n (s) , (10.110)
2 p,s

and .S is the Korteweg stress tensor [91] that can be written as

E
.S = κ ( ps) ∇n ( p) ∇n (s) , (10.111)
p,s

with
E
κ=
. x ( p) x (s) κ ( ps) . (10.112)
p,s

The scalar pressure. Ps differs from the thermodynamic pressure. P, by terms whose
action is restricted to physical regions where the number density .n (s) , .s = 1, . . . , r
have strong gradients, i.e., in the interfaces. In the bulk phases, without gradients,
. Ps = P and .S = 0.
Equation (10.105) could be also written as

∂ (ρv) + ∇ · [ρvv + τ + Ps δ] = ρg(e) − ∇ · S,

. t (10.113)

putting in evidence the role of the non-spherical interface tension tensor .S in the
mechanical properties of interfaces.
The first term on the right-hand side of Eq. (10.109) is a scalar pressure. In
the macroscopic approach, the scalar pressure is made equal to the thermodynamic
pressure . P and the force resulting from the Korteweg stress tensor is considered a
source term in the momentum equation with two components: a radial component,
proportional to the interface tension and to the interface curvature and a tangential
component, the Marangoni force. In the present kinetic approach, these forces are
given in terms of the molecular properties of the involved fluids.
10.5 Multicomponent Systems 365

10.5.2 Using a Reverse Standpoint to Find the Microscopic

Parameters and the Intermolecular Force

The equation of state is a van der Waals equation of state

nkT
. P= − an 2 , (10.114)
1 − bn

with the force parameter .a given by

E
a=
. x ( p) x (s) a ( ps) , (10.115)
p,s

and the volume parameter .b by Eq. (10.93).

Adopting a reverse standpoint, Eq. (10.85) can be used to find the intermolecular
force .g( ps) for any given cubic equation of state . P = P(n, T ). Indeed, the kinetic
model, Eq. (10.103), was derived considering a number of assumptions: molecules
were first considered as material points and although the Enskog’s volume exclu-
sion theory leads to the correct repulsion part of the thermodynamic pressure, the
molecular shapes were not taken into account.
In this case, the equation of state suitable for the problem is supposed to be known
and the inverse problem consists in finding the intermolecular force.g( ps) that retrieves
this equation of state.
Therefore, it is considered that, in the interface, the pressure tensor .P is still given
by Eq. (10.109) as a sum of a scalar pressure . Pδ and the Korteweg stress tensor .S,
but it is supposed that the thermodynamic pressure . P in Eq. (10.110) follows a given
and known equation of state. Therefore, from Eq. (10.85) it results

( ( ) )
E | ( P − 1−bn
nkT
|
. ρ ( p) ( ps)
g = −∇ ) E − ∇ · S.
− 21 ∇ 2 κn 2 − p,s κ ( ps) ∇n ( p) · ∇n (s)
s, p
(10.116)
Consider, for instance, that the mixture is well-described by a cubic equation of
state,

nkT an 2
. P= − , (10.117)
1 − bn f (bn)

where . f (bn) is a dimensionless function of the volume term .bn, with a quadratic
mixing rule for the force parameter .a, as given by Eq. (10.115).
The force parameter .a (ss) for each .s-component is dependent on the temperature
and on an acentric factor .ω(s) which is a measure of the non-sphericity of the .s-
molecules. The cross parameters .a ( ps) , with . p /= s, is given as a geometric average
of the intrinsic force parameters, weighted by a cross interaction-term .ς ( ps) ,
366 10 Non-equilibrium Thermodynamics from a Kinetic Standpoint

√
. a ( ps) = ς ( ps) a ( pp) a (ss) , (10.118)

in such a manner that when . p = s, .ς ( ps) = 1.

It can be easily found the intermolecular force .g( ps) in Eq. (10.103) must satisfy

( )
n ( p) n (s) κ ( ps) | ( p) |
.ρ ( p) g( ps) = a ( ps) ∇ + n ∇∇ 2 n (s) + n (s) ∇∇ 2 n ( p) , (10.119)
f (bn) 2

where the following relationship was used

.n ( p) ∇∇ 2 n (s) + n (s) ∇∇ 2 n ( p)
(| ( ) | )
= ∇ · ∇ 2 n ( p) n (s) − ∇n ( p) · ∇n (s) -δ − 2∇n ( p) ⊗ ∇n (s) . (10.120)

10.5.3 Exercises

1. Consider a system of two components . p and .q. Using Eq. (10.102) and the
relationship given by Eq. (10.120) show that

E
. ρ ( p) g( ps)
p,s
⎛ ( ) ⎞
( an 2 + 21 ∇ 2 κn 2 )
=∇⎝ κ ( pp) ∇n ( p) · ∇n ( p) + 2κ ( pq) ∇n ( p) · ∇n (q) ⎠
− 21 ∇
+κ (qq) ∇n (q) · ∇n (q)
−∇ · S

where
( )2 ( ) ( )2
. a = a ( pp) x ( p) + 2a ( pq) x ( p) x (q) + a (qq) x (q) ,

is the force term in the equation of state,

( )2 ( ) ( )2
κ = κ ( pp) x ( p) + 2κ ( pq) x ( p) x (q) + κ (qq) x (q) ,
.

is the surface force term, and

( ) ( )
S = κ ( pp) ∇n ( p) ∇n ( p) + 2κ ( pq) ∇n ( p) ∇n (q) + κ (qq) ∇n (q) ∇n (q) ,
.

is the Korteweg surface tensor.

10.5 Multicomponent Systems 367

2. By replacing the above expression for

E
. ρ ( p) g( ps) ,
p,s

into Eq. (10.85) show that the scalar pressure is given by Eq. (10.110) and that
the equation of state is the one of van der Waals.
Appendix
The Euler–Lagrange Equations

The problem we wanted to solve in Sect. 6.5.4 was to find .ψexc (n, T ) in such a way
as to minimize the integral
∞
2
κ dn
. exc = ψexc (n, T ) + d x, (A.1)
2 dx
−∞

knowing that .ψexc is null outside the liquid–vapor transition layer or that .ψexc → 0
when .x → ± ∞ and the number density of molecules .n → n vs when .x → − ∞ and
.n → n s when . x → + ∞.

Let us first consider the more general problem of finding the function . y (x) that
makes the integral
b

. S= f y, y , x d x, (A.2)
a

an extremum, when . y (x) is subjected to the conditions . y (a) = ya and . y (b) = yb .

In June 1696, Johann Bernoulli, the father of Daniel Bernoulli, had used the pages
of the Acta Eruditorum Lipsidae to pose a challenge to the international mathematical
community: to find the form of the curve joining two fixed points so that a mass will
slide down along it, under the influence of gravity alone, in the minimum amount of
time. This problem was called the brachistochrone problem, from the Greek words
brákhistos (shortest) and khrónos (time), and was solved by Newton based on the
second law of mechanics.
In 1733, Euler elaborated on the brachistochrone problem using a geometrical
approach but a few years later, in 1755, Joseph-Louis Lagrange, at the age of 19 years
old, studying a problem named the tautochrone problem, which is a variation of the
brachistochrone problem, published a paper explaining an analytical method for ana-
lyzing such a problem. Euler was so impressed that he drops his current method and
switches to Lagrangian method which was purely analytical. In 1756, Euler pub-

©Associação Paranaense de Cultura 2024 369

P. C. Philippi, Thermodynamics, https://doi.org/10.1007/978-3-031-49357-7
370 Appendix: The Euler–Lagrange Equations

lished a paper on the Calculus of Variations based on Lagrange’s research, trying

to extend it to a much wider range of problems. Then, in 1788, Lagrange published
his Mécanique Analytique, a classical mechanics paper that, in contrast with New-
ton’s mechanics, consider mechanical problems as minimization problems, propos-
ing what is presently known as Lagrangian Mechanics. Fifty years later, in 1834,
Willian Rowan Hamilton extended Lagrangian’s approach to problems in electro-
magnetism and quantum theory, launching the foundations of the principle of the
least action, considered by some physicists to be the most fundamental principle of
physics.
Going back to Eq. (A.2), consider a family of curves

. Y (x) = y (x) + g (x) ,

in such a way that .g (a) = g (b) = 0. Therefore, if we replace . y (x) by .Y (x) in

Eq. (A.2), the action . S becomes dependent on . ,

. S( ) = f Y, Y , x d x,
a

and since we want to find the function . y (x) that makes . S an extremum,

dS
. = 0,
d
when . = 0.
On the other hand

b
dS ∂ f ∂Y ∂ f ∂Y
. = + d x,
d ∂Y ∂ ∂Y ∂
a

or
b
dS ∂f ∂f
. = g+ g d x.
d ∂Y ∂Y
a
Appendix: The Euler–Lagrange Equations 371

The second term of the above equation may be integrated by parts

b b
b
∂f ∂f d ∂f
. g dx = g − gd x,
∂Y ∂Y a dx ∂Y
a a

and, since .g (a) = g (b) = 0,

b b
∂f d ∂f
. g dx = − gd x.
∂Y dx ∂Y
a a

Therefore
b
dS ∂f d ∂f
. = − gd x.
d ∂Y dx ∂Y
a

When . = 0, .d S/d = 0 and .Y (x) becomes the function . y (x) we want to find

b
∂f d ∂f
. − gd x = 0.
∂y dx ∂y
a

Since .g (x) is an arbitrary function, this integral is null when and only when

∂f d ∂f
. − =0 (A.3)
∂y dx ∂y

Equation (A.3) is the Euler–Lagrange equation, which is the basis of Lagrangian

mechanics when the integrand . f is replaced by the Lagrangian . L, defined as the
difference between the kinetic . E c and the potential energy . E p of a particle whose
position and velocity are, respectively, given by the generalized coordinates .x and
·
. x = d x/dt in the phase-space,

·
. L = E c x − E p (x) ,
372 Appendix: The Euler–Lagrange Equations

and the independent variable is the time,

b
·
. S= L x, x, t dt.
a

In some cases we do not have an explicite dependence of the integrand . f y, y

on the coordinate .x and the second term on the l.h.s. of Eq. (A.3) may be written as

d ∂f ∂2 f ∂2 f
. = y + y , (A.4)
dx ∂y ∂ y∂ y ∂y 2

because . y and . y are dependent on .x.

Replacing Eq. (A.4) into Eq. (A.3) we obtain

∂f ∂2 f ∂2 f
. − y − y = 0. (A.5)
∂y ∂ y∂ y ∂y 2

Now, consider the function

∂f
. h (x) = y − f.
∂y

When this function is differentiated it gives

dh ∂2 f ∂2 f ∂f
. =y 2 y + y − = 0.
dx ∂y ∂ y∂ y ∂y
=0

Therefore, we get the following equation as a particular case of the Euler–

Lagrange equation

∂f
. y − f = C,
∂y

named Beltrami Identity because it was firstly derived by Eugênio Beltrami in 1868
[120].

References

1. D. G. A. L. Aarts, H. N. W. Lekkerkerker, H. Guo, G. H. Wegdam, and D. Bonn, Hydrody-

namics of droplet coalescence, Physical Review Letters, 95 (2005), p. 164503.
2. A. W. Adamson and A. P. Gast, Physical Chemistry of Surfaces, Wiley, 1997.
Appendix: The Euler–Lagrange Equations 373

3. R. Adhikari and S. Succi, Duality in matrix lattice Boltzmann models, Physical Review E -
Statistical, Nonlinear, and Soft Matter Physics, 78 (2008), pp. 1–9.
4. P. Adler, Porous Media: Geometry and Transports, Butterworth-Heinemann, 1992.
5. T. Andrews, The Bakerian lecture: On the continuity of the gaseous and liquid states of matter,
Philosophical Transactions of the Royal Society of London, 159 (1869), pp. 575–590.
6. C. R. Appoloni, A. Macedo, C. P. Fernandes, and P. C. Philippi, Characterization of porous
microstructure by X-ray microtomography, X-Ray Spectrometry, 31 (2002), pp. 124–127.
7. R. Aris, Vectors, tensors, and the basic equations of fluid mechanics, Dover Publications,
1989.
8. F. Bashfort, An attempt to test the theories of capillary action by comparing the theoretical
and measured forms of drops of fluid, Cambridge University Press, 1883.
9. R. Bazarin, P. Philippi, A. Randles, and L. H. Junior, Moments-based method for boundary
conditions in the lattice Boltzmann framework: A comparative analysis for the lid driven
cavity flow, Computers & Fluids (2021), p. 105142.
10. P. L. Bhatnagar, E. P. Gross, and M. Krook, A model for collision processes in gases. I. Small
amplitude processes in charged and neutral one-component systems, Physical Review, 94
(1954), pp. 511–525.
11. P. Bigdelou, C. Liu, P. Tarey, and P. Ramaprabhu, An efficient ghost fluid method to remove
overheating from material interfaces in compressible multi-medium flows, Computers & Flu-
ids, 233 (2022), p. 105250.
12. J. Black, Lectures on the Elements of Chemistry delivered at the University of Edinburgh,
Mundell and Son, 1803.
13. N. Bogoliubov, “Kinetic equations” (in Russian), Journal of Experimental and Theoretical
Physics, 16 (1946), pp. 691–702.
14. L. Boltzmann, Weitere studien uber das warmegleichgewicht unter gasmolekun, Sitzung-
berichte der Akademie der Wissenschaften zu Wien, mathematisch-naturwissenschaftliche
Klasse, 66 (1872), pp. 275–370.
15. L. Boltzmann, Vorlesungen uber Gas Theory, Part I, J.A. Barth, 1896.
16. L. Boltzmann, Vorlesungen uber Gas Theorie. Part II., J.A. Barth, 1898.
17. M. Born and H. Green. A general kinetic theory of liquids I. The molecular distribution
functions, Proc. Roy. Soc. A, 188 (1946), pp. 10–18.
18. C. Bosanquet, Lv. on the flow of liquids into capillary tubes, The London, Edinburgh, and
Dublin Philosophical Magazine and Journal of Science, 45 (1923), pp. 525–531.
19. R. Boyle, New experiments Physico-Mechanical touching the spring of the air and its effects,
H. Hall, 1660.
20. A. Braslau, M. Deutsch, P. S. Pershan, A. H. Weiss, J. Als-Nielsen, and J. Bohr, Surface
roughness of water measured by X-ray reflectivity, Physical Review Letters, 54 (1985), pp.
114–117.
21. S. Brunauer, P. H. Emmett, and E. Teller, Adsorption of gases in multimolecular layers, Journal
of the American Chemical Society, 60 (1938), pp. 309–319.
22. G. Buresti, A note on Stokes’ hypothesis, Acta Mechanica, 226 (2015), pp. 3555–3559.
23. J. N. Butler and B. H. Bloom, A curve-fitting method for calculating interfacial tension from
the shape of a sessile drop, Surface Science, 4 (1966), pp. 1–17.
24. J. W. Cahn, Critical point wetting, The Journal of Chemical Physics, 66 (1977), p. 3667.
25. J. W. Cahn and J. E. Hilliard, Free energy of a nonuniform system. I. Interfacial free energy,
The Journal of Chemical Physics, 28 (1958), pp. 258–267.
26. C. Caratheodory, Untersuchungen ber die grundlagen der thermodynamik, Mathematische
Annalen, 67 (1909), pp. 355–386.
27. S. Carnot, Réflexions sur la puissance motrice du feu, Bachelier, 1824.
28. H. Cavendish, Three papers, containing experiments on factitious air, by the hon. Henry
Cavendish, f. r. s., Philosophical Transactions of the Royal Society of London, 56 (1766), pp.
141–184.
29. C. Cercignani, Mathematical Methods in Kinetic Theory, London: Macmillan, 1969.
374 Appendix: The Euler–Lagrange Equations

30. S. Chapman and T. Cowling, The Mathematical Theory of Non-uniform Gases, Cambridge
University Press, 1970.
31. H. Chen, R. Zhang, I. Staroselsky, and M. Jhon, Recovery of full rotational invariance in lattice
Boltzmann formulations for high Knudsen number flows, Physica A: Statistical Mechanics
and its Applications, 362 (2006), pp. 125–131.
32. A. K. Chesters, An analytical solution for the profile and volume of a small drop or bubble
symmetrical about a vertical axis, Journal of Fluid Mechanics, 81 (1977), pp. 609–624.
33. B. E. Clapeyron, Memoire sur la puissance motrice de la chaleur, Journal de l’Ecole Royale
Polytechnique, 23 (1834), pp. 153–190.
34. R. Clausius, Ueber die bewegende kraft der warme und die gesetze, welche sich daraus fur die
warmelehre selbst ableiten lassen, Annalen der Physik und Chemie, 155 (1850), pp. 500–524.
35. R. Clausius, Ueber eine veranderte form des zweiten hauptsatzes der mechanischen warmethe-
orie, Annalen der Physik und Chemie, 169 (1854), pp. 481–506.
36. R. Clausius, Abhandlungen uber die mechanische Warmetheorie (On the mechanical theory
of heat), John van Voorst, 1867.
37. J. P. Dalton, Researches on the Joule–Kelvin effect, especially at low temperatures. I. Calcu-
lations for hydrogen, Proceedings of the Royal Netherlands Academy of Arts and Sciences,
11 (1909), pp. 863–873.
38. J. H. de Boer and C. Zwikker, Adsorption als folge von polarization, Z. physik. Chem. B, 3
(1929), pp. 407–418.
39. S. de Groot and P. Mazur, Non-Equilibrium Thermodynamics, Dover Publications, 1984.
40. P. J. Dellar, Incompressible limits of lattice Boltzmann equations using multiple relaxation
times, Journal of Computational Physics, 190 (2003), pp. 351–370.
41. D. D’Humieres, Generalized lattice-Boltzmann equations, 1 1994.
42. D. D’Humieres, I. Ginzburg, M. Krafczyk, P. Lallemand, and L.-S. Luo, Multiple-relaxation-
time lattice Boltzmann models in three dimensions, Philosophical transactions. Series A,
Mathematical, physical, and engineering sciences, 360 (2002), pp. 437–451.
43. L. O. E. dos Santos, F. G. Wolf, and P. C. Philippi, Dynamics of interface displacement in
capillary flow, Journal of Statistical Physics, 121 (2005), pp. 197–207.
44. M. Dreyer, A. Delgado, and H.-J. Path, Capillary rise of liquid between parallel plates under
microgravity, Journal of Colloid and Interface Science, 163 (1994), pp. 158–168.
45. P. Duhem, Le potentiel thermodynamique et ses applications, A. Hermann, 1886.
46. P. Duhem, Josiah Willard Gibbs, á propos de la publication de ses mémoires scientifiques, A.
Hermann, 1908.
47. P. Duhem, Traié d’énergétique ou de thermodynamique générale (2 vols.), Gauthier-Villars,
1911.
48. J. Dunn, Interstitial working and a nonclassical continuum thermodynamics, 1986.
49. C. Eckart, The thermodynamics of irreversible processes. I. The simple fluid, Physical Review,
58 (1940), pp. 267–269.
50. C. Eckart, The thermodynamics of irreversible processes. II. Fluid mixtures, Physical Review,
58 (1940), pp. 269–275.
51. S. Fechter, F. Jaegle, and V. Schleper, Exact and approximate Riemann solvers at phase
boundaries, Computers & Fluids, 75 (2013), pp. 112–126.
52. S. Fechter, C.-D. Munz, C. Rohde, and C. Zeiler, A sharp interface method for compressible
liquid-vapor flow with phase transition and surface tension, Journal of Computational Physics,
336 (2017), pp. 347–374.
53. S. Fechter, C.-D. Munz, C. Rohde, and C. Zeiler, Approximate Riemann solver for compress-
ible liquid vapor flow with phase transition and surface tension, Computers & Fluids, 169
(2018), pp. 169–185.
54. R. P. Fedkiw, T. Aslam, and S. Xu, The ghost fluid method for deflagration and detonation
discontinuities, Journal of Computational Physics, 154 (1999), pp. 393–427.
55. C. Fernandes, F. Magnani, P. Philippi, J. Daïan, C. Fernandas, F. Magnani, P. Philippi, and
J. Daian, Multiscale geometrical reconstruction of porous structures, Physical Review E, 54
(1996), pp. 1734–1741.
Appendix: The Euler–Lagrange Equations 375

56. J. S. Fernandes, C. R. Appoloni, and C. P. Fernandes, Microstructural characterization of

siltstone and sandstone pore space by X-ray microtomography, Soils and Rocks, 42 (2019),
pp. 281–288.
57. A. Fredenslund, R. L. Jones, and J. M. Prausnitz, Group-contribution estimation of activity
coefficients in nonideal liquid mixtures, AIChE Journal, 21 (1975), pp. 1086–1099.
58. L. J. Gay-Lussac, L’expansion des gaz par la chaleur, Annales de Chimie, 43 (1802), p. 137.
NULL.
59. J. Gibbs, On the equilibrium of heterogeneous substances. Part i, Transactions of the Con-
necticut Academy of Arts and Sciences, 3 (1875), pp. 108–248.
60. J. Gibbs, On the equilibrium of heterogeneous substances. Part ii, Transactions of the Con-
necticut Academy of Arts and Sciences, 3 (1877), p. 343–524.
61. J. W. Gibbs, A method of geometrical representation of the thermodynamic properties of
substances by means of surfaces, Transactions of the Connecticut Academy of Arts and
Sciences, 2 (1873), pp. 382–404.
62. E. P. Gross and E. A. Jackson, Kinetic models and the linearized Boltzmann equation, Physics
of Fluids, 2 (1959), pp. 432–441.
63. F. B. Hanson and T. F. Morse, Kinetic models for a gas with internal structure, The Physics
of Fluids, 10 (1967), pp. 345–353.
64. F. Hauksbee, An account of an experiment touching the ascent of water between two glass
planes, in an hyperbolick figure by Mr. Francis Hauksbee, f. r. s., Philosophical Transactions
of the Royal Society of London, 27 (1710), pp. 539–540.
65. R. D. Hazlett, Simulation of capillary-dominated displacements in microtomographic images
of reservoir rocks, Transport in Porous Media, 20 (1995), pp. 21–35.
66. X. He and G. D. Doolen, Thermodynamic foundations of kinetic theory and lattice Boltzmann
models for multi-phase flows, Journal of Statistical Physics, 107 (2002), pp. 309–328.
67. X. He and L.-S. Luo, Theory of the lattice Boltzmann method: From the Boltzmann equation
to the lattice Boltzmann equation, Physical Review E, 56 (1997), pp. 6811–6817.
68. H. Helmholtz, On the conservation of force, Transactions of the Physical Society of Berlin
(1847), pp. 114–162.
69. W. Henry, Experiments on the quantity of gases absorbed by water, at different temperatures,
and under different pressures, Philosophical Transactions of the Royal Society of London, 93
(1803), pp. 29–274.
70. D. J. Holdych, D. Rovas, J. G. Georgiadis, and R. O. Buckius, An improved hydrodynamics
formulation for multiphase flow lattice-Boltzmann models, International Journal of Modern
Physics C, 09 (1998), pp. 1393–1404.
71. I. K. Howard, H is for enthalpy, thanks to Heike Kamerlingh Onnes and Alfred W. Porter,
Journal of Chemical Education, 79 (2002), p. 697.
72. H. Huang, M. Krafczyk, and X. Lu, Forcing term in single-phase and Shan-Chen-type mul-
tiphase lattice Boltzmann models, Physical Review E, 84 (2011), p. 046710.
73. F. S. V. Hurtado and P. C. Philippi, Thermodynamic analysis of the heating-dissipation process
inside a cavity (technical report), 2005.
74. T. Inamuro, N. Konishi, and F. Ogino, A Galilean invariant model of the lattice Boltzmann
method for multiphase fluid flows using free-energy approach, Comput. Phys. Comm., 129
(2000), pp. 32–45.
75. G. Jaumann, Die Grundlagen der Bewegungslehre, Barth Verlag, 1905.
76. G. Jaumann, Geschlossenes System physikalischer und chemischer Diffrerentialgesetze, K. k.
Hof- und Staatsdruckerei, 1911.
77. M. Y. Joshi, A class of stochastic models for porous media, 1974.
78. J. Joule, On the changes of temperature produced by the rarefaction and condensation of air,
Philosophical Magazine Series 3, 26 (1845), pp. 369–383.
79. J. P. Joule, On the heat evolved by metallic conductors of electricity, and in the cells of a
battery during electrolysis, Philosophical Magazine Series 3, 19 (1841), pp. 260–277.
80. J. P. Joule, On the mechanical equivalent of heat, Philosophical Transactions of the Royal
Society of London, 140 (1850), pp. 61–82.
376 Appendix: The Euler–Lagrange Equations

81. L. A. H. Junior, A. Scagliarini, M. Sbragaglia, K. K. Mattila, P. C. Philippi, D. F. Puleri,

J. Gounley, and A. Randles, High-Reynolds-number turbulent cavity flow using the lattice
Boltzmann method, Physical Review E, 98 (2018), p. 043302.
82. J. Jurin, An account of some new experiments, relating to the action of glass tubes upon
water and quicksilver by James Jurin, M. D. Reg. Soc. & Coll. Med. Lond. Soc., Philosophical
Transactions of the Royal Society of London, 30 (1717), pp. 1083–1096.
83. I. V. Karlin, F. Bösch, and S. S. Chikatamarla, textitGibbs’ principle for the lattice-kinetic
theory of fluid dynamics, Physical Review E, 90 (2014), p. 031302.
84. I. V. Karlin, A. Ferrante, and H. C. Öttinger, Perfect entropy functions of the lattice Boltzmann
method, Europhysics Letters (EPL), 47 (1999), pp. 182–188.
85. K. Kinosita and H. Yokota, Temperature dependence of the optical surface thickness of water,
Journal of the Physical Society of Japan, 20 (1965), pp. 1086–1086.
86. J. G. Kirkwood, The statistical mechanical theory of transport processes I. General theory,
The Journal of Chemical Physics, 14 (1946), p. 180.
87. J. G. Kirkwood, The statistical mechanical theory of transport processes II. Transport in
gases, The Journal of Chemical Physics, 15 (1947), p. 72.
88. J. Kloubek, Development of methods for surface free energy determination using contact
angles of liquids on solids, Advances in Colloid and Interface Science, 38 (1992), pp. 99–
142.
89. G. M. Kontogeorgis, I. V. Yakoumis, H. Meijer, E. Hendriks, and T. Moorwood, Multicompo-
nent phase equilibrium calculations for water-methanol-alkane mixtures, Fluid Phase Equi-
libria, 158–160 (1999), pp. 201–209.
90. K. G. Kornev and A. V. Neimark, Spontaneous penetration of liquids into capillaries and
porous membranes revisited, Journal of Colloid and Interface Science, 235 (2001), pp. 101–
113.
91. D. Korteweg, Sur la forme que prennent les équations du mouvement des fluides si l’on
tient compte des forces capillaires causées par des variations de densité considérables mais
continues et sur la théorie de la capillarité dans l’hypothese d’une variation continue de la
densité, Archives Néerlandaises des Sciences Exactes et Naturelles, 2 (1901), pp. 1–24.
92. G. M. Kremer, An Introduction to the Boltzmann Equation and Transport Processes in Gases,
Springer Berlin Heidelberg, 2010.
93. P. Kundu, I. Cohen, and D. Dowling, Fluid Mechanics, Academic Press, 6th ed., 2015.
94. A. Kupershtokh, D. Medvedev, and D. Karpov, On equations of state in a lattice Boltzmann
method, Computers & Mathematics with Applications, 58 (2009), pp. 965–974.
95. A. J. C. Ladd, Numerical simulations of particulate suspensions via a discretized Boltzmann
equation. Part 1. Theoretical foundation, Journal of Fluid Mechanics, 271 (1994), pp. 285–
309.
96. P. Lallemand and L.-S. Luo, Theory of the lattice Boltzmann method: Acoustic and thermal
properties in two and three dimensions, Physical Review E, 68 (2003), p. 036706.
97. I. Langmuir, The constitution and fundamental properties of solids and liquids, Journal of the
American Chemical Society, 39 (1917), pp. 1848–1906.
98. I. Langmuir, The adsorption of gases on plane surfaces of glass, mica and platinum, Journal
of the American Chemical Society, 40 (1918), pp. 1361–1403.
99. P. Laplace, Traite de Mecanique Celeste, Courcier, 1805.
100. J. Latt and B. Chopard, Lattice Boltzmann method with regularized pre-collision distribution
functions, Mathematics and Computers in Simulation, 72 (2006), pp. 165–168.
101. A. Lavoisier, Memoire sur la chaleur lu a l’academie royale des sciences, le 28 juin 1783,
Memoires de l’ Academie Royale des Sciences (1783).
102. A. H. Lefebvre and V. G. McDonell, Atomization and Sprays, CRC Press, 3 2017.
103. A. M. Legendre, Mémoire sur l’intégration de quelques équations aux différences partielles,
Mémoires de l’ Académie des Sciences (1787), pp. 309–351.
104. S. Levine, J. Lowndes, E. J. Watson, and G. Neale, A theory of capillary rise of a liquid in
a vertical cylindrical tube and in a parallel-plate channel, Journal of Colloid and Interface
Science, 73 (1980), pp. 136–151.
Appendix: The Euler–Lagrange Equations 377

105. G. N. Lewis, The law of physico-chemical change, Proceedings of the American Academy of
Arts and Sciences, 37 (1901), p. 49.
106. Z. Liang, C. Fernandes, F. Magnani, and P. Philippi, A reconstruction technique for three-
dimensional porous media using image analysis and Fourier transforms, Journal of Petroleum
Science and Engineering, 21 (1998), pp. 273–283.
107. J. Liouville, Sur la theorie de la variation des constantes arbitraires, Journal de mathématiques
pures et appliquées, 3 (1838), pp. 342–349.
108. L. Luo, Unified theory of lattice Boltzmann models for nonideal gases, Physical Review
Letters, 81 (1998), pp. 1618–1621.
109. O. Malaspinas, Increasing stability and accuracy of the lattice Boltzmann scheme: recursivity
and regularization, arXiv:1505.06900v1 [physics.flu-dyn] 26 May 2015 (2015), pp. 1–31.
110. V. Maltsev, M. Skote, and P. Tsoutsanis, High-order methods for diffuse-interface models in
compressible multi-medium flows: A review, Physics of Fluids, 34 (2022), p. 021301.
111. E. Mariotte, Essais de Physique, ou mémoires pour servir á la science des choses naturelles,
E. Michallet, 1679.
112. M. Matsumoto and Y. Kataoka, Study on liquid-vapor interface of water. I. Simulational
results of thermodynamic properties and orientational structure, The Journal of Chemical
Physics, 88 (1988), p. 3233.
113. K. K. Mattila, L. A. H. Junior, and P. C. Philippi, Investigation of an entropic stabilizer for
the lattice-Boltzmann method, Physical Review E, 91 (2015), p. 063010.
114. K. K. Mattila, P. C. Philippi, and L. A. H. Junior, High-order regularization in lattice-
Boltzmann equations, Physics of Fluids, 29 (2017), p. 046103.
115. J. C. Maxwell, On the dynamical theory of gases, Philosophical transactions of the Royal
Society, 157 (1867), pp. 49–88.
116. J. C. Maxwell, Molecules, Nature (1873), pp. 437–441.
117. J. C. Maxwell, On the dynamical evidence of the molecular constitution of bodies, Nature, 11
(1875), pp. 357–359.
118. A. C. Moreira, C. P. Fernandes, M. V. de Oliveira, A. A. Ribeiro, I. F. Mantovani, L. R.
Medeiros, K. K. Mattila, and R. Lyra, Evaluating the performance of truncated Gaussian
method for three-dimensional reconstruction of the pore microstructure of titanium scaffolds,
Journal of Porous Media, 24 (2021), pp. 31–44.
119. I. Muller, A History of Thermodynamics, Springer-Verlag Berlin Heidelberg, 2007.
120. P. J. Nahim, When Least is Best, Princeton University Press, 2003.
121. L. Natanson, Uber die gezetze nicht umkehrbarer vorgange, Z. phys. Chem. (1896), p. 193.
122. L. Onsager, Reciprocal relations in irreversible processes. i, Physical Review, 37 (1931), pp.
405–426.
123. L. Onsager, Reciprocal relations in irreversible processes. ii, Physical Review, 38 (1931), pp.
2265–2279.
124. H. Orbey and J. H. Vera, Correlation for the third virial coefficient using .Tc , . Pc and .ω as
parameters, AIChE Journal, 29 (1983), pp. 107–113.
125. S. Osher and J. A. Sethian, Fronts propagating with curvature-dependent speed: Algorithms
based on Hamilton-Jacobi formulations, Journal of Computational Physics, 79 (1988), pp.
12–49.
126. J. F. Padday, The profiles of axially symmetric menisci, Philosophical Transactions of the Royal
Society A: Mathematical, Physical and Engineering Sciences, 269 (1971), pp. 265–293.
127. R. Panton, Incompressible Flow, John Wiley and Sons, 2013.
128. D.-Y. Peng and D. B. Robinson, A new two-constant equation of state, Industrial & Engineering
Chemistry Fundamentals, 15 (1976), pp. 59–64.
129. P. Philippi and R. Brun, Kinetic modeling of polyatomic gas mixtures, Physica A: Statistical
Mechanics and its Applications, 105 (1981), pp. 147–168.
130. P. Philippi, L. H. Junior, R. Surmas, D. Siebert, and L. dos Santos, From the Boltzmann to the
lattice-Boltzmann equation: beyond BGK collision models, International Journal of Modern
Physics C, 18 (2007), pp. 556–565.
378 Appendix: The Euler–Lagrange Equations

131. P. C. Philippi, L. O. E. dos Santos, L. A. H. Junior, C. E. P. Ortiz, D. N. Siebert, and R. Surmas,

Thermodynamic consistency in deriving lattice Boltzmann models for describing segregation
in non-ideal mixtures, Philosophical transactions of the Royal Society A, 369 (2011), pp.
2292–300.
132. P. C. Philippi, L. A. H. Júnior, L. D. Santos, and R. Surmas, From the continuous to the lattice
Boltzmann equation: The discretization problem and thermal models, Physical Review E, 73
(2006), p. 056702.
133. P. C. Philippi, K. K. Mattila, D. N. Siebert, L. O. E. dos Santos, L. A. H. Júnior, and R. Surmas,
Lattice-Boltzmann equations for describing segregation in non-ideal mixtures, Journal of
Fluid Mechanics, 713 (2012), pp. 564–587.
134. K. S. Pitzer, The volumetric and thermodynamic properties of fluids. I. Theoretical basis and
virial coefficients 1, Journal of the American Chemical Society, 77 (1955), pp. 3427–3433.
135. K. S. Pitzer, D. Z. Lippmann, R. F. Curl, C. M. Huggins, and D. E. Petersen, The volumetric
and thermodynamic properties of fluids. II. Compressibility factor, vapor pressure and entropy
of vaporization 1, Journal of the American Chemical Society, 77 (1955), pp. 3433–3440.
136. M. Planck, Treatise on Thermodynamics, Dover Publications, 1917.
137. B. E. Poling, J. Prausnitz, and J. P. O’Connell, The properties of gases and liquids, McGraw-
Hill, fifth edition ed., 2001.
138. J. Prausnitz, R. Lichtenthaler, and E. Azevedo, Molecular thermodynamics of fluid-phase
equilibria, Prentice Hall, 1986.
139. J. Priestley, Considerations on the doctrine of phlogiston and the decomposition of water,
Thomas Dobson, 1796.
140. I. Prigogine, Introduction to Thermodynamics of Irreversible Processes, Interscience, 1968.
141. P. Proust and J. H. Vera, PRSV: The Stryjek-Vera modification of the Peng-Robinson equation
of state. Parameters for other pure compounds of industrial interest, The Canadian Journal
of Chemical Engineering, 67 (1989), pp. 170–173.
142. F.-M. Raoult, Recherches expérimentales sur les tensions de vapeur des dissolutions, Journal
de Physique Théorique et Appliquée, 8 (1889), pp. 5–20.
143. L. Rayleigh, On the stability, or instability, of certain fluid motions, Proceedings of the London
Mathematical Society, s1–11 (1879), pp. 57–72.
144. L. J. W. S. Rayleigh, Investigation of the character of the equilibrium of an incompressible
heavy fluid of variable density, Proceedings of the London Mathematical Society, s1–14
(1882), pp. 170–177.
145. O. Redlich and J. N. S. Kwong, On the thermodynamics of solutions. V. An equation of state.
Fugacities of gaseous solutions, Chemical Reviews, 44 (1949), pp. 233–244.
146. C. Rohde and C. Zeiler, A relaxation Riemann solver for compressible two-phase flow with
phase transition and surface tension, Applied Numerical Mathematics, 95 (2015), pp. 267–
279.
147. J. S. Rowlinson, Cohesion: a Scientific History of Intermolecular Forces, Cambridge Univer-
sity Press, 2004.
148. J. S. Rowlinson and B. Widom, Molecular Theory of Capillarity (Dover Books on Chemistry),
Dover Publications, 2003.
149. K. Sankarayanan, X. Shan, I. G. Kevrekidis, and S. Sundaresan, Analysis of drag and virtual
mass forces in bubbly suspensions using an implicit formulation of the lattice Boltzmann
method, Journal of Fluid Mechanics, 452 (2002), pp. 61–96.
150. M. Sbragaglia, R. Benzi, L. Biferale, S. Succi, K. Sugiyama, and F. Toschi, Generalized
lattice Boltzmann method with multirange pseudopotential, Physical Review E, 75 (2007), p.
026702.
151. V. Schleper, A HLL-type Riemann solver for two-phase flow with surface forces and phase
transitions, Applied Numerical Mathematics, 108 (2016), pp. 256–270.
152. X. Shan and H. Chen, Lattice Boltzmann model for simulating flows with multiple phases and
components, Physical Review E, 47 (1993), pp. 1815–1819.
153. X. Shan and H. Chen, Simulation of nonideal gases and liquid-gas phase transitions by the
lattice Boltzmann equation, Physical Review E, 49 (1994), pp. 2941–2948.
Appendix: The Euler–Lagrange Equations 379

154. X. Shan, X.-F. Yuan, and H. Chen, Kinetic theory representation of hydrodynamics: a way
beyond the Navier-Stokes equation, Journal of Fluid Mechanics, 550 (2006), pp. 413–441.
155. C. E. Shannon, A mathematical theory of communication, Bell System Technical Journal, 27
(1948), pp. 379–423.
156. D. Siebert, P. Philippi, and K. Mattila, Consistent lattice Boltzmann equations for phase
transitions, Physical Review E, 90 (2014), p. 053310.
157. D. N. Siebert, Discrete kinetic models for non-ideal fluids and phase transition (doctoral
thesis, in Portuguese), 2013.
158. D. N. Siebert, L. A. Hegele, P. C. Philippi, L. A. H. Júnior, and P. C. Philippi, Lattice Boltzmann
equation linear stability analysis: Thermal and athermal models, Physical Review E, 77
(2008), p. 026707.
159. G. Soave, Equilibrium constants from a modified Redlich-Kwong equation of state, Chemical
Engineering Science, 27 (1972), pp. 1197–1203.
160. B. Song and J. Springer, Determination of interfacial tension from the profile of a pendant drop
using computer-aided image processing: 2. Experimental, Journal of Colloid and Interface
Science, 184 (1996), pp. 77–91.
161. M. Stange, M. E. Dreyer, and H. J. Rath, Capillary driven flow in circular cylindrical tubes,
Physics of Fluids, 15 (2003), pp. 2587–2601.
162. G. G. Stokes, Mathematical and Physical Papers, Vol IV, Cambridge University Press, 1880.
163. R. Stryjek and J. H. Vera, PRSV - an improved Peng-Robinson equation of state with new
mixing rules for strongly nonideal mixtures, The Canadian Journal of Chemical Engineering,
64 (1986), pp. 334–340.
164. R. Stryjek and J. H. Vera, PRSV: An improved Peng-Robinson equation of state for pure
compounds and mixtures, The Canadian Journal of Chemical Engineering, 64 (1986), pp.
323–333.
165. R. Stryjek and J. H. Vera, PRSV2: A cubic equation of state for accurate vapor-liquid equilibria
calculations, The Canadian Journal of Chemical Engineering, 64 (1986), pp. 820–826.
166. M. Sussman, P. Smereka, and S. Osher, A level set approach for computing solutions to
incompressible two-phase flow, Journal of Computational Physics, 114 (1994), pp. 146–159.
167. M. M. R. Swift, E. Orlandini, W. R. W. Osborn, and J. M. Yeomans, Lattice Boltzmann
simulations of liquid-gas and binary fluid systems, Physical Review E, 54 (1996), pp. 5041–
5052.
168. J. Szekely, A. Neumann, and Y. Chuang, The rate of capillary penetration and the applicability
of the Washburn equation, Journal of Colloid and Interface Science, 35 (1971), pp. 273–278.
169. R. S. Taylor, L. X. Dang, and B. C. Garrett, Molecular dynamics simulations of the liquid/vapor
interface of SPC/E water, The Journal of Physical Chemistry, 100 (1996), pp. 11720–11725.
170. S. G. I. Taylor, The instability of liquid surfaces when accelerated in a direction perpendicular
to their planes. I, Proceedings of the Royal Society of London. Series A. Mathematical and
Physical Sciences, 201 (1950), pp. 192–196.
171. W. Thomson, On an absolute thermometric scale founded on Carnot’s theory of the motive
power of heat, and calculated from Regnault’s observations, Proceedings of the Cambridge
Philosophical Society, Philosophical Magazine (1848).
172. W. Thomson, On the dynamical theory of heat, with numerical results deduced from Mr
Joule’s equivalent of a thermal unit, and M. Regnault’s observations on steam, Transactions
of the Royal Society of Edinburgh, 20 (1851), pp. 261–288.
173. W. Thomson, Mathematical and Physical Papers, Cambridge University Press, 1890.
174. W. L. K. Thomson, On the dynamical theory of heat, Mathematical and Physical Papers, 1
(1882), pp. 174–332.
175. R. H. Thurston, History of the Growth of the Steam-Engine, Appleton, 1886.
176. C. Tsalis, Nonextensive statistical mechanics and thermodynamics: Historical background
and present status, 2001.
177. C. Tsonopoulos, An empirical correlation of second virial coefficients, AIChE Journal, 20
(1974), pp. 263–272.
380 Appendix: The Euler–Lagrange Equations

178. C. Tsonopoulos, Second virial coefficients of polar haloalkanes, AIChE Journal, 21 (1975),
pp. 827–829.
179. C. Tsonopoulos, Second virial coefficients of water pollutants, AIChE Journal, 24 (1978), pp.
1112–1115.
180. C. Tsonopoulos and J. L. Heidman, From the virial to the cubic equation of state, Fluid Phase
Equilibria, 57 (1990), pp. 261–276.
181. J. D. van der Waals, Over de continuiteit van den gas en vloeistoftoestand, 1873.
182. H. C. van Ness, H is for enthalpy (letter), Journal of Chemical Education, 80 (2003), p. 486.
183. E. Vogle, On the origins of water wetting terminology, 2001.
184. A. Wagner, Thermodynamic consistency of liquid-gas lattice Boltzmann simulations, Physical
Review E, 74 (2006), p. 056703.
185. C. Wang-Chang and C. Uhlenbeck, University of Michigan engineering research report CM-
681, 1951.
186. F. G. Wolf, Experimental investigation of the effect of the addition of surfactants on the
water-oil interface tension, 2007.
187. F. G. Wolf, L. O. E. dos Santos, and P. C. Philippi, Capillary rise between parallel plates
under dynamic conditions, Journal of colloid and interface science, 344 (2010), pp. 171–179.
188. H. Xu and P. Sagaut, Optimal low-dispersion low-dissipation LBM schemes for computational
aeroacoustics, Journal of Computational Physics, 230 (2011), pp. 5353–5382.
189. T. Young, An essay on the cohesion of fluids, Philosophical transactions of the Royal Society,
95 (1805), pp. 65–87.
190. P. Yuan and L. Schaefer, Equations of state in a lattice Boltzmann model, Physics of Fluids,
18 (2006), p. 042101.
191. J. Yvon, Theorie statistique des fluides et l’equation d’etat, Actes Scientifique et Industrie,
203 (1935).
192. R. Zhang, X. Shan, and H. Chen, Efficient kinetic method for fluid simulation beyond the
Navier-Stokes equation, Physical Review E, 74 (2006), p. 046703.
193. B. Zhmud, F. Tiberg, and K. Hallstensson, Dynamics of capillary rise, Journal of Colloid and
Interface Science, 228 (2000), pp. 263–269.