This action might not be possible to undo. Are you sure you want to continue?
PHYSICS
CHEMISTRY
VERSUS
This page intentionally left blank This page intentionally left blank
Chemical Reactions Near Critical Points
PHYSICS
CHEMISTRY
VERSUS
NE W J E RSE Y • L ONDON • SI NGAP ORE • BE I J I NG • SHANGHAI • HONG KONG • TAI P E I • CHE NNAI
World Scientifc
Moshe Gi tterman
BarIlan University, Israel
British Library CataloguinginPublication Data
A catalogue record for this book is available from the British Library.
For photocopying of material in this volume, please pay a copying fee through the Copyright
Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA. In this case permission to
photocopy is not required from the publisher.
ISBN13 9789814291200
ISBN10 981429120X
All rights reserved. This book, or parts thereof, may not be reproduced in any form or by any means,
electronic or mechanical, including photocopying, recording or any information storage and retrieval
system now known or to be invented, without written permission from the Publisher.
Copyright © 2010 by World Scientific Publishing Co. Pte. Ltd.
Published by
World Scientific Publishing Co. Pte. Ltd.
5 Toh Tuck Link, Singapore 596224
USA office: 27 Warren Street, Suite 401402, Hackensack, NJ 07601
UK office: 57 Shelton Street, Covent Garden, London WC2H 9HE
Printed in Singapore.
CHEMISTRY VERSUS PHYSICS
Chemical Reactions Near Critical Points
ZhangFang  Chemistry vs Physics.pmd 9/10/2009, 7:09 PM 1
September 14, 2009 16:24 World Scientiﬁc Book  9in x 6in Generalnn
Preface
In this book we attempt to trace the connection between chemical reactions
and the physical forces of interaction manifested in critical phenomena.
The physical and chemical descriptions of matter are intimately related. In
fact, the division of forces into “physical” and “chemical” is arbitrary. It
is convenient [1] to distinguish between strong attractive (chemical) forces
leading to the formation of chemical species, and weak attractive (physical)
forces, called van der Waals forces. It should be remembered, therefore,
when one considers the “ideal” ternary mixture A, B and A
m
B
n
, that the
strong chemical bonding interaction between A and B atoms has already
been taken into account via the formation of the chemical complex A
m
B
n
,
and the term “ideal” only means that there are no “physical” forces present.
The growth of clusters in a metastable state is an example of the fuzzy
distinction between physical and chemical forces. In numerical simulations
a rather arbitrary decision has to be made whether a given particle belongs
to a “chemical” cluster.
The usefulness of a “chemical” approach to physical problems can be
seen from the meanﬁeld theory of the phase transition on an Ising lattice
of nonstoichiometric AB alloys [2]. The temperature dependence of the
longrange and shortrange order parameters is found from the “law of mass
action” for the appropriately chosen “chemical reaction.” The latter is the
exchange of position of an atom A from one sublattice and an atom B
from the second sublattice. The change of the interaction energy for such a
transition when both atoms are or are not nearest neighbors determines the
“constant of chemical reaction.” Such an approach allows one to avoid the
calculation of entropy provided that one is interested only in the value of
the critical temperature, rather than in the behavior of all thermodynamic
quantities, which are determined by the same classical critical indices in
v
September 14, 2009 16:24 World Scientiﬁc Book  9in x 6in Generalnn
vi Chemistry versus Physics: Chemical Reactions near Critical Points
all versions of meanﬁeld theory. The latter example is, in fact, a special
case of chapter 15 in [3], where the typical “physical” process of diﬀusion
is considered as a “chemical reaction” in which some amount of substance
A passes from volume element a to b while a diﬀerent amount of B passes
from b to a.
The “chemical” method, in which a given atom with all its neighbors is
considered as the basic group, gives better results than the BraggWilliams
or BethePeierls method. In the BraggWilliams method, each atom is
exposed to the (selfconsistent) average inﬂuence of all other atoms, whereas
in the BethePeierls method, a pair of adjacent atoms is considered as the
basic group.
Even though the border between chemical and physical forces is arbi
trary, one usually considers ﬁrst the “physical” forces in the equation of
state, and then the “chemical” forces in the law of mass action based on
the nonideal equation of state. I shall follow this approach.
One can also trace the common features of phase transitions and chem
ical reactions by analyzing the time evolution of the state variables ψ de
scribed by the equation
dψ/dt = F [ψ, λ] , (0.1)
where λ is the set of internal and external parameters, and F is a nonlinear
functional in ψ. In the case of a phase transition, Eq. (0.1) may be the
LandauGinzburg equation for the order parameter, while in the case of
a chemical reaction — the equation for the rate of reaction. The steady
state of the system is described by ψ
0
(λ) which is the solution of equation
F
ψ
0
, λ
= 0. For nonlinear F, more than one steady state solution is pos
sible, and for some values of λ, at the socalled transition point, bifurcation
may occur, when a system goes from the original steady state to the new
steady state. Such a transition might be of ﬁrst or second order in the case
of a phase transition, and, analogously, the hard or soft transition for a
chemical reaction.
I have kept this book as simple as possible, so that it will be useful
for a wide range of researchers, both physicists and chemists, as well as
teachers and students. No preliminary knowledge is assumed, other than
undergraduate courses in general physics and chemistry. In line with this
approach, I have favored a phenomenological presentation, thus avoiding
the details of both microscopic and numerical approaches. There are a
tremendous number of published articles devoted to this subject, and it
proved impossible to include many of them in this book of small size. I ask
September 14, 2009 16:24 World Scientiﬁc Book  9in x 6in Generalnn
Preface vii
the forgiveness of the authors whose publications were beyond the scope of
this book.
The organization of the book is as follows.
After the Introduction, Chapter 1 contains a short review of phase tran
sitions and chemical reactions and their interconnection, which is needed
to understand the ensuing material. Chapter 2 is devoted to the speciﬁc
changes in a chemical reaction occurring near the critical point. The eﬀect
of pressure and phase transformation on the equilibrium constant and rate
of reaction is the subject of Secs. 2.1–2.2. The hallmark of critical phenom
ena — the slowing down of all processes — leads also to the slowingdown of
the rates of chemical reactions. This phenomenon is described in Sec. 2.3,
while the opposite peculiar phenomenon of speedingup of a chemical re
action near the critical point provides the subject matter for Sec. 2.4. It
is shown that all three types of behavior (slowingdown, speedingup and
unchanged) are possible depending on the given experiment. In Sec. 2.5,
we consider another inﬂuence of criticality — the anomalies in chemical
equilibria including supercritical extraction. The appropriate experiments
are described in Sec. 2.6.
The reverse process — inﬂuence of chemistry on critical phenomena —
forms the content of Chapter 3, including the change in the critical param
eters (Sec. 3.1), critical indices (Sec. 3.2), transport coeﬃcients (Sec. 3.5)
and degree of dissociation (Sec. 3.3). Section 3.4 is devoted to the isotope
exchange reaction in nearcritical systems.
Chapter 4 deals with the problem of the phase separation in reactive
systems. The occurrence of multiple solutions of the law of mass action
is described in Sec. 4.1. The mechanism of phase separation depends on
whether it starts from a metastable or a nonstable state. The former case,
where phase separation takes place through nucleation, and the spinodal
decomposition for the latter case are considered in Secs. 4.2 and 4.3, respec
tively. Section 4.4 is devoted to the special case of a dissociation reaction
in a ternary mixture.
Chapter 5 contains a description of chemical reactions near some speciﬁc
regions of the phase diagram. An account of the supercritical ﬂuids is given
in Sec. 5.1, while the vicinities of the azeotrope, melting and double critical
points are considered in Secs. 5.2, 5.3 and 5.4, respectively. The main ex
perimental methods of analysis of nearcritical ﬂuids — sound propagation
and light scattering — are considered in Chapter 6. Finally, in Chapter 7
we present our conclusions.
September 14, 2009 16:24 World Scientiﬁc Book  9in x 6in Generalnn
Contents
Preface v
1. Criticality and Chemistry 1
1.1 Critical phenomena . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Chemical reactions . . . . . . . . . . . . . . . . . . . . . . 8
1.3 Analogy between critical phenomena and the instability of
chemical reactions . . . . . . . . . . . . . . . . . . . . . . 12
2. Eﬀect of Criticality on Chemical Reaction 15
2.1 The eﬀect of pressure on the equilibrium constant and rate
of reaction . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.2 Eﬀect of phase transformations on chemistry . . . . . . . 17
2.3 Critical slowingdown of chemical reactions . . . . . . . . 19
2.4 Hydrodynamic equations of reactive binary mixture:
piston eﬀect . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.4.1 Heterogeneous reactions in nearcritical systems . 26
2.4.2 Dynamics of chemical reactions . . . . . . . . . . 27
2.4.3 Relaxation time of reactions . . . . . . . . . . . . 29
2.4.4 Hydrodynamic equations of a reactive binary
mixtures . . . . . . . . . . . . . . . . . . . . . . . 32
2.4.5 Hydrodynamic equations with statistically
independent variables . . . . . . . . . . . . . . . 38
2.5 Critical anomalies of chemical equilibria . . . . . . . . . . 46
2.6 Experiment . . . . . . . . . . . . . . . . . . . . . . . . . . 48
3. Eﬀect of Chemistry on Critical Phenomena 53
ix
September 14, 2009 16:24 World Scientiﬁc Book  9in x 6in Generalnn
x Chemistry versus Physics: Chemical Reactions near Critical Points
3.1 Change of critical parameters due to a chemical reaction . 53
3.2 Modiﬁcation of the critical indices . . . . . . . . . . . . . 55
3.3 Singularity in the degree of dissociation near a critical point 60
3.4 Isotope exchange reaction in nearcritical systems . . . . . 62
3.5 Singularities of transport coeﬃcients in reactive systems . 63
3.5.1 Modecoupling analysis . . . . . . . . . . . . . . . 63
3.5.2 Renormalization group methods . . . . . . . . . . 65
4. Phase Separation in Reactive Systems 67
4.1 Multiple solutions of the law of mass action . . . . . . . . 67
4.2 Phase equilibrium in reactive binary mixtures quenched
into a metastable state . . . . . . . . . . . . . . . . . . . . 69
4.2.1 Thermodynamic analysis of reactive binary
mixtures . . . . . . . . . . . . . . . . . . . . . . . 69
4.2.2 Thermodynamic analysis of reactive ternary
mixtures . . . . . . . . . . . . . . . . . . . . . . . 72
4.2.3 Kinetics of phase separation . . . . . . . . . . . . 75
4.3 Phase equilibrium in reactive mixtures quenched into an
unstable state . . . . . . . . . . . . . . . . . . . . . . . . . 81
4.4 Thermodynamics of a threecomponent plasma with a
dissociative chemical reaction . . . . . . . . . . . . . . . . 89
5. Comments on the Geometry of the Phase Diagram of a
Reaction Mixture 93
5.1 Solubility in supercritical ﬂuids . . . . . . . . . . . . . . . 93
5.2 Azeotropic points in reactive manycomponent systems . . 100
5.3 Melting point of reactive binary mixtures . . . . . . . . . 101
5.4 Double critical point . . . . . . . . . . . . . . . . . . . . . 105
6. Sound Propagation and Light Scattering in Chemically
Reactive Systems 111
6.1 Ultrasound attenuation in nearcritical reactive mixtures . 111
6.2 Hydrodynamic analysis of the dispersion relation for sound
waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
6.3 Light scattering from reactive systems . . . . . . . . . . . 118
6.4 Inhomogeneous structure of nearcritical reactive systems 121
7. Conclusions 125
September 14, 2009 16:24 World Scientiﬁc Book  9in x 6in Generalnn
Contents xi
Bibliography 127
Index 135
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Chapter 1
Criticality and Chemistry
1.1 Critical phenomena
Phase transitions occur in Nature in a great variety of systems and un
der a very wide range of conditions. For instance, the paramagnetic
ferromagnetic transition occurs in iron at around 1000 K, the superﬂuid
transition occurs in liquid helium at 2.2 K, and BoseEinstein condensa
tion occurs at 10
−7
K. In addition to this enormous temperature range,
phase transitions occur in a wide variety of substances, including solids,
classical ﬂuids and quantum ﬂuids. Therefore, phase transitions are a very
general phenomenon, associated with the basic properties of manybody
systems. The thermodynamic functions become singular at the phase tran
sition points, and these mathematical singularities lead to many unusual
properties of the system which are called “critical phenomena.” We ﬁrst
consider the diﬀerent types of the phase transition points (“critical points”)
and then we introduce a qualitative method for describing the behavior of
various parameters of the system in the vicinity of critical points.
The liquidgas critical point of an onecomponent ﬂuid is determined by
the condition [4]
∂p
∂ρ
T
=
∂
2
p
∂ρ
2
T
= 0 (1.1)
where p is the pressure, ρ is the density, and T is the temperature. Similarly,
the liquidgas critical points of binary mixtures are characterized by the
vanishing of the ﬁrst and second derivatives of the chemical potential µ
with respect to the concentration x,
∂µ
∂x
T,p
=
∂
2
µ
∂x
2
T,p
= 0. (1.2)
1
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
2 Chemistry versus Physics: Chemical Reactions near Critical Points
Here µ = µ
1
/m
1
− µ
2
/m
2
, where µ
1
, µ
2
and m
1
, m
2
are the chemical
potentials and masses of the two components.
The close relation between (1.1) and (1.2) is evident from the equivalent
form of Eq. (1.2), which can be rewritten as
∂p
∂ρ
T,µ
=
∂
2
p
∂ρ
2
T,µ
= 0. (1.3)
The critical conditions for a binary mixture (1.3) are the same as those
for a pure system (1.1) when the chemical potential is kept constant. Anal
ogously, the critical points for an ncomponent mixture are determined by
the conditions
∂p
∂ρ
T,µ
1
,...,µ
n−1
=
∂
2
p
∂ρ
2
T,µ
1
,...,µ
n−1
= 0 (1.4)
where n −1 chemical potentials are held constant.
In addition to the abovementioned thermodynamic peculiarities, relax
ation processes slow down near the critical points resulting in singularities
in the kinetic coeﬃcients. An example is the slowingdown of diﬀusion near
the critical points of a binary mixture. Nothing happens to the motion of
the separate molecules when one approaches the critical point. It is the rate
of equalization of the concentration gradients by diﬀusion which is reduced
near the critical points. In fact, the excess concentration δx in some part
of a system does not produce diﬀusion by itself. Usually a system has no
diﬃculty in “translating” the change in concentration into a change in the
chemical potential δµ ∼ (∂µ/∂x) δx, which is the driving force for diﬀusion.
However, near the critical point, according to (1.2), ∂µ/∂x is very small,
and the system becomes indiﬀerent to changes in concentration. This is
the simple physical explanation of the slowingdown of diﬀusion near the
critical point.
Since the states of a onecomponent system and a binary mixture are de
ﬁned by the equations of state p = p (T, µ) and µ = µ(T, ρ, x) , respectively,
Eqs. (1.1) and (1.2) deﬁne the isolated critical point for an onecomponent
system, and the line of critical points for a binary mixture. Another dis
tinction between onecomponent systems and binary mixtures is that there
are two types of critical points in the latter: the above considered liquidgas
critical points and liquidliquid critical points, whereas two coexisting liq
uid phases are distinguished by diﬀerent concentrations of the components.
Both critical lines are deﬁned by Eq. (1.2).
Diﬀerent binary liquid mixtures show either concavedown or concave
up coexistence curves in a temperatureconcentration phase diagram (at
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Criticality and Chemistry 3
ﬁxed pressure) or in a pressureconcentration phase diagram (at ﬁxed tem
perature). The consolute point is an extremum in the phase diagram where
the homogeneous liquid mixture ﬁrst begins to separate into two immiscible
liquid layers. For the concavedown diagram, as for a methanolheplan mix
ture, the minimum temperature above which the two liquids are miscible
in all proportions is called the upper critical solution temperature (UCST).
By contrast, for a concaveup diagram, such as the watertriethylamine so
lution, the maximum temperature below which the liquids are miscible in
all proportions is called the lower critical solution temperature (LCST).
Under the assumption of analyticity of the thermodynamic functions at
the critical points, one can obtain the general thermodynamic criterion
for the existence of UCST and LCST [3]. We will discuss this calculation
in the next section when examinating the inﬂuence of chemical reactions
on UCST and LCST. Here we will consider chemical reactions occurring
between solutes near the critical points of the solvent.
The properties of nearcritical ﬂuids range between those of gases and
liquids (see Table 1). Nearcritical ﬂuids combine properties of gases and
liquids. Their densities are lower than those of liquids, but much higher
than the densities of gases, which makes the nearcritical ﬂuids excellent
solvents for a variety of substances.
Table 1. Comparison of some physical properties of gases, liquids
and nearcritical ﬂuids.
Physical Properties Gas Nearcritical ﬂuid Liquid
Density (kg/m
3
) 0.6–2 200–500 600–1000
Kinematic viscosity 5–500 0.02–0.1 0.1–5
(10
−6
m
2
/sec)
Diﬀusion coeﬃcient 10–40 0.07 2 ×10
−4
–
(10
−6
m
2
/ sec) 2 ×10
−3
In Table 2 we list the critical parameters of the solvents in most com
mon use. Water is the most abundant, cheap, safe and environmentally
pure solvent. In spite of its high critical parameters which limits its ap
plication, in addition to the traditional uses, modern applications include
the important problems of solving the environmental pollution problem
and the fabrication of nanocrystalline materials with predictable properties
[5]. Properties of nearcritical water, such as the full mixing with oxygen
and organic compounds, high diﬀusion and mass transfer coeﬃcients, make
water appropriate for eﬃcient treatment of industrial wastes. The use of
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
4 Chemistry versus Physics: Chemical Reactions near Critical Points
nearcritical water for detoxiﬁcation of organic waste using the catalytic
oxidation of pyridine was found [6] to be cheaper than other methods and
also more eﬀective, having almost no limitation on the concentration of the
pyridinecontaining solutions. The eﬃciency of hydrothermal detoxiﬁca
tion of pyridine waste is substantially increased by the addition of a small
amount of heterogeneous catalyst. For instance, the addition of 0.5% of
PtAl
2
O
3
increases the oxidation of pyridine to 99% [7]. Other methods in
clude dechlorination of chlorinate organic compounds, cleaning of polymers
and plastic wastes, hydrolysis of cellulose, and the release of bromine for
polymers and plastics.
Table 2. Critical parameters of ﬂuids which are commonly used
as solvents for chemical reactions.
Solvent T
cr
(C) p
cr
(atm) ρ
cr
(g/mL)
Water (H
2
O) 373.9 220.6 0.322
Carbon dioxide (CO
2
) 30.9 72.9 0.47
Sulfur hexaﬂuoride (SF
6
) 45.5 36.7 0.73
Ammonia (NH
3
) 132.3 113.5 0.235
Methanol (CH
3
OH) 239.4 80.9 0.272
Propane (C
3
H
6
) 96.6 41.9 0.22
Ethane (C
2
H
6
) 32.2 48.2 0.20
Pyridine (C
5
H
5
N) 347 55.6 0.31
Benzene (C
6
H
6
) 289 48.3 0.30
Nanocrystallines (particles whose size is a few interatomic distances)
are new generation materials widely used as sensors, fuel cells, highdensity
ceramics, and semiconductors, among others. Hydrothermal synthesis in
nearcritical water is used to obtain nanocrystalline oxide powders with
speciﬁed particle sizes and phase composition. Many references can be
found [5] dealing with both nanotechnology and environmental problems.
Like water, carbon dioxide (CO
2
) has the advantage of being non
ﬂammable, nontoxic and environment compatible. At the same time, CO
2
has critical parameters more convenient than water, and is, therefore, the
ﬁrst choice for use as a nearcritical solvent. Another advantage of CO
2
lies in the fact that it does not attack enzymes and is therefore suitable
for enzymecatalyzed reactions. Some new applications include the use of
twophase reaction mixtures with high pressure carbon dioxide which are
known as “CO
2
— expanded ﬂuids”. In fact, nearcritical carbon dioxide
is more frequently used in the laboratory and in technology than any other
solvent. Hundreds of examples can be found in recent reviews [8], [9].
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Criticality and Chemistry 5
The inﬁnite increase of the compressibility ρ
−1
(∂ρ/∂p)
T
or (∂x/∂µ)
T,p
as the critical point is approached, leads to a number of peculiarities in the
behavior of a substance near its critical point. The speciﬁc heat at constant
pressure C
p
and the expansion coeﬃcient β = −ρ
−1
(∂ρ/∂T)
p
also increase
near the critical point of a onecomponent system, as follows from Eq. (1.1)
and the appropriate thermodynamic relations.
A sharp increase in the mean square ﬂuctuations of the density (or
concentration) and of the integral of the correlation function g
ρρ
follows
from the wellknown thermodynamic relations
(ρ (r) −ρ)
2
∼
∂ρ
∂p
T
→∞;
g
ρρ
d
3
r ∼
∂ρ
∂p
T
→∞. (1.5)
The large increase of the correlations between the positions of diﬀerent
particles is given by the second expression in (1.5), which is closely con
nected with the ﬁrst expression. In other words, widely separated particles
have to be strongly correlated to cause great changes in density.
The correlation radius ξ, which characterizes the distance over which
correlations are signiﬁcant, increases sharply near the critical temperature
T
C
,
ξ
T→T
C
→∞. (1.6)
According to estimates from scattering experiments, ξ reaches 10
−4
−10
−5
cm near the critical point. Thus, the speciﬁc nature of the critical region
consists of the appearance of a new characteristic distance ξ, satisfying the
condition
a << ξ << R (1.7)
where a is the average distance between particles and R is a characteristic
macroscopic length.
As an illustration of the crucial importance of the new characteristic
length ξ, let us consider the singular part of the transport coeﬃcients near
the critical point for a model ﬂuid consisting of spheres with a characteristic
radius ξ. Particles inside such spheres are strongly correlated, and we can
assume that under the inﬂuence of an external force, they move together
with a mean velocity v and a mean free path ξ. One ﬁnds the following
results [10]:
1. Diﬀusion coeﬃcient [11]. When an external force F is applied, the
spheres move according to Stoke’s law, F ∼ ηξv, where η is the viscosity,
i.e., the mobility b ≡ v/F ∼ (ηξ)
−1
. Using the Einstein relation D = k
B
Tb,
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
6 Chemistry versus Physics: Chemical Reactions near Critical Points
where D is the diﬀusion coeﬃcient and k
B
is the Boltzmann constant, we
have D ∼ (ηξ)
−1
or Dη ∼ ξ
−1
, a result conﬁrmed by the more rigorous
theory and by experiment.
2. Heat conductivity. The usual arguments of molecularkinetic
theory give the heat ﬂux q passing through unit area per unit time,
q ∼ υn(
1
−
2
) . Here, n is the total number of spheres, and
1
−
2
is
the diﬀerence in their energies on two sides of a selected area, arising from
the temperature diﬀerence T
1
−T
2
∼ ξ∇T :
1
−
2
∼ V C
p
ξ∇T, where V is
the total volume of the spheres, so that nV = 1. Thus, q ∼ vC
p
ξ∇T.
One can ﬁnd the velocity v from the estimate for the diﬀusion coeﬃ
cient given above: v ∼ D/ξ ∼ 1/ξ
2
η. Finally, the heat conductivity
λ ∼ q/∇T ∼ vC
p
ξ ∼ C
p
/ηξ. This result is supported by more rigorous
theory and also by experiment.
The qualitative description of the singularity of the quantity a at
the critical point is given by the noninteger critical index x, where
a ∼ T −T
C

x
. If x = 0, this critical index is given by x = ln(a)/ ln T −T
C
,
whereas if x = 0, there are two possibilities. In one case, a becomes constant
at the critical point, with the possibility of diﬀerent values of this constant
on the two sides of the transition, which is called a jump singularity. In the
other case, a exhibits a logarithmic singularity, a ∼ ln T −T
C
.
Although there are general properties which deﬁne the values of the crit
ical indices (dimension d of space, symmetry and the presence of longrange
interactions), according to the universality principle, these values are de
ﬁned by the general statistical properties of the manybody system, rather
than by the details of the microscopic interactions. This implies that dif
ferent isomorphic systems will have the same indices for the appropriate
parameters. Using the language of ﬂuids, the commonly accepted sym
bols α, β, γ, δ, η, and ν describe, respectively, the asymptotic behavior near
the onecomponent liquidgas critical point of the speciﬁc heat at constant
volume, order parameter (deviation of the density from its critical value),
compressibility, pressuredensity relation at the critical isotherm, the cor
relation function at the critical temperature, and the correlation length.
Of these indices, one (β) is determined only below the critical temperature
and two (δ, η) exist only at the critical temperature. The remaining three
indices can be deﬁned for temperatures above critical (α, γ, ν) as well as
below (α
, γ
, ν
). According to scaling theory [12], α = α
, γ = γ
, ν = ν
and the following relations exist between critical indices: α + 2β + γ = 2,
dν = 2 −α, (2 −η) ν = γ, and β (δ −1) = γ.
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Criticality and Chemistry 7
The values of the critical indices depend on the proximity to the critical
point. In the approach to the critical point, the indices have their “classical”
meanﬁeld values, obtained by assuming the analyticity of thermodynamic
functions at the critical points which allows the expansion of these functions
in a power series in the deviation from the critical parameters. However,
the singularity in the critical points makes the series expansion inapplicable
very close to the critical point, leading to “nonclassical” values of the crit
ical indices (noninteger powers, logarithmic dependence, etc.). Since the
diﬀerence between the results of meanﬁeld theory and the exact theory is
due to ﬂuctuations of the order parameter, we expect that the meanﬁeld
approximation will be accurate when these ﬂuctuations are small. Ginzburg
proposed [13] that the mean ﬁeld theory is applicable when the ﬂuctuations
are small compared to the thermodynamic values. The Ginzburg criterion
divides the region near the critical point into two parts, giving the crossover
from classical to nonclassical behavior. An interesting idea has been pro
posed recently [14] of the existence of the second crossover in the immediate
vicinity of the critical point, where the nonclassical critical indices regain
their classical values. Indeed, according to Eq. (1.5), the development of
largescale ﬂuctuations is accompanied by a continuous increase of the sus
Fig. 1.1 Variation of critical exponent for the isothermal compressibility in the imme
diate vicinity of the critical point for CO
2
and SF
6
. The experimental points are taken
from [16], [17]. Reproduced from Ref. [14] with permission, copyright (2002), Springer.
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
8 Chemistry versus Physics: Chemical Reactions near Critical Points
ceptibilities of the critical system, in particular, the susceptibility to varies
external perturbations (gravitational and Coulomb ﬁelds, surface forces,
shear stresses, turbulence, presence of boundaries). As a consequence, clas
sical meanﬁeld behavior must be restored. This transition occurs in the
direction opposite to the Ginzburg criterion direction and deﬁnes the sec
ond crossover. Such behavior was found experimentally as early as 1974
[15] by taking p −V −T measurements of very pure SF
6
in the immediate
vicinity (τ =
T−T
C
T
C
) of the critical point with 2 × 10
−4
K, ±0.01%, and
±0.02% accuracy, respectively in the temperature, dimensionless pressure
and density.
As an example, we show [14] in Fig. 1.1 the crossover to the classical
value of the critical index of the isothermal compressibility in the immedi
ate vicinity of the critical point for CO
2
and SF
6
. The author of [14] states
his belief that the analogous second crossover has been seen in other ex
periments under the inﬂuence of gravity, impurities, and shear ﬂows. This
problem certainly deserves further investigation.
1.2 Chemical reactions
The equilibrium numbers of particles of the diﬀerent substances taking part
in a chemical reaction are connected by the law of mass action. This law
results from a relation between the chemical potentials µ
i
of the various
components [3]. Thus, for the reaction
ν
i
A
i
= 0 , where A
i
are the
chemical symbols of the reagents and ν
i
are positive or negative integers,
the equation for chemical equilibrium has the form
ν
i
µ
i
= 0. For the
simplest case of the isomerization reaction, the reaction equation and the
law of mass action have the form A
1
−A
2
= 0 and µ
1
= µ
2
, respectively.
Thus, binary mixtures undergoing a chemical reaction are characterized
by their concentration, as in a nonreactive mixture. However, according
to the law of mass action, this concentration is a function of other ther
modynamic variables. Therefore, the number of thermodynamic degrees of
freedom of a binary system undergoing a chemical reaction is the same as
for a nonreactive, onecomponent system.
Chemical reactions inﬂuence all properties of manycomponent systems.
As an example, we shall show that the existence of the chemical reaction
may lead to the replacement of UCST by LCST and vice versa [18]. More
over, the ClapeyronClausius equation for a binary mixture is determined
by the chemical reaction, in addition to the latent heat and the volume
diﬀerence between the two phases.
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Criticality and Chemistry 9
Consider a mole of a binary mixture which separates into two phases,
B
and B
. The system can be described by four parameters, p, T, x
2
, and
x
2
, which satisfy the following conditions:
µ
1
(T, p, x
2
) = µ
1
(T, p, x
2
) ; µ
2
(T, p, x
2
) = µ
2
(T, p, x
2
) . (1.8)
One can diﬀerentiate the equilibrium conditions (1.8) along the equilib
rium surface between B
and B
. Using simple thermodynamic relations,
one obtains [3]
∆υ
1
dp −∆h
1
dT/T + x
2
g
2x
dx
2
−x
2
g
2x
dx
2
= 0,
∆υ
2
dp −∆h
2
dT/T + (1 −x
2
) g
2x
dx
2
−
1 −x
2
g
2x
dx
2
= 0 (1.9)
where υ
i
and h
i
are the partial molar volume and enthalpy,
∆υ
i
= υ
i
−υ
i
; ∆h
i
= h
i
−h
i
; g
2x
≡
∂
2
g
∂x
2
2
T,p
; g
2x
≡
∂
2
g
∂x
2
2
T,p
.
(1.10)
The partial derivatives of the intensive variables are given by Eq. (1.9)
at constant pressure (temperature) and at constant concentration [3]. How
ever, there is no need to consider these special sections of the coexis
tence surface when we deal with a reactive system. Indeed, for a reaction
ν
1
A
1
ν
2
A
2
, an additional restriction to (1.9) exists in the form of the
law of mass action
ν
1
µ
1
+ ν
2
µ
2
= 0. (1.11)
Diﬀerentiating the latter equation along the equilibrium surface, yields
(ν
1
υ
1
+ ν
2
υ
2
) dp −(ν
1
h
1
+ ν
2
h
2
) dT/T
+ [ν
1
x
2
g
2x
−ν
2
(1 −x
2
) g
2x
] dx
2
= 0. (1.12)
Combining Eqs. (1.9) and (1.12) yields the slope of the equilibrium line of
a twophase reactive binary mixture,
T
∂p
∂T
chem
=
h
2x
(∆x
2
)
2
−(2∆x
2
/n
) (ν
1
υ
1
+ ν
2
υ
2
)
υ
2x
(∆x
2
)
2
−(2∆x
2
/n
) (ν
1
h
1
+ ν
2
h
2
)
(1.13)
∂T
∂x
2
chem
=
2Tg
2x
∆x
2
+ Tn
υ
2
(∆x
2
)
2
(ν
1
υ
1
+ ν
2
υ
2
)
−1
g
2x
h
2x
(∆x
2
)
2
−υ
2
(∆x
2
) (ν
1
h
1
+ ν
2
h
2
) (ν
1
υ
1
+ ν
2
υ
2
)
−1
(1.14)
where n
≡ ν
1
x
2
− ν
2
(1 −x
2
) and h
≡ (1 −x
2
) h
1
+ x
2
h
2
; υ
≡
(1 −x
2
) υ
1
+x
2
υ
2
are the heat of reaction and the volume change of reaction
in phase B
.
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
10 Chemistry versus Physics: Chemical Reactions near Critical Points
In the absence of a chemical reaction, all terms vanish, except the ﬁrst
term in the denominators and numerators of Eqs. (1.13) and (1.14 ), and
these terms reduce to the wellknown form [3]
T
∂p
∂T
x
2
=
h
2x
υ
2x
(1.15)
∂T
∂x
2
p
=
2Tg
2x
∆x
2
h
2x
(∆x
2
)
2
. (1.16)
Equation (1.13) is the generalized form of the ClapeyronClausius equa
tion (1.15) for a reactive mixture, while Eq. (1.14) determines the criterion
for UCST and LCST. The latter can be obtained in the same way as for a
nonreactive mixture [3].
The “classical” expansion near the critical points g
2x
≈
1
8
g
4x
(∆x
2
)
2
,
yields
∂T
∂x
2
p
=
Tg
4x
(x
2
−x
2
)
4h
2x,cr
. (1.17)
If x
2
> x
2,cr
> x
2
, then (∂x
2
/∂T)
p
is positive if h
2x,cr
≡
∂
2
h/∂x
2
cr
is negative. These signs deﬁne UCST. Analogously, LCST corresponds to
∂
2
h/∂x
2
cr
> 0.
Performing a similar expansion near the single critical point of a reactive
binary mixture, one obtains from (1.13) and (1.14),
T
∂p
∂T
chem
∼
v
1
h
1,cr
+ v
2
h
2,cr
v
1
υ
1,cr
+ v
2
υ
2,cr
(1.18)
∂T
∂x
2
chem
∼ Tg
4x
(x
2
−x
2
)
h
2x,cr
−υ
2x,cr
v
1
h
1,cr
+ v
2
h
2,cr
v
1
υ
1,cr
+ v
2
υ
2,cr
−1
∼ Tg
4x
(x
2
−x
2
) h
2x,cr
1 −
v
1
h
1,cr
+ v
2
h
2,cr
T
c
(v
1
υ
1,cr
+ v
2
υ
2,cr
)
dT
C
/dp
−1
.
(1.19)
Equation (1.15) was used in the last relation in (1.19). It follows from
(1.19) that the existence of a chemical reaction may change the type of
critical point (UCST to LCST and vice versa) if the last bracket in (1.19) is
negative. The ratio of the second derivatives υ
2x,cr
/h
2x,cr
can be replaced
by the ratio of excess volume V
E
and excess enthalpy h
E
at the critical
point.
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Criticality and Chemistry 11
Thus, a chemical reaction will change the nature of the critical point if
the following (equivalent) inequalities are satisﬁed:
v
1
h
1,cr
+ v
2
h
2,cr
v
1
υ
1,cr
+ v
2
υ
2,cr
V
E
h
E
> 1 or
v
1
h
1,cr
+ v
2
h
2,cr
v
1
υ
1,cr
+ v
2
υ
2,cr
1
T
C
dT
C
dp
> 1. (1.20)
One can ﬁnd [19], [20] the form of the critical line near the critical point
dT
C
/dp as well as V
E
and h
E
. Typical examples of positive V
E
and h
E
are
shown in Fig. 1.2.
Fig. 1.2 Typical form of the excess enthalpy h
E
(or excess volume V
E
) for a binary
mixture as a function of concentration. Points a and b correspond to the critical molar
enthalpy of pure substances. Reproduced from Ref. [18] with permission, copyright
(1990), Springer.
Using the deﬁnition of the partial molar quantity y
1
= y −
x
2
(∂y/∂x
2
)
T,p
, where y ≡ (h, υ), one can see that the points a, b in Fig. 1.2
give h
1,cr
and h
2,cr
(or, analogously, υ
1,cr
and υ
2,cr
). For an isomerization
reaction, ν
1
= −ν
2
= 1, and Eq. ( 1.20) becomes
h
1,cr
−h
2,cr
h
E
V
E
υ
1,cr
−υ
2,cr
> 1. (1.21)
There is no physical reason why criterion (1.21) should not be satisﬁed
for some mixtures. Then, the presence of a chemical reaction will change
UCST to LCST and vice versa.
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
12 Chemistry versus Physics: Chemical Reactions near Critical Points
1.3 Analogy between critical phenomena and the instability
of chemical reactions
Critical phenomena describe the behavior of closed thermodynamic systems
whereas chemical reactions occur in the systems open to matter transport
from the environment. The former are described by the wellknown Gibbs
technique, but there is no universal approach to nonequilibrium chemical
reactions which are deﬁned by the equation of the reaction rate. However,
as already mentioned in the Introduction, there is a close analogy between
these two phenomena (in fact, article [21] is entitled “Chemical instabilities
as critical phenomena”).
One distinguishes between phase transitions of ﬁrst and second orders.
Firstorder transitions involve a discontinuity in the state of the system,
and, as a result, a discontinuity in the thermodynamic variables such as
entropy, volume, internal energy (ﬁrst derivatives of the thermodynamic
potential). In secondorder phase transitions, these variables change con
tinuously while their derivatives, which are the second derivatives of the
thermodynamic potentials (speciﬁc heat, thermal expansion, compressibil
ity), are discontinuous. Analogously, in the theory of instability of non
linear diﬀerential equations, which describe the rate of chemical reactions,
one distinguishes between hard and soft transitions. These are similar in
nature to ﬁrstorder and secondorder phase transitions [22].
As an example, consider the following chemical reactions [23]
A + 2X
k
1
k
2
3X; A
k
3
k
4
X. (1.22)
These rate equations describe the conversion of the initial reactant A into
X by two parallel processes: a simple monomolecular degradation or an
autocatalytic trimolecular reaction. Both these reactions are reversible with
reaction constants k
i
, i = 1, . . . , 4. The system is open to interaction with
an external reservoir of reactant A, so that the concentration of A remains
constant. The macroscopic equation for the number of molecules X has the
following form
dX
dt
= −k
2
X
3
+ k
1
AX
2
−k
4
X + k
3
A. (1.23)
The solution of Eq. (1.23) with the initial condition X(0) = X
0
is
X −X
1
X
0
+ X
1
k
3
−k
2
X −X
2
X
0
−X
2
k
1
−k
3
X −X
3
X
0
−X
3
k
2
−k
1
= exp [−k
2
(X
1
− X
2
) (X
2
−X
3
) (X
3
−X
1
) t] (1.24)
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Criticality and Chemistry 13
where X
1
, X
2
and X
3
are the three roots of
k
2
X
3
−k
1
AX
2
+ k
4
X −k
3
A = 0 (1.25)
with X
3
≥ X
2
≥ X
1
. The steady state solutions X
s
of Eq. ( 1.24) are
X
s
= X
1
for X
0
< X
2
;
X
s
= X
2
for X
0
= X
2
; (1.26)
X
s
= X
3
for X
0
> X
2
.
Stability analysis shows that the solution X
2
is unstable with respect to
small perturbations, whereas the solutions X
1
and X
3
are stable. Moreover,
it follows from Eq. (1.26) that hysteresis may occur as X is varied [23]. The
last two results are typical of equilibrium phenomena which are described
near the liquidgas critical point by the classical equation of state, say,
the van der Waals equation of the form (1.25). This establishes the link
between ﬁrstorder phase transitions in equilibrium systems and socalled
hard transitions in reactive systems.
As an example of diﬀerent behavior, consider the chemical reaction
A + C + X A + 2X; X →Y + B (1.27)
where the second reverse reaction is neglected. The rate equation for the
number of molecules X has the following form
dX
dt
= −AX
2
+ (AC −1) X (1.28)
or, introducing τ = At and λ = (AC −1) /A,
dX
dτ
= −X
2
+ λX. (1.29)
The solution of this equation has the typical features of secondorder
phase transitions: the soft transition points X
s
(λ) are continuous at the
transition point λ = 0, but the derivatives are not.
The foregoing equations have to be generalized to include ﬂuctuations
from the steady state. No ﬁrstprinciple microscopic theory exists for ﬂuctu
ations in reactive chemical systems. One usually uses a phenomenological
master equation based on the macroscopic rate equations or a Langevin
equation obtained by adding a stochastic term to the rate equations. The
details can be found in [22]. Here we bring a fascinating example of the
inﬂuence of noise on the chemical reaction [24], which is illustrated by
the socalled ecological model. In this model, one considers two biological
species with densities n
1
(t) and n
2
(t), which decrease with rates a
1
and
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
14 Chemistry versus Physics: Chemical Reactions near Critical Points
a
2
and compete with rates b
1
and b
2
for the same renewable food m. The
birthdeath equations for these species have the following form
dn
1
dt
= (b
1
m−a
1
) n
1
;
dn
2
dt
= (b
2
m−a
2
) n
2
. (1.30)
The amount of food decreases both naturally (at a rate c), and according
to Eq. (1.30). Assume that the amount m of the food increases at rate q,
i.e.,
dm
dt
= q −cm−d
1
n
1
−d
2
n
2
. (1.31)
Assume now that the species n
1
is strong and species n
2
is weak, which
occurs when the ratio a
1
/b
1
is smaller than both a
2
/b
2
and q/c. Under these
conditions, the asymptotic t →∞ solutions of Eqs. (1.30) and (1.31) are
m =
a
1
b
1
; n
2
= 0; n
1
=
qb
1
−ca
1
b
1
d
1
(1.32)
which means that in the long run, the strong species survives and the weak
species becomes extinct. The question arises regarding which changes can
help the weak species to survive. One can easily see [24] that if one al
lows the food growth rate q to ﬂuctuate in time (replacing q in Eq. (1.31)
by q + f (t) with f = 0) or to allow the weak (but not the strong!)
species to be mobile (adding the diﬀusive term D∇
2
n
2
term in the second
of Eqs. (1.30)), the weak species will ﬁnally become extinct. This result
can be formulated as the “ecological theorem”: the longterm coexistence
of two species relying on the same renewable resources is impossible. How
ever, it has been shown [24] that if both these factors occur together with
nottoosmall food growth ﬂuctuations, the corrected equations (1.30) and
(1.31) have nonzero asymptotic solutions for both n
1
and n
2
. This result
has an important ecological interpretation. In a ﬂuctuating environment,
high mobility gives an evolutionary advantage that makes possible the co
existence of weak and strong species. The individuals of the mobile weak
species survive since they can utilize the food growth rate ﬂuctuations more
eﬀectively.
The eﬀect outlined above, which is called a “noiseinduced phase tran
sition” [25], also occurs in systems undergoing chemical reactions [24], as
well as in many other phenomena.
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Chapter 2
Eﬀect of Criticality on Chemical
Reaction
2.1 The eﬀect of pressure on the equilibrium constant and
rate of reaction
Dozens of experimental results for diﬀerent chemical reactions (solutes)
occurring in diﬀerent nearcritical substances (solvents) have been reported
in the chemical literature. One such example is the tautomeric equilibrium
in supercritical 1,1diﬂuoroethane [26], where the equilibrium rate constant
increased fourfold for a small pressure change of 4 MPa. Such a substantial
increase of the reaction rate by small variations of pressure has also been
found [27] in styrene hydroformylation in supercritical CO
2
. One further
example is the decomposition reaction of 2,2isobutyronitrile in mixed CO
2

ethanol solvent [28], where a 70–80% increase of the reaction rate near the
critical point resulted from a small pressure change, whereas outside the
critical region, the reaction rate is nearly independent of pressure.
There are many experimental data concerning the DielsAlder reaction
[29], [30], one of the most interesting and useful reactions in organic chem
istry and often used for the synthesis of sixmembered rings. The discover
ers of this reaction were awarded the Nobel Prize in Chemistry in 1950 “for
their discovery and development of diene synthesis”. In Wikipedia this
reaction is considered the “Mona Lisa” of reactions in organic chemistry
since it requires very little energy to create very useful cyclohexane rings.
The DealsAlder reaction has been studied both near the liquidgas critical
point of water, CO
2
and SF
6
, as well as near the critical point of binary
mixtures, such as CO
2
+ ethanol or CO
2
+ hexane. In Fig. 2.1, we show
[30] the dramatic increase of the reaction rate of the CO
2
+ ethanol mixture
as the pressure approaches the critical point for composition.
From a technological point of view, the CO
2
+ ethanol mixture has the
15
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
16 Chemistry versus Physics: Chemical Reactions near Critical Points
Fig. 2.1 Rate of reaction as a function of pressure for CO
2
ethanol mixture at T =
318.15 K near the critical points. Reproduced from Ref. [30] with permission, copyright
(2005), American Chemical Society.
advantage that the critical points are located at low pressures and tem
peratures. The measurement of the reaction rate of an isoprenemaleic
anhydride reaction near both the liquidgas and liquidliquid critical points
of the hexanenitrobenzene mixture demonstrates [31] anomalous enhance
ment near the liquidgas critical point, in contrast to a very small change
near the liquidliquid critical point.
We now consider the theoretical explanation for the strong inﬂuence
on chemical reactions of changes in pressure. It is not surprising that a
small change in the pressure near the liquidgas critical point of a pure
solvent causes a large change in densitydependent properties, such as the
solubility parameter, refractive index and the dielectric constant. To ac
count for these facts, it is enough to take into account the huge increase of
the compressibility (∂ρ/∂p)
T
near the liquidgas critical point. However,
the detailed inﬂuence on chemical reactions requires a more sophisticated
analysis.
To interpret the pressure eﬀect on reaction kinetics, one uses transition
state theory [32], according to which the activated complex C is formed
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Eﬀect of Criticality on Chemical Reaction 17
through the bimolecular reaction, expressed as
A+ B ←→C →P (2.1)
where A and B are the reactants and P is the product. The pressure
dependence of the rate constant K, expressed in mole fraction units, is
given by
RT
∂ ln K
∂p
= −∆V
∗
≡ V
C
−V
A
−V
B
(2.2)
where ∆V
∗
is the apparent activation volume created from the partial molar
volumes V
i
of component i. Being negative, partial molar volumes dramat
ically increase upon approaching the critical points, peaking at values of
−15, 000 cm
3
/gmol [33] compared with 5 −10 cm
3
/gmol far from the crit
ical point. The large values of the partial molar volumes near the critical
points of maximum compressibility are evident from the thermodynamic
relation
V
i
= υk
T
n
_
∂p
∂n
i
_
T,V,n
j=i
(2.3)
where k
T
is the isothermal compressibility of pure solvent, υ is the molar
volume of the solvent, n
i
is the number of moles of component i, and n is
the total number of moles of the mixture. Moreover, the limiting value V
cr
of the partial molar volumes at the critical point is pathdependent. When
the critical point is approached from above, the coexistence curve of the
binary mixture SF
6
CO
2
, V
cr
= −230 cm
3
/gmol and along the isothermal
isochoric path V
cr
= −40 cm
3
/gmol, whereas V
cr
of pure SF
6
equals 198
cm
3
/gmol.
2.2 Eﬀect of phase transformations on chemistry
One form of the interplay between chemical reactions and phase trans
formations is the change of a slow reaction in a liquid mixture in which
the transition from the twophase to the onephase region results from the
change of composition during the reaction. The reaction rate was deter
mined [34] by measuring the heat production w(t) as a function of time
with a heat ﬂow calorimeter. Thus, the chemical reaction itself serves as
a probe for obtaining information about critical phenomena. There are a
number of requirements for the selection of a suitable chemical reaction:
1. It must start in the twophase region and end in the onephase region
or vice versa.
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
18 Chemistry versus Physics: Chemical Reactions near Critical Points
Fig. 2.2 Heat ﬂow curves for the reaction of acetic anhydride with 1, 2ethanediol with
initial mole fractions x
0
AC
2
O
and x
0
TFA
, respectively. Reproduced from Ref. [34] with
permission, copyright (1989), Elsevier.
2. It must have a slow reaction rate to yield reliable data for the com
position of reactants.
3. The reaction enthalpy must be large enough for heat ﬂow calorimetry.
The authors of [34] have chosen the esteriﬁcation of 1, 2 ethanediol by
acetic anhydride in the presence of a catalytic amount of triﬂuoroacetic
acid. The results are shown in Fig. 2.2, where the diﬀerent curves relate
to diﬀerent initial amounts of reactants. The minimum heat production is
taken as the zero of the time scale so that time t is positive for onephase
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Eﬀect of Criticality on Chemical Reaction 19
regions and negative for twophase regions.
The curves are characterized by the following features:
1. Far from the phase transition, the curves have negative slope due to
the decrease of the reactant concentrations with time.
2. The curves exhibit a 10 − 20% jump in heat production in crossing
the frontier between twophase and onephase regions.
3. The jumps are almost symmetric with respect to the phase transition
point, and fade away as the phase transition occurs more distant from the
consolute point.
2.3 Critical slowingdown of chemical reactions
Experimental studies of chemical reactions near the critical point have a
long history. Although the general idea of the slowingdown of chemical re
actions is wellknown [35], theoretical interest in this problem was rekindled
in 1981 by our work [36]. For the reaction
k
i=1
ν
i
A
i
=
l
j=1
ν
j
A
j
, (2.4)
the observed rate of reaction r is the diﬀerence between the forward and
backward processes,
r =
−→
r −
←−
r = k
f
C
ν
1
1
C
ν
k
k
−k
b
C
ν
1
1
C
ν
l
l
(2.5)
where k
f
and k
b
are the microscopically determined rate coeﬃcients for the
forward and backward reactions, and C
i
are the concentrations (fugacities)
of the ith species in an ideal (nonideal) solution. Using the deﬁnition of
the chemical potentials of the mole of substance i, µ
i
= µ
0
i
+RT ln C
i
, and
the fact that [3]
k
f
k
b
= exp
_
_
_
_
k
i=1
ν
i
µ
0
i
−
l
j=1
ν
j
µ
0
j
_
_
/RT
_
_
(2.6)
one can rewrite Eq. (2.5) as
r = k
f
C
ν
1
1
C
ν
k
k
_
1 −
_
k
b
C
ν
1
1
C
ν
l
l
_
(k
f
C
ν
1
1
C
ν
k
k
)
−1
_
= k
f
C
ν
1
1
C
ν
k
k
[1 −exp (−A/RT)] , (2.7)
where A is the aﬃnity of reaction.
For chlorine dissociation, Cl
2
2Cl, the rate r is deﬁned as r =
k
f
C
Cl
2
−k
b
C
2
Cl
, and, using µ
i
= µ
0
i
+RT ln C
i
, and k
f
/k
b
= exp
ν
i
µ
0
i
/RT,
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
20 Chemistry versus Physics: Chemical Reactions near Critical Points
one gets
r = k
f
C
Cl
2
_
1 −exp
_
−
A
RT
__
. (2.8)
Equation (2.8) implies that, in general, the rate of reaction r is a nonlin
ear function of A. However, at equilibrium, A = 0, and for small deviations
from equilibrium, A/RT << 1, Eq. (2.7) takes the following form,
r = LA = L
_
∂A
∂ξ
_
T, p
_
ξ −ξ
eq
_
+ (2.9)
where L = k
f
C
ν
1
1
C
ν
k
k
/RT. Therefore, for small deviations from equi
librium, at ﬁxed temperature and pressure, one can restrict ourselves to the
ﬁrst term in the expansion of the aﬃnity A(T, p, ξ) in terms of the extent
of reaction ξ.
Another way to obtain Eq. (2.9) is to note that at equilibrium, where
the extent of reaction is equal to ξ
eq
, both dξ/dt and A vanish, and one
gets
dξ
dt
= L
_
∂A
∂ξ
_
T,p
_
ξ −ξ
eq
_
. (2.10)
A detailed analysis of Eq. (2.10) has been performed [37]. Before
discussing this work, one needs to recall some general thermodynamic
ideas. The thermodynamic state of a onecomponent, onephase system
is determined by two variables, say, temperature and pressure. Similarly,
the state of an ncomponent, onephase system is determined by n+1 vari
ables. These could be the temperature T and the chemical potential of each
component µ
i
(i = 1, 2, . . . , n).
As a function of these “ﬁeld” variables, the free energy G has the fol
lowing form
G = G(T, µ
1
...µ
n
) . (2.11)
However, for a reactive system, the chemical potentials are not independent.
According to the law of mass action in equilibrium,
A ≡
n
i=1
ν
i
µ
i
= 0. (2.12)
The unconstrained free energy G can be written as
G(T, µ
1
...µ
n
) = G(T, µ
1
...µ
n
) +ξ
n
i=1
ν
i
µ
i
(2.13)
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Eﬀect of Criticality on Chemical Reaction 21
where ξ is a Lagrange multiplier. By diﬀerentiating Eq. (2.13) with respect
to µ
i
, one obtains the particle number densities
n
i
= n
0
i
+ν
i
ξ. (2.14)
Equation (2.14) implies that for a given system (with initial n
0
i
), the
change in n
i
is completely determined by the extent ξ of the reaction. For
given T and p, the latter quantity deﬁnes the trajectories of all possible
thermodynamic states of a system in ndimensional phase space. However,
the complete stability analysis, at given T and p, requires taking into ac
count all possible changes of n
i
coming both from n
0
i
and ξ in Eq. (2.14).
In multicomponent systems, the stability with respect to diﬀusion is bro
ken before the thermal or mechanical stability. In terms of the Gibbs free
energy G(T, p, n
1
...n
n
) , the condition for diﬀusion stability for given T and
p has the form
d
2
G =
i,j
dn
i
µ
ij
dn
j
> 0; µ
ij
≡ ∂µ
i
/∂n
j
. (2.15)
One can easily show [3] that the matrix µ
ij
is semipositive, i.e., the
determinant of the matrix µ
ij
vanishes, which means that the rank of µ
ij
is equal to n − 1, and all the eigenvalues of µ
ij
but one are positive. Let
N
0
j
be the eigenvector corresponding to eigenvalue zero,
j
µ
ij
N
0
j
= 0. (2.16)
Equation (2.16) is the GibbsDuhem relation, which has the simple phys
ical meaning of invariance with respect to proportional change of all particle
numbers. On substituting (2.14) into (2.15), one obtains
d
2
G =
i,j
_
dn
0
i
µ
ij
dn
0
j
+ 2dn
0
i
µ
ij
ν
j
dξ +ν
i
µ
ij
ν
j
(dξ)
2
_
> 0. (2.17)
The form q ≡
i,j
x
i
µ
ij
x
j
vanishes only when the determinant of the
matrix µ
ij
vanishes (however, the opposite is not true, namely, for zero
determinant of the matrix, q may remain positive). Therefore, the exis
tence of a chemical reaction does not change the stability conditions. The
equivalence of the stability conditions with respect to diﬀusion becomes evi
dent when one considers [3] these conditions as resulting from homogeneous
and nonhomogeneous perturbations of an initially homogeneous system.
On the critical hypersurface, the expression (2.15) vanishes, i.e., new
zero eigenvalues of µ
ij
appear which correspond to the eigenvector N
j
sat
isfying
j
µ
ij
N
j
= 0. (2.18)
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
22 Chemistry versus Physics: Chemical Reactions near Critical Points
Among all points on the critical hypersurface, those worthy of notice occur
where the condition (2.15) is violated, and also
i,j
ν
i
µ
ij
ν
j
= 0. (2.19)
According to Eq. (2.17), this sum equals (dA/dξ)
T,p
, which, according to
Eq. (2.9), determines the slowingdown of chemical reactions.
Let us now return [37] to the main point — the comparison of Eqs.
(2.16), (2.18), and (2.19), which shows that the condition (2.19) for critical
slowingdown will be satisﬁed if and only if
ν
i
= αN
i
+βN
0
i
(2.20)
where α and β are arbitrary constants. There are n − 2 constraints in
(2.20), equal to the rank of the matrix µ
ij
on the critical hypersurface.
Together with two critical conditions, they deﬁne n conditions on n + 1
independent variables. Therefore, there is one free parameter, which is
ﬁxed by the law of mass action. As a result, slowingdown may appear
only at an isolated point on the critical hypersurface provided that all
the constraints are compatible. Each chemical reaction decreases by unity
the dimension of the critical hypersurface. For an ncomponent system
with n −1 chemical reactions (for example, a binary mixture with a single
chemical reaction), the only existing critical point will be where slowing
down occurs. For an ncomponent mixture with fewer than n −1 chemical
reactions, the chemical instability point will be an isolated point on the
critical hypersurface, which makes experimental veriﬁcation very diﬃcult.
However, in addition to a binary mixture with a single chemical reaction,
slowingdown is also very probable [37] for reactions in multicomponent
systems involving the separation of slightlydissolved substances from a
solvent.
Another attack on the slowingdown problem is the general Griﬃths
Wheeler method [38] of estimating the singularity of thermodynamic deriva
tives at the critical points. Indeed, according to Eq. (2.9), slowingdown is
determined by the thermodynamic derivative (dA/dξ)
T,p
, and the question
arises whether this derivative has the singularity at the critical point. Grif
ﬁths and Wheeler [38] divide all thermodynamic variables into two classes:
“ﬁelds”and “densities”. The ﬁelds (temperature, pressure, chemical poten
tials of the components) are the same in all phases coexisting in equilibrium,
whereas the densities (volume, entropy, concentrations of chemical compo
nents, extents of reactions) have diﬀerent values in each coexisting phase.
The power of the singularity of the thermodynamic derivatives depends on
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Eﬀect of Criticality on Chemical Reaction 23
the type of variables which are kept constant under the given experimental
conditions [38]. When the experimental conditions are such that only ﬁeld
variables are ﬁxed, with no ﬁxed density variables, then the derivative of
a ﬁeld with respect to a density, such as (∂A/∂ξ)
eq
, will approach zero
as [T −T
cr
[
x
with a “strong” critical index x of order unity. However, if
among the ﬁxed variables there is one density variable, the critical index
x is “weak” (of order 0.1). Finally, if two or more density variables are
ﬁxed, then x = 0, and the derivatives will be continuous. Applying the
GriﬃthsWheeler rule to chemical reactions near the critical points, one
has to calculate the number of ﬁxed densities. The latter will include all
inert components which are not participating in any chemical reaction.
From the GriﬃthsWheeler division of the thermodynamic variables into
“ﬁeld” and “density”, it follows that the approach to a critical point is path
dependent, as is shown in Fig. 2.3 for the critical point of a onecomponent
ﬂuid [39].
Fig. 2.3 Schematic pressuretemperature and densitypressure diagram for a pure ﬂuid.
Reproduced from Ref. [39] with permission, copyright (1987), International Union of
Pure and Applied Chemistry.
If one approaches the critical point along the critical isotherm, the
isothermal compressibility, (∂ρ/∂p)
T=T
C
, diverges with the strong critical
index x = 1.24. However, approaching the critical point along the criti
cal isochore results in diﬀerent behavior. Neither pressure or temperature
changes rapidly upon approaching the critical point, and the derivative
(∂p/∂T)
υ=υ
C
remains ﬁnite at the critical point. By contrast, the speciﬁc
heat at constant volume, C
v
≡ T (∂S/∂T)
v=v
cr
(derivative of a “density”
with respect to “ﬁeld” with one density held constant), is characterized by
the weak critical index x = 0.11. Similar arguments can be used for the
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
24 Chemistry versus Physics: Chemical Reactions near Critical Points
analysis of the thermodynamic variables near the critical points in many
component systems.
The preceding thermodynamic analysis is not complete, for the following
reason. Only homogeneous changes of the extent of reaction ξ are allowed
by Eq. (2.9). However, if a system is large enough, the k dependent
changes connected with sound, heat or diﬀusion modes become important
[40]. Indeed, the relaxation rates of the latter processes will become smaller
for some small wavenumber k than the ﬁxed k independent rate of the
chemical reaction. The general meanﬁeld approach, which goes back to
[35], is the straightforward generalization of (2.9). On the lefthand side
of the equation, instead of ξ, appears the column matrix of the density
type variables x, and instead of A, their conjugate ﬁelds appear (pressure
density, chemical potentialsconcentrations, temperatureentropy, etc.)
iωx(k, ω) = L(k) X (k, ω) . (2.21)
The ﬁeld variables X are related, in turn, to the density variables x by
the susceptibility matrix χ
−1
by X (k, ω) = χ(k)
−1
x(k, ω) (the constituent
equation), leading to
iωx(k, ω) = L(k) χ(k)
−1
x(k, ω) ≡ Mx(k, ω) . (2.22)
The matrix M = L(k) χ(k)
−1
is the hydrodynamic matrix, which we will
analyze in detail when we consider the hydrodynamic equations for a reac
tive system.
As emphasized in [40], Eq. (2.21) shows that the relaxation of the
extent of reaction ξ arises not only through the chemical processes (with
characteristic time τ ≡
_
L(∂A/∂ξ)
eq
_
−1
away from the critical point),
but also through diﬀusion and heat conductivity (with characteristic times
_
Dk
2
_
−1
and
_
λk
2
/ρC
p
_
−1
, respectively). Therefore, one must compare the
relaxation times or characteristic lengths (inverse wavenumbers) in order to
determine which quantity should be considered constant. As we have seen,
this is of crucial importance for the analysis of slowingdown processes.
Let us deﬁne two characteristic lengths by k
−2
C
= Dτ and k
−2
H
=
λτ/ρC
p
, where, generally, k
−1
H
>> k
−1
C
. Values of the wavenumbers k fall
into three intervals: k < k
H
, k
H
< k < k
C
, and k > k
C
. In the ﬁrst inter
val, the chemical reaction is more rapid than both heat conductivity and
diﬀusion, whereas in the third interval, diﬀusion dominates the chemical
relaxation. Only for processes with wavenumbers located in the “window”
between k
H
and k
C
is the relaxation dominated by chemistry. For typical
values of the heat conductivity, λ/ρC
p
= 0.1 cm
2
/s, the diﬀusion constant
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Eﬀect of Criticality on Chemical Reaction 25
D = 10
−5
cm
2
/s, and for a reaction rate τ
−1
= 120 Hz, we have k
H
= 10
3
cm
−1
and k
C
= 10
5
cm
−1
. Therefore, light scattering experiments with
k ≈ 10
4
cm
−1
at diﬀerent temperatures in the vicinity of the critical tem
perature provides evidence for the slowingdown of chemical reactions near
the critical points.
Note that the “slowingdown” of the chemical reaction does not imply
that the forward or backward reaction is slowed down. It is the measured
rate, which is the net diﬀerence between the forward and backward reac
tions, which is aﬀected by criticality. In fact, condition (2.19), which is re
ally (∂A/∂ξ)
T,p
≈ 0, means that the system becomes indiﬀerent to changes
in the species concentration. In equilibrium, when A = 0, the reaction is
balanced and the measured rate is zero. Usually, a change of ξ from ξ
eq
builds up an aﬃnity A ,= 0 which acts as a driving force to restore the
equilibrium. However, due to the thermodynamic properties in the critical
region, a change in ξ does not create a restoring force (i.e., aﬃnity), and
the reaction continues to be balanced even though ξ ,= ξ
eq
. That is, the
rate continues to be zero. In other words, the physical explanation of the
slowingdown of the chemical reaction is the same as the explanation of dif
fusion near the critical points of nonreactive systems, where the vanishing
of diﬀusion does not mean that the individual particles are slowingdown.
The only diﬀerence is that the chemical perturbation is homogeneous, in
contrast to the inhomogeneous diﬀusion changes.
All the above considerations hold for a twocomponent mixture with a
single chemical reaction. If the system has m reactants and n products, then
there are m+n−2 conserved variables. If all m+n species in the system take
part in the reaction, then the reaction can slowdown strongly for m+n < 3,
weakly for m + n = 3, and not at all for m + n > 3. Therefore, for more
complicated systems than a binary mixture, strong critical slowingdown
is not likely to occur in the usual experimental situation. The additional
factor which has to be taken into account in analyzing an experiment,
is the path of approach to the critical point (at some ﬁxed parameter(s),
along the coexistence curve, etc.), since the behavior of the thermodynamic
derivatives is pathdependent.
2.4 Hydrodynamic equations of reactive binary mixture:
piston eﬀect
The slowingdown of the dynamic processes near the liquidgas critical
points, considered in the previous section, is the hallmark of critical phe
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
26 Chemistry versus Physics: Chemical Reactions near Critical Points
nomena. It is precisely this fact which makes it so diﬃcult to obtain reliable
equilibrium experimental data near the critical points.
However, the very interesting phenomenon of the speedingup of dy
namic processes in critical ﬂuids at constant volume, as opposed to their
slowingdown, was predicted theoretically [41], and afterwards found ex
perimentally [42]. The usual experimental setup is a ﬂuidﬁlled closed cell.
After changing the temperature at the bottom or around the cell, one waits
for the establishment of thermal equilibrium inside the cell. The slow diﬀu
sive heat propagation is responsible for the critical slowingdown. However,
fast thermal equilibrium is established through the thermoacoustic eﬀect
[41], where the temperature change at the surface induces acoustic waves
which lead to a fast change of the pressure, and hence, of the tempera
ture everywhere in the ﬂuid. This is also called the “piston eﬀect”, since
the thermal boundary layer generated near the heated bottom of the cell
produces, like a piston, the pressure on the ﬂuid restricted to a constant
volume.
The speedingup of an interface reaction, which is immediately evident
from the piston eﬀect, has been explained in [43]. There are also other
explanations for the speedingup of chemical reactions near critical points
based on speciﬁc properties, such as appearance of the soft modes in the
vibrational spectrum of solids [44], reduction of the Hbond strength in wa
ter [45], and the solidsolvent clustering mechanism [46]. We shall consider
some special examples.
2.4.1 Heterogeneous reactions in nearcritical systems
The interface reaction that takes place in a supercritical phase may be cited
as an illustration of an immediate inﬂuence of the piston eﬀect on the ﬂux
of matter at the interface, leading to the speedingup of the chemical re
action [43], [47]. Such speedingup manifests itself in the strong corrosion
observed in supercritical ﬂuid containers [48]. Two factors — the sensitivity
of solubility to the small changes of pressure, induced by the piston eﬀect,
and the supercritical hydrodynamics — aﬀect the absorptiondesorption
reaction taking place between a solid interface and an active component in
a dilute binary supercritical mixture. The method of matched asymptotic
expansion has been used [43] for the solution of the simpliﬁed system of
onedimensional hydrodynamic equations with a van der Waals equation of
state and linear mixing rules. The asymptotic analysis shows the coupling
between the piston eﬀect and an interface reaction, which — for some ini
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Eﬀect of Criticality on Chemical Reaction 27
tial conditions — leads to a strong increase of the ﬂux of matter and the
intensiﬁcation of a reaction.
More detailed analysis has been performed [47]. The numerical solution
of the full hydrodynamic equations based on the ﬁnite volume approxima
tion has been carried out for a system involving a small amount of naphtha
lene dissolved in supercritical CO
2
located between two inﬁnite solid plates
made of pure solid naphthalene and able to absorb or desorb naphthalene
from the ﬂuid. The ﬂuid is initially at rest. The temperature of one of the
plates is then gradually increased while the other plate is kept at the initial
temperature. Due to the small amount of naphthalene in solution, for nu
merical simulation one can use the parameters of pure CO
2
. The results of
the analysis show that there are three mechanisms for the increase of the
mass fractions of the naphthalene w at the solid interface:
1. The critical behavior of the solute solubility through the large values
of (∂w/∂p)
T
, and (at the heated plate) of (∂w/∂T)
p
.
2. The strong compressibility of the supercritical ﬂuid through the
derivative (∂w/∂p)
T
, and the piston eﬀect that results through the pressure
variation.
3. The strong dilatability expressed by the large derivative (∂w/∂T)
p
and through the density variation.
The piston eﬀect, coupled with the critical behavior of the solute sol
ubility with respect to pressure, is the leading mechanism governing the
mass fraction of naphthalene at the isothermal plate [47]. Similar simu
lations have been performed with cooling, rather than heating one of the
plates, and a similar phenomenon was observed: the strong decrease of the
pressure caused by the piston eﬀect leads to a strong decrease of the mass
fraction of naphthalene at the isothermal plate.
In the next section, we carry out a phenomenological analysis based on
the full hydrodynamic equations in a reactive mixture [49], [50].
2.4.2 Dynamics of chemical reactions
There are two diﬀerent phenomenological ways to describe chemical reac
tions: the hydrodynamic approach, used by physicists and which we use
here, and the rate equations, used mostly by chemists.
We consider an isomerization reaction between two species which is de
scribed by the equation A B. This simple example of a chemical reaction
can be easily generalized to more complex reactions. The question arises as
to the choice of the thermodynamic variables which characterize the binary
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
28 Chemistry versus Physics: Chemical Reactions near Critical Points
mixture. A onecomponent thermodynamic system is characterized by two
parameters, such as pressure and temperature. For a binary mixture, one
has to add one additional parameter, such as the concentration ξ of one of
the components, which will also describe a chemical reaction. Indeed, for bi
nary mixtures, the chemical potential µ, deﬁned [4] as µ = µ
1
/m
1
−µ
2
/m
2
,
where m
1
and m
2
are the molecular masses of the components, is propor
tional to the aﬃnity A = µ
1
ν
1
− µ
2
ν
2
of the chemical reaction since the
stoichiometric coeﬃcients ν
1
, ν
2
are inversely proportional to the molecular
masses of the species. Thus, one can choose the mass fraction of one com
ponent as the progressive variable ξ of a chemical reaction. The dynamic
behavior of a reactive binary mixture, described by the hydrodynamic equa
tions, will contain a fourth variable, which for an isotropic system can be
chosen as div (v).
As discussed in the previous section, near equilibrium in the linear ap
proximation, one obtains
dξ
dt
= −L
0
A = −L
0
_
dA
dξ
_
eq
_
ξ −ξ
eq
_
. (2.23)
The subscript eq means that the thermodynamic derivative is calcu
lated in equilibrium, and L
0
is the Onsager coeﬃcient. The quantity
_
L
o
(∂A/∂ξ)
eq
_
−1
is the characteristic time of the chemical reaction de
scribed by the phenomenological equation (2.23). In real system, however,
the chemical reaction occurs in some media, and the “chemical” mode in
teracts with other hydrodynamic modes, such as diﬀusive, viscous and heat
conductive modes, i.e., one has to consider the entire system of hydro
dynamic equations. This requires calculating the “renormalized” Onsager
coeﬃcient L due to the interaction with the hydrodynamic modes (“mode
coupling”). In equilibrium statistical mechanics, there exists a general pro
cedure for calculating the thermodynamic quantities, which goes back to
Gibbs. No such systematic procedure exists in nonequilibrium statisti
cal mechanics. However, for nonequilibrium states that are not far from
equilibrium, some general formulae do exist which connect the kinetic coeﬃ
cients with longwavelength limits of the correlation functions. The Onsager
coeﬃcient plays a role analogous to that of transport coeﬃcients in hydro
dynamics, and both are described by the GreenMoriZwanzig relations
[51]–[53]. The “chemical” mode is characterized by the local interactions
whereas the pure hydrodynamic modes depend upon spatial gradients. In
a real ﬂuid, there are couplings between diﬀerent modes, and the Onsager
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Eﬀect of Criticality on Chemical Reaction 29
coeﬃcient L is deﬁned for small wavenumbers k, as the correlation func
tion of the changes of the concentration (∂ξ/∂t)
chem
due to the chemical
reaction [54],
L =
1
κ
B
T
0
∞
_
0
dt
__
∂ξ
∂t
_
chem
(k, t)
_
∂ξ
∂t
_
chem
(−k, 0)
_
. (2.24)
Equation (2.24) provides the connection between the phenomenological
Onsager coeﬃcient and the microscopic time correlation function. First of
all, let us check that in the absence of the hydrodynamic modes, when the
aﬃnity A depends only on the concentration ξ rather than on all thermody
namic variables, the renormalized Onsager coeﬃcient L, deﬁned in (2.24),
reduces to the “bare” Onsager coeﬃcient L
0
. Indeed, the solution of Eq.
(2.23) has the form
ξ −ξ
eq
= exp
_
−L
0
_
dA
dξ
_
eq
t
_
. (2.25)
Substituting (2.23) and (2.25) into (2.24) yields
L = L
2
0
_
dA
dξ
_
2
eq
1
κ
B
T
0
∞
_
0
dt exp
_
−L
0
_
dA
dξ
_
eq
t
_
¸ξ (k, 0) ξ (−k, 0)) .
(2.26)
Performing the integration over t in (2.26), and taking into account
[4] that the equilibrium correlation function ¸ξ (k, 0) ξ (−k, 0)) equals
κ
B
T
0
/ (dA/dξ)
eq
, one immediately obtains that L = L
0
, as requires for
a purely chemical process. In the following section, we will consider the in
ﬂuence of the hydrodynamic modes on the renormalized Onsager coeﬃcient.
2.4.3 Relaxation time of reactions
Consider the isomerization reaction with c
1
and c
2
being the mass con
centrations of substances A and B, and γ
ij
are the transition rates from
component i to component j. The change of the concentration of compo
nent A caused by gain and loss to component B, is described by
dc
1
dt
= γ
21
c
2
−γ
12
c
1
= γ
21
−(γ
12
+γ
21
) c
1
. (2.27)
For a closed system the total number of particles is constant, c
1
+c
2
= 1,
which has been used in Eq. (2.27). This equation contains the characteristic
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
30 Chemistry versus Physics: Chemical Reactions near Critical Points
time of the reaction r
0
which depends on the rate constants γ
12
and γ
21
at
equilibrium,
r
0
=
1
γ
12
+γ
21
. (2.28)
At equilibrium, dc
1
/dt = 0, and
γ
12
c
0
1
= γ
21
c
0
2
. (2.29)
The kinetic processes lead to the renormalization of this characteristic time,
which is described by Eq. (2.24) with a slight change in the constant
coeﬃcient,
r =
1
c
0
1
c
0
2
∞
_
0
dt
__
∂c
∂t
_
chem
(k, t)
_
∂c
∂t
_
chem
(−k, 0)
_
. (2.30)
If a system is open, the particles are able to enter and leave. Then, the
concentrations depend both on chemical transformations and on diﬀusion,
and their time dependence is described by the following equations [55],
∂c
1
/∂t = D
1
∇
2
c
1
−γ
12
c
1
+γ
21
c
2
,
∂c
2
/∂t = D
2
∇
2
c
1
−γ
21
c
2
+γ
12
c
1
.
(2.31)
We use the Fourier and Laplace transforms in space and time, respec
tively,
c
i
(k, s) =
__
drdt exp (−ikr) exp (−st) c
i
(r, t) . (2.32)
The solutions of Eqs. (2.31) are
c
1
(k, s) =
_
s +k
2
D
1
+γ
21
_
c
1
(k, 0) +γ
21
c
2
(k, 0)
∆
,
c
2
(k, s) =
_
s +k
2
D
2
+γ
12
_
c
2
(k, 0) +γ
12
c
1
(k, 0)
∆
,
(2.33)
where
∆ = s
2
+s
_
(D
1
+D
2
) k
2
+γ
21
+γ
12
¸
+k
2
(D
1
γ
12
+D
2
γ
21
) +k
4
D
1
D
2
. (2.34)
Multiplying Eqs. (2.33) by c
1
(−k, 0) and c
2
(−k, 0) and averaging, one
obtains the expressions for the matrix elements ¸c
i
(k, s) c
j
(−k, 0)) , for
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Eﬀect of Criticality on Chemical Reaction 31
i, j = 1, 2, as functions of the equilibrium averages ¸c
i
(k, 0) c
j
(−k, 0)) . For
special systems, such as structurally similar isomers, for which D
1
= D
2
≡
D, one gets
¸c
1
(k, s) c
1
(−k, 0))
=
_
s +k
2
D + γ
21
¸
¸c
1
(k, 0) c
1
(−k, 0)) +γ
21
¸c
2
(k, 0) c
1
(−k, 0))
s
2
+s
_
γ
+
+ 2k
2
D
_
+k
2
γ
+
D +k
4
D
2
, (2.35)
¸c
2
(k, s) c
2
(−k, 0))
=
_
s +k
2
D + γ
12
¸
¸c
2
(k, 0) c
2
(−k, 0)) +γ
12
¸c
1
(k, 0) c
2
(−k, 0))
s
2
+s
_
γ
+
+ 2k
2
D
_
+k
2
γ
+
D +k
4
D
2
, (2.36)
¸c
1
(k, s) c
2
(−k, 0))
=
_
s +k
2
D + γ
21
¸
¸c
1
(k, 0) c
2
(−k, 0)) +γ
21
¸c
2
(k, 0) c
2
(−k, 0))
s
2
+s
_
γ
+
+ 2k
2
D
_
+k
2
γ
+
D +k
4
D
2
, (2.37)
¸c
2
(k, s) c
1
(−k, 0))
=
_
s +k
2
D + γ
12
¸
¸c
2
(k, 0) c
1
(−k, 0)) +γ
12
¸c
1
(k, 0) c
1
(−k, 0))
s
2
+s
_
γ
+
+ 2k
2
D
_
+k
2
γ
+
D +k
4
D
2
, (2.38)
where γ
+
= γ
21
+γ
12
.
Equations (2.35)–(2.38) are simpliﬁed when both components are taken
to be ideal gases. Then, ¸c
i
(k, 0) c
j
(−k, 0)) = c
0
j
δ
ij
, so that
¸c
1
(k, s) c
1
(−k, 0))
= c
0
1
s +k
2
D +γ
21
s
2
+s (γ
21
+γ
12
+ 2k
2
D) +k
2
γ
+
D +k
4
D
2
, (2.39)
¸c
2
(k, s) c
2
(−k, 0))
= c
0
2
_
s +k
2
D +γ
12
¸
s
2
+s (γ
21
+γ
12
+ 2k
2
D) +k
2
γ
+
D +k
4
D
2
, (2.40)
¸c
1
(k, s) c
2
(−k, 0))
= c
0
2
γ
21
s
2
+s (γ
21
+γ
12
+ 2k
2
D) +k
2
γ
+
D +k
4
D
2
, (2.41)
¸c
2
(k, s) c
1
(−k, 0))
= c
0
1
γ
12
s
2
+s (γ
21
+γ
12
+ 2k
2
D) +k
2
γ
+
D +k
4
D
2
. (2.42)
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
32 Chemistry versus Physics: Chemical Reactions near Critical Points
Substituting into (2.30) the expression (∂c/dt)
chem
= γ
21
c
2
−γ
12
c
1
, one
obtains
r =
1
c
0
1
c
0
2
∞
_
0
dt
∞
_
0
ds exp(st) ¦γ
2
12
¸c
1
(k, s) c
1
(−k, 0))
+γ
2
21
¸c
2
(k, s) c
2
(−k, 0))
−γ
12
γ
21
[¸c
1
(k, s) c
2
(−k, 0)) +¸c
2
(k, s) c
1
(−k, 0))]¦ (2.43)
which takes the following form after using Eqs. (2.39)–(2.42),
r =
_
γ
2
12
c
0
1
+γ
2
21
c
0
2
_
c
0
1
c
0
2
∞
_
0
dt
∞
_
0
ds
_
s +k
2
D
_
exp(st)
s
2
+s
_
γ
+
+ 2k
2
D
_
+k
2
γ
+
D +k
4
D
2
. (2.44)
To perform the inverse Laplace transformation on Eq. (2.44), one needs
the roots of the denominator,
s
1
= −k
2
D; s
2
= −k
2
D −γ
+
. (2.45)
Integrating over s yields
r =
γ
2
12
c
0
1
+γ
2
21
c
0
2
c
0
1
c
0
2
(γ
12
+γ
21
)
∞
_
0
exp
_
−
_
k
2
D +γ
+
_
t
¸
dt
=
γ
2
12
c
0
1
+γ
2
21
c
0
2
c
0
1
c
0
2
(γ
12
+γ
21
) (k
2
D + γ
21
+γ
12
)
(2.46)
which can be rewritten by using Eq. (2.29),
r =
1
(k
2
D +γ
21
+γ
12
)
. (2.47)
For D = 0, this solution reduces to Eq. (2.28), as required.
2.4.4 Hydrodynamic equations of a reactive binary
mixtures
The hydrodynamic equations are the conservation laws of mass, momen
tum, concentration and energy. They are the continuity equation,
dρ
dt
= −div (ρv) , (2.48)
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Eﬀect of Criticality on Chemical Reaction 33
the NavierStokes equation,
ρ
dv
i
dt
= −
∂
∂x
j
σ
ij
, (2.49)
the conservation of one of the components,
dξ
dt
= −div (J
ξ
) , (2.50)
and the conservation of energy,
dQ
dt
= ρT
dS
dt
= −div (J
) . (2.51)
Just as ρv in Eq. (2.48) represents the mass ﬂux, the terms σ
ij
, J
ξ
, and
J
in (2.49)–(2.51) describe the stress tensor, ﬂuxes of mass of one of the
components and energy, respectively. The phenomenological forms of these
quantities, which include both hydrodynamic and chemical ﬂows, are based
on the symmetry requirements for vectors J, v and scalars ρ, ξ, S and the
chemical aﬃnity A.
The wellknown expressions for hydrodynamic ﬂuxes are [4]
(σ
ij
)
hydr
= δ
ij
p +η
_
∂v
i
∂x
j
+
∂v
j
∂x
i
−
2
3
δ
ij
div (v)
_
+δ
ij
ζdiv (v) , (2.52)
(J
ξ
)
hydr
= −Dgrad (ξ) −D
_
k
T
T
grad (T) +
k
p
p
grad (p)
_
, (2.53)
(J
)
hydr
= −λgrad (T)−ρD
_
∂A
∂ξ
_
p,T
_
k
T
grad (ξ) +
k
p
p
grad (p)
_
. (2.54)
The variables appearing in (2.48)–(2.54) have obvious meaning. The
coeﬃcient k
T
deﬁnes the Dufour and Soret crosseﬀects, whereas k
p
/p is a
purely thermodynamic quantity deﬁned by
k
p
p
=
(∂A/∂p)
T,ξ
(∂A/∂ξ)
p,T
. (2.55)
For small deviations from chemical equilibrium, the constitutive equa
tion takes the form
_
dξ
dt
_
chem
= −L
0
A. (2.56)
The heat change in the chemical reaction is given by
λ =
∆h
C
(2.57)
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
34 Chemistry versus Physics: Chemical Reactions near Critical Points
where C is the heat capacity, and ∆h is the change of enthalpy per unit
mass of reaction. The heat ﬂux caused by the chemical reaction is
_
dQ
dt
_
chem
= λL
0
A. (2.58)
Since ∆h is proportional to A, the λLA term in Eq. (2.58) is proportional
to A
2
, and can therefore be neglected for small deviations from equilibrium,
as is assumed when using the hydrodynamic theory.
The phenomenological relations (2.56) and (2.52)–(2.54) are based on
symmetry considerations which — in the presence of a chemical reaction —
allow the appearance of an additional viscoreactive cross eﬀects, described
by the L
1
terms. This leads to the following generalization of Eqs. (2.56)
and (2.52),
_
dξ
dt
_
chem
= −L
0
A−L
1
div (v) (2.59)
(σ
ij
)
chem
= δ
ij
L
1
A. (2.60)
Inserting (2.53), (2.59), and (2.60) into (2.49)–(2.51), yields four
linearized hydrodynamic equations,
∂ρ
∂t
= −ρ
0
div (v) (2.61)
ρ
0
∂v
∂t
= −grad (p) +η∇
2
v +
_
η
3
+ζ
_
grad (div (v)) +ρ
0
L
1
grad (A) (2.62)
∂ξ
∂t
= D∇
2
ξ +D
_
k
T
T
∇
2
T +
k
p
p
∇
2
p
_
−L
0
A −L
1
div (v) (2.63)
ρT
∂S
∂t
= λ∇
2
T +ρD
_
∂A
∂ξ
_
p,T
_
k
T
∇
2
ξ +
k
p
p
∇
2
p
_
(2.64)
where ρ, p, T and ξ are the deviations of density, pressure, temperature and
the mass fraction of one of the components from their equilibrium values
denoted by the subscript zero. The transport coeﬃcients in Eqs. (2.61)–
(2.64) are the shear viscosity η, the bulk viscosity ς, the heatconductivity
λ, the diﬀusion coeﬃcient D, the thermaldiﬀusion coeﬃcient k
T
, and the
barodiﬀusion coeﬃcient k
p
.
The rigorous treatment of the correlation functions, which, according
to Eq. (2.24), deﬁne the renormalized Onsager coeﬃcient L, requires that
allowance be made for the symmetry properties of the variables entering the
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Eﬀect of Criticality on Chemical Reaction 35
hydrodynamic equations (2.61)–(2.64). The quantity (∂ξ/∂t)
chem
and the
variable div (v) are odd functions under time reversal and even functions
under parity [36], while the other variables are either even or odd under
both symmetry operations. However, these arguments are not relevant in
the following, since we use simpliﬁed models.
Using Eq. (2.61), one can omit div (v) from Eqs. (2.62 )–(2.63), which
gives
−
∂
2
ρ
∂t
2
= −∇
2
p −
1
ρ
0
_
4
3
η +ζ
_
∇
2
∂ρ
∂t
+ρ
0
L
1
∇
2
A (2.65)
∂ξ
∂t
= D∇
2
ξ +D
_
k
T
T
∇
2
T +
k
p
p
∇
2
p
_
−L
0
A +
L
1
ρ
0
∂ρ
∂t
. (2.66)
To solve Eqs. (2.64)–(2.66), we express A, T and p in terms of the vari
ables ρ, ξ, S. In equilibrium, A = 0. For small deviations from equilibrium,
A =
_
∂A
∂ρ
_
ξ,S
ρ +
_
∂A
∂S
_
ρ,ξ
S +
_
∂A
∂ξ
_
S,ρ
ξ. (2.67)
Our analysis is based on the assumption of local equilibrium, i.e.,
T = (∂T/∂ρ)
ξ,S
ρ + (∂T/∂S)
ρ,ξ
S + (∂T/∂ξ)
S,ρ
ξ,
p = (∂p/∂ρ)
ξ,S
ρ + (∂p/∂S)
ρ,ξ
S + (∂p/∂ξ)
S,ρ
ξ.
(2.68)
Inserting (2.67) and (2.68) into (2.64)–(2.66) yields [56] a quite cumber
some system of three diﬀerential equations in three variables ρ, ξ, S. The
Fourier and Laplace transforms in space and time, respectively, yield the
following solution of these equations,
ξ (k, s) =
f
1
(k, s) ξ (k, 0) +f
2
(k, s) ρ (k, 0) +f
3
(k, s) S (k, 0)
∆
, (2.69)
ρ (k, s) =
f
4
(k, s) ξ (k, 0) +f
5
(k, s) ρ (k, 0) +f
6
(k, s) S (k, 0)
∆
, (2.70)
S (k, s) =
f
7
(k, s) ξ (k, 0) +f
8
(k, s) ρ (k, 0) +f
9
(k, s) S (k, 0)
∆
, (2.71)
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
36 Chemistry versus Physics: Chemical Reactions near Critical Points
where the determinant of the coeﬃcients in the original equations, ∆, and
the functions f
i
(i = 1, . . . , 9) are cumbersome combinations of the thermo
dynamic variables and transport coeﬃcients, which can be found in [56].
The complicated hydrodynamic equations can be simpliﬁed if one ne
glects the higher terms in k
2
. Performing the FourierLaplace transforms
of Eqs. (2.65) and (2.66), written in the variables ρ and ξ, one gets
K
1,ρ
ρ (k, s) + K
1,ξ
ξ (k, s)
=
_
s −
1
ρ
0
_
4
3
η +ζ
_
k
2
_
ρ (k, 0) +∂ρ (k, 0) /∂t,
K
2,ρ
ρ (k, s) + K
2,ξ
ξ (k, s) = ξ (k, 0) −(L
1
/ρ
0
) ρ (k, 0) , (2.72)
where one can set ∂ρ (k, 0) /∂t = 0, and
K
1,ρ
= s
2
+k
2
_
_
∂p
∂ρ
_
ξ,S
+
_
4
3
γ +ζ
_
s
ρ
0
+ρ
0
L
1
_
∂A
∂ρ
_
ξ,S
_
, (2.73)
K
1,ξ
= k
2
_
_
∂p
∂ξ
_
ρ,S
+ρ
0
L
1
_
∂A
∂ξ
_
ρ,S
_
, (2.74)
K
2,ρ
= L
0
_
∂A
∂ρ
_
ξ,S
+Dk
2
_
∂ξ
∂ρ
_
ξ,S
−
sL
1
ρ
0
, (2.75)
K
2,ξ
= s +k
2
D +L
0
_
∂A
∂ξ
_
ρ,S
. (2.76)
The solutions of Eqs. (2.72) are
ρ (k, s)
=
_
K
2,ξ
_
s −
1
ρ
0
_
4
3
η +ζ
_
k
2
_
−K
1,ξ
L
1
ρ
0
_
ρ (k, 0) −K
1,ξ
ξ (k, 0)
K
1,ρ
K
2,ξ
−K
2,ρ
K
1,ξ
(2.77)
ξ (k, s)
=
K
1,ρ
_
ξ (k, 0) −
L
1
ρ
0
ρ (k, 0)
_
−K
2,ρ
_
s −
1
ρ
0
_
4
3
η +ζ
_
k
2
_
ρ (k, 0)
K
1,ρ
K
2,ξ
−K
2,ρ
K
1,ξ
.
(2.78)
To lowest order in k
2
, Eq. (2.77) takes the following form
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Eﬀect of Criticality on Chemical Reaction 37
ρ (k, s) = sρ (k, 0)
_
s
2
+
_
sk
2
/ρ
0
_
_
4
3
γ +ζ
_
+
_
(∂p/∂ρ)
ξ,S
+ρ
0
L
1
(∂A/∂ρ)
ξ,S
_
k
2
+
_
(∂p/∂ξ)
ξ,S
+ρ
0
L
1
∂A/∂ξ
ξ,S
_ _
sL
1
/ρ
0
−L
0
(∂A/∂ξ)
ρ,S
_
_
s +L
0
(∂A/∂ξ)
ρ,S
_
−1
k
2
_
−1
≈ sρ (k, 0)
_
_
s +
_
k
2
/2ρ
0
_
_
4
3
γ +ζ
__
2
+
_
(∂p/∂ρ)
ξ,S
+ρ
0
L
1
(∂A/∂ρ)
ξ,S
_
k
2
_
−1
. (2.79)
The inverse Laplace transform yields
ρ (k, t) = ρ (k, 0) exp
_
−
1
2ρ
0
_
4
3
γ +ζ
_
k
2
t
_
cos
_
_
_
_
_
∂p
∂ρ
_
ξ,S
+L
1
_
∂A
∂ρ
_
ξ,S
_
1/2
kt
_
_
_
. (2.80)
Analogously, to lowest order in k
2
, one gets from Eq. (2.78),
ξ (k, t)
=
_
ξ (k, 0) −
L
1
ρ
0
ρ (k, 0)
_
exp
_
−
_
Dk
2
+L
0
_
∂A
∂ξ
_
ρ,S
_
t
_
. (2.81)
According to Eq. (2.24), the rate of the chemical reaction is deﬁned by
the change of concentration (∂ξ/∂t)
chem
due to the chemical reaction. In
the case under consideration, this change is given by the term −L
0
A in
Eq. (2.66). Using ξ and ρ as independent variables, one gets
L =
_
L
2
0
/k
B
T
0
_
_
∞
0
dt ¸A(k, t) A(−k, 0))
=
_
L
2
0
/k
B
T
0
_
(∂A/∂ξ)
2
ρ,S
_
∞
0
dt[¸ξ (k, t) ξ (−k, 0))
+ (∂ξ/∂ρ)
2
A,S
¸ρ (k, t) ρ (−k, 0))
− (∂ξ/∂ρ)
A,S
(¸ξ (k, t) ρ (−k, 0)) +¸ρ (k, t) ξ (−k, 0)))]. (2.82)
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
38 Chemistry versus Physics: Chemical Reactions near Critical Points
The correlators in Eq. (2.82) can easily be found from Eqs. (2.80) and
(2.81) as functions of equilibrium correlators
¸ρ (k, 0) ρ (−k, 0)) = κT
0
ρ
2
0
(∂p/∂ρ)
−1
ξ,S
,
¸ξ (k, 0) ξ (−k, 0)) = κT
0
(∂A/∂ξ)
−1
ρ,S
,
¸ξ (k, 0) ρ (−k, 0)) = −κT
0
(∂ρ/∂A)
ξ,S
.
(2.83)
Finally, one obtains
L = L
2
0
(∂A/∂ξ)
2
p,S
_
_
_
_
_
L
1
ρ
0
[Dk
2
+L
0
(∂A/∂ξ)
ρ,S
]
−1
+
_
∂ξ
∂ρ
_
A,S
2 (4η/3 +ζ) ρ
0
(4η/3 +ζ)
2
k
2
+ 4
_
(∂p/∂ρ)
ξ,S
+ρ
0
L
1
(∂A/∂ρ)
ξ,S
_
_
_
_
(∂ρ/∂A)
ξ,S
+ρ
2
0
(∂ξ/∂ρ)
A,S
(∂ρ/∂p)
ξ,S
_
_
_
_
. (2.84)
2.4.5 Hydrodynamic equations with statistically
independent variables
The thermodynamic variables x
1
, x
2
, . . . , x
n
are statistically independent
if their crossaverages ¸x
i
x
j
) vanish for i ,= j, so that the probability of
ﬂuctuations is given by
P (δx
1
, δx
2
, . . . , δx
n
) =
exp
_
−
A
1
2
(δx
1
)
2
−
A
2
2
(δx
2
)
2
− −
A
n
2
(δx
n
)
2
_
(2.85)
with
_
(δx
i
)
2
_
= A
−1
i
.
For a onecomponent system, one can choose ρ, T or p, S as statistically
independent variables, since ¸δρδT) = ¸δpδS) = 0. For a manycomponent
system, there are diﬀerent possibilities for choosing the set of the hydrody
namic variables x
1,
x
2
, . . . , which characterize the state of a system. Using
wellknown thermodynamic formulae [4], one can verify that there are dif
ferent sets of statistically independent variables. We have chosen the mass
fraction x, the pressure p, div (v) and the variable φ = C
p,x
T/T
0
−α
T
p/ρ
0
,
where α
T
is the isobaric thermal expansion and C
p,x
is the speciﬁc heat at
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Eﬀect of Criticality on Chemical Reaction 39
constant pressure. Another possible set of statistically independent vari
ables is the mass fraction x, the pressure p, div (v) and the reduced entropy
S
1
= S − (∂S/∂x)
p,T
x, or the temperature T, the pressure p, div (v) and
the variable φ. Note that for a chemically reactive ternary mixture, it is
convenient [57] to use the reduced entropy and the reduced concentration,
S
1
= S −(∂S/∂x)
p,T
x
1
; x = x
1
−(∂x
1
/∂x
2
)
p,T
x
2
. (2.86)
After performing a Fourier transform, the set of linearized hydrody
namic equations can be written symbolically in the form
d
dt
N (k, t) = −M (k, t) N (k, t) (2.87)
where N (k, s) is a column vector with n hydrodynamic variables (n = 4
for the case discussed below), and the form of the n ∗ n matrix
ˆ
M(k, s) is
deﬁned by the hydrodynamic equations. Solutions of Eqs. (2.87) have the
following form
N
i
(k, s) = [det M (k, s)]
−1
j
P
ij
(k, s) N
j
(k) , (2.88)
where the P
ij
are algebraic functions. From Eq. (2.88), one gets the corre
lation functions
¸N
j
(k, s) N
j
(−k)) =
P
ij
(k, s)
det M (k, s)
_
[N
j
(k)[
2
_
, (2.89)
if all N
j
are statistically independent.
An expression for the timedependent correlation function is obtained
by taking the inverse Laplace transform of Eq. (2.89),
¸N
j
(k, t) N
j
(−k)) =
_
[N
j
(k)[
2
_
2π
∞
_
−∞
ds exp (st) P
ij
(k, s)
det M (k, s)
. (2.90)
To perform the integration in (2.90), one has to ﬁnd the roots of
det M (k, s).
Neglecting the viscoreactive terms (setting L
1
= 0 in Eqs. (2.62)–
(2.63)), one can write the complete set of linearized hydrodynamic equa
tions in the variables p, x, and div (v) and the new variable φ = C
p,x
T/T
0
−
α
T
p/ρ
0
. This choice of hydrodynamic variables has been used for non
reactive mixtures [58] , and then extended to reactive mixtures [59], [60].
These equations take the following form [60], [61]
_
∂ρ
∂x
_
p,T
∂x
∂t
+
_
∂ρ
∂p
_
x,T
∂p
∂t
+
T
0
C
p,x
_
∂ρ
∂T
_
x,p
∂φ
∂t
+ ρ
0
div (v) = 0
(2.91)
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
40 Chemistry versus Physics: Chemical Reactions near Critical Points
∂div (v)
∂t
= −∇
2
p +
_
4
3
η +ς
_
∇
2
div (v) ; (2.92)
∂x
dt
= D∇
2
x +D
_
k
p
p
0
+
k
T
α
T
ρ
0
C
p,x
_
∇
2
p +D
k
T
C
p,x
∇
2
φ −L
0
A; (2.93)
ρ
0
T
0
(∂φ/∂t) −ρ
0
k
T
(∂A/∂x)
p,T
(∂x/∂t)
−T
0
_
α
T
−ρ
0
(∂S/∂p)
x,T
_
(∂p/∂t)
= λ
_
(T
0
/C
p,x
) ∇
2
φ +
_
Tα
T
/ρ
0
C
p,x
_
∇
2
p
¸
. (2.94)
The chemical term L
0
A appears in Eq. (2.93). Since the aﬃnity A
vanishes at equilibrium, one can expand A in terms of x, p, and T,
A =
_
∂A
∂x
_
p,T
x +
_
∂A
∂p
_
x,T
p +
_
∂A
∂T
_
x,p
T. (2.95)
Substituting T = T
0
φ/C
p,x
+T
0
α
T
p/ρ
0
C
p,x
into Eq. (2.95) yields
L
0
A = L
0
(∂A/∂x)
p,T
_
x −(∂x/∂p)
T,A
p
− (∂x/∂T)
p,A
(T
0
φ/C
p,x
+T
0
α
T
p/ρ
0
C
p,x
)
_
= L
0
(∂A/∂x)
p,T
_
x −
_
(∂x/∂p)
T,A
+ (∂x/∂T)
p,A
T
0
α
T
/ρ
0
C
p,x
_
p
− (T
0
/C
p,x
) (∂x/∂T)
p,A
φ
_
. (2.96)
Inserting (2.96) into (2.93) gives
∂x/dt =
_
D∇
2
−L
0
(∂A/∂x)
p,T
_
x
+
_
D(k
p
/p
0
+k
T
α
T
/ρ
0
C
p,x
) ∇
2
−L
0
_
(∂A/∂p)
x,T
+ (∂A/∂T)
x,p
T
0
α
T
/ρ
0
C
p,x
__
p
+
_
Dk
T
C
p,x
∇
2
+ (∂A/∂x)
p,T
(∂x/∂T)
p,A
T
0
L
0
/C
p,x
_
φ. (2.97)
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Eﬀect of Criticality on Chemical Reaction 41
Finally, inserting (2.97) into (2.94) yields
∂φ/∂t = (∂A/∂x)
p,T
k
T
/T
0
_
D∇
2
− L
0
(∂A/∂x)
p,T
_
x
+
_
(λ/ρ
0
C
p,x
) ∇
2
+k
T
/T
0
_
(Dk
T
/C
p,x
) (∂A/∂x)
p,T
∇
2
− (T
0
L
0
/C
p,x
) (∂A/∂x)
2
p,T
(∂x/∂T)
p,A
__
φ
+
__
α
T
/ρ
0
−(∂S/∂p)
x,T
_
(∂p/∂t) + (∂A/∂x)
p,T
k
T
/T
0
_
_
D(k
p
/p
0
+k
T
α
T
/ρ
0
C
p,x
) ∇
2
¸
−L
0
_
(∂A/∂p)
x,T
+ (∂A/∂T)
x,p
T
0
α
T
/ρ
0
C
p,x
__
+
_
λT
0
α
T
/ρ
0
C
p,x
_
∇
2
_
p. (2.98)
Equations (2.91), (2.92), (2.97) and (2.98) are the complete system of hy
drodynamic equations.
For a binary mixture without a chemical reaction (L
0
= 0), the roots
of the determinant of the matrix M (k, t) and the appropriate correlation
functions have been found to lowest order in k [61]. In many practical cases,
there exist small parameters, such as
kλ
v
s
ρ
0
C
p,x
;
k (4η/3 +ζ)
3v
s
ρ
0
;
kD
v
s
. (2.99)
To lowest order in these parameters, setting the determinant of the 44
matrix
ˆ
M (k, s) to zero yields two zeros and two sound modes with roots,
s
1,2
= ±iv
s
k. In the next higherorder approximation, there are two sound
modes and two entangled heat conductivitydiﬀusion modes [61],
s
1,2
= ±iv
s
k −Γk
2
, (2.100)
s
3,4
=
k
2
2
_
λ/ρ
0
C
p,x
+D + (Dk
T
/T
0
C
p,x
)
2
(∂A/∂x)
p,T
_
±
k
2
2
_
_
λ/ρ
0
C
p,x
+D + (Dk
T
/T
0
C
p,x
)
2
(∂A/∂x)
p,T
_
2
− 4Dλ/ρ
0
C
p,x
_
1/2
(2.101)
Γ =
1
2ρ
0
_
4
3
η +ζ + (C
p,A
/C
v,x
−1) λ/C
p,x
−
_
(∂ρ/∂x)
p,T
+
k
T
C
p,x
(∂ρ/∂T)
x,p
(∂A/∂x)
p,T
_
2
D(∂p/∂ρ)
x,T
ρ
0
(∂A/∂x)
p,T
_
.
(2.102)
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
42 Chemistry versus Physics: Chemical Reactions near Critical Points
One can readily use the hydrodynamic equation including the chemical
term (L
0
,= 0) and perform the analogous calculations, but this procedure
leads to very cumbersome expressions. Therefore, we consider the inﬂuence
of a chemical reaction for the simpliﬁed case of a noncompressible liquid
(div (v) = 0), neglecting ﬂuctuations of pressure (p = 0) [60]. The compli
cated matrix M (k, t) is then simpliﬁed, and the remaining hydrodynamic
equations (2.97) and (2.98) have the following form
∂x
dt
+a
11
x +a
12
φ = 0 (2.103)
∂φ
dt
+a
21
x +a
22
φ = 0 (2.104)
where
a
11
= Dk
2
+L
0
_
∂A
∂x
_
p,T
,
a
12
= Dk
T
k
2
/C
p,x
−(L
0
T
0
/C
p,x
)
_
∂A
∂x
_
p,T
_
∂x
∂T
_
p,A
,
a
21
=
_
Dk
T
k
2
/T
0
_
_
∂A
∂x
_
p,T
+ (k
T
L
0
/T
0
)
_
∂A
∂x
_
2
p,T
,
a
22
= λk
2
/ρ
0
C
p,x
+
_
k
2
T
/T
0
C
p,x
_
_
∂A
∂x
_
p,T
Dk
2
−(k
T
L
0
/C
p,x
)
_
∂A
∂x
_
2
p,T
_
∂x
∂T
_
p,A
. (2.105)
It is convenient to express these elements by the characteristic times of
diﬀusion τ
d
, heat conductivity τ
h
, and chemical reaction τ
c
, deﬁned by
τ
d
=
_
Dk
2
_
−1
; τ
h
=
_
λ
ρ
0
C
p,x
k
2
_
−1
; τ
c
=
_
L
0
_
∂A
∂x
_
p,T
_
−1
. (2.106)
The matrix elements a
ij
can then be rewritten,
a
11
= τ
−1
d
+τ
−1
c
; a
12
=
k
T
C
p,x
τ
−1
d
−
T
0
C
p,x
_
∂x
∂T
_
p,A
τ
−1
c
,
a
21
=
k
T
T
0
_
∂A
∂x
_
p,T
τ
−1
d
+
k
T
T
0
_
∂A
∂x
_
p,T
τ
−1
c
,
a
22
= τ
−1
h
+
k
2
T
T
0
C
p,x
_
∂A
∂x
_
p,T
τ
−1
d
−
k
T
C
p,x
_
∂A
∂x
_
p,T
_
∂x
∂T
_
p,A
τ
−1
c
.
(2.107)
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Eﬀect of Criticality on Chemical Reaction 43
The determinant ∆ of the 2 2 matrix M
1
of Eqs. (2.103)–(2.104) has
the following form
∆ = (a
11
−s) (a
22
−s) −a
12
a
21
. (2.108)
To solve Eqs. (2.103)–(2.104), one needs the roots s
1
and s
2
of ∆. The
inverse Laplace transforms of the solutions of Eqs. (2.103) and (2.104),
yields
∞
_
0
dt
∞
_
0
ds exp (st) ¸x(k, t) x(−k, 0)) = (a
22
/s
1
s
2
) ¸x(k, 0) x(−k, 0)) ,
∞
_
0
dt
∞
_
0
ds exp (st) ¸φ(k, t) φ(−k, 0)) = (a
11
/s
1
s
2
) ¸φ(k, 0) φ(−k, 0)) ,
∞
_
0
dt
∞
_
0
ds exp (st) ¸x(k, t) φ(−k, 0)) = −(a
12
/s
1
s
2
) ¸φ(k, 0) φ(−k, 0)) ,
∞
_
0
dt
∞
_
0
ds exp (st) ¸φ(k, t) x(−k, 0)) = −(a
21
/s
1
s
2
) ¸x(k, 0) x(−k, 0)) .
(2.109)
The following calculations are analogous to those performed in
Eq. (2.82), but instead of variables ξ and ρ, we use now the statistically
independent variables x and φ. Using Eqs. (2.109), one obtains
L =
_
L
2
0
/κ
B
T
0
_
_
∞
0
dt ¸A(k, t) A(−k.0))
=
1
s
1
s
2
κ
B
T
0
τ
2
c
__
a
21
T
0
C
p,x
_
∂x
∂T
_
p,A
+a
22
_
¸x(k, 0) x(−k, 0))
+
T
0
C
p,x
_
∂x
∂T
_
p,A
_
a
12
+
a
11
T
0
C
p,x
_
∂x
∂T
_
p,A
_
¸φ(k, 0) φ(−k, 0))
_
.
(2.110)
Substituting into Eq. (2.110) the equilibrium correlators
¸x(k, 0) x(−k.0)) =
κ
B
T
0
(∂A/∂x)
p,T
; ¸φ(k, 0) φ(−k.0)) = κ
B
C
p,ξ
(2.111)
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
44 Chemistry versus Physics: Chemical Reactions near Critical Points
one gets
L =
1
s
1
s
2
τ
2
c
__
a
21
T
0
C
p,x
_
∂x
∂T
_
p,A
+a
22
_
(∂A/∂x)
−1
p,T
+
_
∂x
∂T
_
p,A
_
a
12
+
a
11
T
0
C
p,x
_
∂x
∂T
_
p,A
__
. (2.112)
According to Eq. (2.108), the product of the roots s
1
s
2
equals a
11
a
22
−
a
12
a
21
. Substituting the latter into (2.112), and using Eq. (2.107), one
ﬁnally obtains
L =
L
0
τ
c
_
_
_
1
τ
h
+
k
2
T
τ
d
T
0
C
p,x
_
∂A
∂x
_
p,T
_
1 +
T
0
k
T
_
∂x
∂T
_
A,S
_
2
_
_
_
_
1
τ
h
_
1
τ
d
+
1
τ
c
_
+
k
T
τ
d
τ
c
C
p,x
_
_
∂S
∂x
_
p,T
+
k
T
T
0
_
∂A
∂x
_
p,T
_
_
1 +
T
0
k
T
_
∂x
∂T
_
)
A,S
__
−1
. (2.113)
The goal of our analysis was to ﬁnd the mode coupling corrections to
the Onsager coeﬃcient, which deﬁne the change of the characteristic time
of a chemical reaction τ
c
= [L
0
(∂A/∂ξ)]
−1
. Such “mode coupling” is
expressed by replacing the bare Onsager coeﬃcient L
0
by the renormalized
coeﬃcient L, so that the eﬀective characteristic time of a chemical reaction
is [L(∂A/∂ξ)]
−1
. In the general case, such a calculation involves a very
cumbersome analysis. To avoid this, we consider some special cases.
When one neglects the thermal variables, viscous, diﬀusive and chemical
modes appear in Eq. (2.84),
L = L
2
0
_
∂A
∂ξ
_
2
p,S
_
L
1
/ρ
0
Dk
2
+L
0
(∂A/∂ξ)
ρ,S
+
2
_
4
3
η +ζ
_
ρ
0
(∂ξ/∂ρ)
A,S
_
4
3
η +ζ
_
2
k
2
+ 4ρ
0
L
1
(∂A/∂ρ)
ξ,S
_
_
_
∂ρ
∂A
_
ξ,S
+ρ
2
0
_
∂ξ
∂ρ
_
A,S
_
∂ρ
∂p
_
ξ,S
_
. (2.114)
Note that the diﬀusive mode appears in Eq. (2.114) because of the
additional phenomenological viscoreactive coeﬃcient L
1
introduced in
Eqs. (2.59)–(2.60).
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Eﬀect of Criticality on Chemical Reaction 45
It is convenient to consider some limiting cases of Eq. (2.113):
Fast chemical reaction (L
0
_
dA
dx
_
p,T
>>
λ
ρ
0
C
p
k
2
≈ Dk
2
):
L =
L
0
_
1 +
τ
h
τ
d
k
2
T
T
0
C
p,x
_
∂A
∂x
_
p,T
_
_
1 +
T
0
k
T
_
∂x
∂T
_
p,A
_
2
__
1 +
τ
h
τ
d
k
T
C
p,x
_
_
∂S
∂x
_
p,T
+
k
T
T
0
_
∂A
∂x
_
p,T
_ _
1 +
T
0
k
T
_
∂x
∂T
_
p,A
_
. (2.115)
If heat conductivity processes are faster than diﬀusive processes,
(L
0
_
dA
dξ
_
p,T
>>
λ
ρ
0
C
p
,x
k
2
>> Dk
2
), the presence of a chemical reaction
does not change the Onsager coeﬃcient, L = L
0
. In the opposite limit,
(L
0
_
dA
dx
_
p,T
>> Dk
2
>>
λ
ρ
0
C
p,x
k
2
),
L = L
0
(k
T
/T
0
) (∂A/∂x)
p,T
(1 + (T
0
/k
T
) (∂x/∂T)
p,A
)
(∂S/∂x)
p,T
+ (k
T
/T
0
) (∂A/∂x)
p,T
. (2.116)
Fast heat conductivity (
λ
ρ
0
C
p,x
k
2
>> L
0
_
dA
dx
_
p,T
≈ Dk
2
):
L = L
0
L
0
(dA/dx)
p,T
L
0
(dA/dx)
p,T
+Dk
2
. (2.117)
For fast diﬀusion processes (
λ
ρ
0
C
p,x
k
2
>> Dk
2
>> L
0
(dA/dx)
p,T
),
L = L
0
L
0
(dA/dx)
p,T
Dk
2
, (2.118)
whereas in the opposite limit, (
λ
ρ
0
C
p,x
k
2
>> L
0
_
dA
dx
_
p,T
>> Dk
2
), L = L
0
.
Fast diﬀusion (Dk
2
>>
λ
ρ
0
C
p,x
k
2
≈ L
0
_
dA
dx
_
p,T
):
L = L
2
0
_
k
2
T
T
0
C
p,x
__
∂A
∂x
_
2
p,T
_
1 +
T
0
k
T
_
∂x
∂T
_
p,A
_
2
_
λk
2
ρ
0
C
p,x
+L
0
k
T
C
p,x
_
∂A
∂x
_
p,T
_
_
∂S
∂x
_
p,T
+
k
T
T
0
_
∂A
∂x
_
p,T
__
1 +
T
0
k
T
_
∂x
∂T
_
p,A
__
−1
. (2.119)
If Dk
2
>>
λ
ρ
0
C
p
,x
k
2
>> L
0
_
dA
dx
_
p,T
,
L =
L
2
0
_
k
2
T
T
0
C
p,x
_
_
∂A
∂x
_
2
p,T
_
1 +
T
0
k
T
_
∂x
∂T
_
p,A
_
2
_
λk
2
/ρ
0
C
p,x
_
, (2.120)
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
46 Chemistry versus Physics: Chemical Reactions near Critical Points
whereas for Dk
2
>> L
0
_
dA
dx
_
p,T
>>
λ
ρ
0
C
p
k
2
,
L = L
0
k
T
T
0
_
∂A
∂x
_
p,T
_
1 +
T
0
k
T
_
∂x
∂T
_
p,A
_
_
∂S
∂x
_
p,T
+
k
T
T
0
_
∂A
∂x
_
p,T
. (2.121)
Our hydrodynamic approach applies to the regime where k
−1
is larger
than the maximal characteristic size of a system. This is the correlation ra
dius, which increases as the critical point is approached. Therefore, the ﬁrst
important consideration in the analysis of experimental data is the prox
imity to the critical point. Equations (2.114)–(2.121) also contain the ther
modynamic derivatives and the kinetic coeﬃcients, some of which change
drastically near the critical points.
Even without analyzing speciﬁc reactive systems, one can see from
Eqs. (2.114)–(2.121), that these equations contain all three types of be
havior are described in [62]–[65] (slowingdown, speedingup, unchanged).
2.5 Critical anomalies of chemical equilibria
In addition to singularities in the rate of reaction, peculiar behavior of equi
librium concentrations of reagents takes place near the critical point. The
dissociationrecombination reaction of the form B
2
2B near the single
liquidgas critical point of the binary mixture of B
2
and B has been consid
ered in [66], and improved in [67]. Common sense dictates that increasing
(decreasing) the temperature should result in increasing (decreasing) the
concentration of monomers in all dissociationrecombination reactions. In
deed, both energy (due to bond energy) and entropy (due to the number
of particles) considerations seem to lead to the same conclusion. Quite
unexpectedly, the opposite has been observed in qualitative experiments
performed by Krichevskii and his collaborators [68] who studied the equi
librium reaction N
2
O
4
2NO
2
in a dilute solution of CO
2
near the critical
point of the mixture. While another interpretation of these experiments
exists [67], [69], it does not deny the existence of the anomalies in chemi
cal equilibrium near the critical point. Consider varying the temperature
along an equilibrium line of a reactive binary mixture B
2
2B at constant
pressure. Since the aﬃnity A = µ
B
2
− 2µ
B
vanishes for any equilibrium
state, along the equilibrium line one can write
dA =
_
∂A
∂T
_
ξ,p
dT +
_
∂A
∂ξ
_
T,p
dξ = 0 (2.122)
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Eﬀect of Criticality on Chemical Reaction 47
where dξ = dn
B
2
−dn
B
/2 is the change in the extent of the reaction.
It follows from Eq. (2.122) that
_
dξ
dT
_
p, equil.line
= −
(∂H/∂ξ)
T,p
T (∂A/∂ξ)
T,p
(2.123)
where the thermodynamic identity [3]
_
∂A
∂T
_
ξ,p
=
_
∂S
∂ξ
_
T,p
=
A + (∂H/∂ξ)
T,p
T
(2.124)
has been used, and H and S are the enthalpy and entropy, respectively.
The derivative (∂H/∂ξ)
T,p
has no critical singularity. Therefore, the slope
of concentration curves ξ = ξ (T) is inversely proportional to the same
derivative (∂A/∂ξ)
T,p
which determines the critical slowingdown. Hence,
the derivative (dξ/dT)
p, equil.line
exhibits a strong divergence at the critical
point which is characterized [40] by critical indices γ or (δ −1) /δ, depend
ing on whether the critical point is approached at constant critical volume
or pressure.
We used [70] similar arguments to explain the technologically important
phenomenon of “supercritical extraction” — the substantial increase in the
ability of nearcritical ﬂuids to dissolve solids. Fluids near critical points
exist as a single phase having some of the advantageous properties of both
a liquid and a gas. Indeed, they have the dissolving power comparable to
a gas, transport properties intermediate between gas and liquid, and great
sensitivity to small changes in pressure aﬀorded by high compressibility.
Let s and f denote the solid phase and the ﬂuid phase, respectively.
The equality of the chemical potentials in the two phases is a necessary
condition for the coexistence of these phases. Thus, for the solid,
µ
s
(T, p) −µ
f
(T, p, x) ≡ ∆µ = 0 (2.125)
where we used the fact that the concentration x of the solid phase is equal to
unity. Consider now isobaric changes in temperature along the equilibrium
line,
d (∆µ) =
_
∂µ
s
∂T
_
p
dT −
_
∂µ
f
∂T
_
p,x
dT −
_
∂µ
f
∂x
_
T,p
dx = 0. (2.126)
Equation (2.126) can be rewritten as
_
∂x
∂T
_
p
=
s −s
s
(∂µ
f
/∂x)
T,p
(2.127)
where we introduce the molar entropy s of the solid s
s
= −(∂µ
s
/∂T)
p
and
the partial entropy of the solid in the ﬂuid phase s = −
_
∂µ
f
/∂T
_
p,x
.
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
48 Chemistry versus Physics: Chemical Reactions near Critical Points
Equations (2.126)–(2.127), which are analogous to (2.122)–(2.123), ex
plain supercritical extraction. Indeed, near the critical points, the denomi
nator in (2.127) approaches zero, leading to a strong divergence of the slope
of the solubility curve.
The system described above contains a small amount of solid dissolved
in a ﬂuid. One may use the binary mixture as the solvent and the solid as
the solute. We have suggested [71] the use for supercritical extraction in
the neighborhood of a consolute point rather than at the liquidgas critical
point. The former is characterized by much lower pressures and tempera
tures, which has clear technological advantages.
2.6 Experiment
Experiments investigating the interplay between criticality and chemistry
were performed long before the theorists showed any interest in this prob
lem. The ﬁrst publication of the oxidation reaction in the critical region
goes back to 1946 [72]. Mention should be made of a series of qualita
tive experiments performed by the prominent Russian physical chemist
Krichevskii and his collaborators [68]. In two experiments, they irradiated
with light pure Cl
2
and I
2
in CO
2
. In both cases, the irradiation results
in the dissociation of the diatomic molecules into atoms. Under usual con
ditions, when the irradiation stops, rapid recombination restores chemical
equilibrium. However, when the experiment is conducted under critical
conditions (near the liquidgas critical point of pure Cl
2
or of the binary
CO
2
I
2
mixture), the recombination rate is dramatically reduced by one or
two orders of magnitude. When the interpretation of these experiments led
to some doubts [69], this problem came to the attention of the theorists, in
cluding myself. A few interesting experimental results were lately reported
by S. Greer and collaborators. They measured the rate of recombination
of chlorine atoms after thermal and chemical dissociation of chlorine near
its liquidvapor critical point [73]. They found that after a thermal pertur
bation, the recombination was very slow, whereas the recombination after
the photolysis with ultraviolet light was very fast, contrary to the results
obtained in [68]. Additional experiments have been performed [74] near the
liquidliquid critical point of a binary mixture with the aim of ﬁnding the
change in the relative amounts of NO
2
and N
2
O
4
during the dimerization
of NO
2
. Using the static dielectric constant as the measure of the extent
of dimerization, they found 4% increase (and not decrease!) in dissociation
near the critical point.
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Eﬀect of Criticality on Chemical Reaction 49
A review of early development in this ﬁeld can be found in [75]. Special
attention should be paid to the Snyder and Eckert article: “Chemical kinet
ics at a critical point” [76], where, in contrast to the wellknown “slowing
down”, they found “speedingup” (increase by some 40%) of the reaction
rate of the isoprene with maleic anhydride in the vicinity of the upper crit
ical solution point of two liquidliquid mixtures (hexanenitrobenzene and
hexanenitroethane). This result was obtained more than twenty years be
fore the theoretical idea of the “piston eﬀect”, and authors proposed an
explanation related to the dilute solution of the reactants.
Extensive studies of the rate of chemical reactions in the vicinity of the
critical point were carry out by J. K. Baird and collaborators. In their 1998
article [62], they described the measurement of the rate of ﬁve hydrolysis
ﬁrstorder reactions near the consolute point of three binary mixtures (tert
amyl chloride and tertbutyl chloride in isobutyric acidwater, 3chloro3
methylpentane and 4methylbenzyl bromide in 2butoxyethanolwater, tert
butyl bromide in triethylaminewater). All these reactions produce a strong
electrolyte, and the progress of the reaction was followed by measuring
the conductance in the vicinity of the lower and upper critical consolute
points. All measurements showed a slowingdown eﬀect within a few tenths
of a degree on either side of the consolute temperature, in agreement with
previous experiments. The measurement of 2chloro2methylbutane near
the critical point of the mixture isobutyric acidwater, has been repeated
[63]. The more careful, later experiment (system was wellstirred in the two
phase region, and account was taken of the small shift ∆T
c
= 0.2 K of the
critical temperature of the solvent due to the presence of the reactant) yields
diﬀerent results. Above T
C
, the rate of reaction showed slowingdown,
while below T
C
, in the twophase region, the rate of reaction increased
upon approaching the critical point (“speedingup” predicted by the “piston
eﬀect”). To check whether the speedingup of the reaction rate is connected
with the coexistence of two immiscible phases, the experiment was carried
out on 2bromo2methylpropane in triethylaminewater, analogous to that
performed in [63]. In contrast to isobutyric acidwater used in [63], the
latter mixture has a lower critical solution temperature. In spite of this
diﬀerence, the results of the experiments were the same: slowingdown
above the critical temperature and speedingup below, which means that
the “speedingup” eﬀect is associated with the critical point and not with
the number of phases in the system.
Another improvement over the original experiments was that visual de
tection of critical opalescence to determine the critical temperature was
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
50 Chemistry versus Physics: Chemical Reactions near Critical Points
replaced by the association of the critical temperature with the tempera
ture of the onset of the sharp change in electrical conductivity when critical
opalescence occurs [64]. The authors measured the rate of the Menschutkin
reaction between benzyl bromide and triethylamine near the consolute point
of triethylaminewater, which can be compared with the old measurements
[76] of the Menshutkin reaction on ethyl iodide. For temperatures below T
C
,
they could clearly see the critical slowingdown of the reaction rate show
ing the weak singularity, while above T
C
, a few measured points closest to
T
C
gave a hint of speedingup. These results agree with those obtained
earlier [76].
The next question investigated concerned the generality of the eﬀect
found. In particular, does the eﬀect exist for other types of reactions, say,
of secondorder, and not of ﬁrstorder kinetics? To answer this question, the
experiment was performed [65] with the saponiﬁcation of ethyl acetate by
sodium hydroxide near the consolute point of the 2butoxyethanolwater,
which is known to exhibit [77] secondorder kinetics. The experiment [65]
shows slowingdown for temperatures below T
C
. For T > T
C
, the results of
the experiment are not clear: one cannot a rule out speedingup, but only
future experiments will deﬁnitively answer this question.
An important comment has been emphasized in [62], [63] regarding the
quantitative description of the slowingdown and speedingup of the reac
tion rate. The rate of reaction is proportional to thermodynamic deriva
tive of the Gibbs free energy diﬀerence ∆G separating products from re
actance with respect to the extent of reaction ξ. As described in section
2.1, the singularity x of this quantity, (∂∆G/∂ξ)
a,b,...
∼ [(T −T
C
) /T
C
[
x
,
at the critical point depends on number of ﬁxed “density type” variables
a, b (entropy, volume, concentrations of the chemical components, extent
of reaction). Namely, if one density is ﬁxed, then one ﬁnds a “weak” sin
gularity (x = 0.11 or x = 0.33, depending on the thermodynamical path
approaching the critical point — ﬁxed pressure or temperature or along the
coexistence curve). For densities not held ﬁxed, the singularity is “strong”
(x = 1.24 or x = 4.9). Finally, for two or more ﬁxed densities, x = 0 and
there is no singularity of the reaction rate at the critical point. In turn, the
number of ﬁxed densities, i.e., those which do not participate in at least
one chemical reaction, depends on the number of reactions occurring in the
system. Estimating the number of reactions is not trivial, as seen from
the following example [63]. One of the components of a solvent (isobutyric
acid), which does not participate in the hydrolysis reaction, might never
theless react with the alcohols present in the system, thereby changing the
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Eﬀect of Criticality on Chemical Reaction 51
behavior of the reaction rate by changing the number of reactions and the
number of densities held ﬁxed.
In addition to conductance measurements, the same experimental group
used another method to analyzing the progress of a chemical reaction [78],
namely, the measurement of the rate of appearance of a phase that is not
completely miscible with a liquid mixture, such as the gas phase. Three
reactions have been studied: 1) acetone dicarboxylic acid in aniline cat
alyzed decomposition of acetone dicarboxylic acid in isobutyric acidwater,
2) benzene diazonium tetraﬂuoroborate in the decomposition of benzene di
azonium tetraﬂuoroborate in 2butoxyethanolwater, 3) ethyl diazoacetate
in the acid catalyzed decomposition of ethyl diazoacetate in isobutyric acid
water. In all cases, the critical point did not eﬀect the rates of the reactions.
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Chapter 3
Eﬀect of Chemistry on Critical
Phenomena
3.1 Change of critical parameters due to a chemical
reaction
Chemical reactions near the critical points are routinely studied near the
consolute points of a binary mixture which plays the role of a solvent. The
presence of a reactant, as well as impurities, pressure or electric ﬁeld changes
the critical parameters of a binary mixture. There is only limited knowl
edge of the relationship between the shifts of the critical parameters when
the perturbation is applied. For example, it is known [3] that an impurity
will cause a critical temperature shift equal to that of the critical compo
sition only when the solubilities in the two components are roughly equal.
However, it was found that two diﬀerent impurities (water [79] and acetone
[80]), which have quite diﬀerent solubilities in the two components of the
methanolcyclohexane mixture, induce equal relative shifts in critical tem
perature and composition, ∆T
C
/T
C,0
= ∆x
C
/x
C,0
. The same relation has
been observed [81] in a wide range of systems (Fig. 3.1). These data have
been obtained by isobaric measurements upon the addition of impurities.
There are diﬀerent experimental ways of ﬁnding the critical parameters
T
C
and p
C
of a system. The simplest, but less precise method is based
on visual detection of critical opalescence to determine the critical temper
ature. For ionic conductors, the critical temperature is identiﬁed by the
sharp change in electrical conductivity [64]. Alternatively, one can use the
dilatometric method of measuring the height of the liquid in the capillary
sidearm as a function of temperature. A change in the slope of this curve
occurs at the critical temperature [62]. Another method is the acoustic
technique, using the fact that the velocity of sound reaches a minimum at
the critical point [82]. Experiments show that the critical temperature T
C
53
September 14, 2009 16:35 World Scientiﬁc Book  9in x 6in Generalnn
54 Chemistry versus Physics: Chemical Reactions near Critical Points
Fig. 3.1 The shift of critical points T
C,0
and x
C,0
of diﬀerent systems to new values
T
C
and x
C
. The diﬀerent symbols refer to ﬁve diﬀerent systems (for details, see [81]).
Reproduced from Ref. [81] with permission, copyright (1989), American Institute of
Physics.
is the linear function of the initial concentration x of a reactant,
T
C
= T
0
C
+ax. (3.1)
The sign of the constant a depends on whether the chemical reaction
enhances or inhibits the mutual solubility of the original solvent pair. In the
former case, a < 0 for UCST, and a > 0 for LCST, and vice versa for the
latter case. For the equilibrium mixture of benzyl bromide in triethylamine
water
T
0
C
= 291.24 K
, it was found [64] that a = 36.9 K/mole. There are
many experimental results related to diﬀerent manycomponent mixtures
[82]. Measurements of the shift of the critical temperature in the presence of
a chemical reaction have recently been performed for indium oxide dissolved
in the nearcritical isobutyric acidwater mixture [83].
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Eﬀect of Chemistry on Critical Phenomena 55
3.2 Modiﬁcation of the critical indices
In this section, we describe the changes in critical phenomena arising from
the presence of chemical reactions. The singularities of the thermodynamic
and kinetic quantities near the critical points are determined by the critical
indices. The calculations of these indices assumes the constancy of some
intensive variables, say, pressure p or chemical potentials µ, whereas in
practice it is impossible, for example, to ensure the constancy of µ (one has
to vary the concentrations during the course of the experiment). Similarly,
p remains practically unchanged near the liquidliquid critical point since
the experiment is carried out in the presence of saturated vapor. On the
other hand, it is quite diﬃcult to ensure the constancy of p near the liquid
gas critical point where a ﬂuid is highly compressible.
In order to compare theoretical calculations for an ideal onecomponent
system with experimental data obtained for “real” systems, Fisher [84] de
veloped the theory of renormalization of critical indices. The main idea of
renormalization can be explained by the following simple argument. For
a pure substance, mechanical stability is determined by (∂p/∂v)
T
< 0.
The condition for diﬀusion stability in a binary mixture has the form
(∂µ/∂x)
T,p
> 0, which, by a simple thermodynamic transformation, can
be rewritten as (∂p/∂v)
T,µ
< 0. Therefore, at constant chemical poten
tial µ, the critical behavior of a binary mixture, which is deﬁned by the
stability conditions, will be the same as that of a pure substance. Mul
ticomponent systems behave in an analogous manner, namely, the stabil
ity condition for a ncomponent mixture is determined by the condition
(∂p/∂v)
T,µ
1
...µ
n−1
< 0, where n − 1 chemical potentials are held constant.
However, the critical parameters now depend on the variables µ
i
, and in
order to obtain experimentally observable quantities which correspond to
constant concentration x, one has to pass from T (µ) to T (x). As a result,
when one goes from a pure substance to a binary mixture, the critical indices
above and below a critical point are multiplied by ±(1 − α)
−1
[84], where
α is the critical index of the speciﬁc heat at constant volume. The minus
sign refers to the speciﬁc heat at constant volume and the plus sign refers
to all other critical indices. We have discussed [85] the renormalization of
the critical indices associated with the presence of chemical reactions. As
an example, let us again consider a binary ﬂuid mixture A
1
−A
2
with the
single reaction ν
1
A
1
+ ν
2
A
2
= 0 (say, the isomerization reaction). This
equation means that the law of mass action is satisﬁed, i.e., one gets for
the extent of reaction A = ν
1
µ
1
+ ν
2
µ
2
= 0, which, in turn, reduces by
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
56 Chemistry versus Physics: Chemical Reactions near Critical Points
one the number of thermodynamic degrees of freedom in a binary mixture.
Therefore, the reactive binary system has an isolated critical point, and the
critical indices of this system are the same as those of a pure ﬂuid,
_
∂υ
∂p
_
T,A=0
∼ C
p,A=0
∼
_
T −T
C
T
C
_
−γ
;
_
∂v
∂p
_
s,A=0
∼ C
v,A=0
∼
_
T −T
C
T
C
_
−α
, etc.
(3.2)
where υ = 1/ρ is the speciﬁc volume.
It is interesting to compare (3.2) with the case of a frozen chemical
reaction (no catalyst added). The system considered is then a binary mix
ture with a liquidgas critical line. The renormalized critical indices of such
systems are well known [84]
C
p,ξ
∼
_
∂υ
∂p
_
T,ξ
∼
_
T−T
C
T
C
_
−α/(1−α).
;
C
v,ξ
∼
_
∂υ
∂p
_
s,ξ
∼
_
T−T
C
T
C
_
α/(1−α)
.
(3.3)
The correspondence between (3.2) and (3.3) becomes obvious from the
thermodynamic relation [3]
C
p,A=0
= C
p,ξ
−h
2
(∂ξ/∂A)
T,p
(3.4)
where h is the heat of reaction. In fact, the asymptotic behavior of the
lefthand side of Eq. (3.4) on approaching a critical point is determined by
the second term on the righthand side rather than by the ﬁrst term, which
has a weaker singularity.
The occurrence of a chemical reaction in the system under consideration
modiﬁes the critical indices of the observable speciﬁc heats at constant
volume and constant pressure (or inverse velocity of sound) compared to a
system with a frozen chemical reaction. They are changed from α/ (1 −α)
and −α/ (1 −α) to −α and −γ, respectively. For example, the speciﬁc
heat at constant volume has a weak singularity at the critical point when
a chemical reaction occurs, but a ﬁnite, cusped behavior in the absence of
a chemical reaction.
The singularities become weaker for manycomponent mixtures. A com
mon situation is that the solutes undergo various chemical transformations
while the solvent does not. As an example, consider the critical system
containing a reactive binary mixture dissolved in some solvent. Due to the
existence of a neutral third component, the system has a line of critical
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Eﬀect of Chemistry on Critical Phenomena 57
points rather than an isolated critical point, as is the case for a reactive
binary mixture.
Let us consider the frozen chemical reaction. The singularities of the
thermodynamic quantities in a ternary mixture at constant chemical po
tential µ
0
of the solvent are similar to those of a binary mixture. The
liquidliquid critical points depend only slightly on the pressure, so that
the parameter Rρ
C
(dT
C
/dp) (ρ
C
is the critical density) is very small. This
parameter determines the region of renormalization [86]. Therefore, renor
malization is absent, as in the vicinity of the λ line in helium, and
C
p,ξ,µ
0
∼ C
υ,ξ,µ
0
∼ φ(µ
0
)
−α
; φ(µ
0
) ≡
T −T
C
(µ
0
)
T
C
(µ
0
)
. (3.5)
However, measurements are taken at constant number of solvent particles
rather than at µ
0
= const. According to renormalization, near the liquid
liquid critical points, one obtains
C
p,ξ,N
0
∼ C
υ,ξ,N
0
∼
_
T −T
C
T
C
_
α/(1−α)
. (3.6)
Unlike the liquidliquid critical point, there are two renormalizations
in the vicinity of liquidgas critical point; the ﬁrst occurs when passing
from the binary to the ternary mixture, and the second renormalization
takes place when passing from µ
0
= const to N
0
= const. In the region of
renormalization of the ternary mixture, one gets
C
p,ξ,µ
0
∼ φ
−α/(1−α)
; C
υ,ξ,µ
0
∼ φ
α/(1−α)
(3.7)
and for N
0
= const,
C
p,ξ,N
0
∼ φ
α/(1−α)
; C
υ,ξ,N
0
∼ const. (3.8)
Comparing (3.6) and (3.8), one concludes that without a chemical reac
tion, cusplike behavior exists for both speciﬁc heats near the liquidliquid
critical points, but only for the speciﬁc heat at constant pressure near the
liquidgas critical points.
When a chemical reaction takes place, the singularities near the critical
points can be found from the thermodynamic relations, analogous to (3.4),
C
p,N
0
,A=0
= C
p,N
0
,ξ
−T (∂ξ/∂A)
T,p,N
0
(∂A/∂T)
2
p,N
0
,ξ
C
υ,N
0
,A=0
= C
υ,N
0
,ξ
−T (∂ξ/∂A)
υ,T,N
0
(∂A/∂T)
2
v,N
0
,ξ
.
(3.9)
For both types of critical points, the singularities of thermodynamic
quantities in a system undergoing a chemical reaction (A = 0) are deter
mined by the second term on the righthand sides of Eqs. (3.9). The factor
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
58 Chemistry versus Physics: Chemical Reactions near Critical Points
∂A/∂T in the latter terms remains ﬁnite at the critical point, whereas the
second factor has the following asymptotic behavior,
(∂ξ/∂A)
T,p,µ
0
∼ φ
−γ
; (∂ξ/∂A)
υ,T,µ
0
∼ φ
−α
(3.10)
and, after renormalization,
(∂ξ/∂A)
T,p,N
0
∼
_
T −T
C
T
C
_
− α/(1−α)
;
(∂ξ/∂A)
υ,T,N
0
∼
_
T −T
C
T
C
_
α/(1−α)
.
(3.11)
Therefore, for a system undergoing a chemical reaction, one obtains from
(3.9) and (3.11),
C
p,N
0
,A=0
∼
_
∂υ
∂p
_
T,N
0
,A=0
∼ φ
−α/(1−α)
;
C
υ,N
0
,A=0
∼
_
∂υ
∂p
_
v,N
0
,A=0
∼ φ
α/(1−α)
.
(3.12)
Comparing (3.6) and (3.8) with (3.12) yields that the existence of a
chemical reaction leads to a magniﬁcation of the singularities of the speciﬁc
heats at constant pressure for both types of critical points, from cusplike
behavior α/ (1 −α) to a weak singularity −α/ (1 −α). On the other hand,
the speciﬁc heat at constant volume changes its asymptotic behavior from
constant to cusplike only near the liquidgas critical point. In principle, one
can detect such a magniﬁcation experimentally. Experiments are slightly
easier to perform near the liquidliquid critical points, because these points
usually occur at atmospheric pressure and room temperature.
Consider now the singularities of the dielectric constant ε and electri
cal conductivity σ at the critical point of reactive systems. These quan
tities can be obtained by choosing poorly conducting liquid mixtures for
dielectric measurements and strongly conducting mixtures for conductivity
measurements. One can also measure ε and σ simultaneously placing the
ﬂuid between two parallel plates or coaxial cylindrical electrodes with an
alternative current bridge. The ﬂuid then acts in the bridge circuit as a
frequencydependent impedance with balanced capacity and resistive com
ponents [87]. Near the critical points, large ﬂuctuations aﬀect all ther
modynamic and kinetic properties including those which are characterized
by σ and ε. There are hundreds of experimental papers, of which we will
describe only a few. Of special importance are chemical reactions in eco
logical clean nearcritical water. At high temperature and pressure, water
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Eﬀect of Chemistry on Critical Phenomena 59
becomes selfionized with a high conductivity, resembling molten salts [88].
The special case is the dissociation of a weak acid in water near an acid
water liquidliquid critical point,
HA+H
2
O A
−
+H
3
O
+
. (3.13)
Measurements of the conductivity as a function of temperature have
been performed near the liquidliquid critical points for the systems isobu
tyric acidwater [89], [90] and phenolwater [90]. Diﬀerent explanations
have been proposed for the anomaly in the conductivity near the critical
point [91]: 1) The anomaly is due to viscous drag of the acid anion [89]
and, hence, according to the Stokes law, it is proportional to the inverse
of the viscosity η, σ ∼ η
−1
= τ
νz
η
= τ
0.031
, where τ = (T −T
C
) /T
C
, and
ν = 0.63, z
η
= 0.05 are the critical indices for the correlation length and
viscosity, respectively; 2) The anomaly of σ is related to the anomaly in the
protontransfer rate [92], which, in turn [93], has the temperature depen
dence of the nearestneighbor correlation function, σ ∼ τ
1−α
= τ
0.89
, where
α = 0.11 is the critical index for the heat capacity at constant pressure
and composition; 3) A percolation model for conductivity leads [90] to an
anomaly of the form σ ∼ τ
2β
= τ
0.65
, where β = 0.325 is the critical index
describing the coexistence curve; 4) The anomaly of σ is related to that of
the extent of the acid dissociation [94], which leads [95] to σ ∼ τ
1−α
= τ
0.89
.
A detailed analysis of the experimental data obtained in [89], [90] has
been carried out [96], taking account of the background contribution due
to the normal temperature dependence of the dissociation constant and
due to the conﬂuent critical singularity. The leading critical anomaly of
the conductance was found to be characterized by a critical index 1 − α,
consistent with its connection with an anomaly in the extent of the disso
ciation reaction and also for an anomaly in the protontransfer rate.
The static dielectric constant for the dimerization reaction
2NO
2
N
2
O
4
has been measured [74] in a solvent mixture of
perﬂuoromethylcyclohexanecarbon tetrachloride near the liquidliquid crit
ical point of the solvent. The appropriate choice of reaction (no dipole
moment for the N
2
O
4
molecule, and a small moment for NO
2
) and of the
solvent (the components of low polarity make the “background” dipole mo
ment very small) assure the high precision of measurements. The dielectric
constant was found to have a 1−α anomaly near the critical point, the same
as the conductivity. For a mixture of polystyrene and diethyl malonate near
the liquidliquid critical point in the absence of chemical reactions, the di
electric constant was measured [97] as a function of frequency in the range
20 KHz to 1 MHz. No anomaly was found in the dielectric relaxation time.
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
60 Chemistry versus Physics: Chemical Reactions near Critical Points
3.3 Singularity in the degree of dissociation near a
critical point
At the end of the previous section, we considered the anomalies in elec
troconductivity near the critical point, which are caused by the scattering
of charged ions by strongly enhanced concentration ﬂuctuations. However,
apart from the anomalies of scattering processes, an additional factor has
to be taken into account, namely, the density of ions may have a singularity
near the critical point due to the chemical reaction [94]. Assume that the
dissociation reaction
CD C
+
+D
−
(3.14)
occurs between the solutes CD, C
+
and D
−
near the critical point of the
solvent E. The condition for chemical equilibrium has the following form,
A ≡ µ
CD
−µ
C
+ −µ
D
− = 0. (3.15)
The constraint of electroneutrality
N
C
+ = N
D
−, (3.16)
permits one to write the diﬀerential of the Gibbs free energy in the form
dG = −SdT +V dp +µ
E
dN
E
+µ
CD
dN
CD
−(µ
C
+ +µ
D
−) dN
C
+. (3.17)
It is convenient [95] to express the thermodynamics in terms of the total
number N
1
of moles of CD, both dissociated and undissociated, and of total
number N
2
of moles of CD and E,
N
1
= N
CD
+N
C
+; N
2
= N
1
+N
E
. (3.18)
It is also convenient to deﬁne intensive variables,
ξ =
N
C
+
N
2
; χ =
N
1
N
2
; g =
G
N
2
; υ =
V
N
2
; s =
S
N
2
. (3.19)
Using Eqs. (3.15)–(3.19), one may rewrite Eq. (3.17),
dg = −sdT +υdp −Adξ + (µ
CD
−µ
E
) dχ. (3.20)
Note that for a closed system, the variable χ remains constant, indepen
dent of the extent of reaction ξ. Usually the temperature changes during the
experiment while the pressure remains constant. The extent of reaction,
i.e., the number of charges, is deﬁned by the derivative (∂ξ/∂T)
p,χ,A=0
.
This derivative is taken under the condition that two extensive (“ﬁeld”)
variables, p and A, and one intensive (“density”) variable χ are kept ﬁxed.
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Eﬀect of Chemistry on Critical Phenomena 61
According to the general theory [38], this derivative has only a “weak”
singularity at the critical point, deﬁned by the critical index α = 0.12,
_
∂ξ
∂T
_
p,χ,A=0
≈
_
T −T
C
T
C
_
−α
. (3.21)
Thus far, we have considered a onephase system. Let us now consider
a ternary mixture (components 1, 2 and 3) existing in two phases, which
we label α and β. The chemical reaction allowed is of the form 1 2 + 3.
For example, this model describes a onecomponent system (say HI) near
its liquidgas critical point (T
C
= 150 K for HI) dissociated into two species
(2HI H
2
+I
2
). The question is: How will the concentration of the reagents
vary near the critical point of a threecomponent system? For a small
equilibrium constant of reaction, we may neglect the small shift from the
critical point of a onecomponent system.
Denoting by µ
α
i
and µ
β
i
the chemical potentials of the ith component
in each phase, one gets for the equilibrium state,
µ
α
1
= µ
β
1
; µ
α
2
= µ
β
2
; µ
α
3
= µ
β
3
. (3.22)
The condition of chemical equilibrium, similar to Eq. (3.15), has the form
µ
α
1
−µ
α
2
−µ
α
3
= 0. (3.23)
A similar condition in the second phase follows from (3.22), but does not
produce any additional restriction.
We are interested in equilibrium properties. Therefore, the thermody
namic path we choose is a displacement along the equilibrium line, keeping
the pressure constant,
d
_
µ
α
1
−µ
β
1
_
= d
_
µ
α
2
−µ
β
2
_
= d
_
µ
α
3
−µ
β
3
_
= d (µ
α
1
−µ
α
2
−µ
α
3
) = 0. (3.24)
Choosing as independent variables the mole fractions of the ﬁrst and
the second component in each phase, x
α
1
, x
α
2
, x
β
1
, x
β
2
, and the temperature
T, one can rewrite Eqs. (3.24) in the following forms
µ
α
11
dx
α
1
+µ
α
12
dx
α
2
−µ
β
11
dx
β
1
− µ
β
12
dx
β
2
=
_
µ
β
1T
−µ
α
1T
_
dT,
µ
α
21
dx
α
1
+µ
α
22
dx
α
2
−µ
β
21
dx
β
1
− µ
β
22
dx
β
2
=
_
µ
β
2T
−µ
α
2T
_
dT,
µ
α
31∂x
dx
α
1
+µ
α
32
dx
α
2
−µ
β
31
dx
β
1
− µ
β
32
dx
β
2
=
_
µ
β
3T
−µ
α
3T
_
dT,
(µ
α
11
−µ
α
21
−µ
α
31
) dx
α
1
+ (µ
α
12
−µ
α
22
−µ
α
32
) dx
α
2
= −(µ
α
1T
−µ
α
2T
−µ
α
3T
) dT
(3.25)
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
62 Chemistry versus Physics: Chemical Reactions near Critical Points
where
_
∂µ
α,β
i
∂x
α,β
j
_
p,T,x
i
=x
j
≡ µ
α,β
i,j
;
_
∂µ
α,β
i
∂T
_
p,x
j
≡ µ
α,β
i,T
. (3.26)
Solutions of Eqs. (3.25) for dx
α,β
i
/dT are
_
dx
α,β
i
dT
_
p,equil.
=
∆
α,β
i
∆
(3.27)
where the subscript equil refers to the equilibrium line, ∆ is the 4 × 4
determinant of Eqs. (3.25), and ∆
α,β
i
is the determinant with one of the
rows replaced by the row of coeﬃcients from the righthand side of Eqs.
(3.25). Simple analysis shows [94] that ∆ has no singularities at the critical
point, whereas ∆
α,β
i
is proportional to the factor µ
α,β
11
µ
α,β
22
−
_
µ
α,β
2
_
2
which
can be rewritten as
_
∂µ
α,β
12
/∂x
_
p,T,µ
1
(p,T,x
1
)
. The line of critical points is
thus determined by the vanishing of the next derivative of µ
α,β
2
with respect
to x [3], which, having one “density” variable ﬁxed, has a weak singularity
(3.21) at the critical point. Then, the xT curves have an inﬁnite slope at
the critical points.
The coincidence of the ﬁnal results for the two cases results from the
electroneutrality of the fourcomponent mixture which makes it isomorphic
to the threecomponent mixture. If the fourcomponent mixture is com
prised of two phases, each phase can be described by two mole fractions,
say x
1
and x
2
, while x
3
= x
4
= (1 −x
1
−x
2
) /2. Therefore, the equilibrium
equations will have the same form, Eq. (3.25), resulting in Eq. (3.21).
3.4 Isotope exchange reaction in nearcritical systems
The results of experiments performed on isobutyric acid (CO
2
H)  deu
terium oxide (D
2
O) binary mixture are an example of the shift of critical
parameters and the modiﬁcation of critical indices due to a chemical reac
tion. Peculiar to this system is the capability of the proton of the carboxyl
group of isobutyric acid to exchange with the deuteron of deuterium oxide
(isotope exchange reaction). This reaction produces more than two species,
and the concentrations of CO
2
H and D
2
O are not the same because of the
isotope exchange reaction. The experiments include the visual observation
of the phase separation and light scattering [98]. The results have been
compared with those for isobutyric acidwater (CO
2
H/H
2
O) and deuter
ated isobutyric aciddeuterium oxide (CO
2
D/D
2
O) systems. Appreciable
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Eﬀect of Chemistry on Critical Phenomena 63
amounts of additional species due to isotope exchange distort the coexis
tence curve, shifting the critical solution concentration x
C
away from the
concentration x
0
where the maximum phase separation temperature T
max
occurs. This makes the critical indices γ and ν of the correlation length
(ξ = ξ
0
_
T−T
C
T
C
_
−γ
) and isothermal compressibility (k = k
0
_
T−T
C
T
C
_
−ν
)
in the onephase region (T > T
C
) diﬀerent from those of the coexisting
twophase region (T < T
C
). It was found that x
C
(mass fraction of isobu
tyric acid) equals 0.358 at T
C
= 44.90
◦
C, which diﬀers from x
0
= 0.310
and T
max
= 45.11
◦
C. In the onephase region, γ = 1.25, ν = 0.63 and
ξ
0
= 3.13
˚
A in agreement with the renormalization group results, whereas
in the coexistence twophase region, the critical indices appear to be renor
malized with γ = 1.39, ν = 0.76 and ξ
0
= 0.6
˚
A. As result of exchange
equilibria, CO
2
H/D
2
O is eﬀectively a threecomponent system in the co
existence twophase region [98], and the renormalized critical indices are
in agreement with those near the plait point of a ternary liquid mixture
ethanolwaterchloroform [99]. An alternative explanation of these experi
ments is given in [100], [101].
3.5 Singularities of transport coeﬃcients in reactive
systems
In order to determine the inﬂuence of a chemical reaction on the transport
coeﬃcients, one can use the same modemode coupling analysis [102] used
in the previous chapter for ﬁnding the inﬂuence of criticality on the rate of
a chemical reaction. That is, one calculates the correlation functions which,
in turn, deﬁne the change of the transport coeﬃcients. A diﬀerent way of
attacking this problem is by dynamic renormalization group methods [103],
which we will brieﬂy describe in the next section.
3.5.1 Modecoupling analysis
Just as for the reaction rates, the transport coeﬃcients (shear viscosity η
and diﬀusion coeﬃcient D) can be written as time integrals of the correla
tions of appropriate ﬂuxes, which, in turn, can be expanded in powers of the
hydrodynamic variables. In calculating the critical corrections ∆η and ∆D,
one neglects the small critical indices, thus assuming the OrnsteinZernike
form for the susceptibility. The most signiﬁcant contribution to ∆η is the
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
64 Chemistry versus Physics: Chemical Reactions near Critical Points
integral over wavenumber k of the form [36]
∆η ∼
_
k
2
k
1
dk
k +L
. (3.28)
The upper limit k
2
is related to an unimportant cutoﬀ, and the lower limit
k
1
, the inverse correlation length ξ
−1
, vanishes at the critical point. When
there is no reaction (Onsager coeﬃcient L = 0), this integral diverges log
arithmically at the critical point. A more careful calculation of ∆η leads
to a weak power divergence rather than a logarithmic divergence. How
ever, when L = 0, the integral (3.28) does not diverge. It follows that the
viscosity in reactive systems remains ﬁnite at the critical point. Note that
if L is extremely small (extremely slow chemical reaction), then η grows
upon approaching the critical point, but its growth is terminated when the
correlation length becomes of order L
−1
.
The calculation of the modecoupling corrections to the wavenum
ber (frequency)dependent diﬀusion coeﬃcient in reactive binary mixtures
[102], [36] yields
∆D =
κ
B
T
6πρξX
2
η
_
K (X) −
βX
3
Y
3
K (Y )
_
(3.29)
where ξ is the correlation length, X = kξ, Y = X
_
δ/ (1 +X +δ), δ =
L/η, β = δ/ (1 +δ). The function K (X) is given by
K (X) =
3
4
_
1 +X
2
+
_
X
3
−X
−1
_
tan
−1
X
¸
. (3.30)
When L = 0, one regains the wellknown result for nonreactive mixtures
[102]. The inﬂuence of the chemical reaction on the diﬀusion coeﬃcient
can easily be seen in the two limiting cases of low frequencies kξ < 1 (the
“hydrodynamic” region) and high frequencies kξ > 1 (the “critical” region).
a) For kξ < 1, K (X) ∼ X
2
, and
∆D =
κ
B
T
6πρξη
_
1 −β
1/2
_
. (3.31)
The usual result [102] η∆D ∼ ξ
−1
is obtained, but with a renormalized
coeﬃcient. Because D = α(∂A/∂ξ)
T,p
, and since η is not divergent in
reactive systems, α diverges like ξ.
b) For kξ > 1, K (X) ∼ (3π/8) X
2
, and
∆D =
κ
B
T k
16ρ(η +L)
. (3.32)
This result is similar to that obtained for a nonreactive binary mix
ture, except for the renormalization of the coeﬃcient of k. Therefore, the
diﬀusion coeﬃcient in reactive systems still vanishes like ξ
−1
but with a
modiﬁed coeﬃcient of proportionality due to the chemical reaction.
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Eﬀect of Chemistry on Critical Phenomena 65
3.5.2 Renormalization group methods
The hydrodynamic equations (2.61)–(2.64) are written for local space
averaged variables. The renormalization group technique allows one to
average over increasingly large regions, thereby reducing the number of
eﬀective degrees of freedom of the system. Due to the divergence of the
correlation length, such a procedure is of special importance near the crit
ical points, where the increasing number of degrees of freedom becomes
important. Eﬃcient computer programs have been developed for study
ing static critical phenomena [104], and they have been extended to cover
dynamic critical phenomena as well [103].
The starting point is the identiﬁcation of slow modes, whose relaxation
times approach zero at small wavenumbers, and also (if the order parameter
is not conserved) the orderparameter mode. Therefore, one has to assign
each system to the appropriate model A − H [103]. Such an identiﬁca
tion also depends on the regimes of wavenumber which deﬁne the relative
importance of the diﬀerent modes.
The distinctive feature of reactive systems is the existence of a ho
mogeneous chemical mode, which converts the concentration into a non
conserved parameter. Milner and Martin [40] performed a renormalization
group analysis for a reactive binary mixture, improving the result of the
linear analysis of the critical slowingdown described in Sec. 2.1.
They found [40] that the critical slowingdown of a chemical reaction
occurs for k
H
< k < k
C
, where k
C
and k
H
are the inverse length scales
for diﬀusion and heat conductivity. The slowingdown is governed by the
strong critical index γ +α+ην 1.37 rather than by γ ≈ 1.26 obtained in
the linear theory.
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Chapter 4
Phase Separation in Reactive Systems
4.1 Multiple solutions of the law of mass action
Thus far, we have considered singlephase reactive systems near their criti
cal point. Another interesting problem is phase separation in reactive sys
tems as compared to nonreactive systems. Will the presence of a chemical
reaction, induced, say, by a very small amount of catalyst, stimulate or re
strict the process of phase separation? In this chapter we will consider these
problems, but we ﬁrst have to consider the possibility of multiple solutions
of the law of mass action. Two (or more) coexisting phases which have the
same temperature and pressure are diﬀerent in concentrations of diﬀerent
components and, therefore, in conducting chemical reactions. This means
that the law of mass action has more than one solution for the concentration
at given temperature and pressure.
Let us clarify the possibility of multiple solutions of the law of mass
action. The chemical potential of the ith component can be written in the
following form
µ
i
= µ
0
i
(p, T) +κ
B
T ln (γ
i
, x
i
) (4.1)
where the activity γ
i
determines the deviation from the ideal system, for
which γ
i
= 1. Using (4.1), one can rewrite the law of mass action as
x
ν
1
1
, x
ν
2
2
, . . . , x
ν
n
n
= K
id.
(p, T) γ
−ν
1
1
, γ
−ν
2
2
, . . . , γ
−ν
n
n
≡ K. (4.2)
The chemical equilibrium constant for the ideal system K
id
is deter
mined by the functions µ
0
i
in (4.1), i.e., by the properties of individual
nonreactive components, whereas for nonideal systems, K also depends
on the interactions among the components. For ideal systems, all γ
i
= 1,
and Eq. (4.2) has a single set of solutions x
1
, x
2
, . . . , x
n
. This was proved
many years ago [105] for isothermalisochoric systems. Another proof was
67
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
68 Chemistry versus Physics: Chemical Reactions near Critical Points
given for isobaric systems under isothermal [106] and adiabatic [107] con
ditions. A complete analysis has recently been carried out [108].
It worth mentioning that phase separation is possible in ideal, although
slightly artiﬁcial systems, such as that for which n units (atoms, molecules)
of type A can reversibly form an aggregate (molecule, oligomer) of type A
n
.
Then, for large n (in fact, n →∞), the law of mass action has more than one
solution, indicating the possibility of phase separation. It was subsequently
noticed [109] that the latter eﬀect is a special case of the phenomenon of the
enthalpyentropy compensation. This implies that for chemical reactions
exhibiting a linear relationship between enthalpy and entropy, the mag
nitude of change in the Gibbs free energy is less than one might expect.
Mathematically, in the Gibbs free energy equation (∆G = ∆H − T∆S),
the change in enthalpy (∆H) and the change in entropy (∆S) have oppo
site signs. Therefore, ∆G may change very little even if both enthalpy and
entropy increase. However, the existence of the enthalpyentropy compen
sation eﬀect has been doubted by some researchers, as one can see from the
article entitled “Enthalpy–entropy compensation: a phantom phenomenon”
[110].
In general, a system has to be considerably nonideal for the existence of
multiple solutions of the law of mass action (4.2), i.e., the interaction energy
must be of order of the characteristic energy of an individual component.
One possibility is for the interaction energy between components to be high.
Consider a gas consisting of charged particles (plasma, electrolytes, molten
salts, metalammonium solutions, solid state plasma). Another possibility
occurs when the characteristic energies of single particles are small, as in
the case of isomers. Both the isomerization [111] and the dissociation [112]
reactions have been analyzed in detail.
Consider the ionization equilibrium of the chemical reaction of the form
A i +e (dissociationrecombination of neutral particles into positive and
negative charges). Neglecting the complications associated with the inﬁnite
number of bound states, assume that K
id
∼ exp (I
0
/κ
B
T), where I
0
is the
ionization potential of a neutral particle. Assuming Debye screening of the
electrostatic interactions, one readily ﬁnds [112], [113] that
K = K
id
exp [−Φ(x, T) /κ
B
T] ≈ exp {[I
0
−Φ(x, T)] /κ
B
T}
Φ(x, T) ≈ (8πx)
1/2
_
e
2
/κ
B
T
_
3/2
+Bx +· · · (4.3)
where x is the concentration of charges, and the function B, which deter
mines the pair correlation between charges, has been tabulated [106].
At certain T and p, Eq. (4.3) has several solutions for x. This has a
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Phase Separation in Reactive Systems 69
simple physical meaning. The function Φ(x, T) in ( 4.3) diminishes the
ionization potential I
0
as a result of screening. Therefore, the phase with
the larger degree of ionization has higher energy and higher entropy. Hence,
these two phases can have equal chemical potentials, and therefore, can co
exist. The chemical reaction proceeds in diﬀerent ways in the two coexisting
phases. Therefore, the appearance of multiple solutions of the law of mass
action is a necessary condition for phase separation in reactive systems,
where a chemical reaction proceeds in all phases.
4.2 Phase equilibrium in reactive binary mixtures quenched
into a metastable state
4.2.1 Thermodynamic analysis of reactive binary mixtures
Let us assume that two components A
1
and A
2
participate in a chemical
reaction of the form ν
1
A
1
←→ ν
2
A
2
. The law of mass action for this
reaction is
A = ν
1
µ
1
+ν
2
µ
2
= 0 (4.4)
where A is the aﬃnity of reaction. If the system separates into two phases,
their temperatures, pressures and chemical potentials are equal
µ
i
(p, T, x
) = µ
i
(p, T, x
) ; i = 1, 2 (4.5)
where x
and x
denote the concentrations of one of the components in the
two phases.
The kinetics of phase separation from O
2
into two phase, described by
points O
3
and O
4
in Fig. 4.1, proceeds in two clearly distinguishable stages.
During the ﬁrst stage, the system is located at point O
2
, “waiting” for the
appearance of a signiﬁcant number of critical nuclei due to ﬂuctuations
(the duration of this stage is usually called “the lifetime of the metastable
state”). At the end of this stage, the system starts to separate into two
phases, and after some “completion” time, it reaches the twophase state
described by the points O
3
and O
4
.
The existence of a chemical reaction(s) increases the stability of a
metastable state through the appearance of additional constraint(s). More
over, thermodynamic considerations give only necessary but not suﬃcient
conditions for phase separation in reactive mixtures. Even when thermo
dynamics allows phase separation, it might be impossible from the kinetic
aspect. The latter can be achieved in two diﬀerent ways. First, if the for
ward and backward reaction rates are diﬀerent, with a preferred component
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
70 Chemistry versus Physics: Chemical Reactions near Critical Points
Fig. 4.1 The temperatureconcentration phase diagram of a binary mixture. Quenching
is performed from the original homogeneous state O
1
to the metastable state O
2
or to
the nonstable state O
5
. In the former case, phase separation to the ﬁnal stable state,
described by points O
3
and O
4
, occurs through the nucleation process, whereas in the
latter case, through spinodal decomposition. Figure 4.1 describes the upper critical
solution point. For the lower critical solution point, the temperature scale may be
inverted.
corresponding to point O
3
. Then, a chemical reaction leads to an additional
driving force A = 0, shifting the system homogeneously from a quenched
metastable state to the closest stable state O
3
. The points O
3
and O
4
are
separated by a singular line deﬁning the stability boundary. Therefore,
spatially homogeneous relaxation from O
2
to O
4
is impossible. Strictly
speaking, for such reactive systems, the ﬁrst stage of nucleation does not
exist. Immediately after a quench, the system is shifted by chemistry to
the closest state on the coexistence curve at the same temperature. The
second possibility takes place when the forward and backward reactions
rates are equal. If, in addition, the rate of reaction is much faster than that
of phase separation, the chemical reaction, which tends to randomly mix
the mixture, will keep the mixture homogeneous.
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Phase Separation in Reactive Systems 71
These physical arguments are supported by simple model calculations.
Let us start with the simplest model of a strictly regular solution [114],
where the chemical potentials have the following form [3]
µ
1
= µ
0
1
+κ
B
T ln x +ω (1 −x)
2
,
µ
2
= µ
0
2
+κ
B
T ln (1 −x) +ωx
2
.
(4.6)
The coexistence curve is deﬁned by Eq. (4.5). Using (4.6) gives x ≡ x
=
1 − x
for the concentrations of the two coexistence phases [3], following
the symmetric coexistence curve,
ω
κ
B
T
=
ln x −ln (1 −x)
2x −1
. (4.7)
Using (4.4), the law of mass action can be written as
ω
κ
B
T
=
ln(1 −x) −δ ln x
Λ +δ (1 −x)
2
−x
2
(4.8)
where
δ = −
ν
1
ν
2
; Λ =
δµ
0
1
−µ
0
2
ω
. (4.9)
Phase separation in a reactive binary mixture will occur if and only if
Eqs. (4.7) and (4.8) have a common solution. One can immediately see
that Eqs. (4.7) and (4.8) coincide if both Λ = 0 (“symmetric” mixture) and
δ = 1 (isomerization reaction). If δ = 1, Λ = 0, these equations have no
common solution, and the strictly regular reactive binary mixture will not
separate into two phases. In all other cases, one can equate the righthand
sides of Eqs. (4.7) and (4.8) to obtain
(1 −x)
2
ln (1 −x) −x
2
ln x
ln x −ln (1 − x)
=
Λ
δ −1
≡ q. (4.10)
The concentration x ranges from zero to unity. Therefore, Eq. (4.10)
has a solution only for q < (2 ln2 −1) /4 = 0.097. The model of a strictly
regular reactive binary mixture allows phase separation only for substances
and chemical reactions satisfying this inequality. If, for example, Λ = 0.09
and δ = 2, the criterion q < 0.097 is satisﬁed, two phases may coexist, and
their concentrations are given by intersection of the coexistence curve (4.7)
and the law of mass action (4.8). By contrast, for Λ = 0.1 and δ = 2, this
criterion is not satisﬁed, and hence there is no phase separation.
The noncompatibility of Eqs. (4.7) and (4.8) for some range of tem
peratures and pressures, i.e., the nonexistence of a common solution for the
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
72 Chemistry versus Physics: Chemical Reactions near Critical Points
concentrations in the range from zero to unity, may appear in some models
of an ncomponent system with n − 1 independent chemical reactions. In
all other cases, the existence of r chemical reactions will decrease by r the
dimension of the coexistence hypersurface.
Phase separation in binary mixtures has been described by models more
complex than that of the strictly regular solution, such as the van Laar and
Margules models [115].
The calculations performed above for a chemical reaction in a strictly
regular binary mixture can easily be generalized to a threecomponent mix
ture of components A+ B C, described by the activity coeﬃcients
ln γ
1
= ωx
2
(1 −x
1
) ; ln γ
2
= ωx
1
(1 −x
2
) ; ln γ
3
= −ωx
1
x
2
. (4.11)
If there are two independent chemical reactions, a system can only exist
in one phase. However, when there is a single reaction, the situation is very
diﬀerent. A complete analysis has been given for a reversible bimolecular
reaction that involves all three species [116].
4.2.2 Thermodynamic analysis of reactive ternary
mixtures
Another way to ﬁnd the global phase diagram for reactive ternary mixtures
is by topological analysis of the intersection of the critical manifolds of the
model considered and the speciﬁed chemical equilibrium surface. Let us
start with the meanﬁeld model described by Eq. (4.1) with activity coef
ﬁcients given by (4.11) [117]. Phase equilibrium is obtained by demanding
that all three chemical potentials be equal in coexistence phases. The coex
istence curve satisﬁes the following symmetry relations: x
1
= x
1
; x
2
= x
2
;
x
3
= x
3
.
The phase equilibrium surface has the following form,
ω
κ
B
T
=
1
x
1
−x
2
ln
x
1
x
2
(4.12)
and the critical curve is given by the limit x
1
−x
2
→0,
x
1,cr
= x
2,cr
=
κ
B
T
cr
ω
. (4.13)
The constraint A = µ
1
−µ
2
−µ
3
= 0, imposed by the chemical reaction,
takes the following form
x
3
= exp
_
−
∆G
0
κ
B
T
_
x
1
x
2
exp
_
ω (x
1
+x
2
−x
1
x
2
)
κ
B
T
_
(4.14)
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Phase Separation in Reactive Systems 73
where ∆G
0
is the deviation of the standard Gibbs free energy from a hypo
thetical ideal solution state, which can be written as ∆G
0
= ∆H
0
−T∆S
0
,
with ∆H
0
and ∆S
0
being the standard enthalpy and entropy for the given
reaction. The intersection of this chemical equilibrium surface with the
coexistence surface (4.12) deﬁnes a unique coexistence curve in the plane
(κ
B
T/ω, x
1
) for ﬁxed values of ∆G
0
/κ
B
T. Numerical calculations for spe
ciﬁc values of ∆G
0
/ κ
B
T show [117] that A−B repulsion results in phase
separation as the temperature is lowered. Upon further lowering of the
temperature, the formation of C becomes favorable, thereby reducing the
unfavorable mixing, and the phases become miscible again. At low enough
temperatures, the solution consists only of C and either A or B. Hence, the
presence of a chemical reaction might result in the appearance of the lower
critical consolute point, which, together with an existing upper critical con
solute point, results in a closeloop coexistence curve. We will consider this
problem in Sec. 5.4.
Another peculiarity of the phase diagrams in reactive ternary mixtures
has been found [118] in the context of a slightly more complex model for
A+ B C reaction, in which the chemical potentials of the components
have the following form,
µ
1
= µ
0
1
+κ
B
T ln x
1
+bx
3
+cx
2
−W,
µ
2
= µ
0
2
+κ
B
T ln x
2
+ax
3
+cx
1
−W,
µ
3
= µ
0
3
+κ
B
T ln x
3
+ax
2
+bx
1
−W
(4.15)
where W = ax
2
x
3
+ bx
1
x
3
+ cx
1
x
2
. The chemical equilibrium constraint
A = 0 for Eqs. (4.15) can be expressed as follows,
κ
B
T ln
x
3
x
1
x
2
= a (x
3
−x
2
) +b (x
3
−x
1
)
+c (x
1
+x
2
) −W −∆H
0
+T∆S
0
. (4.16)
The intersection of this surface and the critical manifolds following from
(4.15) gives the phase diagram of a reactive ternary mixture. The results
depend on whether the chemical process is enthalpically or entropically
favored. Detailed analysis shows [118] that there are six diﬀerent types of
phase diagrams, depending on the combinations of sign of the parameters
a, b and c which include the triple and quadruple points, closedloop phase
coexistance curves, azeotropelike points, etc.
As an example of homogeneous nucleation in a chemically reactive sys
tem, let us return [119] to the simple bistable chemical reaction (the second
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
74 Chemistry versus Physics: Chemical Reactions near Critical Points
Schlogl trimolecular model) considered in Sec. 1.3
A +X
k
1
k
2
3X; X
k
3
k
4
B (4.17)
with the concentrations x
A
and x
B
of species A and B being held constant.
The kinetic equation for the concentration x of the component X is
dx
dt
= −k
1
x
3
+k
2
x
A
x
2
−k
3
x +k
4
x
B
≡ −
dV (x)
dx
. (4.18)
Equation (4.18) has three solutions. We assume that the values of
the coeﬃcients k
i
provide two stable solutions x
1
and x
3
, and one un
stable solution x
2
such that x
1
< x
2
< x
3
. Moreover, it is suggested that
V (x
1
) < V (x
3
), i.e., the state x
3
is stable while the state x
1
is metastable.
If the reaction (4.17), supplemented by diﬀusion, occurs in a closed one
dimensional vessel, Eq. (4.18) will have the form of the reactiondiﬀusion
equation
∂x(ρ, t)
∂t
= −k
1
x
3
+k
2
x
A
x
2
−k
3
x +k
4
x
B
+D
∂
2
x
∂ρ
2
(4.19)
where ρ is the spatial coordinate and D is the diﬀusion coeﬃcient. Intro
ducing the dimensionless variables u
i
= x
i
/x
1
; τ = tk
1
x
2
1
; D = D/k
1
x
2
1
,
and moving the origin to the metastable state x
1
, ζ = (x −x
1
) /x
1
, one
gets
∂ζ
∂τ
= D
∂
2
ζ
∂ρ
2
−bζ +aζ
2
−ζ
3
(4.20)
with b = (u
2
−1) (u
3
−1) and a = u
2
+ u
3
− 2. If the system is quenched
to the metastable state ζ = 0, its transition to the stable state ζ = ζ
3
, can
be understood as the spontaneous creation of two solutions of Eq. (4.20)
(“kink”“antikink”)
ζ (ρ, t) = ζ
3
{1 + exp [±δ (ρ −ρ
0
±vt)]}
−1
(4.21)
where
ζ
3
=
1
2
_
a +
_
a
2
−4b
_
1/2
_
; δ
2
=
ζ
2
3
2D
; v = −
3b −aζ
3
2δ
. (4.22)
The kinkantikink pair separates with time in opposite direc
tions. When separated by a certain critical length κ, equal to
−δ
−1
ln
_
_
a −3
_
b/2
__
a + 3
_
b/2
_
−1
_
, they create a stationary proﬁle,
called a nucleation nucleus ζ (ρ) . When the kink and the antikink are sep
arated by a distance less than κ, they annihilate, and the concentration
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Phase Separation in Reactive Systems 75
proﬁle returns to the homogeneous metastable state. When the separation
becomes greater than κ, the kinkantikink pair separates, and the system
evolves toward the homogeneous stable state. Equation (4.21) becomes
simpliﬁed in the smallamplitudenucleus limit, when κ → 0 , and the nu
cleation nucleus then becomes
ζ (ρ) =
3b
a
[1 + cosh(δρ)]
−1
=
3b
2a
sech
2
_
δρ
2
_
. (4.23)
Thus far, we have discussed a deterministic system. The inﬂuence of ho
mogeneous external white noise and internal chemical ﬂuctuations has been
considered [119] in the framework of the extended Kramers approach and
the multivariate master equation, respectively. The nucleation due to the
internal ﬂuctuations is limited to a very narrow region near the transition,
and it is diﬃcult to observe this experimentally. The external noise may
be easily greater than the internal noise and, therefore, more important.
In addition to the homogeneous nucleation initiated by the ﬂuctuations
(considered above), the important problem is nonhomogeneous nucleation
initiated, say, by pieces of dust present in solutions.
4.2.3 Kinetics of phase separation
Let us ﬁrst discuss nonreactive systems. All thermodynamic states within
the coexistence curve shown by the solid line in Fig. 4.1 belong to two
phase states. Therefore, after a change in temperature (or pressure) from
point O
1
to O
2
, phase separation will occur and the system will consist of
two phases described by the thermodynamic condition (4.5) and shown by
points O
3
and O
4
in the ﬁgure. All states located within the coexistence
curve and close to it are the socalled metastable states. Although the
energy of these states is larger than that of the corresponding twophase
system, one needs a ﬁnite “push” to pass from the local minimum in energy
to the deeper global minimum. Such a “push” is provided by the thermal
ﬂuctuations in density for a onecomponent system and in concentration
for manycomponent mixtures. Therefore, these ﬂuctuations are of crucial
importance for the analysis of the decay of a metastable state.
The change of the thermodynamic potential associated with the ap
pearance of the nucleus of the new phase (assumed, for simplicity, to be
spherical with radius r) is equal to
∆G(r) = −∆µ
4
3
πr
3
+ 4πσr
2
(4.24)
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
76 Chemistry versus Physics: Chemical Reactions near Critical Points
where ∆µ is the energy gain in a stable state compared to a metastable
state, and σ is the surface tension. The ﬁrst (negative) term in (4.24)
encourages phase separation while the second (positive) term prevents it. It
is clear from (4.24) that for small nuclei, ∆G > 0. That is, small nuclei tend
to shrink because of their high surfacetovolume ratio. Only nuclei larger
than the critical size r
C
are energetically favorable. From ∆G/dr = 0, one
ﬁnds
r
C
=
2σ
∆µ
. (4.25)
The traditional phenomenological approach to the decay of a metastable
state of a pure substance is based on the distribution function W (r, t) of
nuclei of size r at time t. The continuity equation has the form,
∂W
∂t
=
∂
∂r
_
FW +D
∂W
∂r
_
≡ −
∂J
∂t
(4.26)
where the ﬂux J (r, t) of nuclei along the size axis is determined by two
unknown functions F and W. One can reduce [120] the equation(s) of the
critical dynamics to the Langevin equation for the radius of nuclei with
a known random force, which can be transformed to the FokkerPlanck
equation (4.26) with functions F and D of the form
F (r) = D
0
_
1
r
−
1
r
C
_
; D(r) =
D
0
κ
B
T
8πσr
2
(4.27)
where D
0
is the diﬀusion coeﬃcient far from the critical point.
The nucleation process in a binary mixture can be described in analo
gous fashion [121], [18]. The formation energy of a nucleus containing n
1
atoms of component 1 and n
2
atoms of component 2 is
∆G =
_
µ
1,n
−µ
1
_
n
1
+
_
µ
2,n
− µ
2
_
n
2
+ 4πσr
2
(4.28)
where µ
1
and µ
2
are the chemical potentials of the components in the
homogeneous phase, and µ
1,n
and µ
2,n
are the corresponding quantities in
the nuclei. The size r of a nucleus is related to its structure by 4πr
3
/3 =
υ
1
n
1
+υ
2
n
2
, where υ
1
and υ
2
are the volumes per atom.
In contrast to a pure substance, the critical nucleus is deﬁned not only
by its size r, but also by the concentration x = n
1
/ (n
1
+n
2
). Therefore,
the height of the potential barrier is deﬁned by ∂∆G/∂r = ∂∆G/∂x = 0,
which gives the following relation between r
C
and x
C
of the critical nucleus,
r
C
= 2σ
x
C
υ
1
+ (1 −x
C
) υ
2
x
C
∆µ
1
+ (1 −x
C
) ∆µ
2
(4.29)
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Phase Separation in Reactive Systems 77
where ∆µ
i
= µ
i,n
− µ
i
.
Let us now consider reactive systems. The concentration x changes as
a result of a chemical reaction, which shifts the initial quenched state to
the closest equilibrium state on the coexistence curve. Each intermediate
state corresponds to a diﬀerent radius of critical nucleus r
C
(t), increasing
toward the coexistence curve, where r
C
→ ∞. Hence, one has to replace
r
C
by r
C
(t) everywhere. The latter function can be found from Eq. (4.29)
under the assumption of a quasistatic chemical shift. This shift is caused
by a chemical reaction, and can be described in the linear approximation
by
dx
dt
= −
x −x
τ
(4.30)
where τ
−1
is the rate of the chemical reaction.
Inserting the solution of Eq. (4.30), x − x
= (x
0
−x
) exp
_
−
t
τ
_
, into
Eq. (4.29), one obtains
r
C
(t) =
2σ [x
υ
1
+ (1 −x
) υ
2
]
κ
B
T [x
/x
−(1 −x
) / (1 −x
)] (x
−x
)
≡ r
0
exp(t/τ) (4.31)
where we used the coexistence condition (4.5), and retained only the leading
logarithmic part of the chemical potentials.
The next step is to obtain the FokkerPlanck equation for the distri
bution function W (r, t, x) for nuclei of size r and composition x at time
t. We make the approximation that all nuclei which are important for
phase separation have the same composition x
. Therefore, we may omit
the argument x in W (r, t, x). In other words, one assumes that a chemical
reaction brings the path leading to the saddle point closer to that of x = x
.
Substituting (4.27) and (4.31) into the FokkerPlanck equation ( 4.26),
one obtains
∂W
∂t
=
∂
∂r
_
D
0
_
1
r
−
1
r
0
exp
_
−
t
τ
__
W +
D
0
κ
B
T
8πσr
2
∂W
∂r
_
. (4.32)
To ﬁnd the approximate solution of this complicated equation, we ﬁrst
consider a simpliﬁed version, which allows an exact solution. One replaces
functions F (r) and D(r) in (4.27) by simpler functions which incorporate
the main property of (4.27). According to the physical picture, F (r) is
positive for r < r
C
and negative for r > r
C
. Therefore, we approximate
F (r) by
F (r) = B(r
C
−r) . (4.33)
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
78 Chemistry versus Physics: Chemical Reactions near Critical Points
After choosing this form of F (r) , we are no longer free to choose
the second function D(r) . These two functions are not independent be
cause there is no ﬂux in equilibrium, which, according to (4.26), leads
to W
eq
= exp
_
−
_
drF (r) /D(r)
¸
. On the other hand, by deﬁnition,
W = exp(−∆G/k
B
T), and, therefore,
_
F (r)
D(r)
dr =
∆G
k
B
T
(4.34)
where ∆G(r) and r
C
are given by (4.24) and (4.25). A simple calculation
yields
D(r) =
Bk
B
Tr
C
8πσr
≈
Bk
B
T
8πσ
. (4.35)
We keep only the largest term in r −r
C
in the last equality in (4.35). Using
(4.33), (4.31) and (4.35), one obtains the simpliﬁed FokkerPlanck equation,
∂W
∂t
=
∂
∂r
_
B
_
r
0
exp
_
t
τ
_
−r
_
W
_
+D
0
∂
2
W
∂r
2
, (4.36)
where D
0
= Bk
B
T/8πσ. Equation (4.36) can be solved exactly by intro
ducing the characteristic function
W (K, t) =
_
W (r, t) exp(iKr) dr. (4.37)
The original FokkerPlanck equation (4.26)–(4.27) is thus transformed into
a ﬁrstorder linear partial diﬀerential equation for W (K, t) [122],
∂W
∂t
−BK
∂W
∂K
= −
_
D
0
K
2
+iKBr
0
exp(
t
τ
)
_
W. (4.38)
To solve this equation, one uses the method of characteristics, with
initial condition
W (r, t = 0) = δ (r −ξ) (4.39)
where ξ is of order of a single molecule (there are no nuclei immediately
after a quench). Solving Eq. (4.38) and performing the inverse Fourier
transformation, one obtains the Gaussian distribution function
W (r, t) =
1
_
4π∆r
2
exp
_
−
(r −r)
2
2∆r
2
_
(4.40)
with mean value r and variance ∆r
2
given by
r = ξ exp (Bt) −
Br
0
τ
−1
−B
_
exp
_
t
τ
_
−exp (Bt)
_
,
∆r
2
=
D
0
2B
[exp (2Bt) −1] . (4.41)
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Phase Separation in Reactive Systems 79
Two characteristic times, B
−1
and τ, are associated with transient pro
cesses and chemical reactions, respectively. However, the solution (4.40)
diverges with time, so that Eq. (4.36) does not have a steadystate solu
tion. Hence, instead of the exact solution, one has to use the approximate
solution.
One can ﬁnd [122] the appropriate solution of Eq. (4.36), but we prefer
to ﬁnd the solution of the original equations (4.26) and (4.27), assum
ing that B
−1
is very small [123]. This assumption permits us to neglect
transient processes, and, for B
−1
< τ , restrict attention to those solu
tions of (4.36) which supply the quasistatic ﬂux J
qss
(t) independent on
the size of the nuclei. We consider the quasisteadystate regime which
is established after the transients disappear. The quasisteadystate so
lution W
qss
[r, r
C
(t)] does not have an explicit time dependence, so that
∂W
qss
/∂t = 0, and the FokkerPlanck equation (4.32) can be rewritten as
D
0
_
1
r
−
1
r
0
exp
_
−
t
τ
__
W
qss
+
D
0
κ
B
T
8πσr
2
∂W
qss
∂r
≡ J
qss
_
t
τ
_
. (4.42)
The quasisteadystate ﬂux J
qss
reduces to the steadystate ﬂux J
ss
when
the chemical reaction is absent,
J
qss
_
t
τ
= 0
_
= J
ss
. (4.43)
There is no stationary state for the reactive system considered here.
Therefore, one is forced to give a new deﬁnition to the lifetime of a
metastable state in reactive systems. The simplest generalization is the
time required to produce one critical nucleus
T
ch
_
0
dt J
qss
(t) = 1. (4.44)
For the timeindependent case, J
qss
is replaced by J
ss
, according to
(4.43), and T
ch
is replaced by the lifetime of the metastable state in the
nonreactive system, T
0
= J
−1
ss
.
Let us now turn to the solution of Eq. (4.42). The boundary conditions
for this equation are determined by the requirements that the distribution
of nuclei of minimal size ξ will be the equilibrium distribution, and the total
number of nuclei in the system is bounded,
W
qss
(ξ) = W
eq
(ξ) ; W
qss
(r →∞) = 0 (4.45)
where W
eq
corresponds to zero ﬂux.
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
80 Chemistry versus Physics: Chemical Reactions near Critical Points
The solution of Eq. (4.42) which satisﬁes the boundary conditions (4.45)
has the form [123],
W
qss
= W
eq
(r)
_
¸
_
1 −J
qss
r
_
ξ
dz
W
eq
(z) D(z)
_
¸
_
(4.46)
where
J
−1
qss
=
∞
_
ξ
dz
W
eq
(z) D(z)
. (4.47)
The function W
−1
eq
(r) has a sharp maximum at r
C
, which reﬂects the
existence of a barrier to nucleation. Therefore, the integral (4.47) can be
evaluated by the method of steepest descents,
J
qss
=
_
κ
B
T
4π
2
σ
D
0
2r
2
0
exp
_
−
2t
τ
−
∆G(r
0
)
κ
B
T
exp
_
2t
τ
__
(4.48)
where ∆G(r
0
) = 4πσr
3
0
/3 is the minimal work needed to produce the
critical nucleus in the initial state immediately after a quench.
Upon inserting J
ss
, deﬁned by (4.43), into (4.48), one obtains
J
qss
= J
ss
exp
_
−
2t
τ
−u
_
exp
_
2t
τ
_
−1
__
, (4.49)
where u = ∆G(r
0
) /κ
B
T.
Finally, inserting (4.49) and T
0
= J
−1
ss
into (4.44) gives the equation
for the lifetime T
ch
of a metastable state in a reactive system as a function
of the lifetime in the nonreactive system T
0
(the latter depends on the
volume under observation), the extent of quench u, and the rate τ
−1
of the
chemical reaction,
1 =
τu exp(u)
2T
0
uexp(2T
ch
/τ)
_
u
exp (−z)
z
2
dz. (4.50)
Equation (4.50) for T
ch
has been solved numerically. This equation
has no solution if τ is very small, i.e., the chemical reaction is very fast.
Then, although the thermodynamics allows phase separation in a reactive
system, such a separation is impossible from the kinetic point of view. The
system is dragged by the chemical reaction to a homogeneous equilibrium
state on the coexistence curve before the nuclei of the new phase appear.
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Phase Separation in Reactive Systems 81
The minimal τ which allows phase separation for diﬀerent quenches (with
characteristic u and T
0
) is given approximately by the following formula,
_
τ
T
0
_
min
≈ 3.03 + 2.08u. (4.51)
For smaller values of τ, the system will never separate into two phases.
The two possible sets of experiments performed with and without a catalyst
(diﬀerent quenches for a given chemical reaction and diﬀerent chemical
reactions for the same quench) will be discussed in the next section.
4.3 Phase equilibrium in reactive mixtures quenched into
an unstable state
Thus far, we have considered phase separation in a system quenched into
the metastable region of phase diagram (point O
2
in Fig. 4.1) located be
tween the limit of thermodynamic stability (spinodal) and the boundaries
of phase coexistence (binodal). In that case, separation proceeds through
nucleation. Another mechanism for phase separation takes place when a
system is quenched into the unstable region (point O
5
in Fig. 4.1) of the
phase diagram located inside the spinodal curve. In this case, the pro
cess leads to macroscopic phase separation via a complicated labyrinthine
interconnected morphology. The phenomenological description of this pro
cess is based on the CahnHillard equation [124] which, for our case of the
conserved variable, has the following form
∂Ψ
∂t
= Γ∇
2
∂F [Ψ]
∂Ψ
(4.52)
where Ψ = x
1
− x
2
is the concentration diﬀerence (order parameter), Γ is
proportional to the mobility, and F [Ψ] is the GinzburgLandau free energy
functional
F [Ψ] =
_
dx
_
κ  ∇Ψ 
2
+f (Ψ)
¸
, (4.53)
where f (Ψ) is the doublewell potential, f (Ψ) = −aΨ
2
+bΨ
4
, and κ is the
positive constant related to the interaction range. When the A−B mixture
is quenched from a homogeneous state (O
1
in Fig. 4.1) to the nonstable
state O
5
, the system separates into Arich and Brich phases. Numerical
simulation has been performed [125] by placing Eq. (4.52) on a 128 ×128
square lattice with periodic boundary conditions, with the lattice site and
time step both set at unity, and the local part of free energy of the form
df (Ψ)
dΨ
= −1.3 tanhΨ + Ψ. (4.54)
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
82 Chemistry versus Physics: Chemical Reactions near Critical Points
The order parameter Ψ = Ψ
0
+δΨ at each site is given by Ψ
0
= 0.4, and
the random number δΨ chosen in the range (−0.15, 0.15) . For an Arich
mixture (with concentrations of A and B being x
1
= 0.7 and x
2
= 0.3,
respectively), the Brich domains growing in the Arich matrix are shown
in Fig. 4.2.
(a) (b) (c)
Fig. 4.2 Domain patterns for A−B mixture quenched to a nonstable state. From left
to right: (a) after 10
5
steps with no chemical reaction; (b) after t = 1.32 × 10
5
steps
with the chemical reaction (k
f
= k
b
= 0.01) started at t = 10
5
; (c) after t = 8.2 × 10
4
steps with the chemical reaction (k
f
= 0.0013 and k
b
= 0.0007) started at t = 5 ×10
4
.
Reproduced from Ref. [125] with permission, copyright (1997), Physical Society of Japan.
The morphologies of phase separation become much more manifest in
the presence of an ongoing chemical reaction. Unlike the previously con
sidered phase separation from a metastable state in a closed system, we
now assume that our system is open, i.e., far from equilibrium, so that
the chemical reaction is maintained by an external agent, say, light (for a
discussion of this point, see [126], [127]). The simplest way to take into
account the chemical reaction A ↔B is to add to Eq. (4.52) the equation
of the chemical reaction, k
f
x
1
− k
b
x
2
, where k
f
and k
b
are the forward
and backward reaction rates. Thus, we rewrite Eq. (4.52) for the reactive
binary mixture in the following form [128],
∂Ψ
∂t
= Γ∇
2
δF [Ψ]
δΨ
−k
f
Ψ +k
b
(1 −Ψ)
= Γ∇
2
_
∂f
∂Ψ
−2κ∇
2
Ψ
_
−(k
f
+k
b
) Ψ +k
b
. (4.55)
If the chemical reaction starts at t = 10
5
with k
f
= k
b
= 0.01, the
domain patterns shown in Fig. 4.2a gradually change to the concentric pat
terns shown in Fig. 4.2b for t = 1.32×10
5
[125]. The diﬀerence between the
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Phase Separation in Reactive Systems 83
patterns in Figs. 4.2a and 4.2b are connected with the time delay between
the beginning of phase separation after the temperature quench and the
onset of the chemical reaction. If phase separation begins simultaneously
with the chemical reaction, lamellar patterns will form. Still other patterns
are formed when the rates of the forward and back reactions are diﬀerent,
k
f
= k
b
. In the system shown in Fig. 4.2c, the nonsymmetrical chemical
reaction (k
f
= 0.0013 and k
b
= 0.007) started at t = 5 ×10
4
steps after the
quench. As a result, a bicontinuous pattern emerges at t = 8.2 ×10
4
steps
[125]. In this case, even without a chemical reaction, one expects A rich
droplets in the Brich phase because k
b
> k
f
.
The increase in phase separation coupled with a chemical reaction has
been calculated [129] by including hydrodynamic eﬀects. Equation (4.55)
has been coupled with the NavierStokes equation. Numerical simulations
have been performed [129] under conditions similar to those described above
(256 × 256 square lattice, Ψ
0
= 0.1 and δΨ = (−0.025, 0.025)). The inﬂu
ence of hydrodynamic eﬀects is characterized by the viscosity η. When the
viscosity is low (η = 1), hydrodynamics plays an important role, whereas
for low ﬂuidity (η = 100), hydrodynamic eﬀects are not important.
Domain patterns at t = 10
4
are shown in Figs. 4.3–4.4 for η = 1 (left
row), and for η = 100 (right row). Figures 4.3a–4.3c describe the symmetric
reactions (from the top down, k
f
= k
b
equals 10
−4
, 10
−3
and 10
−2
), while
Figs. 4.4a–4.4c show the results for nonsymmetric reactions (from the top
down: k
f
= 1.1 × 10
−4
; 1.1 × 10
−3
; 1.1 × 10
−2
and k
b
= 9 × 10
−5
; 9×
10
−4
; 9 × 10
−3
). These ﬁgures clearly show the competition between hy
drodynamics eﬀects and chemical reactions. As seen in Fig. 4.3a, for a low
reaction rate, isolated domains become more circular under hydrodynamic
ﬂow. For intermediate reaction rates (Fig. 4.3b), hydrodynamic eﬀects lead
to a circular shape for isolated domains since the chemical reaction rate is
not high enough to make discontinuous domains penetrate each other. The
latter eﬀect occurs at higher reaction rates (Fig. 4.3c), where the mixing
process due to the chemical reaction proceeds more quickly than the ther
modynamically induced phase separation. The results are quite similar for
nonsymmetric reactions (k
f
= k
b
), as shown in Fig. 4.4. Some diﬀerence
occurs for intermediate reaction rates (Fig. 4.4b), where a continuous pat
tern emerges along with an isolated A rich phase for η = 100. However,
for η = 1, the disconnected A rich domain appears in the matrix of the
Brich phase, and the Brich phases do not penetrate each other.
The additive contribution of the two ingredients of the process — segre
gation dynamics and chemical reaction — can be demonstrated by starting
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
84 Chemistry versus Physics: Chemical Reactions near Critical Points
Fig. 4.3 Domain patterns at t = 10
4
steps with and without hydrodynamic eﬀects for
symmetric chemical reactions. Arich and Brich regions are shown as white and black,
respectively (see text for explanation). Reproduced from Ref. [129] with permission,
copyright (2003), American Institute of Physics.
from a diﬀerent model [130], in which the dynamics of a quenched system is
described in the context of an Ising model with a spin variable σ
i
at site i.
The two states of the spin1/2 Ising model correspond to the site occupied
by atom 1 or 2. The dynamics of the system are described by spinexchange
or Kawasaki dynamics, whereas the chemical contribution is described by
a master equation for the timedependent probability distribution for N
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Phase Separation in Reactive Systems 85
Fig. 4.4 Domain patterns at t = 10
4
steps with and without hydrodynamic eﬀects
for nonsymmetric chemical reactions. Arich and Brich regions are shown as white
and black, respectively (see text for explanation). Reproduced from Ref. [129] with
permission, copyright (2003), American Institute of Physics
spins, which, after averaging over all conﬁgurations, leads to Eq. (4.55)
[130]. Note that the two last “chemical” terms in Eq. (4.55) can be added
[131] to the free energy functional F [Ψ] in the following form
F [Ψ] →F [Ψ] + (4.56)
+
k
f
+k
b
2
_ _
dzdz
G(z, z
)
_
Ψ(z, t) −Ψ
¸ _
Ψ(z
, t) −Ψ
¸
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
86 Chemistry versus Physics: Chemical Reactions near Critical Points
where Ψ = (k
f
−k
b
) / (k
f
+k
b
) and G(z, z
) is the Green function for the
Laplace equation ∇
2
G(z, z
) = −δ (z −z
) with appropriate boundary con
ditions, so that for d = 3, G(z, z
) = 4π z −z

−1
. Including the chemical
reaction is thus equivalent to including a longrange repulsive interaction,
thereby transforming the problem of a reactive mixture to that of a non
reactive mixture with competing shortrange and longrange interactions.
A linear stability analysis of Eq. (4.55) in the absence of chemical reac
tions predicts exponential growth of the concentration ﬂuctuations with a
growth factor that has a cutoﬀ at large wavenumber k
C
(small wavelength).
Thus, ﬂuctuations with k < k
C
grow while those with k > k
C
decay. It was
shown [128] that the occurrence of a reaction results in shifting the cutoﬀ by
an amount proportional to k
f
+k
b
, i.e., the presence of a chemical reaction
makes a system more stable with respect to phase separation. This result
of linear stability analysis was veriﬁed by numerical simulations of the full
nonlinear equation [128] , as well as for the Kawasaki exchange interaction
model including the chemical reaction [132].
Other interesting results obtained for both models relate to the growth
of the steadystate domains of size R
ss
. For the chemical reaction with
k
f
= k
b
≡ k, R
ss
∼ k
−φ
has the same critical index φ = 0.3 as the growth
index (R(t) ∼ t
φ
) in the absence of a chemical reaction. The latter result
has been reﬁned [133]. A more serious discrepancy was found [134] when,
in contrast to the numerical calculation performed using the Monte Carlo
technique for particles on a square lattice [132], the calculations were carried
out in continuous space using molecular dynamics. Although these two nu
merical techniques usually give the same results for equilibrium properties,
the results are diﬀerent for nonequilibrium dynamics, where the collec
tive hydrodynamics modes are not simulated in the typical singleparticle
MonteCarlo method. Thus, no correction was found for the growth expo
nent in the absence of a chemical reaction and scales with respect to the
reaction rate in reactive systems.
The contributions of critical dynamics (diﬀusion) and the chemical re
action were previously considered independently. Moreover, contrary to
critical dynamics, the chemical rate laws do not take into account the non
ideality of the systems. Such an approach is inconsistent [135] because it is
characteristic of chemical reactions and diﬀusion that the thermodynamic
forces governing these processes are related, with both being functions of
the chemical potentials (in the former case, of the spatial derivatives of
chemical potentials). It is for this reason that for equilibrium systems,
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Phase Separation in Reactive Systems 87
diﬀusional stability automatically insures the stability of chemical equilib
rium (DuhemJoguet theorem [3]). A consistent analysis shows [135] that
the interesting and possible important idea of applications to the freezing
of the phase separation by the use of simultaneous chemical reactions, is
not easily carried out in practice and demands special requirements both
for the systems and for the chemical reactions.
Thus far, we have considered the inﬂuence of the A B chemical
reaction on the phase separation in binary mixtures by the generalization
(4.55) of the LandauGinzburg equation (4.52) in the presence of a chemical
reaction. A similar approach has been applied [136] to the isomerization
reaction between A and B
A
k
1
k
2
B (4.57)
in which C does not participate, and to ternary polymer blends of identical
chains lengths with the chemical reaction [137]
A +B
k
1
k
2
C. (4.58)
The FloryHuggins free energy for the ternary mixture has the following
form [138],
f (x
A
, x
B
) = x
A
ln x
A
+x
B
ln x
B
+ (1 −x
A
−x
B
) ln (1 −x
A
−x
B
) +nχ
AB
x
A
x
B
+nχ
BC
x
B
(1 −x
A
−x
B
) +nχ
AC
x
A
(1 −x
A
−x
B
)
(4.59)
where x
A
, x
B
and x
C
= 1 − x
A
− x
C
are the volume fractions of the
components, χ is the binary interaction parameter, and n is the number
of repeated units in the polymer chain. Two order parameters for these
systems are given by
φ = x
A
−x
B
; η = x
A
+x
B
−ψ
C
(4.60)
where ψ
C
is the critical composition of the phase diagram for phase sepa
ration between C and A−B. Expanding (4.59) in the order parameters η
and φ and their derivatives, one obtains [136]
F {φ, η} =
_
dr [f
η
(η) +f
φ
(φ) +f
int
(η, φ)]
+
D
η
2
_
dr [η (r)]
2
+
D
φ
2
_
dr [φ(r)]
2
(4.61)
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
88 Chemistry versus Physics: Chemical Reactions near Critical Points
where D
η
and D
φ
are phenomenological parameters related to the surface
energy, and
f
η
(η) = −
1
2
c
1
η
2
+
1
4
c
2
η
4
f
φ
(φ) = −
1
2
c
3
φ
2
+
1
4
c
4
φ
4
f
int
(η, φ) = c
5
ηφ −
1
2
c
6
ηφ
2
+
1
2
c
7
η
2
φ
2
(4.62)
where the c
i
are functions of the coeﬃcients χ and ψ
C
(for details, see [136],
[137]). The equation of motion for the order parameters η and φ (Landau
Ginzburg equations generalized to a chemically reactive ternary mixture)
have a slightly diﬀerent form for the chemical reactions (4.57) and (4.58).
In the former case,
∂η
∂t
= M
η
2
δF {φ, η}
δη
,
∂φ
∂t
= M
φ
2
δF {φ, η}
δφ
−(k
1
+k
2
) φ
− (k
1
−k
2
) η −(k
1
−k
2
) ψ
C
(4.63)
whereas in the latter case,
∂η
∂t
= M
η
2
δF{φ,η}
δη
+ 2k
2
ψ
C
−2k
1
x
A
x
B
,
∂φ
∂t
= M
φ
2
δF{φ,η}
δφ
,
(4.64)
where M
φ
and M
η
are the mobility coeﬃcients. For simplicity, the follow
ing forms of the functions f
η
(η) and f
φ
(φ) have been used in numerical
simulations,
df
η
dη
= A
η
tanh η −η;
df
φ
dφ
= A
φ
tanh φ −φ (4.65)
where A
η
and A
φ
are phenomenological parameters which are inversely
proportional to temperature.
Numerical simulations show [136] that for the isomerization reaction
(4.57), a diﬀerence between phase separation in a ternary mixture and in
a binary mixture, related to the availability of the third inert component,
leads to the existence of a multiplephase transition. Otherwise, the dy
namics of phase separation in binary and ternary mixtures is quite similar.
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Phase Separation in Reactive Systems 89
For reaction (4.58), the separation dynamics depends on the competition
between the chemical reaction and phase separation [137]. In the case of a
relatively large reaction rate constant, the eﬀect of a chemical reaction is
dominant, phase separation between C and A − B is inhibited, and phase
separation occurs between A and B. As the phase separation of A and
B proceeds, C decomposes into A and B rapidly, resulting in the phase
pattern of small amount of C distributed along the interface between A
rich and Brich domains. As the reaction rate gradually decreases, the
eﬀect of the chemical reaction diminishes. In the later stages of phase
separation, the concentration of C increases as the rate constant of the
chemical reaction decreases, and the Crich phase changes from dispersed
droplets to a continuous thin phase. In the next section, we consider the
dissociation reaction which is a special type of chemical reaction in a ternary
mixture.
4.4 Thermodynamics of a threecomponent plasma with a
dissociative chemical reaction
Orderdisorder phase transitions and critical phenomena in electrically
neutral systems are due to the competition between repulsive and at
tractive forces. When charged particles appear due to the dissociation
recombination reaction, they give rise to new attractive forces (the Coulomb
interaction between ions, which is, on average, an attractive interaction)
and repulsive forces (say, the hardcore interaction at short distances).
These new competing forces result in an additional ﬁrstorder phase tran
sition in a threecomponent plasma with a critical point. The coexisting
phases for a new phase transition resulting from the chemical reaction are
determined by Eqs. (4.12) and (4.16). It has been proposed [85], [114] that
such arguments can be used to explain some experiments in metallic vapors
at high pressures and in metalammonia solutions.
Metalammonia solutions are a type of system in which a small increase
of the concentration of metal results in a huge increase of electroconduc
tivity, manifesting a transition from nonmetallic to metallic conductivity.
Metallic vapors are another example of the latter behavior. The continuous
decrease in the density of metals with heating above the critical temperature
causes the transition into the nonmetallic state and a signiﬁcant decrease
of the electric conductivity.
The most general explanation of the nature of this transition was given
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
90 Chemistry versus Physics: Chemical Reactions near Critical Points
by Mott [139], who proposed the following elegant physical argument. Mott
argued that changing the number density n of charges interacting by a
screened Coulomb interaction eliminates the bound states when a
B
n
−1/3
≈
0.25, and hence the system will become metallic. Here, a
B
is the Bohr
radius in the medium. Despite the generality of the Mott criterion, not all
metalnonmetal transitions can be explained in this way. Sometimes the
transition takes place at a density lower than that predicted by the Mott
criterion [140]; in other systems, there is more than one transition. Let us
consider two examples in more detail.
The liquidgas critical point of mercury occurs at the density
ρ
C
= 5.77 g/cm
3
(p
C
= 1670 Bar, T
C
= 1750 K). Afterwards, a metal
nonmetal transition takes place at ρ ≈ 9 g/cm
3
. It is generally accepted
(see, however, [141]) that the latter transition is explained by the Mott cri
terion. However, a dielectric anomaly has been found [142] at the density
ρ ≈ 3 g/cm
3
. This was interpreted as a transition from a weakly ionized
mercury plasma to a new inhomogeneous phase of charged droplets. This
transition, probably of ﬁrst order, occurs at densities much lower than those
given by Mott, and therefore requires an explanation.
The second example is even more conclusive. The heat capacity of
sodiumammonia solution has been measured [143] for three diﬀerent mo
lar fractions of metal (X = 0.045, 0.0462, and 0.0631). Starting at low
temperatures, the heat capacity shows a jump across the liquidgas sepa
ration curve of this mixture. Upon further heating, two additional jumps
in the heat capacity occur at two distinct temperatures above the critical
temperature. The changes in slope of the electromotive force take place
at the same temperatures [144] providing additional evidence for a ﬁrst
order phase transition above the liquidgas critical point. Experimental
observation of this new phase transition is very diﬃcult because of the very
small density diﬀerences between the two phases and the existence of slow
diﬀusion processes. The former requires extreme homogeneity of the tem
perature, and the latter results in very long relaxation times. We believe
that these and other experimental problems prevent the observation of this
new thermodynamic phase transition for other systems, while its existence
seems obvious for a broad class of systems. In addition to metalammonia
solutions, one can mention molten salts, weak electrolytes, solutions of met
als in their salts, electronhole plasmas in optically excited semiconductors,
etc.
A ternary mixture with an ionization reaction is another example of
phase separation in reactive systems [114]. Let us consider a system com
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Phase Separation in Reactive Systems 91
posed of neutral particles which are partially dissociated into positive and
negative charges. Assume that the Gibbs free energy has the following
form,
G = N
A
µ
A,0
+N
I
µ
I,0
+N
e
µ
e,0
+N
A
κ
B
T ln x
A
+ 2N
I
κ
B
T ln x
I
+N
I
I +µ
0
N
I
x
α
I
(4.66)
where N
A
is the number of neutral particles, N
I
(N
e
) is the number of
ions (electrons), I is the ionization potential, and x
A
≡ N
A
/ (N
A
+ 2N
I
) =
1 −2x. Electroneutrality requires that N
I
= N
e
.
The ﬁrst three terms in Eq. (4.66) describe the ideal gases of the par
ticles. The next two terms are the entropy, the sixth term describes the
ionization process, and the last term represents the interaction between the
charged particles. We will later make the DebyeH¨ uckel approximation,
α =
1
2
; µ
0
= −
e
3
3κ
B
T
_
p
2ε
3
π
2
_
1/2
. (4.67)
One can formulate the DebyeH¨ uckel approximation using either the
Gibbs or the Helmholtz free energy. If one switches from one free energy to
the other by using the ideal gas equation of state in the correction term, the
DebyeH¨ uckel approximation is slightly diﬀerent. We here use the Gibbs
free energy, and our equation (4.67) follows from Eqs. (94.1) and (75.14)
of Landau and Lifshitz [4].
The chemical potentials of the neutral and charged particles can be
easily found from (4.66),
µ
A
= µ
A,0
+κ
B
T ln x
A
−αµ
0
x
α+1,
µ
q
≡ µ
I
+µ
e
= µ
I,0
+µ
e,0
+2κ
B
T ln x +µ
0
_
(α + 1) x
α
−2αx
α+1
¸
+I.
(4.68)
In order to obtain the coexistence curve, one equates the chemical potentials
in the two phases, analogously to Eq. (4.5). Solving these two equations
for αµ
0
/κ
B
T yields
ln (1 −2x
) −ln (1 −2x
)
(x
)
α+1
−(x
)
α+1
=
2α(ln x
−ln x
)
2α
_
(x
)
α+1
−(x
)
α+1
_
−(α + 1) [(x
)
α
−(x
)
α
]
. (4.69)
One can easily verify that Eq. (4.69) gives the following relation between
the charge concentrations in the coexisting phases,
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
92 Chemistry versus Physics: Chemical Reactions near Critical Points
_
x
x
_
α
=
1 −2x
1 −2x
. (4.70)
For the DebyeH¨ uckel interaction (4.67), the coexistence curve has the
following form,
x
=
_
1 −x
− (2 −3x
)
1/2
(x
)
1/2
_
1/2
,
−
_
e
3
6 (κ
B
T)
2
_
_
p
2ε
3
π
2
_
1/2
=
ln x
−ln x
(x
)
3/2
−(x
)
3/2
. (4.71)
If the dissociation reaction A I + e takes place, one has to add the
restriction stemming from the law of mass action, µ
A
−µ
I
−µ
e
= 0. Using
(4.67) and (4.68), one can write the latter equation in the following form
µ
0
A
−µ
0
I
−µ
0
e
−I +κ
B
T ln
1 −2x
x
2
+
µ
0
2
_
3x
1/2
−x
3/2
_
= 0. (4.72)
Neglecting the diﬀerence between the mass of atom and ion, we put
µ
0
A
= µ
0
I
. Furthermore, using the wellknown expression for an ideal gas of
electrons, one obtains from (4.72),
ln
_
p
κ
B
T
_
2π
2
m
e
κ
B
T
_
3/2
_
+
I
κ
B
T
−ln
_
1 −2x
x
2
_
−
_
e
3
/6 (κ
B
T)
2
_ _
p
2ε
3
π
2
_
1/2
_
3x
1/2
−x
3/2
_
= 0. (4.73)
Equation (4.73) gives the required restriction on the parameters of the
coexisting phases resulting from the chemical reaction. One has to ﬁnd
the simultaneous solutions of Eqs. (4.71) and (4.73). Since one cannot
solve these equations analytically, one adopts the following approximate
procedure: for given I and ε, one starts from some x
, which gives x
and
two equations for T and p (for details, see [114]).
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Chapter 5
Comments on the Geometry of the
Phase Diagram of a Reaction Mixture
5.1 Solubility in supercritical ﬂuids
The properties of ﬂuids slightly above their critical points (“supercritical
ﬂuids”) came to the attention of researchers more than a century ago, be
cause of their many technological applications. These include supercritical
extraction [145], supercritical chromatography [146] , and the strong pres
sure (or temperature) eﬀect on the reaction rate [26]. We concentrate here
on the phenomenon of enhanced solubility of solids in supercritical ﬂuids
(“supercritical extraction”) [70]. The technological advantages of this phe
nomenon have aroused great interest in the chemical engineering commu
nity. Indeed, this interest is reﬂected in the existence of a journal (Journal
of Supercritical Fluids), books ([147], [148], among others), and hundreds
of scientiﬁc and technological articles.
Figure 5.1 displays a typical example of the phenomenon. Depicted is
the mole fraction of naphthalene dissolved in ethylene as a function of the
pressure for various ﬁxed supercritical temperatures [149]. (The critical
temperature and pressure of the solvent are 11
◦
C and 51.2 atm.) The same
eﬀect was found for many other substances. Figure 5.1 shows that for every
isotherm, there is a region of pressure where small changes in pressure result
in a huge increase in solubility. Moreover, the slope of this increase becomes
larger as T approaches T
C
.
The thermodynamic analysis of this phenomenon (similar to that per
formed in Sec. 2.5) must be appropriate to the experimental setup which
allows a solid to equilibrate with a supercritical ﬂuid at a given temper
ature and pressure. Denote the ﬂuid component (solvent) by subscript 1,
and the solid component (solute) by subscript 2. Denoting the solid phase
by superscript s and the ﬂuid phase by f, we have X
s
2
= 1, where X
2
is
93
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
94 Chemistry versus Physics: Chemical Reactions near Critical Points
Fig. 5.1 Pressurenaphthalene composition phase diagram of the system ethylene
naphthalene for diﬀerent temperatures. Reproduced from Ref. [149] with permission,
copyright (1948), American Chemical Society.
the mole fraction of the solid component. In the ﬂuid phase, we deﬁne
X
f
2
≡ X. The chemical potential of the solid component is denoted as µ
2
and, accordingly, µ
s
2
= µ
f
2
for the equilibrium state. We choose T, p, X as
independent variables. Thus, µ
f
2
= µ
f
2
(T, p, X) . For all equilibrium states,
∆µ(T, p, X) = µ
s
2
(T, p) −µ
f
2
(T, p, X) = 0 (5.1)
using the fact that X
s
2
= 1 in the solid phase.
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Comments on the Geometry of the Phase Diagram of a Reaction Mixture 95
Consider isothermal changes of pressure along the equilibrium line,
d [∆µ(T, p, X)] =
_
∂µ
s
2
∂p
_
T
dp
−
_
∂µ
f
2
∂p
_
T,X
dp −
_
∂µ
f
2
∂X
_
T,p
dX = 0 (5.2)
Using the deﬁnition of the molar volume of the solid υ
s
= (∂µ
s
2
/∂p)
T
and
of the partial volume of the solute in the ﬂuid phase υ =
_
∂µ
f
2
/∂p
_
T,X
,
one can rewrite Eq. (5.2) in the following form
_
∂X
∂p
_
T,equil.line
=
υ
s
−υ
_
∂µ
f
2
/∂X
_
T,p
. (5.3)
This result is completely general, and will form the basis of the subse
quent analysis. One can already see from (5.3) that near the critical points,
the denominator goes to zero, leading to the strong divergence of the left
hand side of this equation, which corresponds to the graphs shown in Fig.
5.1.
It is instructive to examine Eq. (5.3) far from criticality. Using the
wellknown form of the chemical potential of dilute solutions,
µ
f
2
= µ
f
2,0
(T, p) + RT ln X, (5.4)
one obtains from (5.3),
_
∂ ln X
∂p
_
T,equil.line
=
υ
s
−υ
RT
. (5.5)
Far from criticality, the ﬂuid mixture is ideal, and the partial volume
of the solute υ is much larger than its molar volume in the solid phase,
υ >> υ
s
. Therefore, (∂ ln X, ∂p)
T,equil.line
< 0, i.e., X decreases as p
increases. In addition,
_
∂
2
ln X
∂p
2
_
T,equil.line
=
1
RT
_
_
∂υ
s
∂p
_
T
−
_
∂υ
∂p
_
T,X
_
(5.6)
because υ does not depend on X in this region of pressure. Therefore,
the derivatives along the equilibrium line coincide with the derivatives at a
point. Since (∂υ
s
/∂p)
T
 <<
¸
¸
¸(∂υ/∂p)
T,X
¸
¸
¸ (i.e., the solid compressibility is
almost zero), the equilibrium isothermal curve X (p) has positive curvature
for low pressures. Since (∂υ/∂p)
T,X
is negative but much larger in absolute
value than (∂υ
s
/∂p)
T
, υ decreases with increasing pressure and a minimum
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
96 Chemistry versus Physics: Chemical Reactions near Critical Points
in the curve X (p) occures when υ = υ
s
. From this point onward, X (p) is
an increasing function unless it reaches the critical region.
Let us now return to the problem of solubility in the critical region.
One can immediately see that the chemical potential of a solute in dilute
solutions cannot have the usual form (5.4) near the critical points, because
one cannot satisfy Eq. (1.2) using the potential of the form (5.4). ln this
sense [150], the critical dilute solution is not “dilute”! In deriving Eq.
(5.4), it was assumed that one can neglect the interaction between solute
particles because of the initial assumption that the solution is very weak.
However, this is incorrect near the critical point because longrange corre
lations (1.5) cause the eﬀective “interaction” between the solute particles
to be important even for dilute solutions.
Let us obtain the correct expression for the chemical potential µ
s
of
a solute in a binary mixture (calculations for manycomponent mixtures
are quite analogous). Diﬀerentiating the GibbsDuhem relation [3] υ =
υ
1
X+υ
2
(1 −X) for the partial volumes υ
1
and υ
2
, with respect to X, and
eliminating υ
2
, yields
υ
1
= υ + (1 −X)
_
∂υ
∂X
_
T,p
. (5.7)
However, the partial volume is related [3] to the chemical potential υ
1
=
(∂µ
s
/∂p)
T,x
. Integrating the latter formula, and using (5.4) for the ideal
gas as a reference system, gives
µ
s
(T, p, X) = RT ln X +
_
p
0
(υ
1
−υ
id
) dp. (5.8)
Diﬀerentiating (5.8) with respect to X yields
_
∂µ
s
∂X
_
T,p
=
RT
X
+
_
p
0
∂υ
1
∂X
dp
=
RT
X
+ (1 −X)
_
X
0
_
∂
2
υ
∂X
2
__
∂p
∂X
_
dX. (5.9)
In the last equality in (5.9), we used the derivative of Eq. (5.7 ) and dp =
(∂p/∂X)
T,v
dX, which follows from the equation of state p = p (T, υ, X).
Integration by parts and using simple thermodynamic transformations
enables one to rewrite Eq. (5.9) in the ﬁnal form,
_
∂µ
s
∂x
_
T,p
=
RT
X
+(1 −X)
(∂p/∂X)
2
T,v
(∂p/∂υ)
T,X
−(1 −X)
_
v
0
_
∂
2
p
∂X
2
_
T,v
dv. (5.10)
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Comments on the Geometry of the Phase Diagram of a Reaction Mixture 97
In contrast to (5.4), Eq. (5.10) describes the critical line of a binary mix
ture, if one assumes that (∂p/∂υ)
T,X
and
_
∂
2
p/∂υ
2
_
T,X
are proportional
to −X. The former condition leads to the cancellation of the singular
ity RT/X in (5.10), while the latter allows one to satisfy the condition
_
∂
2
µ/∂X
2
_
T,p
= 0.
After this general comment, let us consider Eq. (5.3), and its form in
the critical region. We will show that (∂X/∂p)
T,equil.line
diverges at two
diﬀerent critical points in the phase diagram, called the upper and lower
critical end point (UCEP and LCEP), leading to a marked dependence of
X on p in their respective critical regions.
We ﬁrst rewrite Eq. (5.3) in the following form,
_
∂X
∂p
_
T,equil.line
=
υ
s
−
_
υ −(1 −X)
_
∂p
∂X
_
T,v
_
∂p
∂υ
_
−1
T,X
_
_
∂µ
f
2
∂X
_
T,p
(5.11)
where υ is the molar volume, and use has been made of υ = υ +
(1 −X) (∂υ/∂X)
T,p
and (∂υ/∂X)
T,p
= −(∂p/∂X)
T,υ
(∂p/∂υ)
−1
T,X
. Equa
tion (5.11) contains two “dangerous” derivatives. The derivative (∂p/∂υ)
T
vanishes at the critical point of the pure ﬂuid. On the other hand, the
derivative
_
∂µ
f
2
/∂X
_
T,p
vanishes on the liquidgas critical line of the bi
nary mixture, which terminates at the pureﬂuid critical point. However,
the experimental setup considered here is special, and destroys the rele
vance of the critical line. Due to the contact between the ﬂuid phase and
the solid phase, there is a constraint (5.1) on the chemical potential of the
solute in the ﬂuid phase. This constraint decreases the number of degrees
of freedom of the system, and the system has only isolated critical points
rather than a critical line. These critical points can be found from a solution
of Eqs. (5.1) and (1.2) which deﬁne the critical line of a binary mixture.
For typical phase diagrams, these two points are UCEP and LCEP [70]. At
these two points, and nowhere else, the denominator of Eq. (5.11) vanishes.
It was found experimentally that the temperature T
C
of LCEP is lo
cated close to the critical temperature T
C,0
of the pure, more volatile com
ponent. For example, (T
C
−T
C,0
) /T
C,0
< 10
−2
for all the solution of diﬀer
ent compounds in ethylene [151]. Clearly, when (T
C
−T
C,0
) /T
C,0
is small,
we also expect X and (p
C
−p
C,0
) /p
C,0
to be small. In such cases, we are
faced with a critical point of a mixture that is very dilute and, therefore, is
close to the critical point of the solvent, where (∂p/∂υ)
T
= 0. Comparing
with Eq. (5.11), we conclude that there are two reasons for the divergence
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
98 Chemistry versus Physics: Chemical Reactions near Critical Points
of (∂X/∂p) near the LCEP (both (∂p/∂υ) and (∂µ/∂X) tend to zero), but
only one reason for the divergence near the UCEP, where only (∂µ/∂X)
goes to zero.
A qualitative analysis has been performed separately for LCEP and
UCEP [70] under the assumption of analyticity at the critical points. An
other approach to supercritical extraction near UCEP and LCEP is based
on a scaling theory of the critical phenomena [152]. The slope of the solu
bility curve (5.3) is deﬁned by the derivative
_
∂µ
f
2
/∂X
_
T,p
,
_
∂µ
f
2
∂X
_
T,p
= aπ + bτ
γ
+ cX
δ−1
(5.12)
where τ ≡ (T −T
C
) /T
C
, π ≡ (p −p
C
) /p
C
, and a, b, c are constants, while
γ and δ are the critical indices [152].
The temperature dependence of (5.12) is deﬁned by the thermodynamic
path in approaching the critical point. The experiments were performed at
varying concentration. Therefore, it is natural to assume the smoothness
of the chemical potential of the solvent and the solute as a function of
pressure.
Integrating (5.12), one obtains
µ
f
2
= aπX + bτ
γ
X +
c
δ
X
δ
+ b
1
τ + c
1
π (5.13)
where b
1
and c
1
are constants.
For the experimental thermodynamic path π = const, the compatibil
ity between other variables will be of the form τ ∼ X
δ
. Therefore, one
concludes from (5.13) that asymptotically
_
∂µ
f
2
∂X
_
T,p
∼ τ
(δ−1)/δ
. (5.14)
In Eq. (5.12), we neglect the second term having asymptotic behavior τ
γ
compared to the third term proportional to τ
(δ−1)/δ
, since [152] (δ −1) /δ <
γ. In the asymptotic regime, one keeps the term having the lowest power
of τ.
If the LCEP is close to the critical point of the pure solvent, another
singularity appears in the slope of the solubility, that is due to the “weak”
singularity of the partial molar volume
υ ∼ τ
−α
. (5.15)
Equation (5.15) has been veriﬁed experimentally [153].
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Comments on the Geometry of the Phase Diagram of a Reaction Mixture 99
Our analysis shows that for p = p
C
and for diﬀerent isotherms, the
slope of solubility as a function on τ on the loglog scale is a straight
line independent of the system considered, with the value of this slope
depending on the values of critical indices at diﬀerent distances from the
critical point. A straight line with slope 1.05 seems to correspond to the
experiment results. However, we need more precise measurements.
From the large number of results related to supercritical ﬂuids, we bring
here a partial list of possible applications of chemical reactions in super
critical ﬂuids [154]–[156]. There are certain characteristic features of super
critical ﬂuids which make these applications especially convenient, including
their ability to dissolve various components, in particular, solids, the strong
temperature and pressure dependence of the solubility, and the ability to
use cheap inorganic ﬂuids. Those who drink decaﬀeinated coﬀee and beer
probably know that considerable amounts of caﬀeine and hops are currently
produced by CO
2
supercritical extraction. Among many applications in the
food and pharmaceutical industries, one can mention the extraction of oils,
ﬂowers and fragrances. The rapid motion through a nozzle of an expanded
supercritical ﬂuid with dissolved solids leads to the formation of very small
monodisperse particles of prescribed size and morphology, which is useful
in biotechnology and material science [157]. Other applications include
enhanced oil recovery with supercritical CO
2
[158], wetair oxidation of or
ganics in supercritical water [159], coal liquefaction in supercritical toluene
[160], catalytic disproportionation of nearcritical toluene [161], bitumen
extraction from oil shale [162], activated carbon regeneration [163], and
supercritical chromatography [164]. Technological applications (metallur
gical, geochemical, geological) can be added to this list (see [165] for dozens
of references).
The main interest in diﬀusioncontrolled chemical reactions in super
critical ﬂuids is associated with the large reaction rates, which are several
order of magnitude higher than in normal ﬂuids due to the small kinematic
viscosity. Moreover, these rates can be strongly modiﬁed by a change of
thermodynamic parameters. Modern technology also uses another types
of chemical reactions, such as enzymatic reactions in supercritical CO
2
,
oxidation of hazardous materials in supercritical water, and catalyzed het
erogeneous reactions.
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
100 Chemistry versus Physics: Chemical Reactions near Critical Points
5.2 Azeotropic points in reactive manycomponent systems
A typical temperaturecomposition projection of the T −x
−x
equilibrium
surface (at constant pressure) is shown in Fig. 5.2. This mixture forms an
azeotrope point A where [19] “distillation (or condensation) takes place
without change of composition”. Like critical states, azeotropic states do
not exist for ideal nonreactive mixtures.
Fig. 5.2 Liquidgas equilibrium in a binary mixture, showing the azeotrope point A.
Reproduced from Ref. [18] with permission, copyright (1990), Springer.
The situation is quite diﬀerent for reactive mixtures, for which [166] an
azeotrope exists even for ideal mixtures, and it is no longer deﬁned by the
equality of phase composition. For reactive mixtures, the latter condition
for ncomponent twophase systems is replaced by
x
1
−x
1
ν
1
−ν
T
x
1
=
x
i
−x
i
ν
i
−ν
T
x
i
; i = 1, 2, . . . , n −1 (5.16)
where ν
T
=
n
i=1
ν
i
.
A simple proof [166] of Eq. (5.16) is based on the conservation laws
in a closed systems in which the ncomponent liquid mixture is vaporized
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Comments on the Geometry of the Phase Diagram of a Reaction Mixture 101
at constant pressure or at constant temperature. The material balance for
component i gives
d (υ
x
i
)
dt
+
d (υ
x
i
)
dt
= v
i
dξ
dt
; i = 1, 2, . . . , n −1 (5.17)
where υ
and υ
are the molar volumes of the two phases and ξ is the extent
of reaction. The overall material balance
dυ
dt
+
dυ
dt
= v
T
dξ
dt
(5.18)
permits one to rewrite Eq. (5.17) in the following form
υ
dx
i
dt
+ υ
dx
i
dt
= (ν
i
−ν
T
x
i
)
dξ
dt
−(x
i
−x
i
)
dυ
dt
;
i = 1, 2, . . . , n −1.
(5.19)
During the azeotrope transformation, the composition of each phase
remains constant, i.e., (dx
i
/dt) = (dx
i
/dt) = 0, and (5.19) reduces to
x
i
−x
i
ν
i
−ν
T
x
i
=
dξ
dt
_
dυ
dt
_
−1
; i = 1, 2, . . . , n −1 (5.20)
The righthand side of (5.20) does not depend on i, which proves
Eq. (5.16).
5.3 Melting point of reactive binary mixtures
Consider the phase diagram near the melting point for a twophase system
consisting of a solid component AB which dissociates completely on melting
and the liquid of the same composition. The T −x
B
projection of the phase
diagram is shown in Fig. 5.3, where T
+
is the melting point, the vertical
line at x
B
= 1/2 describes a solid phase, and the dashed lines relate to liquid
phases with concentrations larger and smaller than 1/2. The existence of
two branches of the solubility curve can be explained by the lowering of the
melting point by the addition or removal of one component. Therefore, the
solubility lines meet at T = T
+
.
Let us now consider a chemical reaction (dissociation) in the liquid
phase:
AB A + B. (5.21)
The solubility curve will then be rounded at T = T
+
, as shown in Fig. 5.3
by the solid curves. The solubility curve becomes more rounded with an
increase of the degree of dissociation. This may be explained as follows. Let
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
102 Chemistry versus Physics: Chemical Reactions near Critical Points
Fig. 5.3 Liquidsolid solubility curves for compound AB with (solid curve) and without
(dashed curve) chemical reaction. Reproduced from Ref. [18] with permission, copyright
(1990), Springer.
the liquid solution has the total compositions x
A
, x
B
of substances A and B
and detailed compositions y
A
, y
B
, y
AB
of A, B and AB (y
A
+y
B
+y
AB
= 1).
The chemical potential of AB in the solution must be equal to that in
the solid, which plays the role of the chemical potential “bath” for the
solution. The chemical potential of the solid depends only on temperature
and, therefore, one may assume that for small degrees of dissociation, y
AB
is
a function of T.
Let us now introduce the small parameter ξ to describe the left part of
the melting curves (x
B
< 1/2):
ξ = 2
_
1
2
−x
B
_
. (5.22)
The x, y variables are connected by
x
B
=
y
B
+ y
AB
1 + y
AB
. (5.23)
Using Eqs. (5.22)–(5.23), and the relation y
AB
= f (T), one obtains
y
A
=
1 −f (T)
2
+
ξ
2
[1 + f (T)] ;
y
B
=
1 −f (T)
2
−
ξ
2
[1 + f (T)] .
(5.24)
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Comments on the Geometry of the Phase Diagram of a Reaction Mixture 103
If dissociation is slight, than the solution is ideal. Therefore, one can write
the law of mass action for the reaction (5.21) as follows,
y
A
y
B
= K (T) . (5.25)
Inserting (5.24) into (5.25), one can rewrite the latter in the form
ξ
2
=
[1 −f (T)]
2
[1 + f (T)]
2
−
4K (T)
[1 + f (T)]
2
≡ Q(T)
2
−R(T)
2
. (5.26)
At T = T
+
, ξ = 0, i.e., Q(T
+
) = R(T
+
) . Expanding Eq. (5.26) near
T = T
+
yields
ξ
2
= 2Q
_
T
+
_
__
dR
dT
_
−
_
dQ
dT
__
T=T
+
_
T
+
−T
_
+· · · . (5.27)
Equation (5.27) describes a melting line in the presence of a chemical reac
tion (solid curve in Fig. 5.3).
We now turn to the case of a nonreactive mixture. Then, K (T) = 0 ,
and the ﬁrst term in the series expansion of Eq. (5.26) near T = T
+
is of
second order,
ξ
2
=
_
dQ
dT
_
2
T=T
+
_
T
+
−T
_
2
. (5.28)
In this case the melting lines (the dashed curves in Fig. 5.3) meet at a point.
The reason for the diﬀerence between Eqs. (5.27) and (5.28) was ex
plained back in 1892 [167]. If there were any appreciable dissociation of AB
into its constituents, then, in contrast to the nondissociated solution, the
small addition of either component will not change the equilibrium tem
perature. Therefore, the melting point is a true maximum of a solid curve
[Eq. (5.27)] and not a kink [Eq. (5.28)].
Similar results are obtained from more general considerations, includ
ing the more complicated compounds A
m
B
n
[168] and the strictly regular
solution model [169] (instead of an ideal model).
Further development of these ideas has been carried out by Krichevskii
et al. [170], [171]. They considered the inﬂuence of a third component
on the thermodynamics of the threephase equilibrium among a solid phase
(compound AB
m
, a solvent B with a dissolved substance A), a liquid phase
(solution saturated with this component and extremely dilute with respect
to component C), and a gas phase (vapor of the pure solvent B with the
components A and C assumed to be nonvolatile).
For a nondissociated system, the change in fugacity of the solvent B
(which, for the low saturated vapor pressure, coincides with the partial
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
104 Chemistry versus Physics: Chemical Reactions near Critical Points
pressure) due to the addition of a small amount of C at constant pressure
and temperature, is given by [172]
ln
p
B,(ABC)
p
B,(AB)
= −
n
C
n
B
(5.29)
where p
B,(AB)
and p
B,(ABC)
are the partial pressures of the solvent in binary
and ternary solutions at the same temperature, and n is the number of
moles of the components.
Let us now allow the chemical reaction (dissociation) in the solution,
AB
m
A + mB. (5.30)
The chemical potential of the compound AB
m
in solution, which coexists
with a solid, is ﬁxed and depends only on the temperature. The fugacity γ
of the nonvolatile component A can be replaced by the activity a. Therefore,
the law of mass action for the reaction (5.30) can be written as
a
A
γ
m
B
= K (5.31)
or, in diﬀerential form,
d (ln a
A
) + m d ln (γ
B
) = 0. (5.32)
Combining this equation with the GibbsDuhem equation [3]
n
A
d (ln a
A
) + n
B
d (ln a
B
) + n
C
d (ln a
C
) = 0 (5.33)
one obtains
d ln (γ
B
) = −
n
C
d (ln a
C
)
n
B
−mn
A
(5.34)
If the solution is very dilute with respect to C, one can replace the activity
a
C
by the numbers of moles n
C
.
Integrating Eq. (5.34) yields
ln
γ
B,(ABC)
γ
B,(AB)
= ln
p
B,(ABC)
p
B,(AB)
= −
n
C
n
B,0
−mn
A,0
(5.35)
For the chemical reaction of the form (5.30), the diﬀerence n
B
− mn
A
remains constant. Consequently, we replace n
B
and n
A
by the total number
of moles of the components n
B,0
and n
A,0
in both liquid and solid phases.
At m = 0, Eq. (5.35) reduces to Eq. (5.29). At the melting point in Fig.
5.3 (drawn for m = 1), where two branches of the solubility curves meet, the
compositions of the solid and liquid phases are equal, and the denominator
of Eq. (5.35) reduces to zero. Addition of component C to solution must
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Comments on the Geometry of the Phase Diagram of a Reaction Mixture 105
decrease the value of p
B,(ABC)
on the left branch (n
B
> mn
A
) and increase
it on the right branch (n
B
< mn
A
).
The immediate vicinity of the melting point is of special interest. On
approaching the melting point along the left branch, the righthand side of
Eq. (5.35) tends to −∞, whereas approaching along the right branch, it
tends to +∞. Correspondingly, p
B,(ABC)
must decrease to zero and increase
to inﬁnity, respectively. Both these values (zero and inﬁnity) are physically
impossible. In fact, p
B,(ABC)
on the left branch may not be less than the
dissociation pressure of the compound in solution, and on the right branch,
p
B,(ABC)
cannot exceed the saturated vapor pressure of the pure solvent.
5.4 Double critical point
The double critical (or hypercritical or reentrant) point (DCP) results from
merging the phaseseparation region with the upper (UCST) and lower
(LCST) critical solution points. A binary mixture possessing a LCST sep
arates into two phases as the temperature is raised above the LCPT, and
is miscible in all proportions below LCST. In contrast, the mixture is mis
cible in all proportions above UCST. By merging these two points into one
“double critical point”, the coexistence curve has a closedloop form, sep
arating the closed ordered region from the singlephase disordered region.
Therefore, the ordered twophase state exists in the limited temperature
interval between LCST and UCST [173] shown in Fig. 5.4 together with
the DCP at a certain concentration x
0
.
Such phase diagrams have been found in binary mixtures of glycerol
guaiacol [174] and of water with nicotine [175], lutidine [176], and other or
ganic amines and hydroxyethers. The main interest in the phase diagrams of
the type shown in Fig. 5.4 comes from crystals of the technologically impor
tant Rochelle salt group [177]. In multicomponent systems (with three and
more components), there are more than two coexistence phases, creating
thirdorder and higherorder critical points. In some manycomponent mix
tures, such as propanolwaterNaCl, there is a line of double critical points
which divides the phase separation surface into two parts with UCST and
LCST, respectively.
We present here two possible descriptions of the double critical point,
based on the meanﬁeld model and on the phenomenological approach.
Consider [117] a ternary mixture with mole fractions x
A
, x
B
, and x
C
, which
undergoes a chemical reaction A + B C. Assume that A and B form a
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
106 Chemistry versus Physics: Chemical Reactions near Critical Points
Fig. 5.4 Lines of the upper (UCST) and lower (LCST) critical solution points which
merge at concentration x
0
. Reproduced from Ref. [173] with permission, copyright
(1986), Turpion Publications.
highly directional and energetically favorable bond in becoming C with a
positive interaction of mixing W > 0, and all other interactions included in
the standard chemical potentials µ
0
i
(p, T) of each species. In the absence of
a chemical reaction, the coexistence surface for this mixture contains [178]
a line of UCST’s that decreases to lower temperatures as the concentration
of C is increased. However, as will be shown, in the presence of a chemical
reaction, along with UCST, the LCST will appear leading to the closed
loop coexistence curve. As it was described in Sec. 4.2.2, in the meanﬁeld
approximation, the chemical potentials of the species have the following
form
µ
A
= µ
0
A
(p, T) + ln x
A
+ (W/k
B
T) x
B
(1 −x
A
) ,
µ
B
= µ
0
B
(p, T) + ln x
B
+ (W/k
B
T) x
A
(1 −x
B
) ,
µ
C
= µ
0
B
(p, T) + ln x
C
−(W/k
B
T) x
A
x
B
.
(5.36)
The nonlinear terms in (5.36) are symmetric with respect to interchange
of A and B, leading to the coexistence phases x
A
= x
B
, x
A
= x
B
and
x
C
= x
C
. Therefore, for the phase equilibrium surface, one gets
W
k
B
T
=
1
x
A
−x
B
ln
x
A
x
B
. (5.37)
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Comments on the Geometry of the Phase Diagram of a Reaction Mixture 107
Fig. 5.5 Isotherms of the phase equilibrium surface in a meanﬁeld model. Curves
a, b, c, and d correspond to k
B
T/W = 0.45, 0.4, 0.3 and 0.2, respectively. Reproduced
from Ref. [117], copyright (1989), American Institute of Physics.
The set of isotherms in the x
A
−x
B
plane is shown in Fig. 5.5.
With a decrease in temperature, the phase equilibrium surface ap
proaches the x
A
and x
B
axes, whereas with increasing temperature, one
approaches the critical point deﬁned by T
cr
= W/2k
B
, x
A,cr
= x
B,cr
= 1/2,
x
C,cr
= 0. Let us now consider the constraint imposed by the chemical
equilibrium,
µ
A
+ µ
B
−µ
C
= 0 (5.38)
Equations (5.38) and (5.36) deﬁne the equilibrium expression for x
C
,
x
C
= x
A
x
B
exp
_
−
∆H
0
k
B
T
+
∆S
0
k
B
_
exp
_
W
k
B
T
(x
A
+ x
B
−x
A
x
B
)
_
(5.39)
where ∆H
0
and ∆S
0
are the enthalpy and entropy of the chemical reac
tion which are assumed to be constant. The solutions of Eqs. ( 5.37) and
(5.39) deﬁne the coexistence surface of the reactive system in the plane
(k
B
T/W, x
A
) for ﬁxed values of ∆H
0
/W and ∆S
0
/k
B
. These equations
have been solved numerically [117] for ∆H
0
/W = −10 and several values
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
108 Chemistry versus Physics: Chemical Reactions near Critical Points
of ∆S
0
/k
B
. The resulting phase diagram (Fig. 5.6) shows that A − B re
pulsion results in phase separation as the temperature is lowered. However,
upon lowering the temperature still further, the formation of C becomes
favorable, and the phases become miscible again. In this manner, one ob
tains the closedloop coexistence curves created by LCST and UCST with
their ratio decreasing as −∆S
0
/k
B
decreases.
Fig. 5.6 Closedloop coexistence curves for ∆H
0
/W = −10 and
_
−∆S
0
/k
B
_
= 60, 40,
30, 26, 24.5, and 24.1. The loops get smaller with decreasing (−∆S
0
/k
B
). Reproduced
from Ref. [117], copyright (1989), American Institute of Physics.
As usual, the phenomenological description of the vicinity of the DCP
is given by the LandauGinzburg expansion for the free energy Φ in the
order parameters Ψ and its spatial derivatives:
Φ = Φ
0
+ A(T, x) Ψ
2
+ B (T, x) Ψ
4
+ C (∇Ψ)
2
+· · · , (5.40)
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Comments on the Geometry of the Phase Diagram of a Reaction Mixture 109
where the coeﬃcient A changes sign on the line of phase transitions T
C
(x),
with A > 0 for a disordered phase and A < 0 for an ordered state. The
assumption of analyticity, used in (5.40), is usually applied also to the
function A(T, x) , namely, A(T, x) = a (T −T
C
). It is obvious that the
latter assumption is not applicable to the case of DCP, since the non
monotonic behavior of T
C
(x) leads to a nonmonotonic form of A(T) rather
than the linear function [173] (Fig. 5.7).
Fig. 5.7 The nonmonotonic dependence of the ﬁrst coeﬃcient in the LandauGinzburg
expansion as a function of temperature. There are two phase transitions for x < x
0
, and
no phase transitions for x > x
0
. Reproduced from Ref. [173] with permission, copyright
(1986), Turpion Publications.
As one can see from this ﬁgure, for x = x
0
, the curve A(T) is tangent
to the temperature axis at the point T
0
, whereas for x < x
0
, the minimum
of A(T) occurs below the horizontal axis, intersecting it at LCST and at
UCST. Due to the existence of minima, it is convenient to expand the
function A(T, x) around the minimal temperature T
min
and not around T
C
as is usually done,
A(T, x) = A
0
(x) + A
2
(x)
¸
¸
¸
¸
T −T
min
T
0
¸
¸
¸
¸
2
+· · · . (5.41)
The existence of terms in (5.41) quadratic in (T −T
min
) /T
0
, instead
of the usual linear terms, does not change the critical indices, which de
ﬁne the singularities of the compressibility, correlation radius and the
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
110 Chemistry versus Physics: Chemical Reactions near Critical Points
quantities related to them. However, the anomalies of the quantities
containing the derivatives of the coeﬃcient A with respect to tempera
ture, decrease near the DCP. For example, the jump of the heat capacity
∆C = −T∂
2
_
A
2
/2B
_
/∂T
2
at LCST and UCST decreases proportional to
the square of the relative distance between these points. Indeed, a decrease
of several orders of magnitude in the anomalies of the heat capacity was
found in all systems with the DCP. However, there are some discrepancies
between the theory and experiment [173] which makes specially interest
ing the microscopic description of the DCP and the closeloop coexistence
curves which have been carried out in the context of latticegas [179] and
decoratedlatticegas [96] models.
It should be noted that the anomaly of the kinetic coeﬃcients are also
diﬀerent from the usual critical point. For example, near DCP, the mea
surements of the shear viscosity of ternary mixtures of 3methylpyridine,
water and heavy water as a function of temperature show [180] that near
DCP, the exponent describing the powerlaw divergence in temperature of
the viscosity nearly doubles as the concentration approaches the double
criticalpoint concentration. There are also experimental data describing
light scattering and sound propagation near the DCT [181]. However, we
are unaware of any experiments of chemical reactions near DCP, which are
of special interest.
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Chapter 6
Sound Propagation and Light
Scattering in Chemically
Reactive Systems
6.1 Ultrasound attenuation in nearcritical reactive
mixtures
The idea of using the attenuation of sound to study the rate of dissociation
of the reaction N
2
O
4
2NO
2
was proposed in 1920 by Albert Einstein
[182]. The total attenuation per wavelength αλ near the critical point can
be written as
αλ = (αλ)
cr
+ (αλ)
chem
, (6.1)
neglecting the small background contribution due to viscosity and thermal
conductivity. Equation (6.1) contains the assumption of the additive contri
butions of (αλ)
cr
and (αλ)
chem
, which come from the critical ﬂuctuations
and chemical reaction, respectively. The assumption of independence of
these two contributions is based on the following picture [183]: the reaction
takes the system from equilibrium to the steady state, which afterwards
will relax to the initial equilibrium state with characteristic relaxation time
τ
chem
. The latter deﬁnes the sound attenuation due to a chemical reaction
(αλ)
chem
= A
0
ωτ
chem
1 + (ωτ
chem
)
2
(6.2)
which depends on the nearness to the critical point, since τ
chem
is propor
tional to the equilibrium derivative of the extent of reaction ξ with respect
to the aﬃnity A, τ
chem
∼ (∂ξ/∂A)
eq
[36].
The calculation of the critical contribution to the sound attenuation,
expressed by the (αλ)
cr
term in Eq. (2.77 ), has been performed [184].
This theory is based on an assumption going back to Laplace [185], who
corrected Newton calculation showing that the sound velocity in air should
be calculated using the adiabatic rather than the isothermal compressibil
ity. In line with this comment, it was assumed [184] that the change of
111
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
112 Chemistry versus Physics: Chemical Reactions near Critical Points
pressure caused by the propagating sound, produces a shift in the critical
temperature, but not in the critical concentration. Neglecting the changes
in concentration, we are left with two thermodynamic variables p and T,
or p and τ ≡ T −T
C
(p), where T
C
(p) is the pressuretemperature critical
line. Simple thermodynamic analysis shows [184] that the sound veloc
ity u remains constant at the critical point, u (T = T
C
) ≡ u
C
, with small
corrections ∆u near the critical point,
u ≡ u
C
+ ∆u = u
C
+
g
2
u
3
C
2T
C
C
p
(6.3)
where g = T
C
(∂S/∂p)
λline
V
−1
C
is a dimensionless constant, and C
p
is the
speciﬁc heat. The above analysis has been performed at nonzero frequency
ω for an input pressure signal with time dependence exp(−iωt). Therefore,
Eq. (6.3) determines the complex sound velocity u (ω) as a function of the
complex speciﬁc heat C
p
(ω). Dynamic scaling theory has been used to de
termine the frequency dependence of C
p
(ω) from the known temperature
dependence of the thermodynamic heat capacity C
p
. The function C
p
(ω)
can be divided into the noncritical frequencyindependent background com
ponent C
1
and the critical part C (ω), where C
p
= C
1
+ C (0). The low
frequency expansion of C
p
(ω) gives
C (ω) ≈ C (0) + ωC
(0) = C (0) (1 + iωτ) (6.4)
where τ = −iC
1
(0) /C (0) is an eﬀective mean relaxation time. Upon in
serting (6.4) into (6.3) and assuming ωτ << 1, one obtains for the negative
imaginary part of the critical velocity Im (u
C
) (the sound attenuation per
wavelength α
cr
λ),
α
cr
λ = C (0)
u
3
C
2T
C
_
g
C
p
_
2
ωτ. (6.5)
The critical part of the speciﬁc heat C (0) , induced by concentration
ﬂuctuations, has the following thermodynamic temperature dependence
C (0) ≈ t
−α
0
(6.6)
where α
0
is the critical index and t = ∆T/T
C
= (T −T
C
) /T
C
is the re
duced temperature. The relaxation rate of concentration ﬂuctuations of
wavenumber k is k
2
D, where the critical diﬀusion coeﬃcient is D(ξ) =
k
B
T
C
/6πηξ with ξ ≈ t
−ν
and η ≈ t
−z
0
. Therefore, the characteristic relax
ation time for the ﬂuid is
γ (t) = 2Dξ
−2
= γ
0
(∆T)
νz
0
(6.7)
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Sound Propagation and Light Scattering in Chemically Reactive Systems 113
where γ
0
is constant. Equation (6.7) enables us to eliminate t in favour of
γ as the variable that indicates how close the system is to the critical point,
C (γ, ω = 0) = B
_
γ
0
γ
_
α
0
/νz
0
. (6.8)
According to the dynamicscaling requirement [186], the dependence of
Eq. (6.8) on γ disappears when, for γ < ω, γ is replaced by −iω so that
the frequencydependent speciﬁc heat at the critical point is
C (γ = 0, ω) = B
_
γ
0
−iω
_
α
0
/νz
0
= B exp
_
iπα
0
νz
0
_
_
γ
0
ω
_
α
0
/νz
0
. (6.9)
Replacing C (0) in Eq. (6.5) by the imaginary part of C (γ = 0, ω) , and
using α
0
<< νz
0
, one gets ﬁnally
α
cr
λ
π
2
α
0
g
2
u
2
C
B
2T
C
C
2
p
(T
c
)
_
γ
0
2π
_
α
0
/νz
0
f
−α
0
/νz
0
(6.10)
where f = ω/2π is the frequency. Multiplying Eq. (6.10) by (u
C
f)
−1
predicts a linear proportionality between α
cr
f
−2
and f
−1−α
0
/νz
0
. Pre
cisely this dependence had been observed in the measurements [184] of
the sound attenuation in 3methylpentanenitroethane at various frequen
cies ranging from 16.5 to 165 MHz. Agreement between theory and ex
periment was also found for polypropyleneglycol and polyethyleneglycol
mixtures [188], isobutoxyethanolwater system [189], ethanoldodecane
and methanolcyclohexane mixtures [190], cyclohexanenitroethane[191],
methanolhexane [192], and triethylaminewater, methanolcyclohexane
[193].
6.2 Hydrodynamic analysis of the dispersion relation for
sound waves
Sound waves are oscillating movements of small amplitude, which give rise
to small perturbations in the system. Therefore, one can use the linearized
hydrodynamic equations for the analysis of the sound velocity and attenua
tion coeﬃcient in ﬂuids. The linearized system of hydrodynamic equations
for reactive systems was obtained in Chapter 2. We will here use the fol
lowing simpliﬁed version of these equations [194]
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
114 Chemistry versus Physics: Chemical Reactions near Critical Points
∂ρ
∂t
= −ρ
0
div (v) (6.11)
ρ
0
∂v
∂t
= −grad (p) + η∇
2
v
+
_
η
3
+ ζ
_
grad (div (v)) + ρ
0
L
1
grad (A) (6.12)
∂ξ
∂t
= −L
0
A −L
1
div (v) (6.13)
A =
_
∂A
∂ρ
_
ξ,S
ρ +
_
∂A
∂ξ
_
ρ,S
ξ +
_
∂A
∂S
_
ξ,ρ
S (6.14)
p =
_
∂p
∂ρ
_
ξ,S
ρ +
_
∂p
∂ξ
_
ρ,S
ξ +
_
∂p
∂S
_
ξ,ρ
S (6.15)
T =
_
∂T
∂ρ
_
ξ,S
ρ +
_
∂T
∂ξ
_
ρ,S
ξ +
_
∂T
∂S
_
ξ,ρ
S. (6.16)
In contrast to Eqs. (2.61)–(2.64), the independent variables in the above
equations are the velocity v, the deviations from equilibrium values of en
tropy s, density ρ and extent of reaction ξ. Kinetic coeﬃcients are taking
into account only in the NavierStokes equation (6.12).
To derive the dispersion law, we take the Fourier transform of the hy
drodynamic variables
ˆ
X (K, ω) =
_
d
3
r
_
dtx(r, t) exp (iωt −iKr) . (6.17)
The wave vector K is written K = k + iγ, where k is the propagation
vector and γ is the attenuation coeﬃcient.
After some additional simpliﬁcations [194], Eqs. (6.11)–(6.16) yield the
following expressions for k and γ
k =
ω
C
; γ =
ω
2
2ρ
0
G
C
3
(6.18)
where
C
2
≡ −
1
ρ
2
0
_
¸
_
_
∂p
∂υ
_
ξ,S
+
_
∂p
∂ξ
_
v,S
_
∂ξ
∂υ
_
A,S
1 + ω
2
τ
2
+2ρ
0
L
1
_
∂p
∂ξ
_
υ,S
ω
2
τ
2
1 + ω
2
τ
2
_
(6.19)
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Sound Propagation and Light Scattering in Chemically Reactive Systems 115
G ≡
_
∂p
∂ξ
_
υ,S
_
_
∂ξ
∂υ
_
A,s
τ
ρ
0
(1 + ω
2
τ
2
)
+
2τL
1
1 + ω
2
τ
2
_
+
4η
3
+ ζ (6.20)
and the relaxation time τ is given by
τ ≡
_
L
0
_
∂A
∂ξ
_
υ,S
_
−1
. (6.21)
Some limit cases of these general equations has been considered [194].
This theory contains viscoreactive cross coeﬃcients L
1
, which link the
chemical reaction to the momentum ﬂow, which is important for a weak
electrolyte solution [194]. Now let us turn to the slightly diﬀerent hydro
dynamic approach, which allows us to consider the vicinity of the critical
points without the L
1
terms [195]. Starting from the same hydrodynamic
equations obtained in Chapter 2, we replace them by the statistically in
dependent variables as it was done there. However, instead of variables ξ,
p, div (v) and φ = C
p,x
T/T
0
− α
T
p/ρ
0
, we use the set ξ, p, div (v) and
S
1
= S − (∂S/∂ξ) ξ. The hydrodynamic equations (2.91)–(2.94) may then
be rearranged to give
∂ρ
∂t
= −ρ
0
div (v) (6.22)
ρ
0
∂div (v)
∂t
= −∇
2
p +
_
4
3
η + ζ
_
∇
2
div (v) (6.23)
∂ξ
∂t
= D
_
∇
2
ξ +
k
T
T
0
∇
2
T +
k
p
p
0
∇
2
p
_
+
1
τ
T,p
_
ξ −
_
∂ξ
∂T
_
p,A
T −
_
∂ξ
∂p
_
T,A
p
_
(6.24)
∂S
1
∂t
=
λ
ρ
0
T
0
∇
2
T + D
k
T
T
0
_
∂A
∂ξ
_
T,p
_
∇
2
ξ +
k
p
p
0
∇
2
p
_
+D
_
∂S
∂ξ
_
p,T
_
∇
2
ξ +
k
T
T
0
∇
2
T +
k
p
p
0
∇
2
p
_
(6.25)
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
116 Chemistry versus Physics: Chemical Reactions near Critical Points
where τ
T,p
=
_
L(∂A/∂ξ)
T,p
_
−1
is the relaxation time, and all other nota
tion has been explained in Chapter 2.
Since only the ﬂuctuations of concentration and entropy become anoma
lously large near the critical point, we may use the simpliﬁed macroscopic
equations (6.24)–(6.25) in which the pressure is kept constant. Introduc
ing both normalized and statistically independent spatial Fourier transform
variables, one obtains
β
1
(k, t) =
_
ρ
0
k
B
υC
p,ξ
_
1/2
S
1
(k, t) ,
β
2
(k, t) = ρ
0
_
(∂A/∂ξ)
T.p
k
B
Tυ
_
1/2
ξ (K, t) .
(6.26)
It is easy to show [195] that
d
dt
_
β
1
(K, t)
β
2
(K, t)
_
= M
_
β
1
(K, t)
β
2
(K, t)
_
(6.27)
where the 2 ×2 matrix M is given by
M =
_
_
_
_
_
_
λK
2
ρ
0C
p,ξ
+
γ
T
τ
T,p
, k
T
D
_
(
∂A
∂ξ
)
T.p
TC
p,ξ
K
2
−
γ
1/2
T
τ
T,p
k
T
D
_
(
∂A
∂ξ
)
T.p
TC
p,ξ
K
2
−
γ
1/2
T
τ
T,p
, DK
2
+ τ
−1
T,p
_
_
_
_
_
_
(6.28)
and
γ
T
≡
C
p,A
−C
p,ξ
C
p,ξ
. (6.29)
The nondiagonal elements of the matrix M are very small and may be
neglected if the propagation number K satisﬁes the inequality
K >
_
_
C
p,A
−C
p,ξ
_
λ
ρ
0
−DC
p,ξ
_
τ
T,p
_
_
1/2
. (6.30)
Numerical estimates indicate that (6.30) is satisﬁed for K > 10
4
cm
−1
.
The variables β
1
(k, t) and β
2
(k, t) are then eﬀectively decoupled, and
ξ (K, t) ξ (−K, t) = exp
_
−
_
DK
2
+ τ
−1
T,p
_
t
_
ξ (K) ξ (−K) . (6.31)
The modecoupling method has been used [102] in the analysis of the
propagation of sound in a nonreactive binary mixture near the critical
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Sound Propagation and Light Scattering in Chemically Reactive Systems 117
point. The general formula for the complex sound attenuation coeﬃcient
´ α(ω) was obtained in [102],
ˆ α(ω) ≈ ω
2
_
dK
_
∂ ln ξ (K) ξ (−K)
∂T
_
2
×
_
∞
0
dt exp
_
−2
_
DK
2
+ τ
−1
T,p
_
t
_
exp (iωt) . (6.32)
The arguments that lead to Eq. (6.32) remain unchanged when a chem
ical reaction occurs. The existence of a chemical reaction results in an
additional mechanism for the relaxation of the composition back to its
equilibrium value, which results in the appearance of the additional τ
−1
T,p
term in Eq. (6.31). Then, in a reactive system,
ˆ α(ω) ≈
ω
2
2
_
D(K) K
2
+ τ
−1
T,p
(K)
_
−iω
×
_
dK
_
∂ ln < ξ (K) ξ (−K)
∂T
_
2
(6.33)
where the Kdependence of the diﬀusion coeﬃcient, D(K), and of the
relaxation time τ
−1
T,p
(K) = Lχ
−1
K
(K) (which reduces to L(∂A/∂ξ)
T,p
for
K = 0) have been taken into account.
For nonreactive systems, relaxation is associated with diﬀusion. Just
as diﬀusion processes show critical slowingdown with time scale τ
D
∼
_
Dk
2
_
−1
∼ [(T −T
C
) /T
C
]
−3ν
, the sound attenuation will show critical
behavior over the same time scale. On the other hand, if the chemi
cal reaction occurs on a faster time scale than diﬀusion and shows crit
ical slowingdown, then the sound attenuation will relax on a time scale
τ
C
∼
_
(∂A/∂ξ)
T,p
_
−1
∼ [(T −T
C
) /T
C
]
−γ
. Because γ < 3ν, the chemistry
is faster and will dominate the diﬀusion which can be seen in sound exper
iments. However, the diﬀusion coeﬃcient D(k) is wavenumberdependent
and, according to Eq. (3.32), at high frequencies, D ∼ k. The characteristic
diﬀusion time scale will then be τ
D
∼
_
Dk
2
_
−1
∼ k
−3
. This will occur on
a faster time scale than the chemical reaction, and the crossover will occur
by diﬀusioncontrolled sound attenuation.
Finally, by plotting the dimensionless sound attenuation coeﬃcient
α(ω) /ω
2
versus ωτ at various temperatures, one can see whether τ
D
∼
[(T −T
C
) /T
C
]
−3ν
or τ
C
∼ [(T −T
C
) /T
C
]
−γ
yields a better dynamic scal
ing function. The calculated change in sound velocity ∆C/Cω
2
for ωτ << 1
is proportional to (ωτ)
1/2
in the nonreactive case and to ωτ in the reactive
case [195] .
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
118 Chemistry versus Physics: Chemical Reactions near Critical Points
6.3 Light scattering from reactive systems
Light scattering as a probe of chemical dynamics is able to detect sponta
neous ﬂuctuations which are drastically increased near the critical points.
Like sound, this is essentially a nonperturbative method which, by using
the microscopic polarizability of molecules, provides a complimentary tool
to other methods of studying chemical kinetics. Measuring the angular de
pendence and the shift in the frequency of the light scattered by a system
allows one to ﬁnd both the spatial and the time dependence of the ﬂuctu
ations. The electromagnetic wave impinges upon a molecule inducing the
dipole moment proportional to the polarizability. For a macroscopic con
tinuum, the summation over the polarizabilities over individual molecules
gives the macroscopic dielectric constant ε = ε
0
+δε (r, t), where the latter
part comes from the spacetime ﬂuctuations of local thermodynamic vari
ables. Assuming local equilibrium of a reactive binary mixture described
by the reduced entropy S
1
≡ S − (∂S/∂ξ)
T,p
ξ, pressure p, and the extent
of reaction ξ, one obtains
δε (r, t) =
_
∂ε
∂S
_
p,ξ
δS
1
+
_
∂ε
∂p
_
S
1
,ξ
δp +
_
∂ε
∂ξ
_
p,S
1
δξ. (6.34)
Light scattering theory assumes [196], [197] that the intensity I of
scattered light having wavenumber K
0
at point R
0
is described by the
expression
I (r
0
, K, ω) =
NI
0
K
4
0
16π
2
R
2
0
sin
2
φδε (K, ω) δε (−K, 0) (6.35)
where I
0
is the intensity of the incident light, N is the number of molecules
in the scattered volume, ω is the shift in the frequency of the scattered
light, K is the change in the wavenumber in the medium of the scattered
light, and φ is the angle between the electric vector of the incident wave
and the vector R
0
. Inserting the space Fourier transform of Eq. (6.34) into
(6.35) gives the light intensity as a function of timedependent correlators
of the statistical independent thermodynamic variables S
1
(K, t), p (K, t)
and ξ (K, t). The hydrodynamic equations express the latter in terms of
the static thermodynamic ﬂuctuations of S
1
(K) , p (K) and ξ (K).
There are two major mechanisms for the relaxation of ﬂuctuations [197]:
one arises from transport properties, and the other is due to the local
changes induced by chemical reactions. The ﬁrst mechanism depends upon
spatial gradients, i.e., the line width of the scattered light is proportional to
the square of the wavenumber K
2
, whereas for the second mechanism, the
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Sound Propagation and Light Scattering in Chemically Reactive Systems 119
line width is independent on K. Therefore, in principle, one can measure
chemical relaxation times by the extrapolation K →0. However, in a real
ﬂuid, the kinetic and chemical modes are coupled. This coupling can be
described by the hydrodynamic equations (6.22)–(6.25), which, according
to (6.35), have to be written in the ω −K Fourier form,
d
dt
_
_
_
_
β
1
(K, t)
β
2
(K, t)
β
3
(K, t)
β
4
(K, t)
_
_
_
_
= M
_
_
_
_
β
1
(K, t)
β
2
(K, t)
β
3
(K, t)
β
4
(K, t)
_
_
_
_
(6.36)
where the independent variables β
i
(K, t) are div (v) and the normalized
forms of deviations of pressure, reduced entropy, and the mass fraction of
the species with the highest partial speciﬁc enthalpy from their equilibrium
values [198],
β
1
(k, t) =
_
ρ
0
k
B
υC
p,ξ
_
1/2
S
1
(k, t) ,
β
2
(k, t) =
_
ρ
0
(∂A/∂ξ)
T.p
k
B
Tυ
_
1/2
ξ (K, t) ,
β
3
(k, t) =
_
(∂v/∂p)
S,ξ
k
B
Tυ
_
1/2
p (k, t) ,
β
4
(k, t) =
_
ρ
0
k
B
TK
2
υ
_
1/2
div (v)
(6.37)
and the 4 ×4 matrix M has the following form
M
11
=
λ
ρ
0
C
p,ξ
K
2
+ γ
T
τ
−1
T,p
,
M
13
= M
31
= C
1
K
2
+ C
2
τ
−1
T,p
,
M
12
= M
21
= k
T
D
_
(∂A/∂ξ)
T.p
TC
p,ξ
_
1/2
K
2
−γ
1/2
T
τ
−1
T,p
,
M
22
= DK
2
+ τ
−1
T,p
,
M
23
= M
32
= C
3
K
2
+ C
4
τ
−1
T,p
,
M
33
= C
5
K
2
+ (r
M
C
p,A
/C
p,ξ
) τ
−1
T,p
,
M
34
= M
43
= Kc
∞
,
M
44
= C
6
K
2
,
M
14
= M
24
= M
41
= M
42
= 0
(6.38)
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
120 Chemistry versus Physics: Chemical Reactions near Critical Points
where all notations can be found in [198]. However, when the damping
terms in the matrix M are much smaller than the frequency of the sound
wave Kc
∞
,
λ
ρ
0C
p,ξ
K
2
<< Kc
∞
,
DK
2
<< Kc
∞
,
1
ρ
0
_
4
3
η + ς
_
K
2
<< Kc
∞
(6.39)
and the inverse chemical reaction time τ
−1
T,p
satisﬁes
τ
−1
T,p
<< Kc
∞
, (6.40)
then the dispersion relation reduces in zero approximation to Eqs. (6.26)–
(6.27) with additional ﬁrstorder coupling between the pressure ﬂuctuations
and ﬂuctuations in entropy and concentration. The four roots of the dis
persion equation have been found [199]. The two real roots z
1,2
are related
to the nonpropagating modes and deﬁne the Rayleigh line in the scattered
light,
z
1,2
=
1
2
_
(d
1
+ d
2
) ±
_
(d
1
−d
2
)
2
+ 4d
2
12
_
1/2
_
, (6.41)
and the complex roots z
3,4
, relate to two propagating modes deﬁning two
symmetric Brillouin lines,
z
3,4
=
_
ΓK
2
+
1
2
c
2
∞
−c
2
0
c
2
0
τ
−1
T,p
_
±iKc
∞
. (6.42)
The following notation has been introduced in Eq. (6.41),
d
1
=
λ
ρ
0
C
p,ξ
K
2
+
C
p,A
−C
p,ξ
C
p,ξ
τ
T,p
; d
2
= DK
2
+ τ
−1
T,p
,
d
12
=
_
Dk
T
K
2
−
T
τ
T,p
_
∂ξ
∂T
_
p,A
_
_
_
∂A
∂ξ
_
T,p
(TC
p,ξ
)
−1
(6.43)
while Γ in Eq. (6.42) is the same combination of transport coeﬃcients,
which is the damping factor of the Brillouin lines for a twocomponent
nonreactive system.
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Sound Propagation and Light Scattering in Chemically Reactive Systems 121
Let us check the restrictions arising from inequalities (6.39) and ( 6.40).
Conditions (6.39) are always met for suﬃciently small K. For typical magni
tudes of the transport coeﬃcient and the speed of the sound, the conditions
given in Eq. (6.39) are satisﬁed even for the maximum scattering wavenum
bers (K
max
≈ 2 × 10
5
sm
−1
) encountered in light scattering experiments.
To satisfy condition (6.40 ), K must be larger than some minimal value
which depends on τ
T,p
. To be in the region of wavenumbers used in light
scattering experiments, the inverse of the chemical reaction time must be
of order 10
9
s
−1
or smaller.
Spectroscopic techniques permits us to obtain qualitative information
about the local environment around solute molecules. Using pyrene as a
spectroscopic probe, appreciable enhancement up to 2.5 times the bulk
density was seen for supercritical C
2
H
6,
CHF
3
and CO around pyrene
[200].
6.4 Inhomogeneous structure of nearcritical reactive
systems
The high compressibility of a dilute solution near the liquidgas critical
point gives rise to a local density increase, in which the density of solvent
around a solute molecule is higher that the bulk density. The measurement
of partial molar volumes of naphthalene in ethylene yields [201] values that
are extremely large and negative, implying that the local environment near
the solvent molecules is very diﬀerent from the bulk properties of the ﬂuid.
This phenomenon, called “molecular charisma” [201], or “density augmen
tation” [200], or “ﬂuid clustering” [202], has been studied intensively using
spectroscopic measurements. The use of pyrene as a solute is specially con
venient because it is a chromophore, ﬂuorescent, and its photophysics is
well known. Extensive measurements, using steadystate and timeresolved
ﬂuorescence of dissolved pyrene in supercritical CO
2
[203] and critical wa
ter [204], clearly show statedependent local density enhancements in dilute
supercritical ﬂuid solutions. The magnitude of this enhancement was ana
lyzed in terms of the local CO
2
density, based on the Onsager reaction ﬁeld
theory [205]. A solute with a dipole moment µ, contained in a spherical
cavity of radius a, and dissolved in a solvent with static dielectric constant
and refractive index n, polarizes the solvent and produces a reaction ﬁeld.
This reaction ﬁeld lowers the groundstate energy of the solvated solute
relative to the gas phase value. This shift ∆ν between vapor and solution
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
122 Chemistry versus Physics: Chemical Reactions near Critical Points
phase is [206]
∆ν =
µ
2
g
−µ
2
c
a
3
n
2
−1
2n
2
+ 1
+
2µ
c
_
µ
g
−µ
c
_
a
3
_
−1
−2
−
n
2
−1
n
2
+ 2
_
(6.44)
where µ
g
and µ
c
are the dipole moments of the ground state and excited
states, respectively. However, experiment shows that a shift ∆ν between
vapor and solution is not linear in
_
n
2
−1
_
/
_
n
2
+ 2
_
, as predicted by Eq.
(6.44). This demonstrates that the local ﬂuid environment surrounding
the ground state pyrene molecules is diﬀerent from that predicted by Eq.
(6.44), and exceeds the expected value. On the other hand, there is a
relationship [207] between the polarizability term
_
n
2
−1
_
/
_
n
2
+ 2
_
and
the supercritical CO
2
density ρ
loc
,
n
2
−1
n
2
+ 2
= 6.6ρ
loc
+ 1.25ρ
2
loc
−264ρ
3
loc
. (6.45)
Fig. 6.1 Recovered density enhancement ρ
loc
/ρ
bulk
as a function of the dimensionless
ﬂuid density ρ/ρ
crit
for pyrene in CO
2
at temperature T = 45 K. Reproduced from
Ref. [203] with permission, copyright (1995), American Chemical Society.
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Sound Propagation and Light Scattering in Chemically Reactive Systems 123
Equations (6.44) and (6.45) were used [203] to ﬁnd the spectral shift ∆ν
as a function of bulk density ρ
bulk
. The experimental spectral shift ∆ν is
used to compute a local polarizability term in Eq. (6.44), which determines
[203] the local density ρ
loc
from Eq. (6.45). In this manner, the local
density augmentation ρ
loc
/ρ
bulk
was found as a function of the reduced
density ρ/ρ
crit
of CO
2
, as shown in Fig. 6.1, which displays a local density
augmentation up to 170 at about half the critical density. In addition to
this local density enhancement, connected with the ground state of pyrene
molecules, similar experiments have been performed for the pyrene excited
state. The latter turned out to be 1.5 times greater than in the ground
state, due to the increased electrostatic interaction between the excited
state of pyrene and CO
2
. All these results do not depend on the pyrene
concentration, indicating that pyrene molecules are solvated individually
in CO
2
.
The same increase of the solvent density around a solute molecule com
pared with the corresponded bulk density has been obtained [208] using
(N,Ndimethylamino) benzonitrile as a probe instead of pyrene. Additional
veriﬁcation of the inhomogeneity of a nearcritical reactive system has been
obtained by indirect measurements combined with thermodynamic calcu
lations [209], [210]. Finally, the same result follows from numerical simula
tions based on the LennardJones potential [211] and the theory of integral
equations [212].
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Chapter 7
Conclusions
The main body of this book consists of a detailed discussion of the inter
connection between “physical” critical phenomena and chemical reactions.
Although these two phenomena appear to be quite diﬀerent, deep analysis
shows [22], [21] that the instability of diﬀerent nonequilibrium phenomena
(hydrodynamic and chemical instabilities, laser transitions, among others)
show an intimate connection with equilibrium phase transitions and crit
ical phenomena. Indeed, both are characterized by the enhancement of
ﬂuctuations, longrange order and critical slowingdown. Examples of this
connection include the relation which has been established [213] between
the laser threshold and the critical point of a secondorder phase transi
tion, and the analogy between phase transitions of the ﬁrst and second
order and hard and soft transitions of chemical reactions. The theory of
critical phenomena, which is supported experimentally with a precision of
few decimal places, dates from the 1960’s and 1970’s. It is not surprising
that the methods used in this theory (renormalization group, modemode
coupling, etc) have been successfully employed in modern chemistry.
Of three branches of modern physics — theory, experiment and simu
lation — we mostly used the ﬁrst in the phenomenological analysis, with
references to experimental work given in appropriate places. Many refer
ences to numerical methods are given in review articles, such as [165], which
describe numerical methods of phase equilibria in reactive systems.
A nearcritical ﬂuid has some of the advantages of both a liquid and
a gas: high density ensures high dissolving power which is accompanied
by high diﬀusivity of solutes and low viscosity facilitates mass transport.
In addition, high compressibility allows the tuning of the density and
dissolving power by small changes in pressure. That is the reason why
chemical reactions near the critical point are in such wide use in modern
125
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
126 Chemistry versus Physics: Chemical Reactions near Critical Points
science and technology. Many applications can be found in review arti
cles ([156], [165]). We may mention applications in chemistry (supercritical
extraction [145]), metallurgy (the recovery of metals from coal ash [214],
phase diagrams in alloy systems [215]), geochemistry (distribution and ac
tivity of species in aqueous solutions [216], solubility of minerals in brine
solutions [217]), geology (equilibria between aqueous and mineral phases
[218], mineral equilibria [219]), food industry (extraction of caﬀeine from
coﬀee beans [220] and oil from corn, sunﬂower and peanuts [221]), and the
pharmaceutical industry (extraction of vitamin E from soybean oil and its
puriﬁcation [222]).
This subject still attracts great interest, and it is very likely that many
new applications will be found in the future.
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Bibliography
[1] H. G. Harris and J. M. Praushnitz, Ind. Eng. Chem. Found. 8, 180, 1969.
[2] M. Gitterman, Phys. Stat. Sol. 59, 295, 1980.
[3] I. Prigogine and R. Defay, Chemical Thermodynamics, Longmans Green,
New York, 1954
[4] L. D. Landau and E. M. Lifshitz, Statistical Physics, Pergamon, London,
1958; Fluid Dynamics, Pergamon, London, 1987.
[5] A. A. Galkin and V. V. Lunin, Russian Chem. Rev. 74, 21, 2005.
[6] S. N. V. K. Aki and M. A. Abraham, Envir. Prog. 17, 246, 1998.
[7] S. N. V. K. Aki and M. A. Abraham, Chem. Eng. Sci. 54, 3533, 1999.
[8] A. Kruse and H. Vogel, Chem. Eng. Technol. 31, 23, 2008.
[9] A. M. N. da Ponte, J. Supercrit. Fluids 47, 344, 2009.
[10] M. Gitterman, Rev. Mod. Phys. 50, 85, 1978.
[11] G. Arcovito, C. Falocci, M. Riberty, and L. Mistura, Phys. Rev. Lett. 22,
1040, 1969.
[12] H. E. Stanley, Introduction to Phase Transitions and Critical Phenomena,
Oxford University Press, 1993.
[13] V. L. Ginzburg, Sov. Phys.  Solid State 2, 1824, 1960
[14] D. Yu. Ivanov, Doklady Physics 47, 267, 2002
[15] D. Yu. Ivanov, L. A. Makarevich, and O. N. Sokolova, JETP Letters 20,
121, 1974.
[16] W. Wagner, N. Kurzeja, and B. Pieperbeck, Fluid Phase Equilibria 79,
151, 1992.
[17] L. A. Makarevich, O. N. Sokolova, and A. M. Rozen, Sov. Phys.  JETP
40, 305, 1974.
[18] M. Gitterman, J. Stat. Phys. 58, 707, 1990.
[19] J. S. Rowlinson and F. I. Swinton, Liquids and Liquid Mixtures, Butter
worth, London, 1982.
[20] G. M. Schneider, Adv. Chem. Phys. 17, 1, 1970.
[21] A. Nitzan, Phys. Rev. A 17, 1513, 1978.
[22] A. Nitzan, P. Ortoleva, J. Deutch, and J. Ross, J. Chem. Phys. 60, 1056,
1974.
[23] I. Matheson, D. F. Walls, and C. W. Gardiner, J. Stat. Phys. 12, 21, 1975.
127
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
128 Chemistry versus Physics: Chemical Reactions near Critical Points
[24] A. S. Mikhailov, Phys. Lett. A 73, 143, 1979; Z. Phys. B 41, 277, 1981.
[25] W. Horsthemke and R. Levever, Noiseinduced Transitions, Springer,
Berlin, 1984.
[26] D. G. Peck, A. J. Mehta, and K. P. Johnston, J. Phys. Chem. 93, 4297,
1998.
[27] B. Lin and A. Akgerman, Ind. Eng. Chem. Res. 40, 1113, 2001.
[28] H. Li, B. Han, J. Liu,
.
L. Gao, Z. Hou, T. Jiang, Z. Liu, X. Zhang, and J.
He, Chemistry Eur. J. 8, 5593, 2002.
[29] J. Quin, M. T. Timko, A. J. Allen, C. J. Russel, B. Winnik, B. Buckley, J.
I. Steinfeld, and J. W. Tester, J. Am. Chem. Soc. 126, 5465, 2004.
[30] B. Wang, B. Han, T. Jiang, Z. Zhang, Y. Xie, W. Li, and W. Wu, J. Phys.
Chem. B 109, 24203, 2005.
[31] T. Nishikawa, Y. Inoue, M. Sato, Y. Iwai, and Y. Arai, AIChE J. 44, 1706,
1998.
[32] S. Glasstone, K. J. Laidler, and H. Eyring, The Theory of Rate Processes,
1st ed, McGrawHill, New York, 1941.
[33] C. A. Eckert, D. H. Ziger, K. P. Johnston, and S. J. Kim, J. Phys. Chem.
90, 2738, 1986.
[34] C. Muller, A. Steiger, and F. Becker, Thermochim. Acta 151, 131, 1989.
[35] L. Van Hove, Phys. Rev. 95, 1374, 1954.
[36] I. Procaccia and M. Gitterman, Phys. Rev. Lett. 46, 1163, 1981; Phys.
Rev. A 25, 1137, 1982.
[37] A. Z. Patashinskii, V. L. Pokrovskii, and M. V. Feigel’man, Sov. Phys. 
JETP 55, 851, 1982.
[38] R. G. Griﬃths and J. C. Wheeler, Phys. Rev. A 2, 1047, 1970.
[39] M. E. Paulaitis and G. C. Alexander, Pure Appl. Chem. 59, 61, 1987.
[40] S. T. Milner and P. C. Martin, Phys. Rev. A 33, 1996, 1986.
[41] A. Onuki and R. A. Ferrell, Physica A 164, 245, 1989; B. Zappoli, D.
Bailly, Y. Garrabos, R. Le Neindre, P. Guenoun, and D. Beysens, Phys.
Rev. A 41, 2264, 1990; H. I Boukari, J. N. Shaumeyer, M. E. Briggs, and
R. W. Gammon, Phys. Rev. A 41, 2260, 1990.
[42] R. P. Behringer, A. Onuki, and H. Meyer, J. Low Temp. Phys. 81, 71,
1990; H. Meyer and F. Zhong, Compt. Rend. Mecanique 332, 327, 2004.
[43] B. Zappoli and P. Carles, Acta Astron. 38, 39, 1996.
[44] Yu. I. Dakhnovskii, A. A. Ovchinnikov, and V. A. Benderskii, Chem. Phys.
66, 93, 1982.
[45] M. Sato, Y. Ikushima, K. Hatekada, and T. Ikeshoji, Int. J. Chem. Reactor
Eng. 3, A15, 2005.
[46] C. E. Bunker and YaPingSun, J. Am. Chem. Soc. 117, 10865, 1995.
[47] I. Raspo, S. Meradji, and B. Zappoli, Chem. Eng. Sci. 62, 4182, 2007.
[48] S.T. Chung, H.C. Huang, S.J. Pan, W.T. Tsai, P.Y. Lee, C.H. Yang,
and M.B. Wu, Corrosion Science 50, 2614, 2008.
[49] M. Gitterman, Physica A 386, 1, 2007.
[50] M. Gittermam, Physica A 388, 1046, 2009.
[51] M. Green, J. Chem. Phys. 22, 398, 1954.
[52] H. Mori, Progr. Theor. Phys. 34, 399, 1965.
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Bibliography 129
[53] R. Zwanzig, Lecture in Theoretical Physics, vol. 3, Interscience, New York,
1961.
[54] T. Yamamoto, J. Chem. Phys. 33, 281, 1960.
[55] B. J. Berne and R. Pecora, Dynamic Light Scattering, Wiley, NewYork,
1976.
[56] R. M. Velasco and L. S. GarciaColin, Physica 72, 233, 1973.
[57] D. L. Carle, W. G. Laidlaw, and H. N. W. Lekkerkerker, Farad. Soc. Trans.
71, 1448, 1975.
[58] R. D. Mountain and J. M. Deutch, J. Chem. Phys. 50, 1103, 1969.
[59] L. Letamendia, J. P. Vindoula, C. Vaucamps, and G. Nouchi, Phys. Rev.
A 41, 3178, 1990.
[60] C. Allain and P. Lallemand, J. de Chimie Physique 77, 881, 1980.
[61] A. Munster, Statistical Thermodynamics, Springer, 1974.
[62] J. K. Baird and J. C. Clunie, J. Phys. Chem. 102, 6498, 1998.
[63] Y. W. Kim and J. K. Baird, Int. J. Thermophys. 22, 1449, 2001.
[64] C. D. Specker, J. M. Ellis, and J. K. Baird, Int. J. Thermophys. 28, 846,
2007.
[65] Y. W. Kim and J. K. Baird, Int. J. Thermophys. 25, 1025, 2004.
[66] I. Procaccia and M. Gitterman, Phys. Rev. A 27, 555, 1983; ibid. 30, 647,
1984.
[67] J. C. Wheeler and R. G. Petschek, Phys. Rev. A 28, 2442, 1983.
[68] I. R. Krichevskii, Yu. V. Tsekhanskaya, and Z. A. Polyakova, Rus. J. Phys.
Chem. 40, 715, 1966; ibid. 43, 1393, 1969.
[69] G. Morrison, Phys. Rev. A 30, 644, 1984.
[70] M. Gitterman and I. Procaccia, J. Chem. Phys. 78, 2648, 1983; AIChE J.
29, 686, 1983.
[71] I. Procaccia and M. Gitterman, J. Chem. Phys. 78, 5275, 1983.
[72] T. Toroumi, J. Sakai, T. Kawakami, D. Osawa, and M. Azuma, J. Soc. Ind.
Jpn. 49, 1, 1946.
[73] S. C. Greer, Phys. Rev. A 31, 3240, 1985.
[74] J. L. Tveekrem, R. H. Cohn, and S. C. Greer, J. Chem. Phys. 86, 3602,
1987.
[75] R. Narayan and M. J. Antal Jr, J. Am. Chem. Soc. 112, 1927, 1990.
[76] R. B. Snyder and C. A. Eckert, AIChE J. 19, 1126, 1973.
[77] E. S. Gould, Mechanism and Structure of Organic Chemistry, Holt, Rine
hart, and Winston, New York, 1959.
[78] Y. W. Kim and J. K. Baird, J. Phys. Chem. A 107, 8435, 2003; ibid. 109,
4750, 2005.
[79] J. L. Tveekrem and D. T. Jacobs, Phys. Rev. A 27, 2773, 1983.
[80] R. H. Cohn and D. T. Jacobs, J. Chem. Phys. 80, 856, 1984.
[81] D. T. Jacobs, J. Chem. Phys. 91, 560, 1989.
[82] J. Ke, B. Han, M. W. George, H. Yan, and M. Poliakoﬀ, J. Am. Chem.
Soc. 123, 3661, 2001.
[83] B. Hu, R. D. Richey, and J. K. Baird, J. Chem. Eng. Data 54, 1537, 2009.
[84] M. E. Fisher, Phys. Rev. 176, 257, 1968.
[85] M. Gitterman and V. Steinberg, Phys. Rev. A 22, 1287, 1980.
September 14, 2009 16:35 World Scientiﬁc Book  9in x 6in Generalnn
130 Chemistry versus Physics: Chemical Reactions near Critical Points
[86] M. A. Anisimov, A. V. Voronel, and E. E. Gorodetskii, Sov. Phys.  JETP
33, 605, 1971.
[87] J. Thoen, R. Kindt, W. van Dael, M. Merabet, and T. K. Bose, Physica A
156, 92, 1989.
[88] E. U. Frank, in The Physics and Chemistry of Aqueous Ionic Solutions,
eds. M.C. BellissentFuel and G. W. Neilson, Reidel, Dordecht, 1987, p.
337.
[89] A. Stein and G. F. Allen, J. Chem. Phys. 59, 6079, 1973.
[90] C. H. Shaw and W. I. Goldburg, J. Chem. Phys. 65, 4906, 1976.
[91] E. M. Anderson and S. C. Greer, Phys. Rev. A 30, 3129, 1984.
[92] D. Jasnow, W. I. Goldberg, and J. S. Semura, Phys. Rev. A 9, 355, 1974.
[93] M. E. Fisher and J. S. Langer, Phys. Rev. Lett. 20, 665, 1968.
[94] M. Gitterman, Phys. Rev. A 28, 358, 1983.
[95] J. C. Wheeler, Phys. Rev. A 30, 648, 1984.
[96] G. R. Anderson and J. C. Wheeler, J. Chem. Phys. 69, 2082, 3403, 1978;
ibid. 73, 5778, 1980.
[97] J. L. Tveecom, S. C. Greer, and D. T. Jacobs, Macromolecules 21, 147,
1988.
[98] E. Gulari, B. Chu, and D. Woermann, J. Chem. Phys. 73, 2480, 1980;
P. Gansen, T. Janssen, W. Schon, D. Woermann, and H. Schonert, Ber.
Bunsenges. Phys. Chem. 84, 1149, 1980.
[99] K. Ohbayashi and B. Chu, J. Chem. Phys. 68, 5066, 1978.
[100] C. M. Knobler and R. L. Scott, J. Chem. Phys. 76, 2606, 1982.
[101] P. Gansen and D. Woermann, J. Phys. Chem. 88, 2655, 1984.
[102] K. Kawasaki, in Phase Transitions and Critical Phenomena, vol 5a, C.
Domb and M. S. Green, eds., Academic, New York, 1976.
[103] P. C. Hohenberg and B. I. Halpern, Rev. Mod. Phys. 49, 435, 1977.
[104] G. Wilson and J. Kogut, Phys. Rep. C 12, 75, 1974.
[105] Ya. B. Zel’dovich, Zh. Fiz. Khim. 11, 685, 1938 (in Russian).
[106] R. Aris, Arch. Ration. Mech. Anal. 19, 81, 1965.
[107] R. Aris, Arch. Ration. Mech. Anal. 27, 356, 1968.
[108] J. M. Powers and S. Paulucci, Am. J. Phys. 76, 848, 2008.
[109] E. B. Starikov and B. Norden, J. Phys. Chem. B 111, 14431, 2007.
[110] A. L. CornishBowden, J. Biosciences 27, 121, 2002.
[111] M. Gitterman and V. Steinberg, J. Chem. Phys. 69, 2763, 1978; Phys. Rev.
Lett. 35, 1588, 1975.
[112] M. Gitterman and V. Steinberg, Phys. Rev. A 20, 1236, 1979.
[113] W. Ebeling and R. Sandig, Ann. Phys. (Leipzig) 28, 289, 1973.
[114] Y. Albeck and M. Gitterman, Phil. Mag. B 56, 881, 1997.
[115] H. S. Caram and L. E. Scriven, Chem. Ing. Sci. 31, 163, 1976.
[116] G. Othmer, Chem. Ing. Sci. 31, 993, 1976.
[117] L. R. Corrales and J. Wheeler, J. Chem. Phys. 91, 7097, 1989.
[118] V. Talanquer, J. Chem. Phys. 96, 5408, 1992.
[119] D. Borgis and M. Moreau, J. Stat. Phys. 50, 935, 1988
[120] Y. Rabin and M .Gitterman, Phys. Rev. A 29, 1496, 1984.
[121] H. Reiss and M. Shugard, J. Chem. Phys. 65, 5280, 1976.
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Bibliography 131
[122] M. Gitterman, in Nonequilibrium Statistical Mechanics, ed. E. S. Hernan
dez, World Scientiﬁc, 1989, p. 103.
[123] Y. Albeck and M. Gitterman, Phys. Rev. Lett. 60, 588, 1988.
[124] J W. Cahn and J. E. Hillard, J. Chem. Phys. 28, 258, 1958.
[125] M. Motoyama, J. Phys. Soc. Japan 65, 1894, 1996; M. Motoyama and T.
Ohta, ibid. 66, 2715, 1997.
[126] R. Lefever, D. Carati, and N. Hassani, Phys. Rev. Lett. 75, 1674, 1995.
[127] S. C. Glotzer, D. Stauﬀer, and N. Jan, Phys. Rev. Lett. 75, 1675, 1995.
[128] S. C. Glotzer, E. A. DiMarzio, and M. Muthukumar, Phys. Rev. Lett. 74,
2034, 1995.
[129] Y. Huo, X. Jiang, H. Zhang, and Y. Yang, J. Chem. Phys. 118, 9830, 2003.
[130] S. Puri and H. L. Frish, J. Phys. A 27, 6027,1994.
[131] S. C. Glotzer and A. Conigliuo, Phys. Rev. E 50, 4241, 1994.
[132] S. C. Glotzer, D. Stauﬀer, and N. Jan, Phys. Rev. Lett. 72, 4109, 1994.
[133] J. J. Christiansen, K. Elder, and H. C. Fogelby, Phys. Rev. E 54, R2212,
1996.
[134] S. Toxvaerd, Phys. Rev. E 53, 3710, 1996.
[135] D. Carati and R. Lefever, Phys. Rev. E 56, 3127, 1997.
[136] C. Tong and Y. Yang, J. Chem. Phys. 116, 1519, 2002.
[137] B. Liu, C. Tong, and Y. Yang, J. Phys. Chem. B 105, 10091, 2001.
[138] D. Q. He, S. Kwak, and E. B. Nauman, Macromol. Theory Simul. 5, 801,
1996.
[139] N. F. Mott, Rev. Mod. Phys. 40, 677, 1968.
[140] P. Nozieres, Physica B 117, 16, 1983.
[141] L. A. Turkevich and M. H. Cohen, J. Phys. Chem. 88, 3751, 1984.
[142] W. Hefner and F. Hensel, Phys. Rev. Lett. 48, 1026, 1982.
[143] V. Steinberg, A. Voronel, D. Linsky, and U. Schindewolf, Phys. Rev. Lett.
45, 1338, 1980.
[144] U. Schindewolf, J. Phys. Chem. 88, 3820, 1984.
[145] S. Peter, Ber. Bunsenges. Phys. Chem. 88, 875, 1984.
[146] M. D. Palmieri, J. Chem. Edu. 65, A254, 1988; ibid. 66, A141, 1989.
[147] M. A. McHugh and V. J. Krukonis, Supercritical Fluid Extraction, Princi
ples and Practice, Butterworths, Boston, 1986.
[148] L. T. Taylor, Supercritical Fluid Extraction, Wiley, New York, 1996.
[149] G. A. M. Diepen and F. E. C. Scheﬀer, J. Am. Chem. Soc. 70, 4085, 1948.
[150] M. Gitterman, Am. J. Phys. 56, 1000, 1988.
[151] C. A. Van Gunst, F. E. C. Scheﬀer, and G. A. M. Diepen, J. Phys. Chem.
57, 578, 1953.
[152] S. K. Ma, Modern Theory of Critical Phenomena, Benjamin, Reading,
Mass, 1976.
[153] U. van Wasen and G. M. Schneider, J. Phys. Chem. 84, 229, 1980.
[154] M. McHugh and M. E. Paulatis, J. Chem. Eng. Data, 25, 326, 1980.
[155] B. C. Wu, M. T. Klein, and S. I. Sandler, Ind. Eng. Chem. Res. 30, 822,
1991.
[156] R. FernandezPrini and M. L. Japes, Chem. Soc. Rev. 23, 155, 1994.
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
132 Chemistry versus Physics: Chemical Reactions near Critical Points
[157] G. Debenedetti, J. W. Tom, X. Kwauk, and S. D. Yeo, Fluid Phase Equi
libria 83, 311, 1993.
[158] T. M. Doscher and M. ElArabi, Oil Gas J. 80, 144,1982.
[159] R. K. Helling, and J. W. Tester, Envir. Sci. Technol. 22, 1319, 1988.
[160] H. H. Yang and C. A. Eckert, Ind. Eng. Chem. Res. 27, 2009, 1988.
[161] N. A. Collins, P. G, Debenedetti, and S. Sundaresan, AIChE J. 34, 1211,
1988.
[162] W. Eisenbach, P. J. Gottsh, N. Niemann, and K. Zosel, Fluid Phase Equi
libria 10, 315, 1983.
[163] R. P. De Fillipi, V. J. Krukonis, and M. Modell, Environmental Protection
Agency Report No. EPA600/280054, 1980.
[164] C. R. Yonker, R. W. Wright, S. L Frye, and R. D. Smith in Supercritical
Fluids, Squires and Paulaitis, eds, American Chemical Society Symp. Series
329: Washington, 1987, p. 172.
[165] A. E. Mather, Fluid Phase Equilibria 30, 83, 1986.
[166] D. Barbosa and M. F. Doherty, Proc. R. Soc. Lond. A 413, 443, 1987,
[167] Stortenbekar, Z. Physik Chem. 10, 183, 1892.
[168] B. Widom, private communication.
[169] V. A. Rotenberg and G. M. Kuznetsov, Rus. J. Phys. Chem. 48, 2723,
1974.
[170] I. R. Krichevskii, E. E. Sominskaja, and N. L. Nechitallo, Sov. Phys.Dokl.
282, 541, 1985.
[171] I. R. Krichevskii, E. E. Sominskaja, N. L. Nechitallo and G. N. Lukashova,
Sov. Phys.Dokl. 295, 811, 1987.
[172] G. N. Lewis, and M. Rendall, Thermodynamics, McGrawHill, New York,
1961.
[173] A. A. Sobyanin, Sov. Phys.  Uspechi, 29, 570, 1986.
[174] B. C. McEwan, J. Chem. Soc. 123, 2284, 1923.
[175] C. S. Hudson, Z. Phys. Chem. 47, 113, 1904.
[176] R. J. L. Andon and J. D. Cox, J. Chem. Soc. 4601, 1952.
[177] F. Jona and G. Shirane, Ferroelectric Crystals, Pergamon, Oxford, 1962.
[178] J. C. Wheeler and B. Widom, J. Chem. Phys. 52, 5334, 1970.
[179] R. E. Goldstein, J. Chem. Phys. 79, 4439, 1983.
[180] G. A. Larsen and C. M. Sorensen, Phys. Rev. Lett. 54, 343, 1985.
[181] P. K. Khabibullaev and A. A. Saidov, Phase Separation in Soft Matter
Physics, SpringerVerlag, 2003.
[182] A. Einstein, in Physical Acoustics, R. B. Lindsey, ed., Dowden, Hutchison
and Ross, Strasbourg, 1974.
[183] H. Dunker, D. Woermann, and J. K. Bhattacharjee, Ber. Busenges. Phys.
Chem. 87, 591, 1983.
[184] R. A. Ferrell and J. K. Bhattacharjee, Phys. Rev. A 31, 1788, 1985; J. K.
Bhattacharjee and R. A. Ferrell, Phys. Rev. A 24, 1643, 1981.
[185] P. S. LaPlace, Ann. Chim. Phys. 3, 238, 1816.
[186] R. A. Ferrell, N. Menyhard, H. Schmidt, F. Schwabl, and P. Szepfalusy,
Phys. Rev. Lett. 18, 891, 1967.
[187] Y. Harada, Y. Suzuki, and Y. Ishida, J. Phys. Soc. Jpn. 48, 703, 1980.
[188] W. Mayer, S. Hoﬀman, G. Meier, and I. Alig, Phys. Rev. E 55, 3102, 1997.
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Bibliography 133
[189] I. Iwanowski, S. Z. Mirzaev, and U. Kaatze, Phys. Rev. E 73, 061508, 2000.
[190] R. Behrends and U. Kautze, Europhys. Lett. 65, 221, 2004.
[191] G. Sanchez and G. W. Garland, J. Chem. Phys. 79, 3100, 1983.
[192] R. Behrends, T. Telgmann, and U. Kaatze, J. Chem. Phys. 117, 9828,
2002.
[193] R. Behrends, U. Kaatze, and M. Schach, J. Chem. Phys. 119, 7957, 2003.
[194] L. S. GarciaColin and S. M. T. de la Selva, Physica 75, 37, 1974.
[195] H. G. E. Hentschel and I. Procaccia, J. Chem. Phys. 76, 666, 1982
[196] R. D. Mountain, Rev. Mod. Phys. 38, 205, 1966.
[197] L. Blum and Z. W. Zalzburg, J. Chem. Phys. 48, 2292, 1968.
[198] H. N. W. Lekkerkerker and W. G. Laidlaw, Phys. Rev. A 9, 346, 1974.
[199] H. N. W. Lekkerkerker and W. G. Laidlaw, Phys. Chem. Liquids 3, 225,
1972.
[200] J. F. Brennecke, D. L. Tomasko, J. Peshkin, and C. A. Eckert, Ind. Eng.
Chem. Res. 29, 1682, 1990.
[201] C. A. Eckert and C. A. Knutson, Fluid Phase Equilibria 83, 93, 1993.
[202] S. Kim and K. P. Johnston, AIChE J. 33, 1603, 1987.
[203] J. K. Rice, E. D. Niemeyer. R. A. Dunbar, and F. V. Bright, J. Am. Chem.
Soc. 117, 5832, 1995.
[204] E. D. Niemeyer. R. A. Dunbar, and F. V. Bright, Appl. Spectroscopy 51,
1547, 1997.
[205] L. Onsager, J. Am. Chem. Soc. 58, 1486, 1936.
[206] P. A. Geldorf, R. P. H. Rettschnick, and G. J. Hoytnick, Chem. Phys. Lett.
4, 59, 1969.
[207] Y.P. Sun, C. E. Bunker, and N. B. Hamilton, Chem. Phys. Lett. 210, 111,
1993.
[208] O. Kajimoto, M. Futakami, T. Kobayashi, and K. Yamasaki, J. Phys.
Chem. 92, 1347, 1988.
[209] S. Kim and K. P. Johnston, Ind. Eng. Chem. Res. 26, 1206, 1987.
[210] C. R. Yonker and R. D. Smith, J. Phys. Chem. 92, 2374, 1988.
[211] I. B. Petsche and P. G. Debenedetti, J. Chem. Phys. 91, 7075, 1989.
[212] D. Cochran and L. L. Lee, in Supercritical Fluid Science and Technology,
eds. K. P. Johnston and J. M. L. Penninger, ACS Symposium, Ser. 406,
Washington, 1989, Chapter 3.
[213] R. Graham and H. Haken, Z. Phys. 237, 31, 1970.
[214] A. K. Mehrotra, P. R. Bishnoi, and W. V. Svrcek, Can. J. Chem. Eng. 57,
225, 1979.
[215] J. P. Holloway and J. F. Stubbins, Phil. Mag. A 52, 475, 1985.
[216] D. K. Hordstrom and J. L. Munoz, Geochemical Thermodynamics, (second
ed.), Blackwell Scientiﬁc Publications, Boston, 1994.
[217] C. E. Harvier and J. H. Weare, Geochimica Cosmochimica Acta 44, 981,
1980.
[218] H. C. Helgeson, Pure Appl. Chem. 57, 31, 1985.
[219] W. F. Giggenbach, Geochimica Cosmochimica Acta 45, 393, 1981.
[220] J. R. Williams, B. Liu, and A. A. Cliﬀord, Supercritical Fluid Methods and
Protocols, Humana Press, 2000.
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
134 Chemistry versus Physics: Chemical Reactions near Critical Points
[221] L. Phan, H. Brown, J. White, A. Hodgson, and P. C. Jessop, Green Chem.
11, 53, 2009.
[222] G. W. Burton and M. G. Trauber, Ann. Rev. Nutr. 10, 357, 1990.
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
Index
azeotropic point, 100
carbon dioxide, 4
ClapeyronClausius equation, 8
closedloop, 105
complex sound velocity, 112
conductance, 49
consolute point, 3
correlation function, 29
corrosion eﬀect, 26
critical dilute solution, 96
critical phenomena, 1
critical slowingdown, 19
DebyeH¨ uckel approximation, 91
decay of metastable state, 76
degree of dissociation, 60
densityﬁeld variables, 22
dielectric constant, 58
DielsAlder reaction, 15
diﬀusion coeﬃcient, 5
double critical point, 105
ecological theorem, 14
electroneutrality, 60
enthalpyenergy compensation, 68
equilibrium line, 95
extent of reaction, 21
fast chemical reaction, 45
fast diﬀusion, 45
fast heat conductivity, 45
FloryHuggins energy, 87
FokkerPlanck equation, 77
gas phase, 51
Ginzburg criterion, 7
hard and soft transitions, 12
heat production, 17
homogeneous nucleation, 73
inhomogeneous structure, 121
ionization equilibrium, 68
isotope exchange, 62
kinkantikink solution, 74
law of mass action, 8
light scattering, 118
magniﬁcation of singularities, 58
meanﬁeld model, 72
melting point, 101
Menschutkin reaction, 50
metalammonia solutions, 89
metallic vapors, 89
modiﬁcation of critical indices, 55
morphologies of phase separation, 82
Mott transition, 90
multiple solutions, 67
nanocrystalline, 4
noncompressible liquid, 42
135
September 11, 2009 10:7 World Scientiﬁc Book  9in x 6in Generalnn
136 Chemistry versus Physics: Chemical Reactions near Critical Points
nucleation process, 70
Onsager coeﬃcient, 28
patterns domain, 83
phase separation, 67
piston eﬀect, 25
quasisteadystate solutions, 79
rate constants, 30
reactiondiﬀusion equation, 74
scaling theory, 98
second crossover, 7
shift of critical parameters, 53
statistically independent variables, 38
strictly regular solution, 71
supercritical extraction, 47
ternary polymer blends, 87
threecomponent plasma, 89
transitionstate theory, 16
ultrasound attenuation, 111
viscoreactive cross eﬀects, 34
water, 3
CHEMISTRY VERSUS PHYSICS
Chemical Reactions Near Critical Points
This page intentionally left blank
CHEMISTRY VERSUS PHYSICS Chemical Reactions Near Critical Points Moshe Gitterman BarIlan University. Israel World Scientific NEW JERSEY • LONDON • SINGAPORE • BEIJING • SHANGHAI • HONG KONG • TA I P E I • CHENNAI .
electronic or mechanical. including photocopying. This book. Suite 401402. In this case permission to photocopy is not required from the publisher.. Hackensack. London WC2H 9HE British Library CataloguinginPublication Data A catalogue record for this book is available from the British Library. . CHEMISTRY VERSUS PHYSICS Chemical Reactions Near Critical Points Copyright © 2010 by World Scientific Publishing Co. Inc. NJ 07601 UK office: 57 Shelton Street. ISBN13 9789814291200 ISBN10 981429120X Printed in Singapore. 222 Rosewood Drive. MA 01923. Pte. 5 Toh Tuck Link. USA. Pte. Covent Garden. For photocopying of material in this volume. without written permission from the Publisher. Ltd. Danvers. recording or any information storage and retrieval system now known or to be invented.Published by World Scientific Publishing Co. or parts thereof. Singapore 596224 USA office: 27 Warren Street. Ltd. may not be reproduced in any form or by any means. please pay a copying fee through the Copyright Clearance Center. All rights reserved.
the division of forces into “physical” and “chemical” is arbitrary. rather than in the behavior of all thermodynamic quantities. The growth of clusters in a metastable state is an example of the fuzzy distinction between physical and chemical forces. In fact. and weak attractive (physical) forces. In numerical simulations a rather arbitrary decision has to be made whether a given particle belongs to a “chemical” cluster. therefore.” Such an approach allows one to avoid the calculation of entropy provided that one is interested only in the value of the critical temperature. when one considers the “ideal” ternary mixture A.Preface In this book we attempt to trace the connection between chemical reactions and the physical forces of interaction manifested in critical phenomena. called van der Waals forces. It should be remembered. It is convenient [1] to distinguish between strong attractive (chemical) forces leading to the formation of chemical species. The usefulness of a “chemical” approach to physical problems can be seen from the meanﬁeld theory of the phase transition on an Ising lattice of nonstoichiometric AB alloys [2]. The change of the interaction energy for such a transition when both atoms are or are not nearest neighbors determines the “constant of chemical reaction. and the term “ideal” only means that there are no “physical” forces present. The temperature dependence of the longrange and shortrange order parameters is found from the “law of mass action” for the appropriately chosen “chemical reaction.” The latter is the exchange of position of an atom A from one sublattice and an atom B from the second sublattice. The physical and chemical descriptions of matter are intimately related. that the strong chemical bonding interaction between A and B atoms has already been taken into account via the formation of the chemical complex Am Bn . which are determined by the same classical critical indices in v . B and Am Bn .
There are a tremendous number of published articles devoted to this subject. In the BraggWilliams method. one usually considers ﬁrst the “physical” forces in the equation of state. each atom is exposed to the (selfconsistent) average inﬂuence of all other atoms. Such a transition might be of ﬁrst or second order in the case of a phase transition. analogously. The “chemical” method. I shall follow this approach. The steady state of the system is described by ψ 0 (λ) which is the solution of equation F ψ 0 . other than undergraduate courses in general physics and chemistry. and it proved impossible to include many of them in this book of small size. λ = 0. Even though the border between chemical and physical forces is arbitrary. so that it will be useful for a wide range of researchers. in which a given atom with all its neighbors is considered as the basic group. (0. and then the “chemical” forces in the law of mass action based on the nonideal equation of state. while in the case of a chemical reaction — the equation for the rate of reaction.vi Chemistry versus Physics: Chemical Reactions near Critical Points all versions of meanﬁeld theory. and F is a nonlinear functional in ψ. where the typical “physical” process of diﬀusion is considered as a “chemical reaction” in which some amount of substance A passes from volume element a to b while a diﬀerent amount of B passes from b to a. (0. In line with this approach. both physicists and chemists. and. a special case of chapter 15 in [3]. For nonlinear F . I have kept this book as simple as possible. bifurcation may occur. and for some values of λ. at the socalled transition point. In the case of a phase transition. thus avoiding the details of both microscopic and numerical approaches. as well as teachers and students. One can also trace the common features of phase transitions and chemical reactions by analyzing the time evolution of the state variables ψ described by the equation dψ/dt = F [ψ. more than one steady state solution is possible. No preliminary knowledge is assumed. whereas in the BethePeierls method. a pair of adjacent atoms is considered as the basic group. the hard or soft transition for a chemical reaction. I have favored a phenomenological presentation. Eq. The latter example is. in fact.1) where λ is the set of internal and external parameters. λ] . I ask .1) may be the LandauGinzburg equation for the order parameter. gives better results than the BraggWilliams or BethePeierls method. when a system goes from the original steady state to the new steady state.
5. 5. 5. The occurrence of multiple solutions of the law of mass action is described in Sec. Section 4. It is shown that all three types of behavior (slowingdown.1. including the change in the critical parameters (Sec. 2. respectively. Chapter 2 is devoted to the speciﬁc changes in a chemical reaction occurring near the critical point. 2. melting and double critical points are considered in Secs. 3. transport coeﬃcients (Sec. This phenomenon is described in Sec. An account of the supercritical ﬂuids is given in Sec. The hallmark of critical phenomena — the slowing down of all processes — leads also to the slowingdown of the rates of chemical reactions. In Sec. The organization of the book is as follows. Chapter 5 contains a description of chemical reactions near some speciﬁc regions of the phase diagram. speedingup and unchanged) are possible depending on the given experiment. in Chapter 7 we present our conclusions. respectively. we consider another inﬂuence of criticality — the anomalies in chemical equilibria including supercritical extraction. The main experimental methods of analysis of nearcritical ﬂuids — sound propagation and light scattering — are considered in Chapter 6.5. Finally.Preface vii the forgiveness of the authors whose publications were beyond the scope of this book.2. .1–2. and the spinodal decomposition for the latter case are considered in Secs. 2.2 and 4. The former case.3 and 5. The reverse process — inﬂuence of chemistry on critical phenomena — forms the content of Chapter 3.6.4 is devoted to the isotope exchange reaction in nearcritical systems. The appropriate experiments are described in Sec. critical indices (Sec. Chapter 4 deals with the problem of the phase separation in reactive systems. 4. while the vicinities of the azeotrope.4 is devoted to the special case of a dissociation reaction in a ternary mixture.5) and degree of dissociation (Sec. 2. Section 3. while the opposite peculiar phenomenon of speedingup of a chemical reaction near the critical point provides the subject matter for Sec.2). 2. where phase separation takes place through nucleation. 3.4.3. Chapter 1 contains a short review of phase transitions and chemical reactions and their interconnection.2. 3. The eﬀect of pressure and phase transformation on the equilibrium constant and rate of reaction is the subject of Secs. 4.1).3. 3. which is needed to understand the ensuing material.1.3).4. After the Introduction. The mechanism of phase separation depends on whether it starts from a metastable or a nonstable state.
. . . .3 Relaxation time of reactions . . . 2. . . . . 2. . . . . . . . Chemical reactions . . . 2. . . . . . . . . . . . . . . . . . .4. . . . . . . . . . . .Contents Preface 1. . . . . . . . . . . . . . . . . . . . . . . . . . . .2 Dynamics of chemical reactions . . . . . . . Eﬀect of Chemistry on Critical Phenomena ix . . . . . . Critical slowingdown of chemical reactions . . Eﬀect of phase transformations on chemistry . . .1 2. v 1 1 8 12 15 15 17 19 25 26 27 29 32 38 46 48 53 2. . 2. . . . . Criticality and Chemistry 1. . . . . . . .2 2. . . . . . . . . . 2. . Hydrodynamic equations of reactive binary mixture: piston eﬀect . . . Eﬀect of Criticality on Chemical Reaction 2. . . . . Critical anomalies of chemical equilibria . . . Experiment . . . 2. . . . . . . . . .6 3. . . .4.1 1.3 Critical phenomena . . . . . .4. .3 2. . . . . Analogy between critical phenomena and chemical reactions . . . . . . . . . . . . .4 The eﬀect of pressure on the equilibrium constant and rate of reaction .1 Heterogeneous reactions in nearcritical systems . . . the instability of . .4. . . . . . . . . . . . . . . . . . . . . . . . . . . .5 2. . . . . . . . . . . . . .2 1. . . . . . . . . . . . . . .5 Hydrodynamic equations with statistically independent variables .4.4 Hydrodynamic equations of a reactive binary mixtures . . . . .
. Phase equilibrium in reactive binary mixtures quenched into a metastable state . . . Inhomogeneous structure of nearcritical reactive systems 111 111 113 118 121 125 7. . . . . . . . . . . . . . . . . . . . . . 4.2 5. . .5. . . . . . 101 . .4 Ultrasound attenuation in nearcritical reactive mixtures . . . . . . . . . . . . . 3. .1 Modecoupling analysis . . . . . . . . . . . . . . .4 3. . . . . . . . 53 55 60 62 63 63 65 67 67 69 69 72 75 81 89 4. . . . . . . . . . . . . . . . . . . . . . . Sound Propagation and Light Scattering in Chemically Reactive Systems 6.1 5. .2. . . . . . . . 93 . . . . 4. . . . Melting point of reactive binary mixtures . Phase equilibrium in reactive mixtures quenched into an unstable state . . . . . . . . Comments on the Geometry of the Phase Diagram of a Reaction Mixture 5. . . . Singularities of transport coeﬃcients in reactive systems . Thermodynamics of a threecomponent plasma with a dissociative chemical reaction .2 Multiple solutions of the law of mass action . . Singularity in the degree of dissociation near a critical point Isotope exchange reaction in nearcritical systems . . . . 4. 93 . . . . . . . .3 4. . 105 6. . Azeotropic points in reactive manycomponent systems . . . . . Phase Separation in Reactive Systems 4. .2 Thermodynamic analysis of reactive ternary mixtures . . . . .2 3. . . . . . . .3 5. . . Double critical point . . .3 3. . . . . . . .4 5.3 Kinetics of phase separation . .x Chemistry versus Physics: Chemical Reactions near Critical Points 3. .4 Solubility in supercritical ﬂuids .2 Renormalization group methods . . Hydrodynamic analysis of the dispersion relation for sound waves . . . .1 3.1 4. . . . . . . . .2 6. . . . . . . . . . . .3 6. . . . . . . Modiﬁcation of the critical indices . . .2. . . . . Light scattering from reactive systems . . . . . . . 4. 3. . . . . . . .5 Change of critical parameters due to a chemical reaction . . . .2. . . . . . . . . .1 Thermodynamic analysis of reactive binary mixtures . . Conclusions . .5. 100 . . . . .1 6. . . . . . . . . . . . . . . . .
Contents xi Bibliography Index 127 135 .
For instance.p (1. The liquidgas critical point of an onecomponent ﬂuid is determined by the condition [4] ∂p ∂ρ = T ∂2p ∂ρ2 =0 T (1. phase transitions occur in a wide variety of substances. and BoseEinstein condensation occurs at 10−7 K. Therefore.Chapter 1 Criticality and Chemistry 1. phase transitions are a very general phenomenon. and these mathematical singularities lead to many unusual properties of the system which are called “critical phenomena. associated with the basic properties of manybody systems. and T is the temperature.1) where p is the pressure.” We ﬁrst consider the diﬀerent types of the phase transition points (“critical points”) and then we introduce a qualitative method for describing the behavior of various parameters of the system in the vicinity of critical points. ρ is the density. classical ﬂuids and quantum ﬂuids. ∂µ ∂x = T. the liquidgas critical points of binary mixtures are characterized by the vanishing of the ﬁrst and second derivatives of the chemical potential µ with respect to the concentration x. In addition to this enormous temperature range. the paramagneticferromagnetic transition occurs in iron at around 1000 K. Similarly. T. The thermodynamic functions become singular at the phase transition points.2) .2 K. including solids.1 Critical phenomena Phase transitions occur in Nature in a great variety of systems and under a very wide range of conditions. the superﬂuid transition occurs in liquid helium at 2.p ∂2µ ∂x2 1 = 0.
Both critical lines are deﬁned by Eq..µ . Another distinction between onecomponent systems and binary mixtures is that there are two types of critical points in the latter: the above considered liquidgas critical points and liquidliquid critical points.3) The critical conditions for a binary mixture (1.µ (1. where µ1 . the critical points for an ncomponent mixture are determined by the conditions ∂2p ∂p = =0 (1.1) when the chemical potential is kept constant.3) are the same as those for a pure system (1. according to (1..µ 1 n−1 1 n−1 where n − 1 chemical potentials are held constant. Usually a system has no diﬃculty in “translating” the change in concentration into a change in the chemical potential δµ ∼ (∂µ/∂x) δx. This is the simple physical explanation of the slowingdown of diﬀusion near the critical point. However. µ2 and m1 . In addition to the abovementioned thermodynamic peculiarities..2). Eqs.. Nothing happens to the motion of the separate molecules when one approaches the critical point.2) deﬁne the isolated critical point for an onecomponent system. An example is the slowingdown of diﬀusion near the critical points of a binary mixture.2) is evident from the equivalent form of Eq. the excess concentration δx in some part of a system does not produce diﬀusion by itself... near the critical point. x) .4) ∂ρ T. (1.2 Chemistry versus Physics: Chemical Reactions near Critical Points Here µ = µ1 /m1 − µ2 /m2 . ∂µ/∂x is very small. In fact. µ) and µ = µ (T. It is the rate of equalization of the concentration gradients by diﬀusion which is reduced near the critical points. Diﬀerent binary liquid mixtures show either concavedown or concaveup coexistence curves in a temperatureconcentration phase diagram (at .2). and the line of critical points for a binary mixture. Since the states of a onecomponent system and a binary mixture are deﬁned by the equations of state p = p (T.1) and (1.2)..µ ∂2p ∂ρ2 = 0. which can be rewritten as ∂p ∂ρ = T. ρ. T. relaxation processes slow down near the critical points resulting in singularities in the kinetic coeﬃcients. (1..µ . (1. which is the driving force for diﬀusion.µ ∂ρ2 T.1) and (1. whereas two coexisting liquid phases are distinguished by diﬀerent concentrations of the components. Analogously. The close relation between (1. respectively. m2 are the chemical potentials and masses of the two components. and the system becomes indiﬀerent to changes in concentration.
Table 1. Properties of nearcritical water.1–5 (10−6 m2 /sec) Diﬀusion coeﬃcient 10–40 0. modern applications include the important problems of solving the environmental pollution problem and the fabrication of nanocrystalline materials with predictable properties [5]. the maximum temperature below which the liquids are miscible in all proportions is called the lower critical solution temperature (LCST). Their densities are lower than those of liquids. as for a methanolheplan mixture. The use of . The consolute point is an extremum in the phase diagram where the homogeneous liquid mixture ﬁrst begins to separate into two immiscible liquid layers. high diﬀusion and mass transfer coeﬃcients. In spite of its high critical parameters which limits its application. the minimum temperature above which the two liquids are miscible in all proportions is called the upper critical solution temperature (UCST).02–0. Here we will consider chemical reactions occurring between solutes near the critical points of the solvent. safe and environmentally pure solvent. The properties of nearcritical ﬂuids range between those of gases and liquids (see Table 1). For the concavedown diagram. such as the watertriethylamine solution. cheap. Nearcritical ﬂuids combine properties of gases and liquids. but much higher than the densities of gases.07 2 × 10−4 – −6 2 (10 m / sec) 2 × 10−3 In Table 2 we list the critical parameters of the solvents in most common use. for a concaveup diagram.Criticality and Chemistry 3 ﬁxed pressure) or in a pressureconcentration phase diagram (at ﬁxed temperature). By contrast. We will discuss this calculation in the next section when examinating the inﬂuence of chemical reactions on UCST and LCST. such as the full mixing with oxygen and organic compounds. liquids and nearcritical ﬂuids. which makes the nearcritical ﬂuids excellent solvents for a variety of substances. Physical Properties Gas Nearcritical ﬂuid Liquid Density (kg/m3 ) 0. in addition to the traditional uses. Water is the most abundant. Under the assumption of analyticity of the thermodynamic functions at the critical points. one can obtain the general thermodynamic criterion for the existence of UCST and LCST [3].6–2 200–500 600–1000 Kinematic viscosity 5–500 0. Comparison of some physical properties of gases.1 0. make water appropriate for eﬃcient treatment of industrial wastes.
322 Carbon dioxide (CO2 ) 30.9 0. Some new applications include the use of twophase reaction mixtures with high pressure carbon dioxide which are known as “CO2 — expanded ﬂuids”. In fact. The eﬃciency of hydrothermal detoxiﬁcation of pyridine waste is substantially increased by the addition of a small amount of heterogeneous catalyst. Like water.235 Methanol (CH3 OH) 239. CO2 has critical parameters more convenient than water. cleaning of polymers and plastic wastes. [9]. Many references can be found [5] dealing with both nanotechnology and environmental problems.5% of PtAl2 O3 increases the oxidation of pyridine to 99% [7].3 113.4 80.22 Ethane (C2 H6 ) 32. the addition of 0. Other methods include dechlorination of chlorinate organic compounds. nontoxic and environment compatible. Another advantage of CO2 lies in the fact that it does not attack enzymes and is therefore suitable for enzymecatalyzed reactions.47 Sulfur hexaﬂuoride (SF6 ) 45. among others.9 72.2 48.272 Propane (C3 H6 ) 96.6 0. carbon dioxide (CO2 ) has the advantage of being nonﬂammable.6 0.6 41.7 0.20 Pyridine (C5 H5 N) 347 55. Hydrothermal synthesis in nearcritical water is used to obtain nanocrystalline oxide powders with speciﬁed particle sizes and phase composition. and semiconductors. and is. highdensity ceramics. having almost no limitation on the concentration of the pyridinecontaining solutions. At the same time.2 0.9 220.30 Nanocrystallines (particles whose size is a few interatomic distances) are new generation materials widely used as sensors. hydrolysis of cellulose. Table 2. Hundreds of examples can be found in recent reviews [8].73 Ammonia (NH3 ) 132. Solvent Tcr (C) pcr (atm) ρcr (g/mL) Water (H2 O) 373.5 0. Critical parameters of ﬂuids which are commonly used as solvents for chemical reactions.9 0. fuel cells. For instance.31 Benzene (C6 H6 ) 289 48. . the ﬁrst choice for use as a nearcritical solvent. nearcritical carbon dioxide is more frequently used in the laboratory and in technology than any other solvent.3 0.9 0. therefore.4 Chemistry versus Physics: Chemical Reactions near Critical Points nearcritical water for detoxiﬁcation of organic waste using the catalytic oxidation of pyridine was found [6] to be cheaper than other methods and also more eﬀective. and the release of bromine for polymers and plastics.5 36.
The correlation radius ξ. ξ T →TC → ∞. −1 i.5). they move together with a mean velocity v and a mean free path ξ. ξ reaches 10−4 − 10−5 cm near the critical point. (1. satisfying the condition a << ξ << R (1. F ∼ ηξv. As an illustration of the crucial importance of the new characteristic length ξ. (1. The speciﬁc heat at constant pressure Cp and the expansion coeﬃcient β = −ρ−1 (∂ρ/∂T )p also increase near the critical point of a onecomponent system.. . A sharp increase in the mean square ﬂuctuations of the density (or concentration) and of the integral of the correlation function gρρ follows from the wellknown thermodynamic relations (ρ (r) − ρ) ∼ 2 ∂ρ ∂p → ∞.Criticality and Chemistry 5 The inﬁnite increase of the compressibility ρ−1 (∂ρ/∂p)T or (∂x/∂µ)T. as follows from Eq.7) where a is the average distance between particles and R is a characteristic macroscopic length. widely separated particles have to be strongly correlated to cause great changes in density.p as the critical point is approached. which is closely connected with the ﬁrst expression. let us consider the singular part of the transport coeﬃcients near the critical point for a model ﬂuid consisting of spheres with a characteristic radius ξ. T gρρ d3 r ∼ ∂ρ ∂p → ∞. leads to a number of peculiarities in the behavior of a substance near its critical point. and we can assume that under the inﬂuence of an external force. In other words. One ﬁnds the following results [10]: 1. Diﬀusion coeﬃcient [11]. increases sharply near the critical temperature TC . T (1. When an external force F is applied. Thus. Particles inside such spheres are strongly correlated.e.5) The large increase of the correlations between the positions of diﬀerent particles is given by the second expression in (1.6) According to estimates from scattering experiments. the mobility b ≡ v/F ∼ (ηξ) . Using the Einstein relation D = kB T b. the speciﬁc nature of the critical region consists of the appearance of a new characteristic distance ξ. the spheres move according to Stoke’s law.1) and the appropriate thermodynamic relations. which characterizes the distance over which correlations are signiﬁcant. where η is the viscosity.
symmetry and the presence of longrange interactions). which is called a jump singularity. dν = 2 − α. γ = γ .6 Chemistry versus Physics: Chemical Reactions near Critical Points where D is the diﬀusion coeﬃcient and kB is the Boltzmann constant. rather than by the details of the microscopic interactions. η. . q ∼ υn ( 1 − 2 ) . ν ). the commonly accepted symbols α. where a ∼ T −TC x . Although there are general properties which deﬁne the values of the critical indices (dimension d of space. Using the language of ﬂuids. the asymptotic behavior near the onecomponent liquidgas critical point of the speciﬁc heat at constant volume. This result is supported by more rigorous theory and also by experiment. Heat conductivity. a becomes constant at the critical point. Finally. a ∼ ln T − TC . and β (δ − 1) = γ. δ. ν = ν and the following relations exist between critical indices: α + 2β + γ = 2. The usual arguments of molecularkinetic theory give the heat ﬂux q passing through unit area per unit time. Of these indices. one (β) is determined only below the critical temperature and two (δ. and the correlation length. 2. One can ﬁnd the velocity v from the estimate for the diﬀusion coeﬃcient given above: v ∼ D/ξ ∼ 1/ξ 2 η. (2 − η) ν = γ. a exhibits a logarithmic singularity. so that nV = 1. we −1 have D ∼ (ηξ) or Dη ∼ ξ −1 . and ν describe. pressuredensity relation at the critical isotherm. This implies that different isomorphic systems will have the same indices for the appropriate parameters. whereas if x = 0. γ . The remaining three indices can be deﬁned for temperatures above critical (α. compressibility. γ. there are two possibilities. this critical index is given by x = ln(a)/ ln T −TC . a result conﬁrmed by the more rigorous theory and by experiment. the correlation function at the critical temperature. α = α . with the possibility of diﬀerent values of this constant on the two sides of the transition. n is the total number of spheres. β. The qualitative description of the singularity of the quantity a at the critical point is given by the noninteger critical index x. γ. Here. In the other case. these values are deﬁned by the general statistical properties of the manybody system. arising from the temperature diﬀerence T1 − T2 ∼ ξ∇T : 1 − 2 ∼ V Cp ξ∇T . If x = 0. ν) as well as below (α . where V is the total volume of the spheres. according to the universality principle. respectively. q ∼ vCp ξ∇T. According to scaling theory [12]. In one case. Thus. the heat conductivity λ ∼ q/∇T ∼ vCp ξ ∼ Cp /ηξ. order parameter (deviation of the density from its critical value). η) exist only at the critical temperature. and 1 − 2 is the diﬀerence in their energies on two sides of a selected area.
1 Variation of critical exponent for the isothermal compressibility in the immediate vicinity of the critical point for CO2 and SF6 . The experimental points are taken from [16]. obtained by assuming the analyticity of thermodynamic functions at the critical points which allows the expansion of these functions in a power series in the deviation from the critical parameters. the indices have their “classical” meanﬁeld values. [17]. In the approach to the critical point. the singularity in the critical points makes the series expansion inapplicable very close to the critical point.Criticality and Chemistry 7 The values of the critical indices depend on the proximity to the critical point.5).). according to Eq. we expect that the meanﬁeld approximation will be accurate when these ﬂuctuations are small. [14] with permission. . Indeed. etc. An interesting idea has been proposed recently [14] of the existence of the second crossover in the immediate vicinity of the critical point. Springer. logarithmic dependence. (1. Ginzburg proposed [13] that the mean ﬁeld theory is applicable when the ﬂuctuations are small compared to the thermodynamic values. 1. the development of largescale ﬂuctuations is accompanied by a continuous increase of the sus Fig. Reproduced from Ref. giving the crossover from classical to nonclassical behavior. where the nonclassical critical indices regain their classical values. Since the diﬀerence between the results of meanﬁeld theory and the exact theory is due to ﬂuctuations of the order parameter. leading to “nonclassical” values of the critical indices (noninteger powers. The Ginzburg criterion divides the region near the critical point into two parts. However. copyright (2002).
in particular.01%. the number of thermodynamic degrees of freedom of a binary system undergoing a chemical reaction is the same as for a nonreactive. and shear ﬂows. respectively in the temperature. impurities. However.1 the crossover to the classical value of the critical index of the isothermal compressibility in the immediate vicinity of the critical point for CO2 and SF6 . respectively. we shall show that the existence of the chemical reaction may lead to the replacement of UCST by LCST and vice versa [18]. This law results from a relation between the chemical potentials µi of the various components [3]. . This transition occurs in the direction opposite to the Ginzburg criterion direction and deﬁnes the second crossover. The author of [14] states his belief that the analogous second crossover has been seen in other experiments under the inﬂuence of gravity. ±0. shear stresses. For the simplest case of the isomerization reaction. the reaction equation and the law of mass action have the form A1 − A2 = 0 and µ1 = µ2 . Moreover. and TC ±0. this concentration is a function of other thermodynamic variables. classical meanﬁeld behavior must be restored. As a consequence. This problem certainly deserves further investigation. onecomponent system. Chemical reactions inﬂuence all properties of manycomponent systems. Thus.2 Chemical reactions The equilibrium numbers of particles of the diﬀerent substances taking part in a chemical reaction are connected by the law of mass action.8 Chemistry versus Physics: Chemical Reactions near Critical Points ceptibilities of the critical system.02% accuracy. as in a nonreactive mixture. As an example. 1. the equation for chemical equilibrium has the form ν i µi = 0. according to the law of mass action. the ClapeyronClausius equation for a binary mixture is determined by the chemical reaction. we show [14] in Fig. binary mixtures undergoing a chemical reaction are characterized by their concentration. dimensionless pressure and density. Thus. 1. for the reaction ν i Ai = 0 . As an example. turbulence. in addition to the latent heat and the volume diﬀerence between the two phases. Such behavior was found experimentally as early as 1974 [15] by taking p − V − T measurements of very pure SF6 in the immediate vicinity (τ = T −TC ) of the critical point with 2 × 10−4 K. presence of boundaries). where Ai are the chemical symbols of the reagents and ν i are positive or negative integers. surface forces. Therefore. the susceptibility to varies external perturbations (gravitational and Coulomb ﬁelds.
x2 . chem = 2T g2x∆x2 + T n υ 2 (∆x2 ) (ν 1 υ1 + ν 2 υ 2 ) .p (1.9) at constant pressure (temperature) and at constant concentration [3]. υ ≡ (1 − x2 ) υ 1 +x2 υ 2 are the heat of reaction and the volume change of reaction in phase B . an additional restriction to (1. ∆hi = hi − hi . T. and x2 .9) exists in the form of the ν 1 A1 law of mass action ν 1 µ1 + ν 2 µ2 = 0. p. for a reaction ν 2 A2 . Using simple thermodynamic relations. g2x ≡ T. Indeed. B and B . x2 ) . ∆υ 2 dp − ∆h2 dT /T + (1 − x2 ) g2x dx2 − 1 − x2 g2x dx2 = 0 where υ i and hi are the partial molar volume and enthalpy. (1.11) Diﬀerentiating the latter equation along the equilibrium surface. T. The system can be described by four parameters. yields h2x (∆x2 )2 − (2∆x2 /n ) (ν 1 υ 1 + ν 2 υ2 ) υ 2x (∆x2 ) − (2∆x2 /n ) (ν 1 h1 + ν 2 h2 ) 2 −1 2 2 (1. (ν 1 υ 1 + ν 2 υ 2 ) dp − (ν 1 h1 + ν 2 h2 ) dT /T + [ν 1 x2 g2x − ν 2 (1 − x2 ) g2x ] dx2 = 0. one obtains [3] ∆υ 1 dp − ∆h1 dT /T + x2 g2x dx2 − x2 g2x dx2 = 0.8) One can diﬀerentiate the equilibrium conditions (1.9) ∂2g ∂x2 2 .12) yields the slope of the equilibrium line of a twophase reactive binary mixture.9) and (1.10) The partial derivatives of the intensive variables are given by Eq. (1.14) where n ≡ ν 1 x2 − ν 2 (1 − x2 ) and h ≡ (1 − x2 ) h1 + x2 h2 . µ2 (T. (1. p.8) along the equilibrium surface between B and B . p. (1.Criticality and Chemistry 9 Consider a mole of a binary mixture which separates into two phases. there is no need to consider these special sections of the coexistence surface when we deal with a reactive system. x2 ) = µ2 (T. which satisfy the following conditions: µ1 (T.12) Combining Eqs. p. T ∂T ∂x2 ∂p ∂T = chem (1. x2 ) . p. ∆υ i = υ i − υ i . g2x ≡ ∂2g ∂x22 . x2 ) = µ1 (T.13) g2x h2x (∆x2 ) − υ 2 (∆x2 ) (ν 1 h1 + ν 2 h2 ) (ν 1 υ 1 + ν 2 υ 2 )−1 (1. However.p (1.
4h2x.cr ) .15) was used in the last relation in (1.cr + v2 h2.cr v1 υ 1.cr (1. T ∂p ∂T ∼ chem v1 h1. (1.17) If x2 > x2. except the ﬁrst term in the denominators and numerators of Eqs.cr −1 (1. (1.cr can be replaced by the ratio of excess volume VE and excess enthalpy hE at the critical point. (1.15) for a reactive mixture.10 Chemistry versus Physics: Chemical Reactions near Critical Points In the absence of a chemical reaction. Analogously. LCST corresponds to ∂ 2 h/∂x2 cr > 0. . and these terms reduce to the wellknown form [3] T ∂p ∂T = x2 h2x υ 2x 2.cr chem ∼ T g4x (x2 − x2 ) h2x. (1.14 ).19) Equation (1.13) and (1.cr /h2x.19) is negative.14).cr + v2 υ2.cr dTC /dp Tc (v1 υ 1.cr − υ2x. Performing a similar expansion near the single critical point of a reactive binary mixture. then (∂x2 /∂T )p is positive if h2x. The latter can be obtained in the same way as for a nonreactive mixture [3].cr ≡ ∂ 2 h/∂x2 cr is negative. while Eq. These signs deﬁne UCST.cr + v2 h2. The ratio of the second derivatives υ 2x.cr > x2 .16) Equation (1.19).18) −1 ∂T ∂x2 ∼ T g4x (x2 − x2 ) h2x. It follows from (1. one obtains from (1.cr + v2 h2. all terms vanish.cr v1 υ 1.15) ∂T ∂x2 = p 2T g2x∆x2 h2x (∆x2 ) (1. 2 The “classical” expansion near the critical points g2x ≈ 1 g4x (∆x2 ) .14) determines the criterion for UCST and LCST.cr + v2 υ 2.cr v1 h1. 8 yields ∂T ∂x2 = p T g4x (x2 − x2 ) .13) and (1.cr + v2 υ 2.cr 1 − v1 h1.19) that the existence of a chemical reaction may change the type of critical point (UCST to LCST and vice versa) if the last bracket in (1.13) is the generalized form of the ClapeyronClausius equation (1.
Reproduced from Ref. Springer.Criticality and Chemistry 11 Thus.cr TC dp One can ﬁnd [19].2 Typical form of the excess enthalpy hE (or excess volume VE ) for a binary mixture as a function of concentration.20) becomes VE h1.21) hE υ 1. Typical examples of positive VE and hE are shown in Fig.2.20) > 1 or v1 υ 1. [18] with permission. υ). 1. and Eq.cr 1 dTC v1 h1. ( 1. the presence of a chemical reaction will change UCST to LCST and vice versa.cr − υ 2. [20] the form of the critical line near the critical point dTC /dp as well as VE and hE . a chemical reaction will change the nature of the critical point if the following (equivalent) inequalities are satisﬁed: v1 h1. one can see that the points a.p .cr + v2 υ 2.cr hE v1 υ 1. 1.cr and h2.21) should not be satisﬁed for some mixtures. where y ≡ (h.2 give h1.cr and υ 2. Points a and b correspond to the critical molar enthalpy of pure substances.cr − h2. 1.cr + v2 h2. ν 1 = −ν 2 = 1.cr ). For an isomerization reaction. copyright (1990).cr There is no physical reason why criterion (1. Then. υ 1.cr + v2 υ 2.cr VE > 1. (1. Fig. (1.cr + v2 h2.cr (or.cr > 1. Using the deﬁnition of the partial molar quantity y1 = y − x2 (∂y/∂x2 )T. analogously. b in Fig. .
as already mentioned in the Introduction. which describe the rate of chemical reactions. Analogously. As an example. 4. . consider the following chemical reactions [23] A + 2X k1 k2 3X. article [21] is entitled “Chemical instabilities as critical phenomena”). but there is no universal approach to nonequilibrium chemical reactions which are deﬁned by the equation of the reaction rate.12 Chemistry versus Physics: Chemical Reactions near Critical Points 1. so that the concentration of A remains constant. compressibility). internal energy (ﬁrst derivatives of the thermodynamic potential). Both these reactions are reversible with reaction constants ki . One distinguishes between phase transitions of ﬁrst and second orders. and. In secondorder phase transitions. . a discontinuity in the thermodynamic variables such as entropy. (1. thermal expansion. in the theory of instability of nonlinear diﬀerential equations. Firstorder transitions involve a discontinuity in the state of the system. The macroscopic equation for the number of molecules X has the following form dX = −k2 X 3 + k1 AX 2 − k4 X + k3 A.23) dt The solution of Eq. However. These are similar in nature to ﬁrstorder and secondorder phase transitions [22]. A k3 k4 X. which are the second derivatives of the thermodynamic potentials (speciﬁc heat. there is a close analogy between these two phenomena (in fact. (1. are discontinuous. The system is open to interaction with an external reservoir of reactant A.22) These rate equations describe the conversion of the initial reactant A into X by two parallel processes: a simple monomolecular degradation or an autocatalytic trimolecular reaction. these variables change continuously while their derivatives.23) with the initial condition X(0) = X0 is X − X1 X0 + X1 k3 −k2 X − X2 X0 − X2 k1 −k3 X − X3 X0 − X3 k2 −k1 = exp [−k2 (X1 − X2 ) (X2 − X3 ) (X3 − X1 ) t] (1. (1. . The former are described by the wellknown Gibbs technique.3 Analogy between critical phenomena and the instability of chemical reactions Critical phenomena describe the behavior of closed thermodynamic systems whereas chemical reactions occur in the systems open to matter transport from the environment. volume. one distinguishes between hard and soft transitions. as a result. . i = 1.24) .
No ﬁrstprinciple microscopic theory exists for ﬂuctuations in reactive chemical systems. one considers two biological species with densities n1 (t) and n2 (t). X →Y +B (1. which decrease with rates a1 and . Here we bring a fascinating example of the inﬂuence of noise on the chemical reaction [24].24) are Xs = X1 for X0 < X2 . (1. (1. Stability analysis shows that the solution X2 is unstable with respect to small perturbations. Xs = X3 for X0 > X2 .Criticality and Chemistry 13 where X1 .25) with X3 ≥ X2 ≥ X1 . say. As an example of diﬀerent behavior. The rate equation for the number of molecules X has the following form dX = −AX 2 + (AC − 1) X dt or. introducing τ = At and λ = (AC − 1) /A. Xs = X2 for X0 = X2 . whereas the solutions X1 and X3 are stable. but the derivatives are not. consider the chemical reaction A+C+X A + 2X.26) where the second reverse reaction is neglected. Moreover.27) (1. which is illustrated by the socalled ecological model. This establishes the link between ﬁrstorder phase transitions in equilibrium systems and socalled hard transitions in reactive systems.29) dτ The solution of this equation has the typical features of secondorder phase transitions: the soft transition points Xs (λ) are continuous at the transition point λ = 0. In this model. (1.28) dX = −X 2 + λX. ( 1.25).26) that hysteresis may occur as X is varied [23]. One usually uses a phenomenological master equation based on the macroscopic rate equations or a Langevin equation obtained by adding a stochastic term to the rate equations. The details can be found in [22]. The steady state solutions Xs of Eq. the van der Waals equation of the form (1. it follows from Eq. X2 and X3 are the three roots of k2 X 3 − k1 AX 2 + k4 X − k3 A = 0 (1. The foregoing equations have to be generalized to include ﬂuctuations from the steady state. The last two results are typical of equilibrium phenomena which are described near the liquidgas critical point by the classical equation of state.
30) and (1. it has been shown [24] that if both these factors occur together with nottoosmall food growth ﬂuctuations. The birthdeath equations for these species have the following form dn1 dn2 = (b1 m − a1 ) n1 . which occurs when the ratio a1 /b1 is smaller than both a2 /b2 and q/c.30)). (1. . i. This result can be formulated as the “ecological theorem”: the longterm coexistence of two species relying on the same renewable resources is impossible. The question arises regarding which changes can help the weak species to survive. also occurs in systems undergoing chemical reactions [24]. Under these conditions. (1. n1 = qb1 − ca1 b1 d1 (1. (1. (1. However. This result has an important ecological interpretation.30) dt dt The amount of food decreases both naturally (at a rate c). (1. as well as in many other phenomena.31) by q + f (t) with f = 0) or to allow the weak (but not the strong!) species to be mobile (adding the diﬀusive term D∇2 n2 term in the second of Eqs.32) which means that in the long run. In a ﬂuctuating environment.31) have nonzero asymptotic solutions for both n1 and n2 . high mobility gives an evolutionary advantage that makes possible the coexistence of weak and strong species.30). dm = q − cm − d1 n1 − d2 n2 . One can easily see [24] that if one allows the food growth rate q to ﬂuctuate in time (replacing q in Eq. the strong species survives and the weak species becomes extinct. the corrected equations (1.14 Chemistry versus Physics: Chemical Reactions near Critical Points a2 and compete with rates b1 and b2 for the same renewable food m.e. the weak species will ﬁnally become extinct.31) dt Assume now that the species n1 is strong and species n2 is weak. Assume that the amount m of the food increases at rate q.31) are m= a1 . which is called a “noiseinduced phase transition” [25]. = (b2 m − a2 ) n2 .30) and (1. the asymptotic t → ∞ solutions of Eqs. (1. b1 n2 = 0. and according to Eq. The individuals of the mobile weak species survive since they can utilize the food growth rate ﬂuctuations more eﬀectively. The eﬀect outlined above..
we show [30] the dramatic increase of the reaction rate of the CO2 + ethanol mixture as the pressure approaches the critical point for composition. such as CO2 + ethanol or CO2 + hexane. One further example is the decomposition reaction of 2.1. 2. the reaction rate is nearly independent of pressure. whereas outside the critical region. [30]. where a 70–80% increase of the reaction rate near the critical point resulted from a small pressure change. One such example is the tautomeric equilibrium in supercritical 1. There are many experimental data concerning the DielsAlder reaction [29]. From a technological point of view. In Fig. Such a substantial increase of the reaction rate by small variations of pressure has also been found [27] in styrene hydroformylation in supercritical CO2 . the CO2 + ethanol mixture has the 15 . as well as near the critical point of binary mixtures.Chapter 2 Eﬀect of Criticality on Chemical Reaction 2. CO2 and SF6 . one of the most interesting and useful reactions in organic chemistry and often used for the synthesis of sixmembered rings. The discoverers of this reaction were awarded the Nobel Prize in Chemistry in 1950 “for their discovery and development of diene synthesis”.1 The eﬀect of pressure on the equilibrium constant and rate of reaction Dozens of experimental results for diﬀerent chemical reactions (solutes) occurring in diﬀerent nearcritical substances (solvents) have been reported in the chemical literature. In Wikipedia this reaction is considered the “Mona Lisa” of reactions in organic chemistry since it requires very little energy to create very useful cyclohexane rings. The DealsAlder reaction has been studied both near the liquidgas critical point of water.1diﬂuoroethane [26].2isobutyronitrile in mixed CO2 ethanol solvent [28]. where the equilibrium rate constant increased fourfold for a small pressure change of 4 MPa.
such as the solubility parameter. [30] with permission. refractive index and the dielectric constant. one uses transitionstate theory [32]. the detailed inﬂuence on chemical reactions requires a more sophisticated analysis. American Chemical Society.15 K near the critical points. However. The measurement of the reaction rate of an isoprenemaleic anhydride reaction near both the liquidgas and liquidliquid critical points of the hexanenitrobenzene mixture demonstrates [31] anomalous enhancement near the liquidgas critical point. To account for these facts. 2. To interpret the pressure eﬀect on reaction kinetics.16 Chemistry versus Physics: Chemical Reactions near Critical Points Fig. in contrast to a very small change near the liquidliquid critical point. Reproduced from Ref. We now consider the theoretical explanation for the strong inﬂuence on chemical reactions of changes in pressure. it is enough to take into account the huge increase of the compressibility (∂ρ/∂p)T near the liquidgas critical point. copyright (2005). It is not surprising that a small change in the pressure near the liquidgas critical point of a pure solvent causes a large change in densitydependent properties.1 Rate of reaction as a function of pressure for CO2 ethanol mixture at T = 318. advantage that the critical points are located at low pressures and temperatures. according to which the activated complex C is formed .
Thus. whereas Vcr of pure SF6 equals 198 cm3 /gmol.V. is given by ∂ ln K = −∆V ∗ ≡ VC − VA − VB (2. There are a number of requirements for the selection of a suitable chemical reaction: 1. Moreover.2) RT ∂p where ∆V ∗ is the apparent activation volume created from the partial molar volumes Vi of component i. The reaction rate was determined [34] by measuring the heat production w (t) as a function of time with a heat ﬂow calorimeter. When the critical point is approached from above. . the coexistence curve of the binary mixture SF6 CO2 . the chemical reaction itself serves as a probe for obtaining information about critical phenomena. The pressure dependence of the rate constant K.2 Eﬀect of phase transformations on chemistry One form of the interplay between chemical reactions and phase transformations is the change of a slow reaction in a liquid mixture in which the transition from the twophase to the onephase region results from the change of composition during the reaction. expressed as A + B ←→ C → P (2. The large values of the partial molar volumes near the critical points of maximum compressibility are evident from the thermodynamic relation ∂p Vi = υkT n (2. expressed in mole fraction units. the limiting value Vcr of the partial molar volumes at the critical point is pathdependent.Eﬀect of Criticality on Chemical Reaction 17 through the bimolecular reaction. υ is the molar volume of the solvent. 000 cm3 /gmol [33] compared with 5 − 10 cm3 /gmol far from the critical point. and n is the total number of moles of the mixture. Being negative. partial molar volumes dramatically increase upon approaching the critical points. peaking at values of −15. Vcr = −230 cm3 /gmol and along the isothermalisochoric path Vcr = −40 cm3 /gmol. ni is the number of moles of component i. It must start in the twophase region and end in the onephase region or vice versa.1) where A and B are the reactants and P is the product.3) ∂ni T.nj=i where kT is the isothermal compressibility of pure solvent. 2.
The results are shown in Fig. Reproduced from Ref. 2ethanediol with initial mole fractions x0 2 O and x0 F A . The authors of [34] have chosen the esteriﬁcation of 1. 3. where the diﬀerent curves relate to diﬀerent initial amounts of reactants. The minimum heat production is taken as the zero of the time scale so that time t is positive for onephase .18 Chemistry versus Physics: Chemical Reactions near Critical Points Fig.2 Heat ﬂow curves for the reaction of acetic anhydride with 1. 2. 2. 2 ethanediol by acetic anhydride in the presence of a catalytic amount of triﬂuoroacetic acid. It must have a slow reaction rate to yield reliable data for the composition of reactants. 2.2. The reaction enthalpy must be large enough for heat ﬂow calorimetry. copyright (1989). Elsevier. [34] with AC T permission. respectively.
5) as ν ν ν ν r = kf C1 1 · · · Ck k 1 − kb C1 1 · · · Cl l (kf C1 1 · · · Ck k ) ν ν −1 ν ν = kf C1 1 · · · Ck k [1 − exp (−A/RT )] . the curves have negative slope due to the decrease of the reactant concentrations with time. (2.4) the observed rate of reaction r is the diﬀerence between the forward and backward processes. using µi = µ0 +RT ln Ci . and fade away as the phase transition occurs more distant from the consolute point. The jumps are almost symmetric with respect to the phase transition point.5) where kf and kb are the microscopically determined rate coeﬃcients for the forward and backward reactions. theoretical interest in this problem was rekindled in 1981 by our work [36]. 3. Although the general idea of the slowingdown of chemical reactions is wellknown [35].Eﬀect of Criticality on Chemical Reaction 19 regions and negative for twophase regions. and Ci are the concentrations (fugacities) of the ith species in an ideal (nonideal) solution. (2. Far from the phase transition. For the reaction k l ν i Ai = i=1 j=1 ν j Aj . 2. 2. 2Cl. The curves are characterized by the following features: 1. Using the deﬁnition of the chemical potentials of the mole of substance i. ν ν ν ν → − r = − − ← = kf C1 1 · · · Ck k − kb C1 1 · · · Cl l r r (2. and i the fact that [3] kf = exp ν i µ0 − ν j µ0 /RT i j kb i=1 j=1 one can rewrite Eq. and kf /kb = exp ν i µ0 /RT . and. Cl2 2 kf CCl2 −kb CCl . The curves exhibit a 10 − 20% jump in heat production in crossing the frontier between twophase and onephase regions. µi = µ0 + RT ln Ci . i i k l (2.3 Critical slowingdown of chemical reactions Experimental studies of chemical reactions near the critical point have a long history. (2.6) .7) where A is the aﬃnity of reaction. the rate r is deﬁned as r = For chlorine dissociation.
at equilibrium.p (2. for small deviations from equilibrium..11) However. both dξ/dt and A vanish. one needs to recall some general thermodynamic ideas. Another way to obtain Eq.10) has been performed [37]. in general. n)..20 Chemistry versus Physics: Chemical Reactions near Critical Points one gets r = kf CCl2 1 − exp − A RT . . the rate of reaction r is a nonlinear function of A. 2. A = 0. µ1 . p. As a function of these “ﬁeld” variables.9) is to note that at equilibrium. Similarly. temperature and pressure. µ1 . and one gets dξ =L dt ∂A ∂ξ ξ − ξ eq . onephase system is determined by two variables. the free energy G has the following form G = G (T. (2. at ﬁxed temperature and pressure.µn ) = G (T. for a reactive system. These could be the temperature T and the chemical potential of each component µi (i = 1.µn ) + ξ i=1 ν i µi (2. (2. and for small deviations from equilibrium. ξ) in terms of the extent of reaction ξ.10) A detailed analysis of Eq. (2. However.8) Equation (2.. (2. (2. Therefore. say. . the chemical potentials are not independent. onephase system is determined by n + 1 variables. .. p (2. . r = LA = L ∂A ∂ξ ξ − ξ eq + · · · T.. n A≡ i=1 ν i µi = 0.8) implies that. µ1 . According to the law of mass action in equilibrium. The thermodynamic state of a onecomponent. Before discussing this work. where the extent of reaction is equal to ξ eq . T.µn ) . one can restrict ourselves to the ﬁrst term in the expansion of the aﬃnity A (T. the state of an ncomponent. Eq.. A/RT << 1.9) ν ν where L = kf C1 1 · · · Ck k /RT .12) The unconstrained free energy G can be written as n G (T.7) takes the following form. (2.13) .
the condition for diﬀusion stability for given T and p has the form d2 G = i. (2. (2.Eﬀect of Criticality on Chemical Reaction 21 where ξ is a Lagrange multiplier. However. j (2.. for zero determinant of the matrix. q may remain positive). the complete stability analysis.15) vanishes.18) . i.j xi µij xj vanishes only when the determinant of the matrix µij vanishes (however.j dni µij dnj > 0. i. For given T and p. p.14). On the critical hypersurface. the existence of a chemical reaction does not change the stability conditions.j dn0 µij dn0 + 2dn0 µij ν j dξ + ν i µij ν j (dξ)2 > 0. On substituting (2. new zero eigenvalues of µij appear which correspond to the eigenvector Nj satisfying µij Nj = 0.17) The form q ≡ i. (2. the expression (2.. namely.. which means that the rank of µij is equal to n − 1.16) is the GibbsDuhem relation. the stability with respect to diﬀusion is broken before the thermal or mechanical stability.14) n0 ). Let 0 Nj be the eigenvector corresponding to eigenvalue zero. the opposite is not true.16) Equation (2. i j i (2. j (2. requires taking into account all possible changes of ni coming both from n0 and ξ in Eq.14) into (2. The equivalence of the stability conditions with respect to diﬀusion becomes evident when one considers [3] these conditions as resulting from homogeneous and nonhomogeneous perturbations of an initially homogeneous system. i In multicomponent systems. By diﬀerentiating Eq. i (2.13) with respect to µi . 0 µij Nj = 0. and all the eigenvalues of µij but one are positive. i the Equation (2.15). n1 . In terms of the Gibbs free energy G (T. one obtains the particle number densities ni = n0 + ν i ξ.e.e. at given T and p. Therefore. which has the simple physical meaning of invariance with respect to proportional change of all particle numbers. one obtains d2 G = i.15) One can easily show [3] that the matrix µij is semipositive.nn ) . the latter quantity deﬁnes the trajectories of all possible thermodynamic states of a system in ndimensional phase space. the determinant of the matrix µij vanishes.14) implies that for a given system (with initial change in ni is completely determined by the extent ξ of the reaction. µij ≡ ∂µi /∂nj ..
chemical potentials of the components) are the same in all phases coexisting in equilibrium. Therefore. in addition to a binary mixture with a single chemical reaction.18). pressure.9). a binary mixture with a single chemical reaction).j (2. However. entropy.19) According to Eq. the only existing critical point will be where slowingdown occurs. whereas the densities (volume. Grifﬁths and Wheeler [38] divide all thermodynamic variables into two classes: “ﬁelds”and “densities”.16). (2. the chemical instability point will be an isolated point on the critical hypersurface. There are n − 2 constraints in (2.19) for critical slowingdown will be satisﬁed if and only if ν i = αNi + βNi0 (2. and (2. (2. The ﬁelds (temperature. Another attack on the slowingdown problem is the general GriﬃthsWheeler method [38] of estimating the singularity of thermodynamic derivatives at the critical points. slowingdown is determined by the thermodynamic derivative (dA/dξ)T. For an ncomponent system with n − 1 chemical reactions (for example.p . As a result. which. (2. The power of the singularity of the thermodynamic derivatives depends on . determines the slowingdown of chemical reactions. Let us now return [37] to the main point — the comparison of Eqs. equal to the rank of the matrix µij on the critical hypersurface.22 Chemistry versus Physics: Chemical Reactions near Critical Points Among all points on the critical hypersurface. which shows that the condition (2. slowingdown may appear only at an isolated point on the critical hypersurface provided that all the constraints are compatible. For an ncomponent mixture with fewer than n − 1 chemical reactions. and the question arises whether this derivative has the singularity at the critical point. Together with two critical conditions.p .19). they deﬁne n conditions on n + 1 independent variables. which is ﬁxed by the law of mass action. this sum equals (dA/dξ)T.9).15) is violated.20) where α and β are arbitrary constants. Indeed.17). extents of reactions) have diﬀerent values in each coexisting phase. which makes experimental veriﬁcation very diﬃcult. (2. those worthy of notice occur where the condition (2. and also ν i µij ν j = 0. (2. there is one free parameter. according to Eq. slowingdown is also very probable [37] for reactions in multicomponent systems involving the separation of slightlydissolved substances from a solvent.20). Each chemical reaction decreases by unity the dimension of the critical hypersurface. concentrations of chemical components. i. according to Eq.
3 Schematic pressuretemperature and densitypressure diagram for a pure ﬂuid. one has to calculate the number of ﬁxed densities. Cv ≡ T (∂S/∂T )v=vcr (derivative of a “density” with respect to “ﬁeld” with one density held constant). then the derivative of a ﬁeld with respect to a density.24. The latter will include all inert components which are not participating in any chemical reaction. the speciﬁc heat at constant volume. and the derivative (∂p/∂T )υ=υC remains ﬁnite at the critical point. International Union of Pure and Applied Chemistry. and the derivatives will be continuous. [39] with permission. Reproduced from Ref. Fig. then x = 0. (∂ρ/∂p)T =TC .1). However. will approach zero x as T − Tcr  with a “strong” critical index x of order unity. From the GriﬃthsWheeler division of the thermodynamic variables into “ﬁeld” and “density”. By contrast. Finally.11. the critical index x is “weak” (of order 0. diverges with the strong critical index x = 1. with no ﬁxed density variables. If one approaches the critical point along the critical isotherm. if among the ﬁxed variables there is one density variable. the isothermal compressibility. as is shown in Fig.3 for the critical point of a onecomponent ﬂuid [39]. approaching the critical point along the critical isochore results in diﬀerent behavior. When the experimental conditions are such that only ﬁeld variables are ﬁxed. 2. if two or more density variables are ﬁxed. Applying the GriﬃthsWheeler rule to chemical reactions near the critical points. is characterized by the weak critical index x = 0. Neither pressure or temperature changes rapidly upon approaching the critical point. Similar arguments can be used for the . such as (∂A/∂ξ)eq . it follows that the approach to a critical point is pathdependent. copyright (1987). 2. However.Eﬀect of Criticality on Chemical Reaction 23 the type of variables which are kept constant under the given experimental conditions [38].
which we will analyze in detail when we consider the hydrodynamic equations for a reactive system. As emphasized in [40]. the diﬀusion constant −1 . The preceding thermodynamic analysis is not complete. ω) ≡ M x (k.9). ω) (the constituent equation). characteristic time τ ≡ L (∂A/∂ξ)eq but also through diﬀusion and heat conductivity (with characteristic times −1 −1 and λk 2 /ρCp . Eq. λ/ρCp = 0. kH >> kC . ω) . this is of crucial importance for the analysis of slowingdown processes. Indeed. respectively). Only for processes with wavenumbers located in the “window” between kH and kC is the relaxation dominated by chemistry. −2 −2 Let us deﬁne two characteristic lengths by kC = Dτ and kH = −1 −1 λτ /ρCp .1 cm2 /s. ω) = L (k) χ (k) −1 −1 x (k. temperatureentropy.24 Chemistry versus Physics: Chemical Reactions near Critical Points analysis of the thermodynamic variables near the critical points in manycomponent systems. and k > kC . for the following reason. the relaxation rates of the latter processes will become smaller for some small wavenumber k than the ﬁxed k independent rate of the chemical reaction. ω) = χ (k) x (k. diﬀusion dominates the chemical relaxation. (2. the chemical reaction is more rapid than both heat conductivity and diﬀusion. (2. In the ﬁrst interval. Therefore. Only homogeneous changes of the extent of reaction ξ are allowed by Eq. (2. (2. The general meanﬁeld approach. etc. kH < k < kC . heat or diﬀusion modes become important [40]. if a system is large enough. which goes back to [35]. ω) = L (k) X (k. to the density variables x by −1 the susceptibility matrix χ−1 by X (k.21) shows that the relaxation of the extent of reaction ξ arises not only through the chemical processes (with away from the critical point). On the lefthand side of the equation. For typical values of the heat conductivity. and instead of A. As we have seen.22) The matrix M = L (k) χ (k) is the hydrodynamic matrix. the k dependent changes connected with sound. where. However. their conjugate ﬁelds appear (pressuredensity. one must compare the Dk 2 relaxation times or characteristic lengths (inverse wavenumbers) in order to determine which quantity should be considered constant. in turn. is the straightforward generalization of (2. chemical potentialsconcentrations. instead of ξ. whereas in the third interval. leading to iωx (k.9).21) The ﬁeld variables X are related. appears the column matrix of the densitytype variables x. Values of the wavenumbers k fall into three intervals: k < kH . ω) . generally.) iωx (k.
a change of ξ from ξ eq builds up an aﬃnity A = 0 which acts as a driving force to restore the equilibrium. The only diﬀerence is that the chemical perturbation is homogeneous. since the behavior of the thermodynamic derivatives is pathdependent. for more complicated systems than a binary mixture.4 Hydrodynamic equations of reactive binary mixture: piston eﬀect The slowingdown of the dynamic processes near the liquidgas critical points. In equilibrium. That is. considered in the previous section. If the system has m reactants and n products. we have kH = 103 cm−1 and kC = 105 cm−1 . light scattering experiments with k ≈ 104 cm−1 at diﬀerent temperatures in the vicinity of the critical temperature provides evidence for the slowingdown of chemical reactions near the critical points. etc.p ≈ 0. strong critical slowingdown is not likely to occur in the usual experimental situation. The additional factor which has to be taken into account in analyzing an experiment. then there are m+n−2 conserved variables. 2. is the hallmark of critical phe .. All the above considerations hold for a twocomponent mixture with a single chemical reaction.e. Note that the “slowingdown” of the chemical reaction does not imply that the forward or backward reaction is slowed down. a change in ξ does not create a restoring force (i. Usually.19). along the coexistence curve.). when A = 0. in contrast to the inhomogeneous diﬀusion changes. It is the measured rate. and the reaction continues to be balanced even though ξ = ξ eq . where the vanishing of diﬀusion does not mean that the individual particles are slowingdown. In fact. aﬃnity). and not at all for m + n > 3. In other words. However. then the reaction can slowdown strongly for m+n < 3. the physical explanation of the slowingdown of the chemical reaction is the same as the explanation of diffusion near the critical points of nonreactive systems. If all m+n species in the system take part in the reaction. the rate continues to be zero. and for a reaction rate τ −1 = 120 Hz. due to the thermodynamic properties in the critical region. the reaction is balanced and the measured rate is zero. which is aﬀected by criticality. is the path of approach to the critical point (at some ﬁxed parameter(s). Therefore. which is really (∂A/∂ξ)T.Eﬀect of Criticality on Chemical Reaction 25 D = 10−5 cm2 /s. which is the net diﬀerence between the forward and backward reactions. Therefore. weakly for m + n = 3. condition (2. means that the system becomes indiﬀerent to changes in the species concentration.
such as appearance of the soft modes in the vibrational spectrum of solids [44]. There are also other explanations for the speedingup of chemical reactions near critical points based on speciﬁc properties.26 Chemistry versus Physics: Chemical Reactions near Critical Points nomena. and hence. However. Such speedingup manifests itself in the strong corrosion observed in supercritical ﬂuid containers [48]. and afterwards found experimentally [42]. like a piston. After changing the temperature at the bottom or around the cell. of the temperature everywhere in the ﬂuid. and the solidsolvent clustering mechanism [46]. The asymptotic analysis shows the coupling between the piston eﬀect and an interface reaction. and the supercritical hydrodynamics — aﬀect the absorptiondesorption reaction taking place between a solid interface and an active component in a dilute binary supercritical mixture. The slow diﬀusive heat propagation is responsible for the critical slowingdown. which — for some ini . The method of matched asymptotic expansion has been used [43] for the solution of the simpliﬁed system of onedimensional hydrodynamic equations with a van der Waals equation of state and linear mixing rules. This is also called the “piston eﬀect”. where the temperature change at the surface induces acoustic waves which lead to a fast change of the pressure. However. which is immediately evident from the piston eﬀect. since the thermal boundary layer generated near the heated bottom of the cell produces. was predicted theoretically [41]. We shall consider some special examples. leading to the speedingup of the chemical reaction [43]. The usual experimental setup is a ﬂuidﬁlled closed cell.1 Heterogeneous reactions in nearcritical systems The interface reaction that takes place in a supercritical phase may be cited as an illustration of an immediate inﬂuence of the piston eﬀect on the ﬂux of matter at the interface. fast thermal equilibrium is established through the thermoacoustic eﬀect [41]. the pressure on the ﬂuid restricted to a constant volume. 2. Two factors — the sensitivity of solubility to the small changes of pressure. It is precisely this fact which makes it so diﬃcult to obtain reliable equilibrium experimental data near the critical points. as opposed to their slowingdown. induced by the piston eﬀect. the very interesting phenomenon of the speedingup of dynamic processes in critical ﬂuids at constant volume. has been explained in [43]. The speedingup of an interface reaction. reduction of the Hbond strength in water [45]. [47].4. one waits for the establishment of thermal equilibrium inside the cell.
Eﬀect of Criticality on Chemical Reaction 27 tial conditions — leads to a strong increase of the ﬂux of matter and the intensiﬁcation of a reaction. we carry out a phenomenological analysis based on the full hydrodynamic equations in a reactive mixture [49]. The ﬂuid is initially at rest. coupled with the critical behavior of the solute solubility with respect to pressure. The strong dilatability expressed by the large derivative (∂w/∂T )p and through the density variation. is the leading mechanism governing the mass fraction of naphthalene at the isothermal plate [47]. This simple example of a chemical reaction can be easily generalized to more complex reactions. rather than heating one of the plates. The strong compressibility of the supercritical ﬂuid through the derivative (∂w/∂p)T . for numerical simulation one can use the parameters of pure CO2 . 2. and the piston eﬀect that results through the pressure variation. 2. The numerical solution of the full hydrodynamic equations based on the ﬁnite volume approximation has been carried out for a system involving a small amount of naphthalene dissolved in supercritical CO2 located between two inﬁnite solid plates made of pure solid naphthalene and able to absorb or desorb naphthalene from the ﬂuid. The temperature of one of the plates is then gradually increased while the other plate is kept at the initial temperature. The question arises as to the choice of the thermodynamic variables which characterize the binary . and (at the heated plate) of (∂w/∂T )p . The critical behavior of the solute solubility through the large values of (∂w/∂p)T . 3. The results of the analysis show that there are three mechanisms for the increase of the mass fractions of the naphthalene w at the solid interface: 1. We consider an isomerization reaction between two species which is described by the equation A B. used mostly by chemists. used by physicists and which we use here. and a similar phenomenon was observed: the strong decrease of the pressure caused by the piston eﬀect leads to a strong decrease of the mass fraction of naphthalene at the isothermal plate. Due to the small amount of naphthalene in solution. [50]. The piston eﬀect. Similar simulations have been performed with cooling.2 Dynamics of chemical reactions There are two diﬀerent phenomenological ways to describe chemical reactions: the hydrodynamic approach. and the rate equations. In the next section. More detailed analysis has been performed [47].4.
In a real ﬂuid. some general formulae do exist which connect the kinetic coeﬃcients with longwavelength limits of the correlation functions. described by the hydrodynamic equations. Indeed. The “chemical” mode is characterized by the local interactions whereas the pure hydrodynamic modes depend upon spatial gradients. which for an isotropic system can be chosen as div (v). This requires calculating the “renormalized” Onsager coeﬃcient L due to the interaction with the hydrodynamic modes (“mode coupling”). The Onsager coeﬃcient plays a role analogous to that of transport coeﬃcients in hydrodynamics. A onecomponent thermodynamic system is characterized by two parameters.23) The subscript eq means that the thermodynamic derivative is calculated in equilibrium. the chemical potential µ. such as pressure and temperature. where m1 and m2 are the molecular masses of the components. eq (2. In equilibrium statistical mechanics. one has to add one additional parameter. deﬁned [4] as µ = µ1 /m1 − µ2 /m2 . The dynamic behavior of a reactive binary mixture.23). such as diﬀusive.e. which goes back to Gibbs. The quantity −1 is the characteristic time of the chemical reaction deLo (∂A/∂ξ)eq scribed by the phenomenological equation (2. for nonequilibrium states that are not far from equilibrium. Thus. will contain a fourth variable. and both are described by the GreenMoriZwanzig relations [51]–[53]. one has to consider the entire system of hydrodynamic equations. one can choose the mass fraction of one component as the progressive variable ξ of a chemical reaction. and the “chemical” mode interacts with other hydrodynamic modes. is proportional to the aﬃnity A = µ1 ν 1 − µ2 ν 2 of the chemical reaction since the stoichiometric coeﬃcients ν 1 . near equilibrium in the linear approximation. No such systematic procedure exists in nonequilibrium statistical mechanics. As discussed in the previous section. such as the concentration ξ of one of the components. there exists a general procedure for calculating the thermodynamic quantities. In real system. however. However. For a binary mixture. there are couplings between diﬀerent modes. the chemical reaction occurs in some media. viscous and heat conductive modes.. i. which will also describe a chemical reaction. for binary mixtures. one obtains dξ = −L0 A = −L0 dt dA dξ ξ − ξ eq .28 Chemistry versus Physics: Chemical Reactions near Critical Points mixture. and the Onsager . and L0 is the Onsager coeﬃcient. ν 2 are inversely proportional to the molecular masses of the species.
26) Performing the integration over t in (2. t) chem ∂ξ ∂t (−k. one immediately obtains that L = L0 .24) yields L= L2 0 dA dξ 2 eq 1 κB T0 ∞ dt exp −L0 0 dA dξ t eq ξ (k. (2. the renormalized Onsager coeﬃcient L. let us check that in the absence of the hydrodynamic modes. 0) ξ (−k. 0) equals κB T0 / (dA/dξ)eq . 0) . 2. Indeed. reduces to the “bare” Onsager coeﬃcient L0 . 0) . First of all.4. is described by dc1 = γ 21 c2 − γ 12 c1 = γ 21 − (γ 12 + γ 21 ) c1 . (2. which has been used in Eq. (2. 1 L= κB T 0 ∞ dt 0 ∂ξ ∂t (k. 0) ξ (−k. eq (2. The change of the concentration of component A caused by gain and loss to component B. we will consider the inﬂuence of the hydrodynamic modes on the renormalized Onsager coeﬃcient. c1 +c2 = 1.24) Equation (2.Eﬀect of Criticality on Chemical Reaction 29 coeﬃcient L is deﬁned for small wavenumbers k. and taking into account [4] that the equilibrium correlation function ξ (k. This equation contains the characteristic .23) and (2. (2. In the following section.24) provides the connection between the phenomenological Onsager coeﬃcient and the microscopic time correlation function. chem (2. when the aﬃnity A depends only on the concentration ξ rather than on all thermodynamic variables.3 Relaxation time of reactions Consider the isomerization reaction with c1 and c2 being the mass concentrations of substances A and B. and γ ij are the transition rates from component i to component j. the solution of Eq.27).24).27) dt For a closed system the total number of particles is constant. deﬁned in (2.26).25) into (2.25) Substituting (2. as the correlation function of the changes of the concentration (∂ξ/∂t)chem due to the chemical reaction [54].23) has the form ξ − ξ eq = exp −L0 dA dξ t . as requires for a purely chemical process.
0) + γ 12 c1 (k. ci (k. t) chem ∂c ∂t (−k. Then.33) by c1 (−k. s) = ∆ where ∆ = s2 + s (D1 + D2 ) k 2 + γ 21 + γ 12 + k 2 (D1 γ 12 + D2 γ 21 ) + k 4 D1 D2 . (2. ∂c2 /∂t = D2 ∇2 c1 − γ 21 c2 + γ 12 c1 . 1 r= 0 0 c1 c2 ∞ dt 0 ∂c ∂t (k. and their time dependence is described by the following equations [55].31) We use the Fourier and Laplace transforms in space and time.30 Chemistry versus Physics: Chemical Reactions near Critical Points time of the reaction r0 which depends on the rate constants γ 12 and γ 21 at equilibrium. s) cj (−k.24) with a slight change in the constant coeﬃcient.34) Multiplying Eqs.32) The solutions of Eqs. and γ 12 c0 = γ 21 c0 . respectively. 0) and averaging. r0 = At equilibrium. 0) . s) = s + k 2 D1 + γ 21 c1 (k. ∆ s + k 2 D2 + γ 12 c2 (k. 0) .33) (2. chem (2.29) 1 . 0) + γ 21 c2 (k. γ 12 + γ 21 (2. (2.31) are c1 (k. dc1 /dt = 0. t) . (2. for . 0) . 0) and c2 (−k. 1 2 (2. s) = drdt exp (−ikr) exp (−st) ci (r.28) The kinetic processes lead to the renormalization of this characteristic time. c2 (k.30) If a system is open. which is described by Eq. (2. (2. ∂c1 /∂t = D1 ∇2 c1 − γ 12 c1 + γ 21 c2 . the concentrations depend both on chemical transformations and on diﬀusion. (2. one obtains the expressions for the matrix elements ci (k. 0) . the particles are able to enter and leave.
For special systems. 2. as functions of the equilibrium averages ci (k. j = 1. 0) .36) s2 + s γ + + 2k 2 D + k 2 γ + D + k 4 D2 s + k 2 D + γ 21 c1 (k. 0) c1 (−k. 0) . s) c1 (−k. s) c1 (−k. 0) s + k 2 D + γ 21 c1 (k. 0) s + k 2 D + γ 12 c2 (k. (2. Equations (2. s) c2 (−k. 0) . 0) + γ 21 c2 (k. 0) c2 (−k.41) c2 (k. + s (γ 21 + γ 12 + 2k 2 D) + k 2 γ + D + k 4 D2 (2. (2. 0) c1 (−k. 0) = s + k 2 D + γ 12 c2 (k.40) c1 (k. s) c2 (−k.Eﬀect of Criticality on Chemical Reaction 31 i. 0) c1 (−k. 0) = c0 1 s + k 2 D + γ 21 . 0) = c0 2 γ 21 . 0) c2 (−k. 0) . 0) c2 (−k. 0) = c0 2 s2 s + k 2 D + γ 12 . 0) c2 (−k.37) s2 + s γ + + 2k 2 D + k 2 γ + D + k 4 D2 c1 (k. 0) + γ 12 c1 (k.38) = s2 + s γ + + 2k 2 D + k 2 γ + D + k 4 D2 where γ + = γ 21 + γ 12 .38) are simpliﬁed when both components are taken to be ideal gases. s) c1 (−k. 0) cj (−k. 0) cj (−k. ci (k.39) c2 (k. 0) . 0) = c2 (k. 0) = c0 δ ij . 0) + γ 21 c2 (k. + s (γ 21 + γ 12 + 2k 2 D) + k 2 γ + D + k 4 D2 (2.35) = s2 + s γ + + 2k 2 D + k 2 γ + D + k 4 D2 c2 (k.42) . 0) = c0 1 s2 γ 12 . s2 + s (γ 21 + γ 12 + 2k 2 D) + k 2 γ + D + k 4 D2 (2. (2. for which D1 = D2 ≡ D. s) c2 (−k. 0) c1 (−k. s2 + s (γ 21 + γ 12 + 2k 2 D) + k 2 γ + D + k 4 D2 (2. so that j c1 (k.35)–(2. one gets c1 (k. such as structurally similar isomers. (2. 0) + γ 12 c1 (k. s) c1 (−k. s) c2 (−k. Then.
0) 12 + γ 2 c2 (k. 0) + c2 (k. dt (2. one obtains 1 r= 0 0 c1 c2 ∞ ∞ dt 0 0 ds exp (st) {γ 2 c1 (k.42).4 Hydrodynamic equations of a reactive binary mixtures The hydrodynamic equations are the conservation laws of mass. one needs the roots of the denominator. (2. + γ 21 + γ 12 ) (2. 2.47) For D = 0. They are the continuity equation. s) c1 (−k.28). s) c2 (−k. dρ = −div (ρv) . Integrating over s yields γ 2 c0 + γ 2 c0 21 2 r = 0 12 1 c1 c0 (γ 12 + γ 21 ) 2 = γ 2 c0 12 1 ∞ s2 = −k 2 D − γ + .44). (2. this solution reduces to Eq.39)–(2.30) the expression (∂c/dt)chem = γ 21 c2 − γ 12 c1 . 0) 21 − γ 12 γ 21 [ c1 (k. s) c2 (−k. s1 = −k 2 D.43) × 0 dt 0 ds s + k 2 D exp (st) . r= γ 2 c0 + γ 2 c0 12 1 21 2 c0 c0 1 2 ∞ ∞ (2.44) To perform the inverse Laplace transformation on Eq.4.46) which can be rewritten by using Eq.29). momentum. s2 + s γ + + 2k 2 D + k 2 γ + D + k 4 D2 (2. (2. concentration and energy. r= (k 2 D (2. s) c1 (−k. as required. (2.45) exp − k 2 D + γ + t dt 0 + γ 2 c0 21 2 c0 c0 (γ 12 + γ 21 ) (k 2 D + γ 21 + γ 12 ) 1 2 1 . (2. 0) ]} which takes the following form after using Eqs.48) .32 Chemistry versus Physics: Chemical Reactions near Critical Points Substituting into (2.
which include both hydrodynamic and chemical ﬂows.53) kp grad (p) . ξ. (2. Jξ .Eﬀect of Criticality on Chemical Reaction 33 the NavierStokes equation. ∂xj ∂xi 3 kT kp grad (T ) + grad (p) . ρ ∂ dvi =− σ ij .56) dt chem The heat change in the chemical reaction is given by λ= ∆h C (2. (2.51) dt dt Just as ρv in Eq. the constitutive equation takes the form dξ = −L0 A.49)–(2. (2.55) For small deviations from chemical equilibrium. S and the chemical aﬃnity A. (2.T (2. ﬂuxes of mass of one of the components and energy. the terms σ ij .48)–(2. whereas kp /p is a purely thermodynamic quantity deﬁned by (∂A/∂p)T. dt ∂xj (2. The phenomenological forms of these quantities. dξ = −div (Jξ ) .T (2. T p kT grad (ξ) + p. respectively.50) dQ dS = ρT = −div (J ) .48) represents the mass ﬂux.51) describe the stress tensor. v and scalars ρ.ξ kp = .57) . dt and the conservation of energy. The wellknown expressions for hydrodynamic ﬂuxes are [4] (σ ij )hydr = δ ij p + η ∂vj 2 ∂vi + − δ ij div (v) + δ ij ζdiv (v) .54) p The variables appearing in (2.54) have obvious meaning.49) the conservation of one of the components. The coeﬃcient kT deﬁnes the Dufour and Soret crosseﬀects. and J in (2.52) (Jξ )hydr = −Dgrad (ξ) − D (J )hydr = −λgrad (T ) − ρD ∂A ∂ξ (2. are based on the symmetry requirements for vectors J. p (∂A/∂ξ)p. (2.
according to Eq.59).54) are based on symmetry considerations which — in the presence of a chemical reaction — allow the appearance of an additional viscoreactive cross eﬀects. chem (2. (2.60) into (2.58) Since ∆h is proportional to A. T and ξ are the deviations of density.52). The rigorous treatment of the correlation functions. The heat ﬂux caused by the chemical reaction is dQ dt = λL0 A. temperature and the mass fraction of one of the components from their equilibrium values denoted by the subscript zero. and can therefore be neglected for small deviations from equilibrium. yields four linearized hydrodynamic equations.24). (2. (2.64) are the shear viscosity η. deﬁne the renormalized Onsager coeﬃcient L. and (2. requires that allowance be made for the symmetry properties of the variables entering the .59) (2. dξ dt = −L0 A − L1 div (v) chem (2. and ∆h is the change of enthalpy per unit mass of reaction.58) is proportional to A2 . The phenomenological relations (2.34 Chemistry versus Physics: Chemical Reactions near Critical Points where C is the heat capacity. Inserting (2.53).56) and (2. (2.T (2.64) where ρ. the heatconductivity λ. the diﬀusion coeﬃcient D.60) (σ ij )chem = δ ij L1 A. the thermaldiﬀusion coeﬃcient kT . p. and the barodiﬀusion coeﬃcient kp . the λLA term in Eq. which.62) ∂t 3 ∂ξ = D∇2 ξ + D ∂t ρT kT 2 kp ∇ T + ∇2 p − L0 A − L1 div (v) T p ∂A ∂ξ kT ∇2 ξ + p.49)–(2. described by the L1 terms.63) ∂S = λ∇2 T + ρD ∂t kp 2 ∇ p p (2.51). This leads to the following generalization of Eqs.61) η ∂v = −grad (p) + η∇2 v + + ζ grad (div (v)) + ρ0 L1 grad (A) (2. The transport coeﬃcients in Eqs.56) and (2. the bulk viscosity ς. as is assumed when using the hydrodynamic theory. (2. ∂ρ = −ρ0 div (v) ∂t ρ0 (2.52)–(2. pressure.61)– (2.
while the other variables are either even or odd under both symmetry operations. 0) .ρ ξ. The Fourier and Laplace transforms in space and time. T = (∂T /∂ρ)ξ. (2.ξ S + (∂p/∂ξ)S.65) ∂ξ = D∇2 ξ + D ∂t kp L1 ∂ρ kT 2 ∇ T + ∇ 2 p − L0 A + . A = 0. 0) + f6 (k.68) into (2. s) ρ (k.71) . yield the following solution of these equations.ξ S + (∂T /∂ξ)S. s) ξ (k. s) ρ (k. s) S (k. For small deviations from equilibrium.64)–(2. s) = (2. 0) + f3 (k. which gives − 1 ∂2ρ = −∇2 p − 2 ∂t ρ0 4 ∂ρ η + ζ ∇2 + ρ0 L1 ∇2 A 3 ∂t (2. s) ρ (k.S ∂A ∂S S+ ρ. (2. 0) + f2 (k.61).ρ ξ. In equilibrium. 0) . we express A.70) S (k. 0) . T and p in terms of the variables ρ.. S. ξ.62 )–(2.69) ρ (k. s) ξ (k. The quantity (∂ξ/∂t)chem and the variable div (v) are odd functions under time reversal and even functions under parity [36].S ρ + (∂p/∂S)ρ.68) Inserting (2.S ρ + (∂T /∂S)ρ. (2.67) and (2. these arguments are not relevant in the following. respectively. S. ∆ f4 (k.ρ (2. Using Eq. 0) + f8 (k. ξ (k.ξ ∂A ∂ξ ξ. i. ξ. s) ξ (k.Eﬀect of Criticality on Chemical Reaction 35 hydrodynamic equations (2. ∆ (2.64).66) To solve Eqs. However.e. S.61)–(2. s) = (2.63).66).64)–(2. 0) + f5 (k. 0) + f9 (k. s) S (k. (2. since we use simpliﬁed models. T p ρ0 ∂t (2.67) Our analysis is based on the assumption of local equilibrium. ∆ f7 (k.66) yields [56] a quite cumbersome system of three diﬀerential equations in three variables ρ. one can omit div (v) from Eqs. A= ∂A ∂ρ ρ+ ξ. s) S (k. s) = f1 (k. p = (∂p/∂ρ)ξ.
ρ = L0 + Dk 2 ξ.ρ ξ (k. 0) + ∂ρ (k. (2.78) K1.ξ To lowest order in k 2 . and K1.S sL1 . s) = K1.76) The solutions of Eqs. 9) are cumbersome combinations of the thermodynamic variables and transport coeﬃcients. s) + K2. 0) − K1. 0) − L1 ρ0 ρ (k.S K2.ξ = s + k 2 D + L0 ∂A ∂ξ (2.ξ = k 2 ∂p ∂ξ ∂A ∂ρ ∂p ∂ρ + ξ.ρ K2.ξ s − 1 ρ0 4 3η + ζ k 2 − K1. . 0) − (L1 /ρ0 ) ρ (k.S − ξ.ξ L1 ρ (k. and the functions fi (i = 1. where one can set ∂ρ (k.ξ − K2. Performing the FourierLaplace transforms of Eqs. 0) .S (2. one gets K1.74) ρ. ξ.66). ρ.ξ − K2. s) + K1.S s + ρ0 L 1 ρ0 . ∂A ∂ρ . 0) . .ξ ξ (k. s) = s− 1 ρ0 4 η + ζ k 2 ρ (k.ρ ρ (k.ρ ρ (k. (2.S (2. written in the variables ρ and ξ.ρ s − 1 ρ0 4 3η + ζ k 2 ρ (k. 0) /∂t = 0. ρ0 (2.65) and (2.72) are ρ (k.ρ K1.ρ K2.73) + ρ0 L 1 ρ. (2. Eq. ∆.S 4 γ+ζ 3 ∂A ∂ξ ∂ξ ∂ρ .ξ (2.77) ξ (k.ξ ξ (k.77) takes the following form . (2. . .72) K2.75) K2. s) = K2. 3 (2. 0) − K2. s) = ξ (k. 0) /∂t.ρ = s2 + k 2 K1.36 Chemistry versus Physics: Chemical Reactions near Critical Points where the determinant of the coeﬃcients in the original equations.ξ ξ (k.ρ K1. The complicated hydrodynamic equations can be simpliﬁed if one neglects the higher terms in k 2 . which can be found in [56]. 0) ρ 0 K1.
S k 2 + (∂p/∂ξ)ξ.S k2 −1 4 γ+ζ 3 2 + (∂p/∂ρ)ξ.78). 0) s2 + sk 2 /ρ0 4 γ+ζ 3 + (∂p/∂ρ)ξ. 0) − (∂ξ/∂ρ)A. t) A (−k.S (2.81) According to Eq. one gets L = L2 /kB T0 0 ∞ 0 dt A (k. ρ. (2.S k 2 The inverse Laplace transform yields ρ (k.S ( ξ (k. (2.S ξ. t) ρ (−k.S + ρ0 L1 (∂A/∂ρ)ξ.82) .80) Analogously.S × s + L0 (∂A/∂ξ)ρ.S ≈ sρ (k.S 0 2 dt[ ξ (k. 0) s + k 2 /2ρ0 −1 sL1 /ρ0 − L0 (∂A/∂ξ)ρ.S kt . 0) + (∂ξ/∂ρ)A. one gets from Eq. (2. t) = ρ (k. (2. t) = ξ (k. 0) − L1 ρ (k. 0) )]. t) ξ (−k. 0) 2 ∞ 0 = L2 /kB T0 (∂A/∂ξ)ρ. t) ξ (−k.S + ρ0 L1 (∂A/∂ρ)ξ. to lowest order in k 2 . s) = sρ (k. Using ξ and ρ as independent variables. 0) exp − Dk 2 + L0 ρ0 ∂A ∂ξ t .S + ρ0 L1 ∂A/∂ξ ξ. In the case under consideration. ξ (k.24).S ρ (k. 0) exp − ∂p ∂ρ 1 2ρ0 −1 .Eﬀect of Criticality on Chemical Reaction 37 ρ (k. the rate of the chemical reaction is deﬁned by the change of concentration (∂ξ/∂t)chem due to the chemical reaction. this change is given by the term −L0 A in Eq.66). 0) + ρ (k. (2. (2.79) 4 γ + ζ k2 t 3 + L1 ∂A ∂ρ 1/2 × cos ξ. t) ρ (−k.
x T /T0 − αT p/ρ0 . 0) = κT0 ρ2 (∂p/∂ρ)ξ. .84) 2. S as statistically independent variables.S (∂ρ/∂p)ξ. We have chosen the mass fraction x. (2. .38 Chemistry versus Physics: Chemical Reactions near Critical Points The correlators in Eq. div (v) and the variable φ = Cp.83) (4η/3 + ζ) k 2 + 4 (∂p/∂ρ)ξ. . the pressure p. where αT is the isobaric thermal expansion and Cp. . 0) ξ (−k. . . .5 Hydrodynamic equations with statistically independent variables The thermodynamic variables x1 .4. For a manycomponent system. x2 . i For a onecomponent system. . 0) = −κT0 (∂ρ/∂A)ξ. Using wellknown thermodynamic formulae [4].x is the speciﬁc heat at . . one can verify that there are different sets of statistically independent variables. . 0 ξ (k. xn are statistically independent if their crossaverages xi xj vanish for i = j. T or p. 0) = κT0 (∂A/∂ξ)ρ. .S .S . which characterize the state of a system.S + ρ2 (∂ξ/∂ρ)A. one obtains L 1 2 L = L2 (∂A/∂ξ)p.S .S [Dk 2 + L0 (∂A/∂ξ)ρ. x2 . 2 × (∂ρ/∂A)ξ.S −1 −1 (2. δx2 . 0) ρ (−k. one can choose ρ. .85) with (δxi ) = A−1 .S . so that the probability of ﬂuctuations is given by P (δx1 . ξ (k. (2.S 0 (2. since δρδT = δpδS = 0.S ]−1 0 ρ0 + ∂ξ ∂ρ 2 (4η/3 + ζ) ρ0 A. Finally.81) as functions of equilibrium correlators ρ (k. δxn ) = exp − 2 A1 A2 An 2 2 2 (δx1 ) − (δx2 ) − · · · − (δxn ) 2 2 2 (2.82) can easily be found from Eqs.S + ρ0 L1 (∂A/∂ρ)ξ.80) and (2. 0) ρ (−k. there are diﬀerent possibilities for choosing the set of the hydrodynamic variables x1.
88) where the Pij are algebraic functions.Eﬀect of Criticality on Chemical Reaction 39 constant pressure. These equations take the following form [60]. and then extended to reactive mixtures [59]. one has to ﬁnd the roots of det M (k. s) is deﬁned by the hydrodynamic equations. s).T x1 . t) = −M (k. s) is a column vector with n hydrodynamic variables (n = 4 ˆ for the case discussed below). and the form of the n ∗ n matrix M (k. S1 = S − (∂S/∂x)p. Nj (k) Nj (k.90) To perform the integration in (2. div (v) and the reduced entropy S1 = S − (∂S/∂x)p. the set of linearized hydrodynamic equations can be written symbolically in the form d N (k. Note that for a chemically reactive ternary mixture. the pressure p. s) (2.T ∂x + ∂t ∂ρ ∂p x.x T /T0 − αT p/ρ0 . (2. s)] −1 j Pij (k. the pressure p.89) Nj (k) .88).91) .63)). s) Nj (−k) = det M (k. s) = [det M (k. and div (v) and the new variable φ = Cp. t) (2. s) . From Eq.T x. or the temperature T. (2. Another possible set of statistically independent variables is the mass fraction x.x ∂ρ ∂T x.87) have the following form Ni (k. (2. t) N (k. one gets the correlation functions Pij (k. [60]. s) 2 (2.87) dt where N (k. t) Nj (−k) = 2π 2 ∞ −∞ ds exp (st) Pij (k.86) After performing a Fourier transform. it is convenient [57] to use the reduced entropy and the reduced concentration. Solutions of Eqs. Nj (k. s) if all Nj are statistically independent. one can write the complete set of linearized hydrodynamic equations in the variables p. An expression for the timedependent correlation function is obtained by taking the inverse Laplace transform of Eq.T T0 ∂p + ∂t Cp.89). (2. x = x1 − (∂x1 /∂x2 )p.p ∂φ + ρ0 div (v) = 0 ∂t (2. div (v) and the variable φ. det M (k. This choice of hydrodynamic variables has been used for nonreactive mixtures [58] . (2.T x2 .62)– (2. x. (2. s) Nj (k) .90). Neglecting the viscoreactive terms (setting L1 = 0 in Eqs. [61] ∂ρ ∂x p.
A + (∂x/∂T )p.x ) = L0 (∂A/∂x)p.x ) (∂x/∂T )p. and T .x .93) gives ∂x/dt = D∇2 − L0 (∂A/∂x)p. x.T (∂x/∂T )p.x ρ0 T0 (∂φ/∂t) − ρ0 kT (∂A/∂x)p.A T0 αT /ρ0 Cp. one can expand A in terms of x.x p − (T0 /Cp.T x − (∂x/∂p)T.x + T0 αT p/ρ0 Cp.94) The chemical term L0 A appears in Eq.p T0 αT /ρ0 Cp.x + T0 αT p/ρ0 Cp.A (T0 φ/Cp.x ) ∇2 φ + T αT /ρ0 Cp.96) into (2.40 Chemistry versus Physics: Chemical Reactions near Critical Points ∂div (v) = −∇2 p + ∂t ∂x = D∇2 x + D dt 4 η + ς ∇2 div (v) .T T.93). (2.T x − (∂x/∂p)T.x + p (2.T (∂x/∂t) − T0 αT − ρ0 (∂S/∂p)x.A φ .T + (∂A/∂T )x. (2.x (2. (2.93) kp kT αT + p0 ρ0 Cp. ∂A ∂x ∂A ∂p ∂A ∂T A= x+ p.x ) ∇2 − L0 (∂A/∂p)x. Cp.p (2.97) DkT 2 ∇ + (∂A/∂x)p. 3 ∇2 p + D kT 2 ∇ φ − L0 A. (2.95) yields L0 A = L0 (∂A/∂x)p.A T0 L0 /Cp.x into Eq.T (∂p/∂t) = λ (T0 /Cp. Cp.x ∇2 p .95) Substituting T = T0 φ/Cp.96) Inserting (2.92) (2. p.T p+ x. Since the aﬃnity A vanishes at equilibrium.x φ.A p − (∂x/∂T )p.T x + D (kp /p0 + kT αT /ρ0 Cp.
such as k (4η/3 + ζ) kD kλ . s1. For a binary mixture without a chemical reaction (L0 = 0).T ρ0 (∂A/∂x)p.x Equations (2. In many practical cases.100) λ/ρ0 Cp.T kT /T0 D∇2 − L0 (∂A/∂x)p.A p. s1.98) + (∂A/∂T )x.T (∂p/∂t) + (∂A/∂x)p.T + × αT /ρ0 − (∂S/∂p)x.x ) (∂A/∂x)2 (∂x/∂T )p. .x Γ= 1 2ρ0 4 η + ζ + (Cp.x 2 (2. s3.x ) ∇2 + kT /T0 (DkT /Cp.T . setting the determinant of the 4×4 ˆ matrix M (k.T Cp. s) to zero yields two zeros and two sound modes with roots.T 1/2 2 k2 2 − 4Dλ/ρ0 Cp. In the next higherorder approximation. there are two sound modes and two entangled heat conductivitydiﬀusion modes [61]. inserting (2.2 = ±ivs k.x − 1) λ/Cp. (2.p (∂A/∂x)p.x + D + (DkT /T0 Cp.4 = k 2 ± 2 (2.94) yields ∂φ/∂t = (∂A/∂x)p.Eﬀect of Criticality on Chemical Reaction 41 Finally. (2.A /Cv.T λ/ρ0 Cp.x )2 (∂A/∂x)p.T kT /T0 D (kp /p0 + kT αT /ρ0 Cp.2 = ±ivs k − Γk 2 .x ∇2 p.102) . the roots of the determinant of the matrix M (k.T ∇2 φ − (T0 L0 /Cp.97) into (2.T + λT0 αT /ρ0 Cp.101) − (∂ρ/∂x)p. (2.x )2 (∂A/∂x)p.x ) (∂A/∂x)p. (2.x 3vs ρ0 vs To lowest order in these parameters.T x + (λ/ρ0 Cp. (2. .x + D + (DkT /T0 Cp.98) are the complete system of hydrodynamic equations.92).p T0 αT /ρ0 Cp.x ) ∇2 − L0 (∂A/∂p)x. t) and the appropriate correlation functions have been found to lowest order in k [61].x 3 kT (∂ρ/∂T )x. there exist small parameters.T + D (∂p/∂ρ)x.99) vs ρ0 Cp.97) and (2.91).
x + kT /T0 Cp. (2.A (2. p.97) and (2.A τ −1 .T ∂x ∂T τ −1 . p.A (2. c d a21 = kT T0 ∂A ∂x a12 = τ −1 + d p.T . p.x kT T0 ∂A ∂x τ −1 − d p.104) a12 = DkT k 2 /Cp. Therefore.T ∂x ∂T ∂A ∂x Dk 2 2 . deﬁned by τ d = Dk 2 −1 .T 2 a22 = λk 2 /ρ0 Cp. and chemical reaction τ c .x ∂A ∂x kT Cp.98) have the following form ∂x + a11 x + a12 φ = 0 (2. t) is then simpliﬁed.x d Cp.107) .T . The complicated matrix M (k.T (2. p. neglecting ﬂuctuations of pressure (p = 0) [60].A + (kT L0 /T0 ) p.T ∂x ∂T τ −1 . we consider the inﬂuence of a chemical reaction for the simpliﬁed case of a noncompressible liquid (div (v) = 0). but this procedure leads to very cumbersome expressions.42 Chemistry versus Physics: Chemical Reactions near Critical Points One can readily use the hydrodynamic equation including the chemical term (L0 = 0) and perform the analogous calculations.106) p.x ) a21 = DkT k 2 /T0 ∂A ∂x ∂A ∂x p.x − (L0 T0 /Cp.x −1 . τh = λ k2 ρ0 Cp.103) dt ∂φ + a21 x + a22 φ = 0 dt where a11 = Dk 2 + L0 ∂A ∂x .x ∂A ∂x p. a11 = τ −1 + τ −1 . c p. c p. τ c = L0 ∂A ∂x −1 .T kT −1 T0 τ − Cp.x ) ∂A ∂x 2 p.T a22 = τ −1 + h 2 kT T0 Cp. c p.T The matrix elements aij can then be rewritten.x ∂A ∂x ∂x ∂T p. and the remaining hydrodynamic equations (2.105) It is convenient to express these elements by the characteristic times of diﬀusion τ d . heat conductivity τ h .T − (kT L0 /Cp.
82). 0) x (−k. (2. t) φ (−k. 0) x (−k. (2.A x (k.104). (2. 0) . φ (k. one needs the roots s1 and s2 of ∆. 0) φ (−k. 0) φ (−k. t) x (−k. we use now the statistically independent variables x and φ. ∞ dt 0 ∞ 0 ds exp (st) x (k.108) To solve Eqs. t) x (−k. but instead of variables ξ and ρ.A .103)–(2.0) = κB T 0 .x ∂x ∂T φ (k.104) has the following form ∆ = (a11 − s) (a22 − s) − a12 a21 . yields ∞ ∞ dt 0 ∞ 0 ds exp (st) x (k. 0) φ (−k.ξ (∂A/∂x)p. ∞ dt 0 ∞ 0 ds exp (st) φ (k. (2. 0) = − (a12 /s1 s2 ) φ (k. 0) . 0) = − (a21 /s1 s2 ) x (k. (2.104).0) a21 T0 Cp.A a11 T0 Cp. t) A (−k. 0) x (−k. Using Eqs. (2. 0) a12 + p. 0) p.111) . 0) . one obtains L = L2 /κB T0 0 = 1 s1 s2 κ B T 0 τ 2 c + T0 Cp. (2. 0) φ (−k. ∞ dt 0 0 ds exp (st) φ (k. (2.103) and (2.0) = κB Cp. 0) x (−k. 0) = (a22 /s1 s2 ) x (k. (2. 0) = (a11 /s1 s2 ) φ (k. t) φ (−k. The inverse Laplace transforms of the solutions of Eqs. 0) .x ∂x ∂T + a22 p.109).110) Substituting into Eq.Eﬀect of Criticality on Chemical Reaction 43 The determinant ∆ of the 2 × 2 matrix M1 of Eqs.103)–(2.109) The following calculations are analogous to those performed in Eq.x ∂x ∂T ∞ 0 dt A (k.T (2.110) the equilibrium correlators x (k.
one ﬁnally obtains 2 2 ∂A kT T0 ∂x L0 1 + 1+ L= τc τh τ d T0 Cp.A (2. . Such “mode coupling” is expressed by replacing the bare Onsager coeﬃcient L0 by the renormalized coeﬃcient L.T kT T0 ∂A ∂x p.S A.A a12 + p. so that the eﬀective characteristic time of a chemical reaction −1 is [L (∂A/∂ξ)] . (2. which deﬁne the change of the characteristic time −1 of a chemical reaction τ c = [L0 (∂A/∂ξ)] . we consider some special cases. In the general case. viscous.T T0 ∂x .112).114) because of the additional phenomenological viscoreactive coeﬃcient L1 introduced in Eqs.x ∂x ∂T + a22 (∂A/∂x)−1 p.S k + 4ρ0 L1 (∂A/∂ρ)ξ. To avoid this. × 1+ L = L2 0 ∂A ∂ξ 2 4 3η 2 p.x −1 ∂S ∂x + p.A a11 T0 Cp. the product of the roots s1 s2 equals a11 a22 − a12 a21 . (2.T p. such a calculation involves a very cumbersome analysis. (2.114) ξ.44 Chemistry versus Physics: Chemical Reactions near Critical Points one gets L= 1 s1 s2 τ 2 c + ∂x ∂T a21 T0 Cp. p. (2.112) According to Eq. (2.x ∂x p.107). ξ.113) )A.x ∂x ∂T .T kT ∂T A. and using Eq.S Note that the diﬀusive mode appears in Eq.S + ρ2 0 ∂ξ ∂ρ ∂ρ ∂p .108).S + +ζ 4 3η + ζ 2 2 ρ0 (∂ξ/∂ρ)A. diﬀusive and chemical modes appear in Eq.S × ∂ρ ∂A (2. (2. When one neglects the thermal variables.S kT ∂T The goal of our analysis was to ﬁnd the mode coupling corrections to the Onsager coeﬃcient.60).84).S Dk 2 L1 /ρ0 + L0 (∂A/∂ξ)ρ.59)–(2. Substituting the latter into (2.S × 1 τh 1 1 + τd τc + kT τ d τ c Cp.
T × p. ( ρ Fast diﬀusion (Dk 2 >> L = L2 0 × 2 kT T0 Cp.T L0 (dA/dx)p. In the opposite limit.T p.x L = L0 (kT /T0 ) (∂A/∂x)p.T + (kT /T0 ) (∂A/∂x)p. For fast diﬀusion processes ( ρ L = L0 whereas in the opposite limit.119) If Dk 2 >> λ 2 ρ0 Cp . (2.T ): 2 ∂A ∂x 2 1+ p. the presence of a chemical reaction dξ p. (2. (2.T T0 1+ kT . dx 0 Cp.x k 2 >> Dk 2 >> L0 (dA/dx)p.T ≈ Dk 2 ): (2.T ). 1+ T0 kT ∂x ∂T −1 .T ∂x ∂T p.x Dk 2 k 2 >> L0 (2.x k >> L0 2 kT T0 Cp.T >> ρ λ k 2 ≈ Dk 2 ): dx 0 Cp L0 1 + L= 1+ τ h kT τ d Cp. (L0 dA p.Eﬀect of Criticality on Chemical Reaction 45 It is convenient to consider some limiting cases of Eq.T + Dk 2 λ 0 Cp. (L0 dA >> ρ λ .x 0 Cp. dA dx p.120) . (2.118) >> Dk 2 ).117) L0 (dA/dx)p. k 2 ≈ L0 T0 kT dA dx p.116) Fast heat conductivity ( ρ L = L0 k 2 >> L0 dA dx p.A kT λk 2 + L0 ρ0 Cp.A (2.A T0 kT 2 ∂S ∂x p.A 2 λk 2 /ρ0 Cp.x 2 τ h kT τ d T0 Cp.x .115) If heat conductivity processes are faster than diﬀusive processes.x dA dx p. p.113): Fast chemical reaction (L0 dA p.T λ λ ρ0 Cp. L = L0 .x ∂S ∂x kT + T0 ∂A ∂x ∂A ∂x p. L = L0 .x Cp.x ∂A ∂x p.T (1 + (T0 /kT ) (∂x/∂T )p.T ∂x ∂T p.T >> Dk 2 >> ρ λ k 2 ).x k 2 >> Dk 2 ).T 0 Cp does not change the Onsager coeﬃcient.T T0 kT ∂x ∂T p.x .T λ 0 Cp.T L= L2 0 ∂A 2 ∂x p.x .T 1+ ∂A ∂x p.A .T + kT T0 1+ ∂x ∂T p.T L0 (dA/dx)p.A ) (∂S/∂x)p. .
T L = L0 ∂A ∂x p. speedingup.5 Critical anomalies of chemical equilibria In addition to singularities in the rate of reaction.121). Even without analyzing speciﬁc reactive systems. Indeed.114)–(2. (2. peculiar behavior of equilibrium concentrations of reagents takes place near the critical point. one can see from Eqs.46 Chemistry versus Physics: Chemical Reactions near Critical Points dA dx p. 2. the ﬁrst important consideration in the analysis of experimental data is the proximity to the critical point. both energy (due to bond energy) and entropy (due to the number of particles) considerations seem to lead to the same conclusion.p (2. Consider varying the temperature 2B at constant along an equilibrium line of a reactive binary mixture B2 pressure.A ∂A ∂x p.T 1+ + kT T0 .p ∂A ∂ξ dξ = 0 T. While another interpretation of these experiments exists [67]. and improved in [67]. T0 kT ∂x ∂T p.T ∂S ∂x p.114)–(2.121) Our hydrodynamic approach applies to the regime where k −1 is larger than the maximal characteristic size of a system. which increases as the critical point is approached. the opposite has been observed in qualitative experiments performed by Krichevskii and his collaborators [68] who studied the equi2NO2 in a dilute solution of CO2 near the critical librium reaction N2 O4 point of the mixture. Equations (2. Common sense dictates that increasing (decreasing) the temperature should result in increasing (decreasing) the concentration of monomers in all dissociationrecombination reactions. The 2B near the single dissociationrecombination reaction of the form B2 liquidgas critical point of the binary mixture of B2 and B has been considered in [66].122) . some of which change drastically near the critical points. Since the aﬃnity A = µB2 − 2µB vanishes for any equilibrium state. This is the correlation radius. [69]. (2. Therefore.T kT T0 λ ρ0 Cp whereas for Dk 2 >> L0 >> k2 . that these equations contain all three types of behavior are described in [62]–[65] (slowingdown. Quite unexpectedly.121) also contain the thermodynamic derivatives and the kinetic coeﬃcients. unchanged). it does not deny the existence of the anomalies in chemical equilibrium near the critical point. along the equilibrium line one can write dA = ∂A ∂T dT + ξ.
Eﬀect of Criticality on Chemical Reaction 47 where dξ = dnB2 − dnB /2 is the change in the extent of the reaction.p (2. p. depending on whether the critical point is approached at constant critical volume or pressure. equil. Hence.line (∂H/∂ξ)T. they have the dissolving power comparable to a gas. Fluids near critical points exist as a single phase having some of the advantageous properties of both a liquid and a gas. Thus. Let s and f denote the solid phase and the ﬂuid phase. T. (2. The equality of the chemical potentials in the two phases is a necessary condition for the coexistence of these phases. d (∆µ) = ∂µs ∂T dT − p ∂µf ∂T dT − p. We used [70] similar arguments to explain the technologically important phenomenon of “supercritical extraction” — the substantial increase in the ability of nearcritical ﬂuids to dissolve solids. It follows from Eq. for the solid. respectively.p T (∂A/∂ξ)T.126) can be rewritten as ∂x ∂T = p s − ss (∂µf /∂x)T.x ∂µf ∂x dx = 0.p T (2. µs (T. the slope of concentration curves ξ = ξ (T ) is inversely proportional to the same derivative (∂A/∂ξ)T.p has no critical singularity. x) ≡ ∆µ = 0 (2. equil.p (2.126) Equation (2.line exhibits a strong divergence at the critical point which is characterized [40] by critical indices γ or (δ − 1) /δ.124) has been used. respectively.123) where the thermodynamic identity [3] = ξ.p which determines the critical slowingdown. Consider now isobaric changes in temperature along the equilibrium line. . and H and S are the enthalpy and entropy. and great sensitivity to small changes in pressure aﬀorded by high compressibility.p ∂S ∂ξ = T.127) where we introduce the molar entropy s of the solid ss = − (∂µs /∂T )p and the partial entropy of the solid in the ﬂuid phase s = − ∂µf /∂T p. The derivative (∂H/∂ξ)T.125) where we used the fact that the concentration x of the solid phase is equal to unity.x . transport properties intermediate between gas and liquid.p (2. Therefore.122) that dξ dT ∂A ∂T =− p. Indeed. p) − µf (T.p A + (∂H/∂ξ)T. the derivative (dξ/dT )p.
We have suggested [71] the use for supercritical extraction in the neighborhood of a consolute point rather than at the liquidgas critical point. Mention should be made of a series of qualitative experiments performed by the prominent Russian physical chemist Krichevskii and his collaborators [68]. 2. The system described above contains a small amount of solid dissolved in a ﬂuid. Using the static dielectric constant as the measure of the extent of dimerization.48 Chemistry versus Physics: Chemical Reactions near Critical Points Equations (2. Under usual conditions. A few interesting experimental results were lately reported by S. the recombination rate is dramatically reduced by one or two orders of magnitude. The former is characterized by much lower pressures and temperatures. which has clear technological advantages. rapid recombination restores chemical equilibrium. They found that after a thermal perturbation. this problem came to the attention of the theorists. which are analogous to (2. However. In two experiments. the irradiation results in the dissociation of the diatomic molecules into atoms. near the critical points. contrary to the results obtained in [68]. including myself. explain supercritical extraction. they found 4% increase (and not decrease!) in dissociation near the critical point. When the interpretation of these experiments led to some doubts [69].127). In both cases. They measured the rate of recombination of chlorine atoms after thermal and chemical dissociation of chlorine near its liquidvapor critical point [73].122)–(2.126)–(2. Additional experiments have been performed [74] near the liquidliquid critical point of a binary mixture with the aim of ﬁnding the change in the relative amounts of NO2 and N2 O4 during the dimerization of NO2 . One may use the binary mixture as the solvent and the solid as the solute. leading to a strong divergence of the slope of the solubility curve.123).6 Experiment Experiments investigating the interplay between criticality and chemistry were performed long before the theorists showed any interest in this problem. Indeed. when the irradiation stops. The ﬁrst publication of the oxidation reaction in the critical region goes back to 1946 [72]. Greer and collaborators. when the experiment is conducted under critical conditions (near the liquidgas critical point of pure Cl2 or of the binary CO2 I2 mixture).127) approaches zero. the recombination was very slow. they irradiated with light pure Cl2 and I2 in CO2 . whereas the recombination after the photolysis with ultraviolet light was very fast. . the denominator in (2.
the rate of reaction showed slowingdown. Another improvement over the original experiments was that visual detection of critical opalescence to determine the critical temperature was .2 K of the critical temperature of the solvent due to the presence of the reactant) yields diﬀerent results. This result was obtained more than twenty years before the theoretical idea of the “piston eﬀect”. later experiment (system was wellstirred in the twophase region. All these reactions produce a strong electrolyte. the latter mixture has a lower critical solution temperature. and the progress of the reaction was followed by measuring the conductance in the vicinity of the lower and upper critical consolute points. analogous to that performed in [63]. they found “speedingup” (increase by some 40%) of the reaction rate of the isoprene with maleic anhydride in the vicinity of the upper critical solution point of two liquidliquid mixtures (hexanenitrobenzene and hexanenitroethane). the experiment was carried out on 2bromo2methylpropane in triethylaminewater. Special attention should be paid to the Snyder and Eckert article: “Chemical kinetics at a critical point” [76]. while below TC . they described the measurement of the rate of ﬁve hydrolysis ﬁrstorder reactions near the consolute point of three binary mixtures (tertamyl chloride and tertbutyl chloride in isobutyric acidwater. In their 1998 article [62]. In spite of this diﬀerence. To check whether the speedingup of the reaction rate is connected with the coexistence of two immiscible phases. 3chloro3methylpentane and 4methylbenzyl bromide in 2butoxyethanolwater. The more careful. The measurement of 2chloro2methylbutane near the critical point of the mixture isobutyric acidwater. and account was taken of the small shift ∆Tc = 0. which means that the “speedingup” eﬀect is associated with the critical point and not with the number of phases in the system. in the twophase region. Extensive studies of the rate of chemical reactions in the vicinity of the critical point were carry out by J. has been repeated [63]. K. the results of the experiments were the same: slowingdown above the critical temperature and speedingup below.Eﬀect of Criticality on Chemical Reaction 49 A review of early development in this ﬁeld can be found in [75]. where. and authors proposed an explanation related to the dilute solution of the reactants. in agreement with previous experiments. in contrast to the wellknown “slowingdown”. All measurements showed a slowingdown eﬀect within a few tenths of a degree on either side of the consolute temperature. Above TC . In contrast to isobutyric acidwater used in [63]. Baird and collaborators. the rate of reaction increased upon approaching the critical point (“speedingup” predicted by the “piston eﬀect”). tertbutyl bromide in triethylaminewater).
In turn. i. For T > TC . for two or more ﬁxed densities. concentrations of the chemical components. then one ﬁnds a “weak” singularity (x = 0. as seen from the following example [63]. For densities not held ﬁxed..e. which is known to exhibit [77] secondorder kinetics. ∼ (T − TC ) /TC  .33. (∂∆G/∂ξ)a. and not of ﬁrstorder kinetics? To answer this question.24 or x = 4. but only future experiments will deﬁnitively answer this question. the number of ﬁxed densities. An important comment has been emphasized in [62].50 Chemistry versus Physics: Chemical Reactions near Critical Points replaced by the association of the critical temperature with the temperature of the onset of the sharp change in electrical conductivity when critical opalescence occurs [64]. a few measured points closest to TC gave a hint of speedingup. thereby changing the . the singularity is “strong” (x = 1. [63] regarding the quantitative description of the slowingdown and speedingup of the reaction rate. the experiment was performed [65] with the saponiﬁcation of ethyl acetate by sodium hydroxide near the consolute point of the 2butoxyethanolwater. those which do not participate in at least one chemical reaction. Finally.1. The experiment [65] shows slowingdown for temperatures below TC .. As described in section x 2. might nevertheless react with the alcohols present in the system. The rate of reaction is proportional to thermodynamic derivative of the Gibbs free energy diﬀerence ∆G separating products from reactance with respect to the extent of reaction ξ.b.. In particular. One of the components of a solvent (isobutyric acid). while above TC . if one density is ﬁxed. The next question investigated concerned the generality of the eﬀect found. the results of the experiment are not clear: one cannot a rule out speedingup.11 or x = 0.9). say. depends on the number of reactions occurring in the system. Estimating the number of reactions is not trivial. volume. For temperatures below TC . at the critical point depends on number of ﬁxed “density type” variables a. x = 0 and there is no singularity of the reaction rate at the critical point. Namely. they could clearly see the critical slowingdown of the reaction rate showing the weak singularity.. depending on the thermodynamical path approaching the critical point — ﬁxed pressure or temperature or along the coexistence curve). which can be compared with the old measurements [76] of the Menshutkin reaction on ethyl iodide. extent of reaction). which does not participate in the hydrolysis reaction. the singularity x of this quantity. of secondorder. b (entropy. The authors measured the rate of the Menschutkin reaction between benzyl bromide and triethylamine near the consolute point of triethylaminewater. does the eﬀect exist for other types of reactions. These results agree with those obtained earlier [76].
3) ethyl diazoacetate in the acid catalyzed decomposition of ethyl diazoacetate in isobutyric acidwater. the same experimental group used another method to analyzing the progress of a chemical reaction [78]. such as the gas phase. In addition to conductance measurements. the critical point did not eﬀect the rates of the reactions. 2) benzene diazonium tetraﬂuoroborate in the decomposition of benzene diazonium tetraﬂuoroborate in 2butoxyethanolwater. .Eﬀect of Criticality on Chemical Reaction 51 behavior of the reaction rate by changing the number of reactions and the number of densities held ﬁxed. namely. the measurement of the rate of appearance of a phase that is not completely miscible with a liquid mixture. In all cases. Three reactions have been studied: 1) acetone dicarboxylic acid in aniline catalyzed decomposition of acetone dicarboxylic acid in isobutyric acidwater.
The presence of a reactant. which have quite diﬀerent solubilities in the two components of the methanolcyclohexane mixture. There is only limited knowledge of the relationship between the shifts of the critical parameters when the perturbation is applied. These data have been obtained by isobaric measurements upon the addition of impurities. Alternatively. pressure or electric ﬁeld changes the critical parameters of a binary mixture. 3. The simplest. For example. Another method is the acoustic technique. There are diﬀerent experimental ways of ﬁnding the critical parameters TC and pC of a system. as well as impurities. ∆TC /TC. induce equal relative shifts in critical temperature and composition. For ionic conductors.1 Change of critical parameters due to a chemical reaction Chemical reactions near the critical points are routinely studied near the consolute points of a binary mixture which plays the role of a solvent. Experiments show that the critical temperature TC 53 . but less precise method is based on visual detection of critical opalescence to determine the critical temperature. using the fact that the velocity of sound reaches a minimum at the critical point [82]. However. it is known [3] that an impurity will cause a critical temperature shift equal to that of the critical composition only when the solubilities in the two components are roughly equal. the critical temperature is identiﬁed by the sharp change in electrical conductivity [64].1). The same relation has been observed [81] in a wide range of systems (Fig. A change in the slope of this curve occurs at the critical temperature [62].Chapter 3 Eﬀect of Chemistry on Critical Phenomena 3. it was found that two diﬀerent impurities (water [79] and acetone [80]). one can use the dilatometric method of measuring the height of the liquid in the capillary sidearm as a function of temperature.0 = ∆xC /xC.0 .
Measurements of the shift of the critical temperature in the presence of a chemical reaction have recently been performed for indium oxide dissolved in the nearcritical isobutyric acidwater mixture [83].54 Chemistry versus Physics: Chemical Reactions near Critical Points Fig. There are many experimental results related to diﬀerent manycomponent mixtures [82].1) The sign of the constant a depends on whether the chemical reaction enhances or inhibits the mutual solubility of the original solvent pair. Reproduced from Ref. and vice versa for the latter case. is the linear function of the initial concentration x of a reactant.24 K . The diﬀerent symbols refer to ﬁve diﬀerent systems (for details. see [81]). it was found [64] that a = 36. and a > 0 for LCST. copyright (1989). a < 0 for UCST. American Institute of Physics.0 of diﬀerent systems to new values TC and xC . 0 TC = TC + ax.0 and xC.1 The shift of critical points TC. (3. . In the former case. 3. [81] with permission. For the equilibrium mixture of benzyl bromide in triethylamine0 water TC = 291.9 K/mole.
and in order to obtain experimentally observable quantities which correspond to constant concentration x. for example. As an example. which is deﬁned by the stability conditions.µ1 . the isomerization reaction). As a result. i. one gets for the extent of reaction A = ν 1 µ1 + ν 2 µ2 = 0.. Therefore. The condition for diﬀusion stability in a binary mixture has the form (∂µ/∂x)T. where n − 1 chemical potentials are held constant. The singularities of the thermodynamic and kinetic quantities near the critical points are determined by the critical indices. the critical behavior of a binary mixture. the critical indices above and below a critical point are multiplied by ± (1 − α)−1 [84].. we describe the changes in critical phenomena arising from the presence of chemical reactions. The minus sign refers to the speciﬁc heat at constant volume and the plus sign refers to all other critical indices. On the other hand. in turn. to ensure the constancy of µ (one has to vary the concentrations during the course of the experiment). The calculations of these indices assumes the constancy of some intensive variables. p remains practically unchanged near the liquidliquid critical point since the experiment is carried out in the presence of saturated vapor. namely. mechanical stability is determined by (∂p/∂v)T < 0. In order to compare theoretical calculations for an ideal onecomponent system with experimental data obtained for “real” systems. pressure p or chemical potentials µ. which. the stability condition for a ncomponent mixture is determined by the condition (∂p/∂v)T.µn−1 < 0. the critical parameters now depend on the variables µi . Multicomponent systems behave in an analogous manner. which. where α is the critical index of the speciﬁc heat at constant volume.2 Modiﬁcation of the critical indices In this section. However. reduces by . one has to pass from T (µ) to T (x).p > 0. For a pure substance.µ < 0. can be rewritten as (∂p/∂v)T. Similarly. will be the same as that of a pure substance. when one goes from a pure substance to a binary mixture.Eﬀect of Chemistry on Critical Phenomena 55 3. We have discussed [85] the renormalization of the critical indices associated with the presence of chemical reactions.. say. by a simple thermodynamic transformation.e. let us again consider a binary ﬂuid mixture A1 − A2 with the single reaction ν 1 A1 + ν 2 A2 = 0 (say. Fisher [84] developed the theory of renormalization of critical indices. it is quite diﬃcult to ensure the constancy of p near the liquidgas critical point where a ﬂuid is highly compressible. whereas in practice it is impossible. This equation means that the law of mass action is satisﬁed. at constant chemical potential µ. The main idea of renormalization can be explained by the following simple argument.
∼ Cv. Therefore. For example. As an example.4) where h is the heat of reaction.2) and (3. the speciﬁc heat at constant volume has a weak singularity at the critical point when a chemical reaction occurs. The occurrence of a chemical reaction in the system under consideration modiﬁes the critical indices of the observable speciﬁc heats at constant volume and constant pressure (or inverse velocity of sound) compared to a system with a frozen chemical reaction. It is interesting to compare (3.A=0 ∼ T.3) becomes obvious from the thermodynamic relation [3] Cp. but a ﬁnite. cusped behavior in the absence of a chemical reaction.ξ The correspondence between (3.56 Chemistry versus Physics: Chemical Reactions near Critical Points one the number of thermodynamic degrees of freedom in a binary mixture.ξ ∼ ∂υ ∂p ∂υ ∂p ∼ T.ξ T −TC TC T −TC TC −α/(1−α).2) . The system considered is then a binary mixture with a liquidgas critical line. consider the critical system containing a reactive binary mixture dissolved in some solvent. The renormalized critical indices of such systems are well known [84] Cp. α/(1−α) (3. Due to the existence of a neutral third component. .ξ − h2 (∂ξ/∂A)T. The singularities become weaker for manycomponent mixtures. the system has a line of critical . ∼ s.p (3. (3. the reactive binary system has an isolated critical point. A common situation is that the solutes undergo various chemical transformations while the solvent does not. etc.3) . In fact.4) on approaching a critical point is determined by the second term on the righthand side rather than by the ﬁrst term.A=0 = Cp.ξ ∼ Cv.A=0 ∼ s. respectively. and the critical indices of this system are the same as those of a pure ﬂuid. which has a weaker singularity.2) with the case of a frozen chemical reaction (no catalyst added). −α (3.A=0 T − TC TC T − TC TC −γ . ∂υ ∂p ∂v ∂p ∼ Cp. They are changed from α/ (1 − α) and −α/ (1 − α) to −α and −γ. the asymptotic behavior of the lefthand side of Eq.A=0 where υ = 1/ρ is the speciﬁc volume.
µ0 ∼ φ (µ0 ) −α .N0 .T.9). the singularities of thermodynamic quantities in a system undergoing a chemical reaction (A = 0) are determined by the second term on the righthand sides of Eqs.ξ − T (∂ξ/∂A)T.6) and (3. there are two renormalizations in the vicinity of liquidgas critical point.ξ . the singularities near the critical points can be found from the thermodynamic relations.N0 (∂A/∂T )v. as in the vicinity of the λ line in helium.N0 .ξ. one obtains Cp. one concludes that without a chemical reaction. one gets Cp. φ (µ0 ) ≡ T − TC (µ0 ) . Cυ. the ﬁrst occurs when passing from the binary to the ternary mixture.µ0 ∼ Cυ.8).A=0 = Cp. The singularities of the thermodynamic quantities in a ternary mixture at constant chemical potential µ0 of the solvent are similar to those of a binary mixture. so that the parameter RρC (dTC /dp) (ρC is the critical density) is very small.ξ. Cp. and for N0 = const. Let us consider the frozen chemical reaction.9) For both types of critical points.N0 ∼ T − TC TC α/(1−α) .ξ. near the liquidliquid critical points. (3.N0 ∼ φα/(1−α) .ξ. renormalization is absent. (3. as is the case for a reactive binary mixture. The liquidliquid critical points depend only slightly on the pressure. but only for the speciﬁc heat at constant pressure near the liquidgas critical points. TC (µ0 ) (3.ξ. According to renormalization.4).ξ.ξ − T (∂ξ/∂A)υ. (3.N0 .6) Unlike the liquidliquid critical point. analogous to (3.N0 (∂A/∂T )p. and Cp. The factor .8) Cυ. and the second renormalization takes place when passing from µ0 = const to N0 = const. measurements are taken at constant number of solvent particles rather than at µ0 = const.ξ. 2 2 (3.N0 .p.5) However.N0 . When a chemical reaction takes place. cusplike behavior exists for both speciﬁc heats near the liquidliquid critical points.ξ Cυ. Therefore.N0 ∼ const.Eﬀect of Chemistry on Critical Phenomena 57 points rather than an isolated critical point.N0 ∼ Cυ.µ0 ∼ φα/(1−α) (3. This parameter determines the region of renormalization [86]. Cp. In the region of renormalization of the ternary mixture.A=0 = Cυ.ξ.7) Comparing (3.N0 .µ0 ∼ φ−α/(1−α) .
(∂ξ/∂A)T. Therefore.12) yields that the existence of a chemical reaction leads to a magniﬁcation of the singularities of the speciﬁc heats at constant pressure for both types of critical points.A=0 ∼ Cυ.A=0 (3.A=0 Comparing (3. At high temperature and pressure. because these points usually occur at atmospheric pressure and room temperature. T. after renormalization.10) .9) and (3. large ﬂuctuations aﬀect all thermodynamic and kinetic properties including those which are characterized by σ and ε. and. One can also measure ε and σ simultaneously placing the ﬂuid between two parallel plates or coaxial cylindrical electrodes with an alternative current bridge.µ0 ∼ φ−γ . The ﬂuid then acts in the bridge circuit as a frequencydependent impedance with balanced capacity and resistive components [87]. (∂ξ/∂A)T.11) .p. Near the critical points. On the other hand.T. water .8) with (3. Consider now the singularities of the dielectric constant ε and electrical conductivity σ at the critical point of reactive systems.µ0 ∼ φ−α (3.A=0 ∼ ∂υ ∂p ∂υ ∂p ∼ φ−α/(1−α) . whereas the second factor has the following asymptotic behavior.N0 . for a system undergoing a chemical reaction. Cp.11). Experiments are slightly easier to perform near the liquidliquid critical points.12) ∼ φα/(1−α) . α/(1−α) (3.N0 ∼ T − TC TC T − TC TC − α/(1−α) (∂ξ/∂A)υ. the speciﬁc heat at constant volume changes its asymptotic behavior from constant to cusplike only near the liquidgas critical point. from cusplike behavior α/ (1 − α) to a weak singularity −α/ (1 − α). There are hundreds of experimental papers.p. Of special importance are chemical reactions in ecological clean nearcritical water.6) and (3. In principle. These quantities can be obtained by choosing poorly conducting liquid mixtures for dielectric measurements and strongly conducting mixtures for conductivity measurements. one obtains from (3.N0 . of which we will describe only a few.T.N0 .N0 ∼ (∂ξ/∂A)υ.58 Chemistry versus Physics: Chemical Reactions near Critical Points ∂A/∂T in the latter terms remains ﬁnite at the critical point. one can detect such a magniﬁcation experimentally. v.N0 .
respectively. The static dielectric constant for the dimerization reaction N2 O4 has been measured [74] in a solvent mixture of 2NO2 perﬂuoromethylcyclohexanecarbon tetrachloride near the liquidliquid critical point of the solvent.11 is the critical index for the heat capacity at constant pressure and composition. has the temperature dependence of the nearestneighbor correlation function. The special case is the dissociation of a weak acid in water near an acidwater liquidliquid critical point. σ ∼ τ 1−α = τ 0. The dielectric constant was found to have a 1−α anomaly near the critical point. which leads [95] to σ ∼ τ 1−α = τ 0.89 . it is proportional to the inverse of the viscosity η. . and ν = 0.63.05 are the critical indices for the correlation length and viscosity. For a mixture of polystyrene and diethyl malonate near the liquidliquid critical point in the absence of chemical reactions.Eﬀect of Chemistry on Critical Phenomena 59 becomes selfionized with a high conductivity. the same as the conductivity. [90] has been carried out [96].65 . A− +H3 O+ . 4) The anomaly of σ is related to that of the extent of the acid dissociation [94]. No anomaly was found in the dielectric relaxation time. hence. 3) A percolation model for conductivity leads [90] to an anomaly of the form σ ∼ τ 2β = τ 0.89 . taking account of the background contribution due to the normal temperature dependence of the dissociation constant and due to the conﬂuent critical singularity. resembling molten salts [88]. in turn [93]. [90] and phenolwater [90]. which. the dielectric constant was measured [97] as a function of frequency in the range 20 KHz to 1 MHz. consistent with its connection with an anomaly in the extent of the dissociation reaction and also for an anomaly in the protontransfer rate. where α = 0. where β = 0.13) HA+H2 O Measurements of the conductivity as a function of temperature have been performed near the liquidliquid critical points for the systems isobutyric acidwater [89]. where τ = (T − TC ) /TC .325 is the critical index describing the coexistence curve. The leading critical anomaly of the conductance was found to be characterized by a critical index 1 − α. A detailed analysis of the experimental data obtained in [89]. σ ∼ η −1 = τ νzη = τ 0. zη = 0. according to the Stokes law. (3. and a small moment for NO2 ) and of the solvent (the components of low polarity make the “background” dipole moment very small) assure the high precision of measurements. 2) The anomaly of σ is related to the anomaly in the protontransfer rate [92]. Diﬀerent explanations have been proposed for the anomaly in the conductivity near the critical point [91]: 1) The anomaly is due to viscous drag of the acid anion [89] and. The appropriate choice of reaction (no dipole moment for the N2 O4 molecule.031 .
e.3 Singularity in the degree of dissociation near a critical point At the end of the previous section. (3. namely. p and A. υ= . The condition for chemical equilibrium has the following form. an additional factor has to be taken into account.χ.15) permits one to write the diﬀerential of the Gibbs free energy in the form dG = −SdT + V dp + µE dNE + µCD dNCD − (µC + + µD− ) dNC + . and of total number N2 of moles of CD and E. and one intensive (“density”) variable χ are kept ﬁxed.17).18) It is also convenient to deﬁne intensive variables. (3. χ= . The constraint of electroneutrality NC + = ND − .15)–(3. (3. which are caused by the scattering of charged ions by strongly enhanced concentration ﬂuctuations. the density of ions may have a singularity near the critical point due to the chemical reaction [94].20) Note that for a closed system. N2 N2 N2 N2 N2 (3. Usually the temperature changes during the experiment while the pressure remains constant. g= . A ≡ µCD − µC + − µD− = 0. the number of charges. the variable χ remains constant. The extent of reaction. we considered the anomalies in electroconductivity near the critical point.19).16) (3. is deﬁned by the derivative (∂ξ/∂T )p. dg = −sdT + υdp − Adξ + (µCD − µE ) dχ. N1 = NCD + NC + . (3. Assume that the dissociation reaction CD C+ +D− (3. i. However.19) Using Eqs. (3.60 Chemistry versus Physics: Chemical Reactions near Critical Points 3. s= .14) occurs between the solutes CD. C+ and D− near the critical point of the solvent E.A=0 . (3. This derivative is taken under the condition that two extensive (“ﬁeld”) variables. one may rewrite Eq.17) It is convenient [95] to express the thermodynamics in terms of the total number N1 of moles of CD. both dissociated and undissociated. N2 = N1 + NE . apart from the anomalies of scattering processes. ξ= N1 G V S NC + . . independent of the extent of reaction ξ..
24) in the following forms µα dxα + µα dxα − µβ dxβ − µβ dxβ = µβ − µα dT. 31∂x 1 32 2 3T 31 1 32 2 3T (µα − µα − µα ) dxα + (µα − µα − µα ) dxα = − (µα − µα − µα ) dT 11 21 31 1 12 22 32 2 1T 2T 3T (3. which we label α and β.χ. deﬁned by the critical index α = 0. Let us now consider a ternary mixture (components 1. keeping the pressure constant. The question is: How will the concentration of the reagents vary near the critical point of a threecomponent system? For a small equilibrium constant of reaction. 2 2 µα = µβ . d µα − µβ = d µα − µβ = d µα − µβ 1 2 3 1 2 3 = d (µα − µα − µα ) = 0. (3. The chemical reaction allowed is of the form 1 2 + 3.Eﬀect of Chemistry on Critical Phenomena 61 According to the general theory [38]. similar to Eq. xα . (3. but does not produce any additional restriction.A=0 T − TC TC −α . one gets for the equilibrium state. 1 2 3 A similar condition in the second phase follows from (3.22) (3. we may neglect the small shift from the critical point of a onecomponent system. 11 1 12 2 1T 11 1 12 2 1T µα dxα + µα dxα − µβ dxβ − µβ dxβ = µβ − µα dT. xβ . For example. has the form µα − µα − µα = 0. this model describes a onecomponent system (say HI) near its liquidgas critical point (TC = 150 K for HI) dissociated into two species (2HI H2 +I2 ).24) Choosing as independent variables the mole fractions of the ﬁrst and the second component in each phase. Therefore. 1 1 µα = µβ . 1 2 3 (3. xα .22).15).21) Thus far. 21 1 22 2 2T 21 1 22 2 2T µα dxα + µα dxα − µβ dxβ − µβ dxβ = µβ − µα dT. Denoting by µα and µβ the chemical potentials of the ith component i i in each phase.23) The condition of chemical equilibrium. µα = µβ . ∂ξ ∂T ≈ p. 3 3 (3. 2 and 3) existing in two phases. We are interested in equilibrium properties.25) . we have considered a onephase system. (3. one can rewrite Eqs. this derivative has only a “weak” singularity at the critical point.12. and the temperature 1 2 1 2 T. xβ . the thermodynamic path we choose is a displacement along the equilibrium line.
Then. Peculiar to this system is the capability of the proton of the carboxyl group of isobutyric acid to exchange with the deuteron of deuterium oxide (isotope exchange reaction).β . ∆ is the 4 × 4 determinant of Eqs.β j p. which. Simple analysis shows [94] that ∆ has no singularities at the critical point.β /∂x 12 p. 3.β i ∂T ≡ µα. each phase can be described by two mole fractions. (3.25). having one “density” variable ﬁxed. The line of critical points is thus determined by the vanishing of the next derivative of µα. whereas ∆α. (3. has a weak singularity (3. the xT curves have an inﬁnite slope at the critical points. Appreciable . resulting in Eq.x1 ) 2 which . If the fourcomponent mixture is comprised of two phases.β with respect 2 to x [3].27) where the subscript equil refers to the equilibrium line.25).β − µα.T p.β 11 22 2 i can be rewritten as ∂µα.xj (3. The coincidence of the ﬁnal results for the two cases results from the electroneutrality of the fourcomponent mixture which makes it isomorphic to the threecomponent mixture.T.4 Isotope exchange reaction in nearcritical systems The results of experiments performed on isobutyric acid (CO2 H) .26) Solutions of Eqs.β /dT are i ∆α.21) at the critical point.j ∂µα.25). This reaction produces more than two species.β is proportional to the factor µα. The experiments include the visual observation of the phase separation and light scattering [98].β is the determinant with one of the i rows replaced by the row of coeﬃcients from the righthand side of Eqs. (3. (3.β .T. while x3 = x4 = (1 − x1 − x2 ) /2.β dxα. i.62 Chemistry versus Physics: Chemical Reactions near Critical Points where ∂µα. The results have been compared with those for isobutyric acidwater (CO2 H/H2 O) and deuterated isobutyric aciddeuterium oxide (CO2 D/D2 O) systems. Eq.T. and the concentrations of CO2 H and D2 O are not the same because of the isotope exchange reaction.deuterium oxide (D2 O) binary mixture are an example of the shift of critical parameters and the modiﬁcation of critical indices due to a chemical reaction.equil. i.β i = i dT ∆ p.µ1 (p.β µα.xi =xj ≡ µα. (3. and ∆α.β i ∂xα. say x1 and x2 . (3.25) for dxα. the equilibrium equations will have the same form.21). Therefore.
Eﬀect of Chemistry on Critical Phenomena 63 amounts of additional species due to isotope exchange distort the coexistence curve. deﬁne the change of the transport coeﬃcients.5 Singularities of transport coeﬃcients in reactive systems In order to determine the inﬂuence of a chemical reaction on the transport coeﬃcients. one calculates the correlation functions which.5.13 ˚ in agreement with the renormalization group results. That is. can be expanded in powers of the hydrodynamic variables. whereas in the coexistence twophase region.310 and Tmax = 45. thus assuming the OrnsteinZernike form for the susceptibility. the critical indices appear to be renorA malized with γ = 1.1 Modecoupling analysis Just as for the reaction rates. 3. which diﬀers from x0 = 0. in turn. and the renormalized critical indices are in agreement with those near the plait point of a ternary liquid mixture ethanolwaterchloroform [99]. In the onephase region. This makes the critical indices γ and ν of the correlation length (ξ = ξ 0 T −TC TC −γ ) and isothermal compressibility (k = k0 T −TC TC −ν ) in the onephase region (T > TC ) diﬀerent from those of the coexisting twophase region (T < TC ). one can use the same modemode coupling analysis [102] used in the previous chapter for ﬁnding the inﬂuence of criticality on the rate of a chemical reaction.39. the transport coeﬃcients (shear viscosity η and diﬀusion coeﬃcient D) can be written as time integrals of the correlations of appropriate ﬂuxes. shifting the critical solution concentration xC away from the concentration x0 where the maximum phase separation temperature Tmax occurs.358 at TC = 44. In calculating the critical corrections ∆η and ∆D. It was found that xC (mass fraction of isobutyric acid) equals 0. ν = 0. 3. in turn. An alternative explanation of these experiments is given in [100]. A diﬀerent way of attacking this problem is by dynamic renormalization group methods [103].76 and ξ 0 = 0. As result of exchange equilibria. one neglects the small critical indices. which.90◦C.25.6 ˚. ν = 0.11◦C. The most signiﬁcant contribution to ∆η is the .63 and A ξ 0 = 3. γ = 1. [101]. CO2 H/D2 O is eﬀectively a threecomponent system in the coexistence twophase region [98]. which we will brieﬂy describe in the next section.
The inﬂuence of the chemical reaction on the diﬀusion coeﬃcient can easily be seen in the two limiting cases of low frequencies kξ < 1 (the “hydrodynamic” region) and high frequencies kξ > 1 (the “critical” region). δ = L/η. . and κB T k . and since η is not divergent in reactive systems.p . when L = 0. Note that if L is extremely small (extremely slow chemical reaction). β = δ/ (1 + δ). this integral diverges logarithmically at the critical point.28) does not diverge.28) ∆η ∼ k+L k1 The upper limit k2 is related to an unimportant cutoﬀ. However. Because D = α (∂A/∂ξ)T. the inverse correlation length ξ −1 . a) For kξ < 1. one regains the wellknown result for nonreactive mixtures [102].32) ∆D = 16ρ (η + L) This result is similar to that obtained for a nonreactive binary mixture. The function K (X) is given by 3 K (X) = (3. but with a renormalized coeﬃcient. The calculation of the modecoupling corrections to the wavenumber (frequency)dependent diﬀusion coeﬃcient in reactive binary mixtures [102].31) 1 − β 1/2 . Y = X δ/ (1 + X + δ).30) 1 + X 2 + X 3 − X −1 tan−1 X . then η grows upon approaching the critical point. α diverges like ξ.64 Chemistry versus Physics: Chemical Reactions near Critical Points integral over wavenumber k of the form [36] k2 dk . K (X) ∼ X 2 . but its growth is terminated when the correlation length becomes of order L−1 . 4 When L = 0. and κB T (3. It follows that the viscosity in reactive systems remains ﬁnite at the critical point.29) K (X) − ∆D = 2η 6πρξX Y3 where ξ is the correlation length. except for the renormalization of the coeﬃcient of k. and the lower limit k1 . Therefore. b) For kξ > 1. the diﬀusion coeﬃcient in reactive systems still vanishes like ξ −1 but with a modiﬁed coeﬃcient of proportionality due to the chemical reaction. the integral (3. X = kξ. [36] yields βX 3 κB T K (Y ) (3. (3. ∆D = 6πρξη The usual result [102] η∆D ∼ ξ −1 is obtained. vanishes at the critical point. A more careful calculation of ∆η leads to a weak power divergence rather than a logarithmic divergence. K (X) ∼ (3π/8) X 2 . (3. When there is no reaction (Onsager coeﬃcient L = 0).
26 obtained in the linear theory.2 Renormalization group methods The hydrodynamic equations (2. whose relaxation times approach zero at small wavenumbers. 2.37 rather than by γ ≈ 1. The renormalization group technique allows one to average over increasingly large regions.5. Such an identiﬁcation also depends on the regimes of wavenumber which deﬁne the relative importance of the diﬀerent modes. one has to assign each system to the appropriate model A − H [103]. Milner and Martin [40] performed a renormalization group analysis for a reactive binary mixture.1. The starting point is the identiﬁcation of slow modes. thereby reducing the number of eﬀective degrees of freedom of the system. Due to the divergence of the correlation length. Eﬃcient computer programs have been developed for studying static critical phenomena [104]. They found [40] that the critical slowingdown of a chemical reaction occurs for kH < k < kC . . improving the result of the linear analysis of the critical slowingdown described in Sec. and they have been extended to cover dynamic critical phenomena as well [103].64) are written for local spaceaveraged variables. where kC and kH are the inverse length scales for diﬀusion and heat conductivity.61)–(2. and also (if the order parameter is not conserved) the orderparameter mode. where the increasing number of degrees of freedom becomes important. which converts the concentration into a nonconserved parameter. The slowingdown is governed by the strong critical index γ + α + ην 1. The distinctive feature of reactive systems is the existence of a homogeneous chemical mode. Therefore. such a procedure is of special importance near the critical points.Eﬀect of Chemistry on Critical Phenomena 65 3.
stimulate or restrict the process of phase separation? In this chapter we will consider these problems. γ −ν n ≡ K.1 Multiple solutions of the law of mass action Thus far. . This was proved many years ago [105] for isothermalisochoric systems. xi ) i (4.2) The chemical equilibrium constant for the ideal system Kid is determined by the functions µ0 in (4. . T ) γ −ν 1 . x2 . .2) has a single set of solutions x1 .e. The chemical potential of the ith component can be written in the following form µi = µ0 (p. . i. but we ﬁrst have to consider the possibility of multiple solutions of the law of mass action. Another interesting problem is phase separation in reactive systems as compared to nonreactive systems. for which γ i = 1.1). one can rewrite the law of mass action as xν 1 .Chapter 4 Phase Separation in Reactive Systems 4. by the properties of individual i nonreactive components. This means that the law of mass action has more than one solution for the concentration at given temperature and pressure. . in conducting chemical reactions. γ −ν 2 . n n 1 2 1 2 (4. . we have considered singlephase reactive systems near their critical point. xn . Let us clarify the possibility of multiple solutions of the law of mass action. K also depends on the interactions among the components. and Eq. Another proof was 67 . (p. xν 2 .. . induced. Using (4.1) where the activity γ i determines the deviation from the ideal system. all γ i = 1. (4. . Will the presence of a chemical reaction. whereas for nonideal systems. Two (or more) coexisting phases which have the same temperature and pressure are diﬀerent in concentrations of diﬀerent components and. xν n = Kid. T ) + κB T ln (γ i . . For ideal systems. therefore. . . . say. by a very small amount of catalyst.1).
assume that Kid ∼ exp (I0 /κB T ). a system has to be considerably nonideal for the existence of multiple solutions of the law of mass action (4.68 Chemistry versus Physics: Chemical Reactions near Critical Points given for isobaric systems under isothermal [106] and adiabatic [107] conditions. the law of mass action has more than one solution. This has a . electrolytes.. T ) ≈ (8πx) 1/2 e2 /κB T 3/2 + Bx + · · · (4. although slightly artiﬁcial systems. the change in enthalpy (∆H) and the change in entropy (∆S) have opposite signs. Both the isomerization [111] and the dissociation [112] reactions have been analyzed in detail. [113] that K = Kid exp [−Φ (x. This implies that for chemical reactions exhibiting a linear relationship between enthalpy and entropy. Consider a gas consisting of charged particles (plasma. molten salts. which determines the pair correlation between charges. solid state plasma). the existence of the enthalpyentropy compensation eﬀect has been doubted by some researchers.2). T ) /κB T ] ≈ exp {[I0 − Φ (x. molecules) of type A can reversibly form an aggregate (molecule. However. It worth mentioning that phase separation is possible in ideal. the magnitude of change in the Gibbs free energy is less than one might expect. ∆G may change very little even if both enthalpy and entropy increase.e. Another possibility occurs when the characteristic energies of single particles are small. One possibility is for the interaction energy between components to be high. A complete analysis has recently been carried out [108]. has been tabulated [106]. for large n (in fact. as in the case of isomers. It was subsequently noticed [109] that the latter eﬀect is a special case of the phenomenon of the enthalpyentropy compensation. Then. and the function B. where I0 is the ionization potential of a neutral particle.3) where x is the concentration of charges. Neglecting the complications associated with the inﬁnite number of bound states. such as that for which n units (atoms. (4. metalammonium solutions. n → ∞). T )] /κB T } Φ (x. indicating the possibility of phase separation. At certain T and p. as one can see from the article entitled “Enthalpy–entropy compensation: a phantom phenomenon” [110]. oligomer) of type An . in the Gibbs free energy equation (∆G = ∆H − T ∆S).3) has several solutions for x. Eq. Therefore. the interaction energy must be of order of the characteristic energy of an individual component. In general. one readily ﬁnds [112]. Consider the ionization equilibrium of the chemical reaction of the form A i + e (dissociationrecombination of neutral particles into positive and negative charges). Mathematically. Assuming Debye screening of the electrostatic interactions. i.
these two phases can have equal chemical potentials. The kinetics of phase separation from O2 into two phase. The existence of a chemical reaction(s) increases the stability of a metastable state through the appearance of additional constraint(s). the system is located at point O2 . x ) = µi (p. x ) . At the end of this stage. the system starts to separate into two phases.1. i = 1. T ) in ( 4.2 Phase equilibrium in reactive binary mixtures quenched into a metastable state Thermodynamic analysis of reactive binary mixtures 4. First. with a preferred component . thermodynamic considerations give only necessary but not suﬃcient conditions for phase separation in reactive mixtures.5) where x and x denote the concentrations of one of the components in the two phases. The law of mass action for this reaction is A = ν 1 µ1 + ν 2 µ2 = 0 (4.3) diminishes the ionization potential I0 as a result of screening. The chemical reaction proceeds in diﬀerent ways in the two coexisting phases. If the system separates into two phases. described by points O3 and O4 in Fig. proceeds in two clearly distinguishable stages. Moreover. Hence.2. T. The function Φ (x. 2 (4.Phase Separation in Reactive Systems 69 simple physical meaning. and therefore. 4. and after some “completion” time. the phase with the larger degree of ionization has higher energy and higher entropy. 4. pressures and chemical potentials are equal µi (p. “waiting” for the appearance of a signiﬁcant number of critical nuclei due to ﬂuctuations (the duration of this stage is usually called “the lifetime of the metastable state”). can coexist. if the forward and backward reaction rates are diﬀerent.4) where A is the aﬃnity of reaction.1 Let us assume that two components A1 and A2 participate in a chemical reaction of the form ν 1 A1 ←→ ν 2 A2 . The latter can be achieved in two diﬀerent ways. Therefore. Even when thermodynamics allows phase separation. it might be impossible from the kinetic aspect. During the ﬁrst stage. Therefore. where a chemical reaction proceeds in all phases. T. their temperatures. the appearance of multiple solutions of the law of mass action is a necessary condition for phase separation in reactive systems. it reaches the twophase state described by the points O3 and O4 .
corresponding to point O3 . For the lower critical solution point. the system is shifted by chemistry to the closest state on the coexistence curve at the same temperature. If. through spinodal decomposition. the ﬁrst stage of nucleation does not exist. Quenching is performed from the original homogeneous state O1 to the metastable state O2 or to the nonstable state O5 . The second possibility takes place when the forward and backward reactions rates are equal. will keep the mixture homogeneous. . In the former case. occurs through the nucleation process. the rate of reaction is much faster than that of phase separation. Then. 4.70 Chemistry versus Physics: Chemical Reactions near Critical Points Fig. the temperature scale may be inverted. Strictly speaking. phase separation to the ﬁnal stable state. a chemical reaction leads to an additional driving force A = 0. shifting the system homogeneously from a quenched metastable state to the closest stable state O3 . in addition.1 The temperatureconcentration phase diagram of a binary mixture. The points O3 and O4 are separated by a singular line deﬁning the stability boundary. which tends to randomly mix the mixture. for such reactive systems.1 describes the upper critical solution point. described by points O3 and O4 . Therefore. the chemical reaction. whereas in the latter case. Figure 4. spatially homogeneous relaxation from O2 to O4 is impossible. Immediately after a quench.
where the chemical potentials have the following form [3] µ1 = µ0 + κB T ln x + ω (1 − x) . The noncompatibility of Eqs. (4.8) have a common solution. and the strictly regular reactive binary mixture will not separate into two phases. two phases may coexist. By contrast.7) Phase separation in a reactive binary mixture will occur if and only if Eqs.7) and (4. 1 µ2 = µ0 + κB T ln (1 − x) + ωx2 . following the symmetric coexistence curve.4). Λ = 0.7) and (4. and their concentrations are given by intersection of the coexistence curve (4.09 and δ = 2.6) gives x ≡ x = 1 − x for the concentrations of the two coexistence phases [3].9) (4.7) and the law of mass action (4. (4.5). ln x − ln (1 − x) δ−1 2 (4. these equations have no common solution. for example.8) to obtain Λ (1 − x) ln (1 − x) − x2 ln x = ≡ q. In all other cases. one can equate the righthand sides of Eqs.8) (4. the criterion q < 0. ν2 Λ= δµ0 − µ0 1 2 .8) coincide if both Λ = 0 (“symmetric” mixture) and δ = 1 (isomerization reaction).7) and (4.e.Phase Separation in Reactive Systems 71 These physical arguments are supported by simple model calculations. the nonexistence of a common solution for the . If. Using (4. and hence there is no phase separation.6) The coexistence curve is deﬁned by Eq. (4.10) has a solution only for q < (2 ln 2 − 1) /4 = 0. (4. (4. κB T 2x − 1 Using (4.097 is satisﬁed.7) and (4. this criterion is not satisﬁed. 2 2 (4. Let us start with the simplest model of a strictly regular solution [114]. The model of a strictly regular reactive binary mixture allows phase separation only for substances and chemical reactions satisfying this inequality.. If δ = 1. Λ = 0. Eq. One can immediately see that Eqs. ln x − ln (1 − x) ω = .10) The concentration x ranges from zero to unity.097.8). for Λ = 0.1 and δ = 2. the law of mass action can be written as ω ln(1 − x) − δ ln x = 2 κB T Λ + δ (1 − x) − x2 where δ=− ν1 . Therefore. i. ω (4. (4.8) for some range of temperatures and pressures.
Let us start with the meanﬁeld model described by Eq. 1 ω x1 = ln κB T x1 − x2 x2 and the critical curve is given by the limit x1 − x2 → 0. The coexistence curve satisﬁes the following symmetry relations: x1 = x1 . when there is a single reaction.14) (4.11) [117].2.cr = x2.13) ω The constraint A = µ1 − µ2 − µ3 = 0. such as the van Laar and Margules models [115]. a system can only exist in one phase. κB Tcr . In all other cases. may appear in some models of an ncomponent system with n − 1 independent chemical reactions.cr = x3 = exp − ∆G0 κB T x1 x2 exp ω (x1 + x2 − x1 x2 ) κB T (4.11) If there are two independent chemical reactions. (4. the situation is very diﬀerent. Phase equilibrium is obtained by demanding that all three chemical potentials be equal in coexistence phases. x2 = x2 .12) . A complete analysis has been given for a reversible bimolecular reaction that involves all three species [116]. The calculations performed above for a chemical reaction in a strictly regular binary mixture can easily be generalized to a threecomponent mixture of components A+ B C.2 Thermodynamic analysis of reactive ternary mixtures Another way to ﬁnd the global phase diagram for reactive ternary mixtures is by topological analysis of the intersection of the critical manifolds of the model considered and the speciﬁed chemical equilibrium surface. ln γ 2 = ωx1 (1 − x2 ) . (4. However. ln γ 3 = −ωx1 x2 . the existence of r chemical reactions will decrease by r the dimension of the coexistence hypersurface.72 Chemistry versus Physics: Chemical Reactions near Critical Points concentrations in the range from zero to unity. Phase separation in binary mixtures has been described by models more complex than that of the strictly regular solution. The phase equilibrium surface has the following form. (4. takes the following form x1. 4. x3 = x3 . imposed by the chemical reaction.1) with activity coefﬁcients given by (4. described by the activity coeﬃcients ln γ 1 = ωx2 (1 − x1 ) .
(4. which can be written as ∆G0 = ∆H0 − T ∆S0 .15) gives the phase diagram of a reactive ternary mixture. Detailed analysis shows [118] that there are six diﬀerent types of phase diagrams. 1 µ2 = µ0 + κB T ln x2 + ax3 + cx1 − W. results in a closeloop coexistence curve. closedloop phase coexistance curves.Phase Separation in Reactive Systems 73 where ∆G0 is the deviation of the standard Gibbs free energy from a hypothetical ideal solution state. 5. κB T ln x3 = a (x3 − x2 ) + b (x3 − x1 ) x1 x2 + c (x1 + x2 ) − W − ∆H0 + T ∆S0 . thereby reducing the unfavorable mixing. the solution consists only of C and either A or B. As an example of homogeneous nucleation in a chemically reactive system. together with an existing upper critical consolute point. x1 ) for ﬁxed values of ∆G0 /κB T .15) where W = ax2 x3 + bx1 x3 + cx1 x2 .15) can be expressed as follows. Upon further lowering of the temperature. etc.4. azeotropelike points. which. in which the chemical potentials of the components have the following form. b and c which include the triple and quadruple points. Another peculiarity of the phase diagrams in reactive ternary mixtures has been found [118] in the context of a slightly more complex model for A+ B C reaction. We will consider this problem in Sec.16) The intersection of this surface and the critical manifolds following from (4. and the phases become miscible again. The intersection of this chemical equilibrium surface with the coexistence surface (4. the presence of a chemical reaction might result in the appearance of the lower critical consolute point. The chemical equilibrium constraint A = 0 for Eqs. (4. 2 µ3 = µ0 3 + κB T ln x3 + ax2 + bx1 − W (4. The results depend on whether the chemical process is enthalpically or entropically favored. Numerical calculations for speciﬁc values of ∆G0 / κB T show [117] that A − B repulsion results in phase separation as the temperature is lowered. with ∆H0 and ∆S0 being the standard enthalpy and entropy for the given reaction. At low enough temperatures. Hence.12) deﬁnes a unique coexistence curve in the plane (κB T /ω. let us return [119] to the simple bistable chemical reaction (the second . the formation of C becomes favorable. depending on the combinations of sign of the parameters a. µ1 = µ0 + κB T ln x1 + bx3 + cx2 − W.
the state x3 is stable while the state x1 is metastable. they create a stationary proﬁle. The kinetic equation for the concentration x of the component X is dx dV (x) = −k1 x3 + k2 xA x2 − k3 x + k4 xB ≡ − . If the system is quenched to the metastable state ζ = 0. i. ζ = (x − x1 ) /x1 . called a nucleation nucleus ζ (ρ) . they annihilate. δ2 = 3 . Eq. t) = −k1 x3 + k2 xA x2 − k3 x + k4 xB + D 2 ∂t ∂ρ (4. 1. supplemented by diﬀusion.17) with the concentrations xA and xB of species A and B being held constant. (4.18) has three solutions.. and the concentration . Moreover. We assume that the values of the coeﬃcients ki provide two stable solutions x1 and x3 . (4.20) with b = (u2 − 1) (u3 − 1) and a = u2 + u3 − 2. it is suggested that V (x1 ) < V (x3 ).19) where ρ is the spatial coordinate and D is the diﬀusion coeﬃcient.21) b/2 a + 3 b/2 . (4. one gets ∂ 2ζ ∂ζ = D 2 − bζ + aζ 2 − ζ 3 ∂τ ∂ρ (4.18) will have the form of the reactiondiﬀusion equation ∂2x ∂x (ρ. D = D/k1 x2 . t) = ζ 3 {1 + exp [±δ (ρ − ρ0 ± vt)]} where 3b − aζ 3 1 ζ2 1/2 .20) (“kink”“antikink”) ζ (ρ. Introducing the dimensionless variables ui = xi /x1 . If the reaction (4. τ = tk1 x2 .e.22) 2 2D 2δ The kinkantikink pair separates with time in opposite directions.74 Chemistry versus Physics: Chemical Reactions near Critical Points Schlogl trimolecular model) considered in Sec. equal to ζ3 = −δ −1 ln a−3 −1 −1 (4. can be understood as the spontaneous creation of two solutions of Eq.18) dt dx Equation (4. When separated by a certain critical length κ. 1 1 and moving the origin to the metastable state x1 . v = − a + a2 − 4b .3 A+X k1 k2 3X. occurs in a closed onedimensional vessel. X k3 k4 B (4. and one unstable solution x2 such that x1 < x2 < x3 .17). When the kink and the antikink are separated by a distance less than κ. (4. its transition to the stable state ζ = ζ 3 .
these ﬂuctuations are of crucial importance for the analysis of the decay of a metastable state. more important. respectively.21) becomes simpliﬁed in the smallamplitudenucleus limit. Although the energy of these states is larger than that of the corresponding twophase system. In addition to the homogeneous nucleation initiated by the ﬂuctuations (considered above). we have discussed a deterministic system. Equation (4. The external noise may be easily greater than the internal noise and. when κ → 0 . and the nucleation nucleus then becomes ζ (ρ) = 3b 3b [1 + cosh (δρ)]−1 = sech2 a 2a δρ 2 . the important problem is nonhomogeneous nucleation initiated. When the separation becomes greater than κ. All states located within the coexistence curve and close to it are the socalled metastable states. by pieces of dust present in solutions.3 Kinetics of phase separation Let us ﬁrst discuss nonreactive systems. therefore. for simplicity. one needs a ﬁnite “push” to pass from the local minimum in energy to the deeper global minimum. to be spherical with radius r) is equal to 4 ∆G (r) = −∆µ πr3 + 4πσr2 3 (4. (4. phase separation will occur and the system will consist of two phases described by the thermodynamic condition (4.1 belong to twophase states. The inﬂuence of homogeneous external white noise and internal chemical ﬂuctuations has been considered [119] in the framework of the extended Kramers approach and the multivariate master equation.Phase Separation in Reactive Systems 75 proﬁle returns to the homogeneous metastable state.5) and shown by points O3 and O4 in the ﬁgure. after a change in temperature (or pressure) from point O1 to O2 . the kinkantikink pair separates. The change of the thermodynamic potential associated with the appearance of the nucleus of the new phase (assumed. 4. and the system evolves toward the homogeneous stable state. 4.24) . The nucleation due to the internal ﬂuctuations is limited to a very narrow region near the transition. Therefore. and it is diﬃcult to observe this experimentally. All thermodynamic states within the coexistence curve shown by the solid line in Fig. Such a “push” is provided by the thermal ﬂuctuations in density for a onecomponent system and in concentration for manycomponent mixtures.2. say.23) Thus far. Therefore.
small nuclei tend to shrink because of their high surfacetovolume ratio. The formation energy of a nucleus containing n1 atoms of component 1 and n2 atoms of component 2 is ∆G = µ1. ∆G > 0.26) with functions F and D of the form F (r) = D0 1 1 − r rC . t) of nuclei of size r at time t. ∂ ∂W ∂W = FW + D ∂t ∂r ∂r ≡− ∂J ∂t (4.n are the corresponding quantities in the nuclei. one ﬁnds 2σ . From ∆G/dr = 0.29) . the height of the potential barrier is deﬁned by ∂∆G/∂r = ∂∆G/∂x = 0.n and µ2. which can be transformed to the FokkerPlanck equation (4. The ﬁrst (negative) term in (4. which gives the following relation between rC and xC of the critical nucleus. One can reduce [120] the equation(s) of the critical dynamics to the Langevin equation for the radius of nuclei with a known random force. It is clear from (4.n − µ2 n2 + 4πσr2 (4. where υ1 and υ 2 are the volumes per atom.27) where D0 is the diﬀusion coeﬃcient far from the critical point. but also by the concentration x = n1 / (n1 + n2 ). and µ1.76 Chemistry versus Physics: Chemical Reactions near Critical Points where ∆µ is the energy gain in a stable state compared to a metastable state. (4.26) where the ﬂux J (r.24) that for small nuclei. The nucleation process in a binary mixture can be described in analogous fashion [121]. In contrast to a pure substance.n − µ1 n1 + µ2. rC = 2σ xC υ 1 + (1 − xC ) υ 2 xC ∆µ1 + (1 − xC ) ∆µ2 (4. D (r) = D 0 κB T 8πσr2 (4. Only nuclei larger than the critical size rC are energetically favorable.25) rC = ∆µ The traditional phenomenological approach to the decay of a metastable state of a pure substance is based on the distribution function W (r.28) where µ1 and µ2 are the chemical potentials of the components in the homogeneous phase. [18].24) encourages phase separation while the second (positive) term prevents it. Therefore. The size r of a nucleus is related to its structure by 4πr3 /3 = υ 1 n1 + υ2 n2 . That is. t) of nuclei along the size axis is determined by two unknown functions F and W . and σ is the surface tension. The continuity equation has the form. the critical nucleus is deﬁned not only by its size r.
we approximate F (r) by F (r) = B (rC − r) . (4. we may omit the argument x in W (r. In other words. Each intermediate state corresponds to a diﬀerent radius of critical nucleus rC (t). Substituting (4.27).30). According to the physical picture. The next step is to obtain the FokkerPlanck equation for the distribution function W (r.5). increasing toward the coexistence curve. The concentration x changes as a result of a chemical reaction.30) dt τ where τ −1 is the rate of the chemical reaction. We make the approximation that all nuclei which are important for phase separation have the same composition x . and retained only the leading logarithmic part of the chemical potentials. which allows an exact solution. Therefore. into Eq. Let us now consider reactive systems.26). where rC → ∞. we ﬁrst consider a simpliﬁed version.Phase Separation in Reactive Systems 77 where ∆µi = µi.27) by simpler functions which incorporate the main property of (4. (4. t. one obtains rC (t) = 2σ [x υ 1 + (1 − x ) υ 2 ] κB T [x /x − (1 − x ) / (1 − x )] (x − x ) (4. (4. Hence.29). x − x = (x0 − x ) exp − τ . which shifts the initial quenched state to the closest equilibrium state on the coexistence curve. t Inserting the solution of Eq. Therefore.27) and (4. one obtains 1 ∂ 1 t ∂W D0 κB T ∂W = − exp − D0 W+ . t. (4.29) under the assumption of a quasistatic chemical shift.31) ≡ r0 exp (t/τ ) where we used the coexistence condition (4. One replaces functions F (r) and D (r) in (4. and can be described in the linear approximation by x−x dx =− (4.31) into the FokkerPlanck equation ( 4. one has to replace rC by rC (t) everywhere.33) . F (r) is positive for r < rC and negative for r > rC . The latter function can be found from Eq. one assumes that a chemical reaction brings the path leading to the saddle point closer to that of x = x . x). x) for nuclei of size r and composition x at time t. (4. This shift is caused by a chemical reaction.32) ∂t ∂r r r0 τ 8πσr2 ∂r To ﬁnd the approximate solution of this complicated equation.n − µi .
∆G F (r) dr = (4. ∂W t ∂W − BK = − D0 K 2 + iKBr0 exp( ) W. according to (4. leads to Weq = exp − drF (r) /D (r) .26)–(4. (4. t) exp (iKr) dr. t) = W (r.38) ∂t ∂K τ To solve this equation. (4.33). by deﬁnition.27) is thus transformed into a ﬁrstorder linear partial diﬀerential equation for W (K.41) . 2B − exp (Bt) . therefore. (4. we are no longer free to choose the second function D (r) .39) where ξ is of order of a single molecule (there are no nuclei immediately after a quench). Solving Eq. with initial condition W (r. Equation (4.38) and performing the inverse Fourier transformation.78 Chemistry versus Physics: Chemical Reactions near Critical Points After choosing this form of F (r) .35) D (r) = 8πσr 8πσ We keep only the largest term in r − rC in the last equality in (4.37) The original FokkerPlanck equation (4. (4.36) B r0 exp − r W + D0 ∂t ∂r τ ∂r2 where D0 = BkB T /8πσ.24) and (4.35). A simple calculation yields BkB T BkB T rC ≈ . one obtains the Gaussian distribution function W (r. t) [122]. These two functions are not independent because there is no ﬂux in equilibrium.35). and. (4. W = exp (−∆G/kB T ).25).40) with mean value r and variance ∆r2 given by t Br0 r = ξ exp (Bt) − −1 exp τ −B τ ∆r2 = D0 [exp (2Bt) − 1] .36) can be solved exactly by introducing the characteristic function W (K. one uses the method of characteristics. t = 0) = δ (r − ξ) (4. (4. On the other hand. Using (4. (4. which. t ∂ ∂2W ∂W = .31) and (4.34) D (r) kB T where ∆G (r) and rC are given by (4. t) = 1 4π∆r2 exp − (r − r) 2∆r2 2 (4. one obtains the simpliﬁed FokkerPlanck equation.26).
36) which supply the quasistatic ﬂux Jqss (t) independent on the size of the nuclei. (4. restrict attention to those solutions of (4. 0 (4. instead of the exact solution. and. The simplest generalization is the time required to produce one critical nucleus Tch dt Jqss (t) = 1. and Tch is replaced by the lifetime of the metastable state in the −1 nonreactive system. are associated with transient processes and chemical reactions. The quasisteadystate solution Wqss [r. Jqss t =0 τ = Jss .Phase Separation in Reactive Systems 79 Two characteristic times.36). However. Therefore.26) and (4. rC (t)] does not have an explicit time dependence. according to (4.44) For the timeindependent case.40) diverges with time. We consider the quasisteadystate regime which is established after the transients disappear. one is forced to give a new deﬁnition to the lifetime of a metastable state in reactive systems. so that Eq. T0 = Jss .27).32) can be rewritten as D0 1 1 t − exp − r r0 τ Wqss + D0 κB T ∂Wqss ≡ Jqss 8πσr2 ∂r t τ .42). Jqss is replaced by Jss . the solution (4. This assumption permits us to neglect transient processes. (4. Let us now turn to the solution of Eq. Hence.43). assuming that B −1 is very small [123]. for B −1 < τ . and the FokkerPlanck equation (4. (4. One can ﬁnd [122] the appropriate solution of Eq. (4. The boundary conditions for this equation are determined by the requirements that the distribution of nuclei of minimal size ξ will be the equilibrium distribution. but we prefer to ﬁnd the solution of the original equations (4.43) There is no stationary state for the reactive system considered here. so that ∂Wqss /∂t = 0. Wqss (ξ) = Weq (ξ) . B −1 and τ . where Weq corresponds to zero ﬂux.36) does not have a steadystate solution. respectively. (4. and the total number of nuclei in the system is bounded.42) The quasisteadystate ﬂux Jqss reduces to the steadystate ﬂux Jss when the chemical reaction is absent. Wqss (r → ∞) = 0 (4. one has to use the approximate solution.45) .
46) dz . Upon inserting Jss .50) Equation (4. which reﬂects the existence of a barrier to nucleation. u exp(2Tch /τ ) τ u exp (u) 1= 2T0 u exp (−z) dz. This equation has no solution if τ is very small.e. deﬁned by (4.49) and T0 = Jss into (4. the integral (4. The system is dragged by the chemical reaction to a homogeneous equilibrium state on the coexistence curve before the nuclei of the new phase appear. one obtains Jqss = Jss exp − 2t − u exp τ 2t τ −1 .49) where u = ∆G (r0 ) /κB T . . such a separation is impossible from the kinetic point of view.45) has the form [123]. = Weq (r) 1 − Jqss ξ r dz Weq (z) D (z) (4.48). Therefore. (4. the chemical reaction is very fast. Then.43).. and the rate τ −1 of the chemical reaction. z2 (4.47) Jqss = κB T D0 2t ∆G (r0 ) exp 2 exp − τ − 4π 2 σ 2r0 κB T 2t τ (4.47) can be evaluated by the method of steepest descents. (4.42) which satisﬁes the boundary conditions (4. −1 Finally. i. although the thermodynamics allows phase separation in a reactive system. Weq (z) D (z) (4.48) 3 where ∆G (r0 ) = 4πσr0 /3 is the minimal work needed to produce the critical nucleus in the initial state immediately after a quench.44) gives the equation for the lifetime Tch of a metastable state in a reactive system as a function of the lifetime in the nonreactive system T0 (the latter depends on the volume under observation). inserting (4. the extent of quench u. Wqss where ∞ −1 Jqss = ξ −1 The function Weq (r) has a sharp maximum at rC .80 Chemistry versus Physics: Chemical Reactions near Critical Points The solution of Eq. into (4.50) for Tch has been solved numerically.
4.54) dΨ . (4. we have considered phase separation in a system quenched into the metastable region of phase diagram (point O2 in Fig.03 + 2. 4. and F [Ψ] is the GinzburgLandau free energy functional F [Ψ] = dx κ  ∇Ψ 2 +f (Ψ) . and κ is the positive constant related to the interaction range. In this case. Numerical simulation has been performed [125] by placing Eq. the system will never separate into two phases.1) of the phase diagram located inside the spinodal curve. 4. The phenomenological description of this process is based on the CahnHillard equation [124] which. (4.51) T0 min For smaller values of τ .3 Phase equilibrium in reactive mixtures quenched into an unstable state Thus far.3 tanh Ψ + Ψ.1) located between the limit of thermodynamic stability (spinodal) and the boundaries of phase coexistence (binodal). f (Ψ) = −aΨ2 + bΨ4 . 4.08u. for our case of the conserved variable. In that case. (4.53) where f (Ψ) is the doublewell potential. The two possible sets of experiments performed with and without a catalyst (diﬀerent quenches for a given chemical reaction and diﬀerent chemical reactions for the same quench) will be discussed in the next section. separation proceeds through nucleation. has the following form ∂F [Ψ] ∂Ψ = Γ∇2 (4.Phase Separation in Reactive Systems 81 The minimal τ which allows phase separation for diﬀerent quenches (with characteristic u and T0 ) is given approximately by the following formula.52) ∂t ∂Ψ where Ψ = x1 − x2 is the concentration diﬀerence (order parameter). Γ is proportional to the mobility.52) on a 128 × 128 square lattice with periodic boundary conditions.1) to the nonstable state O5 . and the local part of free energy of the form df (Ψ) = −1. the process leads to macroscopic phase separation via a complicated labyrinthine interconnected morphology. τ ≈ 3. the system separates into Arich and Brich phases. When the A− B mixture is quenched from a homogeneous state (O1 in Fig. (4. with the lattice site and time step both set at unity. Another mechanism for phase separation takes place when a system is quenched into the unstable region (point O5 in Fig.
4.0013 and kb = 0. and the random number δΨ chosen in the range (−0.. so that the chemical reaction is maintained by an external agent. [127]).7 and x2 = 0. say. see [126]. (b) after t = 1. i.55) ∂Ψ 5 If the chemical reaction starts at t = 10 with kf = kb = 0. 0. The simplest way to take into account the chemical reaction A ↔ B is to add to Eq. (4. δF [Ψ] ∂Ψ = Γ∇2 − kf Ψ + kb (1 − Ψ) ∂t δΨ ∂f − 2κ∇2 Ψ − (kf + kb ) Ψ + kb . The diﬀerence between the = Γ∇2 . For an Arich mixture (with concentrations of A and B being x1 = 0.52) for the reactive binary mixture in the following form [128]. kf x1 − kb x2 .2. The morphologies of phase separation become much more manifest in the presence of an ongoing chemical reaction. 4. (a) (b) (c) Fig. we rewrite Eq.52) the equation of the chemical reaction. (4. From left to right: (a) after 105 steps with no chemical reaction.82 Chemistry versus Physics: Chemical Reactions near Critical Points The order parameter Ψ = Ψ0 + δΨ at each site is given by Ψ0 = 0. the domain patterns shown in Fig. [125] with permission.e. 4. far from equilibrium.2 Domain patterns for A − B mixture quenched to a nonstable state. 4. (c) after t = 8. 4. light (for a discussion of this point.01) started at t = 105 . (4. we now assume that our system is open.0007) started at t = 5 × 104 .15) .2a gradually change to the concentric patterns shown in Fig. Physical Society of Japan. respectively).01. the Brich domains growing in the Arich matrix are shown in Fig.3.2b for t = 1. Reproduced from Ref. Unlike the previously considered phase separation from a metastable state in a closed system. Thus. copyright (1997). where kf and kb are the forward and backward reaction rates.32×105 [125].32 × 105 steps with the chemical reaction (kf = kb = 0.2 × 104 steps with the chemical reaction (kf = 0.15.
4.025. kf = kb equals 10−4 .3–4. 4.3b). As a result. 4.1 and δΨ = (−0.Phase Separation in Reactive Systems 83 patterns in Figs. 9 × 10−3 ). 4. as shown in Fig.3c).3a–4. If phase separation begins simultaneously with the chemical reaction. As seen in Fig. Ψ0 = 0. hydrodynamics plays an important role. Domain patterns at t = 104 are shown in Figs.1 × 10−4 .2a and 4. Numerical simulations have been performed [129] under conditions similar to those described above (256 × 256 square lattice. where the mixing process due to the chemical reaction proceeds more quickly than the thermodynamically induced phase separation. 4. Equation (4. 4. The results are quite similar for nonsymmetric reactions (kf = kb ). while Figs. lamellar patterns will form.55) has been coupled with the NavierStokes equation. The increase in phase separation coupled with a chemical reaction has been calculated [129] by including hydrodynamic eﬀects. 1. one expects A rich droplets in the Brich phase because kb > kf . In the system shown in Fig. Figures 4. 4. In this case. These ﬁgures clearly show the competition between hydrodynamics eﬀects and chemical reactions. and for η = 100 (right row). The additive contribution of the two ingredients of the process — segregation dynamics and chemical reaction — can be demonstrated by starting . and the Brich phases do not penetrate each other.3a. even without a chemical reaction.4.1 × 10−2 and kb = 9 × 10−5 .007) started at t = 5 × 104 steps after the quench. the nonsymmetrical chemical reaction (kf = 0.4c show the results for nonsymmetric reactions (from the top down: kf = 1.4 for η = 1 (left row). Some diﬀerence occurs for intermediate reaction rates (Fig.2 × 104 steps [125]. for η = 1.2c.4a–4. kf = kb .0013 and kb = 0.3c describe the symmetric reactions (from the top down.025)). whereas for low ﬂuidity (η = 100). where a continuous pattern emerges along with an isolated A rich phase for η = 100. for a low reaction rate. The latter eﬀect occurs at higher reaction rates (Fig.2b are connected with the time delay between the beginning of phase separation after the temperature quench and the onset of the chemical reaction. a bicontinuous pattern emerges at t = 8. isolated domains become more circular under hydrodynamic ﬂow. the disconnected A rich domain appears in the matrix of the Brich phase. hydrodynamic eﬀects are not important. The inﬂuence of hydrodynamic eﬀects is characterized by the viscosity η.4b). 4. However.1 × 10−3 . 1. 4. 9× 10−4 . For intermediate reaction rates (Fig. 0. Still other patterns are formed when the rates of the forward and back reactions are diﬀerent. hydrodynamic eﬀects lead to a circular shape for isolated domains since the chemical reaction rate is not high enough to make discontinuous domains penetrate each other. 10−3 and 10−2 ). When the viscosity is low (η = 1).
Arich and Brich regions are shown as white and black. Reproduced from Ref. 4. whereas the chemical contribution is described by a master equation for the timedependent probability distribution for N . The dynamics of the system are described by spinexchange or Kawasaki dynamics. respectively (see text for explanation). copyright (2003). from a diﬀerent model [130].3 Domain patterns at t = 104 steps with and without hydrodynamic eﬀects for symmetric chemical reactions.84 Chemistry versus Physics: Chemical Reactions near Critical Points Fig. The two states of the spin1/2 Ising model correspond to the site occupied by atom 1 or 2. American Institute of Physics. [129] with permission. in which the dynamics of a quenched system is described in the context of an Ising model with a spin variable σ i at site i.
4 Domain patterns at t = 104 steps with for nonsymmetric chemical reactions. respectively (see text for explanation). which. permission. t) − Ψ Ψ (z .56) . [129] with Physics spins. t) − Ψ (4. 4.55) can be added [131] to the free energy functional F [Ψ] in the following form F [Ψ] → F [Ψ] + kf + kb + 2 dzdz G (z. (4.Phase Separation in Reactive Systems 85 Fig. leads to Eq. Note that the two last “chemical” terms in Eq. American Institute of and without hydrodynamic eﬀects Brich regions are shown as white Reproduced from Ref. Arich and and black. after averaging over all conﬁgurations. (4. copyright (2003).55) [130]. z ) Ψ (z.
It is for this reason that for equilibrium systems. Including the chemical reaction is thus equivalent to including a longrange repulsive interaction.. Although these two numerical techniques usually give the same results for equilibrium properties. (4. The contributions of critical dynamics (diﬀusion) and the chemical reaction were previously considered independently. ﬂuctuations with k < kC grow while those with k > kC decay. so that for d = 3. z ) is the Green function for the Laplace equation ∇2 G (z. no correction was found for the growth exponent in the absence of a chemical reaction and scales with respect to the reaction rate in reactive systems. Rss ∼ k −φ has the same critical index φ = 0. i. contrary to critical dynamics. It was shown [128] that the occurrence of a reaction results in shifting the cutoﬀ by an amount proportional to kf + kb . This result of linear stability analysis was veriﬁed by numerical simulations of the full nonlinear equation [128] .86 Chemistry versus Physics: Chemical Reactions near Critical Points where Ψ = (kf − kb ) / (kf + kb ) and G (z. the presence of a chemical reaction makes a system more stable with respect to phase separation. the chemical rate laws do not take into account the nonideality of the systems. z ) = −δ (z − z ) with appropriate boundary con−1 ditions. . Thus. Other interesting results obtained for both models relate to the growth of the steadystate domains of size Rss .55) in the absence of chemical reactions predicts exponential growth of the concentration ﬂuctuations with a growth factor that has a cutoﬀ at large wavenumber kC (small wavelength). thereby transforming the problem of a reactive mixture to that of a nonreactive mixture with competing shortrange and longrange interactions. A more serious discrepancy was found [134] when. where the collective hydrodynamics modes are not simulated in the typical singleparticle MonteCarlo method. with both being functions of the chemical potentials (in the former case. z ) = 4π z − z  . The latter result has been reﬁned [133]. the results are diﬀerent for nonequilibrium dynamics. Thus. of the spatial derivatives of chemical potentials). A linear stability analysis of Eq. Moreover. For the chemical reaction with kf = kb ≡ k. the calculations were carried out in continuous space using molecular dynamics. Such an approach is inconsistent [135] because it is characteristic of chemical reactions and diﬀusion that the thermodynamic forces governing these processes are related.e.3 as the growth index (R (t) ∼ tφ ) in the absence of a chemical reaction. in contrast to the numerical calculation performed using the Monte Carlo technique for particles on a square lattice [132]. G (z. as well as for the Kawasaki exchange interaction model including the chemical reaction [132].
A consistent analysis shows [135] that the interesting and possible important idea of applications to the freezing of the phase separation by the use of simultaneous chemical reactions.52) in the presence of a chemical reaction.55) of the LandauGinzburg equation (4. Thus far. xB ) = xA ln xA + xB ln xB + (1 − xA − xB ) ln (1 − xA − xB ) + nχAB xA xB + nχBC xB (1 − xA − xB ) + nχAC xA (1 − xA − xB ) (4.59) where xA . Two order parameters for these systems are given by φ = xA − xB . A similar approach has been applied [136] to the isomerization reaction between A and B A k1 k2 B (4.Phase Separation in Reactive Systems 87 diﬀusional stability automatically insures the stability of chemical equilibrium (DuhemJoguet theorem [3]).58) The FloryHuggins free energy for the ternary mixture has the following form [138]. (4. η = xA + xB − ψ C (4. is not easily carried out in practice and demands special requirements both for the systems and for the chemical reactions. Expanding (4.61) . φ)] Dη 2 dr [ η (r)] + 2 Dφ 2 dr [ φ (r)] 2 (4. we have considered the inﬂuence of the A B chemical reaction on the phase separation in binary mixtures by the generalization (4.60) where ψ C is the critical composition of the phase diagram for phase separation between C and A − B.57) in which C does not participate. one obtains [136] F {φ. η} = + dr [fη (η) + fφ (φ) + fint (η. and n is the number of repeated units in the polymer chain. f (xA . χ is the binary interaction parameter. and to ternary polymer blends of identical chains lengths with the chemical reaction [137] A+B k1 k2 C. xB and xC = 1 − xA − xC are the volume fractions of the components.59) in the order parameters η and φ and their derivatives.
88 Chemistry versus Physics: Chemical Reactions near Critical Points where Dη and Dφ are phenomenological parameters related to the surface energy. φ) = c5 ηφ − c6 ηφ2 + c7 η 2 φ2 2 2 (4.η} . a diﬀerence between phase separation in a ternary mixture and in a binary mixture. η} − (k1 + k2 ) φ δφ (4. Numerical simulations show [136] that for the isomerization reaction (4. (4. The equation of motion for the order parameters η and φ (LandauGinzburg equations generalized to a chemically reactive ternary mixture) have a slightly diﬀerent form for the chemical reactions (4.57) and (4. δφ where Mφ and Mη are the mobility coeﬃcients. In the former case. and 1 1 fη (η) = − c1 η 2 + c2 η 4 2 4 1 1 fφ (φ) = − c3 φ2 + c4 φ4 2 4 1 1 fint (η. Otherwise. η} .63) 2 δF − (k1 − k2 ) η − (k1 − k2 ) ψ C whereas in the latter case. ∂η = Mη ∂t ∂φ = Mφ ∂t 2 δF {φ. For simplicity. δη {φ.58).η} δη + 2k2 ψ C − 2k1 xA xB . ∂η = Mη ∂t ∂φ = Mφ ∂t 2 δF {φ.57). the following forms of the functions fη (η) and fφ (φ) have been used in numerical simulations. related to the availability of the third inert component. [137]). .62) where the ci are functions of the coeﬃcients χ and ψ C (for details. the dynamics of phase separation in binary and ternary mixtures is quite similar.64) 2 δF {φ.65) where Aη and Aφ are phenomenological parameters which are inversely proportional to temperature. leads to the existence of a multiplephase transition. see [136]. dη dfφ = Aφ tanh φ − φ dφ (4. dfη = Aη tanh η − η.
and phase separation occurs between A and B. the separation dynamics depends on the competition between the chemical reaction and phase separation [137]. the concentration of C increases as the rate constant of the chemical reaction decreases. In the later stages of phase separation. In the next section. manifesting a transition from nonmetallic to metallic conductivity. the hardcore interaction at short distances). (4. Metalammonia solutions are a type of system in which a small increase of the concentration of metal results in a huge increase of electroconductivity. they give rise to new attractive forces (the Coulomb interaction between ions.4 Thermodynamics of a threecomponent plasma with a dissociative chemical reaction Orderdisorder phase transitions and critical phenomena in electrically neutral systems are due to the competition between repulsive and attractive forces. These new competing forces result in an additional ﬁrstorder phase transition in a threecomponent plasma with a critical point. we consider the dissociation reaction which is a special type of chemical reaction in a ternary mixture. As the phase separation of A and B proceeds. which is. As the reaction rate gradually decreases. It has been proposed [85]. on average. The most general explanation of the nature of this transition was given .16). phase separation between C and A − B is inhibited. resulting in the phase pattern of small amount of C distributed along the interface between Arich and Brich domains. The coexisting phases for a new phase transition resulting from the chemical reaction are determined by Eqs. In the case of a relatively large reaction rate constant. When charged particles appear due to the dissociationrecombination reaction. the eﬀect of a chemical reaction is dominant.12) and (4. Metallic vapors are another example of the latter behavior. [114] that such arguments can be used to explain some experiments in metallic vapors at high pressures and in metalammonia solutions. an attractive interaction) and repulsive forces (say.Phase Separation in Reactive Systems 89 For reaction (4. The continuous decrease in the density of metals with heating above the critical temperature causes the transition into the nonmetallic state and a signiﬁcant decrease of the electric conductivity. C decomposes into A and B rapidly. the eﬀect of the chemical reaction diminishes.58). and the Crich phase changes from dispersed droplets to a continuous thin phase. 4.
Afterwards. This was interpreted as a transition from a weakly ionized mercury plasma to a new inhomogeneous phase of charged droplets. Let us consider two examples in more detail. a dielectric anomaly has been found [142] at the density ρ ≈ 3 g/cm3 . and hence the system will become metallic. etc. TC = 1750 K). Sometimes the transition takes place at a density lower than that predicted by the Mott criterion [140]. who proposed the following elegant physical argument. Let us consider a system com . The liquidgas critical point of mercury occurs at the density ρC = 5. weak electrolytes. probably of ﬁrst order. in other systems. However. solutions of metals in their salts. It is generally accepted (see. Here. electronhole plasmas in optically excited semiconductors. aB is the Bohr radius in the medium.90 Chemistry versus Physics: Chemical Reactions near Critical Points by Mott [139].0631). not all metalnonmetal transitions can be explained in this way. and the latter results in very long relaxation times. A ternary mixture with an ionization reaction is another example of phase separation in reactive systems [114]. Despite the generality of the Mott criterion. there is more than one transition. two additional jumps in the heat capacity occur at two distinct temperatures above the critical temperature. and 0. however. The second example is even more conclusive. the heat capacity shows a jump across the liquidgas separation curve of this mixture. Mott argued that changing the number density n of charges interacting by a screened Coulomb interaction eliminates the bound states when aB n−1/3 ≈ 0. The changes in slope of the electromotive force take place at the same temperatures [144] providing additional evidence for a ﬁrstorder phase transition above the liquidgas critical point.045. and therefore requires an explanation. one can mention molten salts.25.0462. [141]) that the latter transition is explained by the Mott criterion. Upon further heating. The former requires extreme homogeneity of the temperature. This transition. In addition to metalammonia solutions.77 g/cm3 (pC = 1670 Bar. 0. Experimental observation of this new phase transition is very diﬃcult because of the very small density diﬀerences between the two phases and the existence of slow diﬀusion processes. while its existence seems obvious for a broad class of systems. The heat capacity of sodiumammonia solution has been measured [143] for three diﬀerent molar fractions of metal (X = 0. We believe that these and other experimental problems prevent the observation of this new thermodynamic phase transition for other systems. occurs at densities much lower than those given by Mott. Starting at low temperatures. a metalnonmetal transition takes place at ρ ≈ 9 g/cm3 .
.Phase Separation in Reactive Systems 91 posed of neutral particles which are partially dissociated into positive and negative charges.0 + Ne µe. u 1/2 p e3 1 .66). µq ≡ µI + µe = µI.0 + κB T ln xA − αµ0 xα+1. and xA ≡ NA / (NA + 2NI ) = 1 − 2x. The next two terms are the entropy. I is the ionization potential. The ﬁrst three terms in Eq. α= µA = µA. In order to obtain the coexistence curve. We will later make the DebyeH¨ ckel approximation. (4. If one switches from one free energy to the other by using the ideal gas equation of state in the correction term.14) of Landau and Lifshitz [4].1) and (75. and the last term represents the interaction between the charged particles. one equates the chemical potentials in the two phases.67) follows from Eqs. analogously to Eq. NI (Ne ) is the number of ions (electrons). the DebyeH¨ ckel approximation is slightly diﬀerent. and our equation (4.69) One can easily verify that Eq. Assume that the Gibbs free energy has the following form. Solving these two equations for αµ0 /κB T yields ln (1 − 2x ) − ln (1 − 2x ) (x ) = 2α (x ) α+1 − (x ) α+1 2α (ln x − ln x ) α+1 − (x ) α+1 − (α + 1) [(x ) − (x ) ] α α .0 + 2κB T ln x + µ0 (α + 1) x − 2αx α α+1 (4. We here use the Gibbs u free energy. the sixth term describes the ionization process.0 + NI µI. Electroneutrality requires that NI = Ne .5).68) + I. (4.66) describe the ideal gases of the particles.0 + NA κB T ln xA + 2NI κB T ln xI + NI I + µ0 NI xα I (4. µ0 = − . (4. (4. (94.0 + µe.66) where NA is the number of neutral particles.69) gives the following relation between the charge concentrations in the coexisting phases. (4.67) 2 3κB T 2ε3 π 2 One can formulate the DebyeH¨ckel approximation using either the u Gibbs or the Helmholtz free energy. The chemical potentials of the neutral and charged particles can be easily found from (4. G = NA µA.
67) and (4. one can write the latter equation in the following form 1 − 2x µ0 (4. the coexistence curve has the u following form. Equation (4. µ0 − µ0 − µ0 − I + κB T ln A I e ln p κB T 2π 2 me κ B T 2 3/2 + p 2ε3 π 2 I − ln κB T 1 − 2x x2 (4. one adopts the following approximate procedure: for given I and ε.73).70) For the DebyeH¨ ckel interaction (4.72).73) gives the required restriction on the parameters of the coexisting phases resulting from the chemical reaction.72) 3x1/2 − x3/2 = 0. One has to ﬁnd the simultaneous solutions of Eqs. (4.67). x = 1 − x − (2 − 3x ) − e3 6 (κB T ) 2 1/2 (x ) 1/2 1/2 1/2 .73) − e3 /6 (κB T ) 1/2 3x1/2 − x3/2 = 0. which gives x and two equations for T and p (for details. one obtains from (4. . µA − µI − µe = 0. one has to add the restriction stemming from the law of mass action.92 Chemistry versus Physics: Chemical Reactions near Critical Points α x x = 1 − 2x . one starts from some x .71) and (4. 1 − 2x (4. Furthermore. . see [114]).71) p 2ε3 π 2 = ln x − ln x (x ) 3/2 − (x ) 3/2 If the dissociation reaction A I + e takes place. (4.68). using the wellknown expression for an ideal gas of A I electrons. Using (4. we put µ0 = µ0 . + x2 2 Neglecting the diﬀerence between the mass of atom and ion. Since one cannot solve these equations analytically.
) The same eﬀect was found for many other substances. Indeed. Depicted is the mole fraction of naphthalene dissolved in ethylene as a function of the pressure for various ﬁxed supercritical temperatures [149]. Figure 5. among others). Denote the ﬂuid component (solvent) by subscript 1. Denoting the solid phase s by superscript s and the ﬂuid phase by f. books ([147]. and the solid component (solute) by subscript 2. because of their many technological applications. The technological advantages of this phenomenon have aroused great interest in the chemical engineering community. there is a region of pressure where small changes in pressure result in a huge increase in solubility.1 shows that for every isotherm.1 Solubility in supercritical ﬂuids The properties of ﬂuids slightly above their critical points (“supercritical ﬂuids”) came to the attention of researchers more than a century ago. The thermodynamic analysis of this phenomenon (similar to that performed in Sec.1 displays a typical example of the phenomenon.Chapter 5 Comments on the Geometry of the Phase Diagram of a Reaction Mixture 5. this interest is reﬂected in the existence of a journal (Journal of Supercritical Fluids). where X2 is 93 . Figure 5. the slope of this increase becomes larger as T approaches TC .2 atm. [148]. 2. and hundreds of scientiﬁc and technological articles. (The critical temperature and pressure of the solvent are 11◦ C and 51. We concentrate here on the phenomenon of enhanced solubility of solids in supercritical ﬂuids (“supercritical extraction”) [70]. Moreover. supercritical chromatography [146] . we have X2 = 1. These include supercritical extraction [145].5) must be appropriate to the experimental setup which allows a solid to equilibrate with a supercritical ﬂuid at a given temperature and pressure. and the strong pressure (or temperature) eﬀect on the reaction rate [26].
we deﬁne f X2 ≡ X. p. p) − µf (T. Thus.94 Chemistry versus Physics: Chemical Reactions near Critical Points Fig. In the ﬂuid phase. The chemical potential of the solid component is denoted as µ2 and. 5.1) . 2 2 ∆µ (T. American Chemical Society. accordingly.1 Pressurenaphthalene composition phase diagram of the system ethylenenaphthalene for diﬀerent temperatures. p. We choose T. X as 2 2 independent variables. µs = µf for the equilibrium state. X) = 0 2 2 s using the fact that X2 = 1 in the solid phase. [149] with permission. p. (5. µf = µf (T. the mole fraction of the solid component. Reproduced from Ref. copyright (1948). X) . X) = µs (T. p. For all equilibrium states.
υ decreases with increasing pressure and a minimum . X decreases as p increases. (5. In addition. leading to the strong divergence of the lefthand side of this equation. (5. the solid compressibility is almost zero).3) This result is completely general.e.3) far from criticality.X (i. the ﬂuid mixture is ideal.X ∂µs 2 ∂p ∂µf 2 ∂X dp T dX = 0 T.equil. and will form the basis of the subsequent analysis.line < 0. d [∆µ (T.Comments on the Geometry of the Phase Diagram of a Reaction Mixture 95 Consider isothermal changes of pressure along the equilibrium line.2) in the following form ∂X ∂p = T. and the partial volume of the solute υ is much larger than its molar volume in the solid phase.p (5. p) + RT ln X. RT (5.2) Using the deﬁnition of the molar volume of the solid υs = (∂µs /∂p)T and 2 of the partial volume of the solute in the ﬂuid phase υ = ∂µf /∂p 2 one can rewrite Eq. υ >> υ s . (5. µf = µf (T. Since (∂υ s /∂p)T  << (∂υ/∂p)T.e.3) that near the critical points. the equilibrium isothermal curve X (p) has positive curvature for low pressures. υs − υ ∂µf /∂X 2 T.line T. p.4) υs − υ . ∂p)T.equil.. Therefore.X is negative but much larger in absolute value than (∂υs /∂p)T .0 one obtains from (5.6) T. Since (∂υ/∂p)T.equil..5) Far from criticality.X because υ does not depend on X in this region of pressure.X . Using the wellknown form of the chemical potential of dilute solutions. the derivatives along the equilibrium line coincide with the derivatives at a point. ∂ ln X ∂p = T. 5.line 1 RT ∂υ s ∂p − T ∂υ ∂p (5.1. One can already see from (5.line (5. ∂ 2 ln X ∂p2 = T.p . X)] = − ∂µf 2 ∂p dp − T. (∂ ln X. Therefore. i. the denominator goes to zero.equil. 2 2. It is instructive to examine Eq.3). which corresponds to the graphs shown in Fig.
10) .9) X ∂X 2 ∂X 0 In the last equality in (5. it was assumed that one can neglect the interaction between solute particles because of the initial assumption that the solution is very weak. X (p) is an increasing function unless it reaches the critical region. (5.7) ∂X T. ln this sense [150].p However. (5. gives p µs (T. Integrating the latter formula. Let us obtain the correct expression for the chemical potential µs of a solute in a binary mixture (calculations for manycomponent mixtures are quite analogous).p X = + (1 − X) (∂p/∂X)2 T. However. yields ∂υ υ 1 = υ + (1 − X) . (5.5) cause the eﬀective “interaction” between the solute particles to be important even for dilute solutions. One can immediately see that the chemical potential of a solute in dilute solutions cannot have the usual form (5. υ. X) = RT ln X + 0 (υ1 − υ id ) dp. and using (5. ∂µs RT = ∂x T. X). which follows from the equation of state p = p (T. this is incorrect near the critical point because longrange correlations (1.x .9).4) near the critical points. Let us now return to the problem of solubility in the critical region.X − (1 − X) v 0 ∂2p ∂X 2 dv. the partial volume is related [3] to the chemical potential υ 1 = (∂µs /∂p)T.p X 0 ∂X X RT ∂2υ ∂p + (1 − X) dX.8) with respect to X yields p ∂µs ∂υ 1 RT + dp = ∂X T.2) using the potential of the form (5. and eliminating υ2 . because one cannot satisfy Eq. we used the derivative of Eq.4). (5. Integration by parts and using simple thermodynamic transformations enables one to rewrite Eq.4) for the ideal gas as a reference system. (1. (5.v dX.v (∂p/∂υ)T.4).v (5. From this point onward. (5.9) in the ﬁnal form. the critical dilute solution is not “dilute”! In deriving Eq.8) Diﬀerentiating (5. T. p. with respect to X.7 ) and dp = (∂p/∂X)T.96 Chemistry versus Physics: Chemical Reactions near Critical Points in the curve X (p) occures when υ = υ s . Diﬀerentiating the GibbsDuhem relation [3] υ = υ 1 X + υ 2 (1 − X) for the partial volumes υ1 and υ 2 .
called the upper and lower critical end point (UCEP and LCEP). Comparing with Eq. and use has been made of υ = υ + −1 (1 − X) (∂υ/∂X)T. the denominator of Eq.p = 0.0 < 10−2 for all the solution of diﬀerent compounds in ethylene [151].11). where (∂p/∂υ)T = 0. Equation (5.1) and (1. there is a constraint (5.υ (∂p/∂υ)T. (TC − TC. Clearly.1) on the chemical potential of the solute in the ﬂuid phase. (5.3).0 to be small.p nary mixture. if one assumes that (∂p/∂υ)T.line ∂µf 2 ∂X ∂p ∂X T. we are faced with a critical point of a mixture that is very dilute and. Due to the contact between the ﬂuid phase and the solid phase.Comments on the Geometry of the Phase Diagram of a Reaction Mixture 97 In contrast to (5. while the latter allows one to satisfy the condition ∂ 2 µ/∂X 2 T.X .3) in the following form.p where υ is the molar volume. ∂X ∂p υ s − υ − (1 − X) = T. These critical points can be found from a solution of Eqs. This constraint decreases the number of degrees of freedom of the system. the derivative ∂µf /∂X vanishes on the liquidgas critical line of the bi2 T. We will show that (∂X/∂p)T.0 ) /TC.0 ) /pC.X and ∂ 2 p/∂υ2 T. (5. The former condition leads to the cancellation of the singularity RT /X in (5. The derivative (∂p/∂υ)T vanishes at the critical point of the pure ﬂuid.11) T. (5. and nowhere else. which terminates at the pureﬂuid critical point. let us consider Eq. On the other hand. (5. leading to a marked dependence of X on p in their respective critical regions. when (TC − TC.v ∂p ∂υ −1 T. At these two points.0 ) /TC. more volatile component.10).equil.10) describes the critical line of a binary mixture. the experimental setup considered here is special. It was found experimentally that the temperature TC of LCEP is located close to the critical temperature TC. However. and destroys the relevance of the critical line. and the system has only isolated critical points rather than a critical line.0 is small.2) which deﬁne the critical line of a binary mixture. (5. For example. and its form in the critical region.p and (∂υ/∂X)T.X are proportional to −X. (5. therefore.line diverges at two diﬀerent critical points in the phase diagram. these two points are UCEP and LCEP [70]. For typical phase diagrams. we conclude that there are two reasons for the divergence .X (5. Eq. In such cases.0 of the pure.4).11) vanishes. After this general comment. We ﬁrst rewrite Eq.equil. we also expect X and (pC − pC.p = − (∂p/∂X)T. is close to the critical point of the solvent.11) contains two “dangerous” derivatives.
98
Chemistry versus Physics: Chemical Reactions near Critical Points
of (∂X/∂p) near the LCEP (both (∂p/∂υ) and (∂µ/∂X) tend to zero), but only one reason for the divergence near the UCEP, where only (∂µ/∂X) goes to zero. A qualitative analysis has been performed separately for LCEP and UCEP [70] under the assumption of analyticity at the critical points. Another approach to supercritical extraction near UCEP and LCEP is based on a scaling theory of the critical phenomena [152]. The slope of the solu, bility curve (5.3) is deﬁned by the derivative ∂µf /∂X 2
T,p
∂µf 2 ∂X
= aπ + bτ γ + cX δ−1
T,p
(5.12)
where τ ≡ (T − TC ) /TC , π ≡ (p − pC ) /pC , and a, b, c are constants, while γ and δ are the critical indices [152]. The temperature dependence of (5.12) is deﬁned by the thermodynamic path in approaching the critical point. The experiments were performed at varying concentration. Therefore, it is natural to assume the smoothness of the chemical potential of the solvent and the solute as a function of pressure. Integrating (5.12), one obtains c (5.13) µf = aπX + bτ γ X + X δ + b1 τ + c1 π 2 δ where b1 and c1 are constants. For the experimental thermodynamic path π = const, the compatibility between other variables will be of the form τ ∼ X δ . Therefore, one concludes from (5.13) that asymptotically ∂µf 2 ∂X ∼ τ (δ−1)/δ .
T,p
(5.14)
In Eq. (5.12), we neglect the second term having asymptotic behavior τ γ compared to the third term proportional to τ (δ−1)/δ , since [152] (δ − 1) /δ < γ. In the asymptotic regime, one keeps the term having the lowest power of τ . If the LCEP is close to the critical point of the pure solvent, another singularity appears in the slope of the solubility, that is due to the “weak” singularity of the partial molar volume υ ∼ τ −α . Equation (5.15) has been veriﬁed experimentally [153]. (5.15)
Comments on the Geometry of the Phase Diagram of a Reaction Mixture
99
Our analysis shows that for p = pC and for diﬀerent isotherms, the slope of solubility as a function on τ on the loglog scale is a straight line independent of the system considered, with the value of this slope depending on the values of critical indices at diﬀerent distances from the critical point. A straight line with slope 1.05 seems to correspond to the experiment results. However, we need more precise measurements. From the large number of results related to supercritical ﬂuids, we bring here a partial list of possible applications of chemical reactions in supercritical ﬂuids [154]–[156]. There are certain characteristic features of supercritical ﬂuids which make these applications especially convenient, including their ability to dissolve various components, in particular, solids, the strong temperature and pressure dependence of the solubility, and the ability to use cheap inorganic ﬂuids. Those who drink decaﬀeinated coﬀee and beer probably know that considerable amounts of caﬀeine and hops are currently produced by CO2 supercritical extraction. Among many applications in the food and pharmaceutical industries, one can mention the extraction of oils, ﬂowers and fragrances. The rapid motion through a nozzle of an expanded supercritical ﬂuid with dissolved solids leads to the formation of very small monodisperse particles of prescribed size and morphology, which is useful in biotechnology and material science [157]. Other applications include enhanced oil recovery with supercritical CO2 [158], wetair oxidation of organics in supercritical water [159], coal liquefaction in supercritical toluene [160], catalytic disproportionation of nearcritical toluene [161], bitumen extraction from oil shale [162], activated carbon regeneration [163], and supercritical chromatography [164]. Technological applications (metallurgical, geochemical, geological) can be added to this list (see [165] for dozens of references). The main interest in diﬀusioncontrolled chemical reactions in supercritical ﬂuids is associated with the large reaction rates, which are several order of magnitude higher than in normal ﬂuids due to the small kinematic viscosity. Moreover, these rates can be strongly modiﬁed by a change of thermodynamic parameters. Modern technology also uses another types of chemical reactions, such as enzymatic reactions in supercritical CO2 , oxidation of hazardous materials in supercritical water, and catalyzed heterogeneous reactions.
100
Chemistry versus Physics: Chemical Reactions near Critical Points
5.2
Azeotropic points in reactive manycomponent systems
A typical temperaturecomposition projection of the T −x −x equilibrium surface (at constant pressure) is shown in Fig. 5.2. This mixture forms an azeotrope point A where [19] “distillation (or condensation) takes place without change of composition”. Like critical states, azeotropic states do not exist for ideal nonreactive mixtures.
Fig. 5.2 Liquidgas equilibrium in a binary mixture, showing the azeotrope point A. Reproduced from Ref. [18] with permission, copyright (1990), Springer.
The situation is quite diﬀerent for reactive mixtures, for which [166] an azeotrope exists even for ideal mixtures, and it is no longer deﬁned by the equality of phase composition. For reactive mixtures, the latter condition for ncomponent twophase systems is replaced by x − xi x1 − x1 = i ; i = 1, 2, . . . , n − 1 (5.16) ν 1 − ν T x1 ν i − ν T xi where ν T = i=1 ν i . A simple proof [166] of Eq. (5.16) is based on the conservation laws in a closed systems in which the ncomponent liquid mixture is vaporized
n
(dxi /dt) = (dxi /dt) = 0. . which proves Eq. The existence of two branches of the solubility curve can be explained by the lowering of the melting point by the addition or removal of one component.20) The righthand side of (5. . the composition of each phase remains constant.Comments on the Geometry of the Phase Diagram of a Reaction Mixture 101 at constant pressure or at constant temperature. dt dt dt dt i = 1. 2.3 Melting point of reactive binary mixtures Consider the phase diagram near the melting point for a twophase system consisting of a solid component AB which dissociates completely on melting and the liquid of the same composition. Therefore. . .19) During the azeotrope transformation. 5. Let us now consider a chemical reaction (dissociation) in the liquid phase: AB A + B. the vertical line at xB = 1/2 describes a solid phase. .18) dt dt dt permits one to rewrite Eq. . n − 1 (5. (5. This may be explained as follows. (5.19) reduces to xi − xi dξ = ν i − ν T xi dt dυ dt −1 . The material balance for component i gives dξ d (υ xi ) d (υ xi ) + = vi . 2. . (5. The solubility curve becomes more rounded with an increase of the degree of dissociation.3 by the solid curves. . i.e.20) does not depend on i. .21) The solubility curve will then be rounded at T = T + . where T + is the melting point. 5. 2. Let . and the dashed lines relate to liquid phases with concentrations larger and smaller than 1/2. as shown in Fig.17) in the following form υ dxi dxi dξ dυ +υ = (ν i − ν T xi ) − (xi − xi ) . n − 1. The overall material balance dυ dξ dυ + = vT (5. n − 1 (5. i = 1. the solubility lines meet at T = T + . The T − xB projection of the phase diagram is shown in Fig. i = 1. 5. . .16).17) dt dt dt where υ and υ are the molar volumes of the two phases and ξ is the extent of reaction. . (5.3.. and (5.
therefore. xB of substances A and B and detailed compositions yA . yB . 5. one obtains yA = 1 − f (T ) ξ yB = − [1 + f (T )] . Reproduced from Ref.22)–(5. The chemical potential of the solid depends only on temperature and. 2 2 (5. 2 (5.22) The x. yAB of A.3 Liquidsolid solubility curves for compound AB with (solid curve) and without (dashed curve) chemical reaction. [18] with permission. B and AB (yA +yB +yAB = 1).23) Using Eqs. y variables are connected by yB + yAB xB = . Springer. (5. which plays the role of the chemical potential “bath” for the solution.24) . The chemical potential of AB in the solution must be equal to that in the solid. yAB is a function of T . the liquid solution has the total compositions xA . Let us now introduce the small parameter ξ to describe the left part of the melting curves (xB < 1/2): ξ=2 1 − xB .23). copyright (1990). 2 2 (5. and the relation yAB = f (T ). one may assume that for small degrees of dissociation. 1 + yAB 1 − f (T ) ξ + [1 + f (T )] .102 Chemistry versus Physics: Chemical Reactions near Critical Points Fig.
(5. the change in fugacity of the solvent B (which. then. (5. 5. one can write the law of mass action for the reaction (5. yA yB = K (T ) . and a gas phase (vapor of the pure solvent B with the components A and C assumed to be nonvolatile). 2 2 (5. Similar results are obtained from more general considerations.26) near T = T + yields ξ 2 = 2Q T + dR dT − dQ dT T =T + T+ − T + ··· .28) In this case the melting lines (the dashed curves in Fig. ξ = 0. Q (T + ) = R (T + ) . including the more complicated compounds Am Bn [168] and the strictly regular solution model [169] (instead of an ideal model). Therefore. one can rewrite the latter in the form ξ2 = − 4K (T ) [1 + f (T )] 2 ≡ Q (T ) − R (T ) . (5. for the low saturated vapor pressure. 5.27) and (5. The reason for the diﬀerence between Eqs. coincides with the partial . the small addition of either component will not change the equilibrium temperature. i. K (T ) = 0 . Further development of these ideas has been carried out by Krichevskii et al. (5. [1 − f (T )]2 [1 + f (T )] 2 (5.21) as follows.3). (5. If there were any appreciable dissociation of AB into its constituents. Then. a solvent B with a dissolved substance A). [171]. the melting point is a true maximum of a solid curve [Eq. ξ2 = dQ dT 2 T =T + T+ − T 2 .. We now turn to the case of a nonreactive mixture.27) describes a melting line in the presence of a chemical reaction (solid curve in Fig.3) meet at a point. Expanding Eq.28) was explained back in 1892 [167]. (5. (5. than the solution is ideal. For a nondissociated system. a liquid phase (solution saturated with this component and extremely dilute with respect to component C). They considered the inﬂuence of a third component on the thermodynamics of the threephase equilibrium among a solid phase (compound ABm .26) near T = T + is of second order.25) Inserting (5.26) At T = T + . and the ﬁrst term in the series expansion of Eq. Therefore.27)] and not a kink [Eq.27) Equation (5.Comments on the Geometry of the Phase Diagram of a Reaction Mixture 103 If dissociation is slight. in contrast to the nondissociated solution.25).e.28)].24) into (5. [170].
and the denominator of Eq. (5. Combining this equation with the GibbsDuhem equation [3] nA d (ln aA ) + nB d (ln aB ) + nC d (ln aC ) = 0 one obtains d ln (γ B ) = − nC d (ln aC ) nB − mnA (5. ABm A + mB.(ABC) nC = ln =− (5. the compositions of the solid and liquid phases are equal. (5.0 and nA. Eq.33) (5.(AB) nB where pB. (5. Addition of component C to solution must .35) reduces to Eq. in diﬀerential form. Therefore. the law of mass action for the reaction (5.30) The chemical potential of the compound ABm in solution. At m = 0.0 For the chemical reaction of the form (5.34) yields γ B. (5. which coexists with a solid.(ABC) are the partial pressures of the solvent in binary and ternary solutions at the same temperature. At the melting point in Fig. (5. Consequently.34) (5. is given by [172] pB.(ABC) nC =− (5. d (ln aA ) + m d ln (γ B ) = 0.0 − mnA. where two branches of the solubility curves meet.35) ln γ B.104 Chemistry versus Physics: Chemical Reactions near Critical Points pressure) due to the addition of a small amount of C at constant pressure and temperature.29). one can replace the activity aC by the numbers of moles nC . we replace nB and nA by the total number of moles of the components nB.30).29) ln pB. and n is the number of moles of the components.(ABC) pB.(AB) pB.30) can be written as aA γ m = K B or. 5. is ﬁxed and depends only on the temperature.32) (5. Integrating Eq. the diﬀerence nB − mnA remains constant. Let us now allow the chemical reaction (dissociation) in the solution.35) reduces to zero.(AB) nB.(AB) and pB.3 (drawn for m = 1).0 in both liquid and solid phases.31) If the solution is very dilute with respect to C. The fugacity γ of the nonvolatile component A can be replaced by the activity a.
5. Assume that A and B form a .35) tends to −∞. pB. there is a line of double critical points which divides the phase separation surface into two parts with UCST and LCST. respectively.(ABC) must decrease to zero and increase to inﬁnity. there are more than two coexistence phases. and on the right branch. and is miscible in all proportions below LCST. The main interest in the phase diagrams of the type shown in Fig. 5. On approaching the melting point along the left branch. Such phase diagrams have been found in binary mixtures of glycerolguaiacol [174] and of water with nicotine [175]. xB .(ABC) cannot exceed the saturated vapor pressure of the pure solvent. the righthand side of Eq. the ordered twophase state exists in the limited temperature interval between LCST and UCST [173] shown in Fig.4 Double critical point The double critical (or hypercritical or reentrant) point (DCP) results from merging the phaseseparation region with the upper (UCST) and lower (LCST) critical solution points. and other organic amines and hydroxyethers. separating the closed ordered region from the singlephase disordered region. We present here two possible descriptions of the double critical point. A binary mixture possessing a LCST separates into two phases as the temperature is raised above the LCPT. In contrast. respectively.(ABC) on the left branch may not be less than the dissociation pressure of the compound in solution.4 together with the DCP at a certain concentration x0 . In some manycomponent mixtures. the mixture is miscible in all proportions above UCST. pB. Both these values (zero and inﬁnity) are physically impossible.Comments on the Geometry of the Phase Diagram of a Reaction Mixture 105 decrease the value of pB. such as propanolwaterNaCl.(ABC) on the left branch (nB > mnA ) and increase it on the right branch (nB < mnA ). it tends to +∞. whereas approaching along the right branch. creating thirdorder and higherorder critical points. 5.4 comes from crystals of the technologically important Rochelle salt group [177]. By merging these two points into one “double critical point”. The immediate vicinity of the melting point is of special interest. the coexistence curve has a closedloop form. In fact. Correspondingly. (5. and xC . based on the meanﬁeld model and on the phenomenological approach. pB. Therefore. lutidine [176]. Consider [117] a ternary mixture with mole fractions xA . In multicomponent systems (with three and more components). which undergoes a chemical reaction A + B C.
2. T ) of each species.106 Chemistry versus Physics: Chemical Reactions near Critical Points Fig. the LCST will appear leading to the closedloop coexistence curve. in the presence of a chemical reaction. the coexistence surface for this mixture contains [178] a line of UCST’s that decreases to lower temperatures as the concentration of C is increased. T ) + ln xB + (W/kB T ) xA (1 − xB ) . [173] with permission. copyright (1986). Reproduced from Ref. highly directional and energetically favorable bond in becoming C with a positive interaction of mixing W > 0. 4. As it was described in Sec. A µB = µ0 (p. Turpion Publications. for the phase equilibrium surface. However.37) kB T xA − xB xB .36) (p. T ) + ln xC − (W/kB T ) xA xB . (5. as will be shown. B µ0 B (5. along with UCST. Therefore.2. in the meanﬁeld approximation. and all other interactions included in the standard chemical potentials µ0 (p.36) are symmetric with respect to interchange of A and B. µC = The nonlinear terms in (5. T ) + ln xA + (W/kB T ) xB (1 − xA ) . 5. In the absence of i a chemical reaction. the chemical potentials of the species have the following form µA = µ0 (p. xA = xB and xC = xC . one gets 1 W xA = ln . leading to the coexistence phases xA = xB .4 Lines of the upper (UCST) and lower (LCST) critical solution points which merge at concentration x0 .
one approaches the critical point deﬁned by Tcr = W/2kB .39) deﬁne the coexistence surface of the reactive system in the plane (kB T /W.39) where ∆H 0 and ∆S 0 are the enthalpy and entropy of the chemical reaction which are assumed to be constant. Let us now consider the constraint imposed by the chemical equilibrium.4. [117]. and d correspond to kB T /W = 0.37) and (5. The set of isotherms in the xA − xB plane is shown in Fig.3 and 0. Curves a. 5. These equations have been solved numerically [117] for ∆H 0 /W = −10 and several values . 5. xC. With a decrease in temperature. whereas with increasing temperature.cr = xB. b. 0.38) and (5.cr = 0. xC = xA xB exp − (5.36) deﬁne the equilibrium expression for xC . xA. c.5 Isotherms of the phase equilibrium surface in a meanﬁeld model.5. American Institute of Physics. copyright (1989).45. 0. respectively. µA + µB − µC = 0 W ∆S 0 ∆H 0 + (xA + xB − xA xB ) exp kB T kB kB T (5.38) Equations (5.Comments on the Geometry of the Phase Diagram of a Reaction Mixture 107 Fig. Reproduced from Ref. xA ) for ﬁxed values of ∆H 0 /W and ∆S 0 /kB .cr = 1/2. the phase equilibrium surface approaches the xA and xB axes. The solutions of Eqs. ( 5.2.
The loops get smaller with decreasing (−∆S 0 /kB ). Fig. 40. upon lowering the temperature still further. the formation of C becomes favorable. 26.108 Chemistry versus Physics: Chemical Reactions near Critical Points of ∆S 0 /kB . [117]. and the phases become miscible again. 5.1. and 24. 5. x) Ψ4 + C (∇Ψ) + · · · . 2 (5. one obtains the closedloop coexistence curves created by LCST and UCST with their ratio decreasing as −∆S 0 /kB decreases. However. American Institute of Physics.5. x) Ψ2 + B (T. Reproduced from Ref. As usual.40) . In this manner. 24. 30. the phenomenological description of the vicinity of the DCP is given by the LandauGinzburg expansion for the free energy Φ in the order parameters Ψ and its spatial derivatives: Φ = Φ0 + A (T.6) shows that A − B repulsion results in phase separation as the temperature is lowered. copyright (1989).6 Closedloop coexistence curves for ∆H 0 /W = −10 and −∆S 0 /kB = 60. The resulting phase diagram (Fig.
As one can see from this ﬁgure. 5. Fig. whereas for x < x0 . 5. correlation radius and the . is usually applied also to the function A (T. does not change the critical indices. (5. The assumption of analyticity.41) quadratic in (T − Tmin) /T0 . with A > 0 for a disordered phase and A < 0 for an ordered state. x) = a (T − TC ).41) A (T.Comments on the Geometry of the Phase Diagram of a Reaction Mixture 109 where the coeﬃcient A changes sign on the line of phase transitions TC (x). instead of the usual linear terms. [173] with permission. it is convenient to expand the function A (T. 2 T − Tmin + ··· . the minimum of A (T ) occurs below the horizontal axis. There are two phase transitions for x < x0 . x) = A0 (x) + A2 (x) T0 The existence of terms in (5. namely. Due to the existence of minima. copyright (1986). and no phase transitions for x > x0 . Reproduced from Ref. intersecting it at LCST and at UCST. x) . since the nonmonotonic behavior of TC (x) leads to a nonmonotonic form of A (T ) rather than the linear function [173] (Fig.7). the curve A (T ) is tangent to the temperature axis at the point T0 .40).7 The nonmonotonic dependence of the ﬁrst coeﬃcient in the LandauGinzburg expansion as a function of temperature. It is obvious that the latter assumption is not applicable to the case of DCP. which deﬁne the singularities of the compressibility. used in (5. Turpion Publications. for x = x0 . x) around the minimal temperature Tmin and not around TC as is usually done. A (T.
we are unaware of any experiments of chemical reactions near DCP. the exponent describing the powerlaw divergence in temperature of the viscosity nearly doubles as the concentration approaches the doublecriticalpoint concentration. a decrease of several orders of magnitude in the anomalies of the heat capacity was found in all systems with the DCP. the anomalies of the quantities containing the derivatives of the coeﬃcient A with respect to temperature. However.110 Chemistry versus Physics: Chemical Reactions near Critical Points quantities related to them. There are also experimental data describing light scattering and sound propagation near the DCT [181]. Indeed. decrease near the DCP. water and heavy water as a function of temperature show [180] that near DCP. For example. there are some discrepancies between the theory and experiment [173] which makes specially interesting the microscopic description of the DCP and the closeloop coexistence curves which have been carried out in the context of latticegas [179] and decoratedlatticegas [96] models. However. the measurements of the shear viscosity of ternary mixtures of 3methylpyridine. which are of special interest. the jump of the heat capacity ∆C = −T ∂ 2 A2 /2B /∂T 2 at LCST and UCST decreases proportional to the square of the relative distance between these points. It should be noted that the anomaly of the kinetic coeﬃcients are also diﬀerent from the usual critical point. . For example. However. near DCP.
The total attenuation per wavelength αλ near the critical point can be written as αλ = (αλ)cr + (αλ)chem .2) 2 1 + (ωτ chem ) which depends on the nearness to the critical point. expressed by the (αλ)cr term in Eq. This theory is based on an assumption going back to Laplace [185]. (2.1) contains the assumption of the additive contributions of (αλ)cr and (αλ)chem . has been performed [184].77 ). which afterwards will relax to the initial equilibrium state with characteristic relaxation time τ chem . The calculation of the critical contribution to the sound attenuation.1 Ultrasound attenuation in nearcritical reactive mixtures The idea of using the attenuation of sound to study the rate of dissociation 2NO2 was proposed in 1920 by Albert Einstein of the reaction N2 O4 [182]. The latter deﬁnes the sound attenuation due to a chemical reaction ωτ chem (αλ)chem = A0 (6.Chapter 6 Sound Propagation and Light Scattering in Chemically Reactive Systems 6. (6. The assumption of independence of these two contributions is based on the following picture [183]: the reaction takes the system from equilibrium to the steady state. which come from the critical ﬂuctuations and chemical reaction. In line with this comment. it was assumed [184] that the change of 111 . since τ chem is proportional to the equilibrium derivative of the extent of reaction ξ with respect to the aﬃnity A. respectively.1) neglecting the small background contribution due to viscosity and thermal conductivity. τ chem ∼ (∂ξ/∂A)eq [36]. Equation (6. who corrected Newton calculation showing that the sound velocity in air should be calculated using the adiabatic rather than the isothermal compressibility.
produces a shift in the critical temperature.4) into (6. but not in the critical concentration. Simple thermodynamic analysis shows [184] that the sound velocity u remains constant at the critical point. The lowfrequency expansion of Cp (ω) gives C (ω) ≈ C (0) + ωC (0) = C (0) (1 + iωτ ) (6. and Cp is the speciﬁc heat.3) and assuming ωτ << 1. has the following thermodynamic temperature dependence C (0) ≈ t−α0 (6.3) −1 where g = TC (∂S/∂p)λline VC is a dimensionless constant. The above analysis has been performed at nonzero frequency ω for an input pressure signal with time dependence exp (−iωt).3) determines the complex sound velocity u (ω) as a function of the complex speciﬁc heat Cp (ω). induced by concentration ﬂuctuations. where Cp = C1 + C (0). where TC (p) is the pressuretemperature critical line.7) .112 Chemistry versus Physics: Chemical Reactions near Critical Points pressure caused by the propagating sound. one obtains for the negative imaginary part of the critical velocity Im (uC ) (the sound attenuation per wavelength αcr λ).6) where α0 is the critical index and t = ∆T /TC = (T − TC ) /TC is the reduced temperature. or p and τ ≡ T − TC (p). where the critical diﬀusion coeﬃcient is D (ξ) = kB TC /6πηξ with ξ ≈ t−ν and η ≈ t−z0 . with small corrections ∆u near the critical point. (6. u ≡ uC + ∆u = uC + g 2 u3 C 2TC Cp (6. αcr λ = C (0) u3 C 2TC g Cp 2 ωτ . the characteristic relaxation time for the ﬂuid is γ (t) = 2Dξ −2 = γ 0 (∆T ) νz0 (6.4) where τ = −iC1 (0) /C (0) is an eﬀective mean relaxation time. Eq. The relaxation rate of concentration ﬂuctuations of wavenumber k is k 2 D. Therefore. The function Cp (ω) can be divided into the noncritical frequencyindependent background component C1 and the critical part C (ω). (6. Neglecting the changes in concentration.5) The critical part of the speciﬁc heat C (0) . Therefore. Upon inserting (6. Dynamic scaling theory has been used to determine the frequency dependence of Cp (ω) from the known temperature dependence of the thermodynamic heat capacity Cp . we are left with two thermodynamic variables p and T . u (T = TC ) ≡ uC .
ω) . (6. the dependence of Eq.9) Replacing C (0) in Eq. for γ < ω. We will here use the following simpliﬁed version of these equations [194] . Agreement between theory and experiment was also found for polypropyleneglycol and polyethyleneglycol mixtures [188]. (6.10) by (uC f ) predicts a linear proportionality between αcr f −2 and f −1−α0 /νz0 .10) −1 where f = ω/2π is the frequency. cyclohexanenitroethane[191]. γ is replaced by −iω so that the frequencydependent speciﬁc heat at the critical point is C (γ = 0. ethanoldodecane and methanolcyclohexane mixtures [190]. Precisely this dependence had been observed in the measurements [184] of the sound attenuation in 3methylpentanenitroethane at various frequencies ranging from 16. and using α0 << νz0 . (6.8) According to the dynamicscaling requirement [186]. Equation (6. one can use the linearized hydrodynamic equations for the analysis of the sound velocity and attenuation coeﬃcient in ﬂuids. (6. Multiplying Eq. and triethylaminewater. ω = 0) = B γ0 γ α0 /νz0 . C (γ. 6. ω) = B γ0 −iω α0 /νz0 = B exp iπα0 νz0 γ0 ω α0 /νz0 .Sound Propagation and Light Scattering in Chemically Reactive Systems 113 where γ 0 is constant. which give rise to small perturbations in the system. (6.2 Hydrodynamic analysis of the dispersion relation for sound waves Sound waves are oscillating movements of small amplitude. Therefore. The linearized system of hydrodynamic equations for reactive systems was obtained in Chapter 2. isobutoxyethanolwater system [189].5) by the imaginary part of C (γ = 0.8) on γ disappears when. one gets ﬁnally αcr λ π 2 α0 g 2 u2 B C 2 2TC Cp (Tc ) γ0 2π α0 /νz0 f −α0 /νz0 (6.5 to 165 MHz. methanolhexane [192].7) enables us to eliminate t in favour of γ as the variable that indicates how close the system is to the critical point. methanolcyclohexane [193].
14) p= ρ+ ξ.19) υ.15) T = ρ+ ξ.16) In contrast to Eqs.61)–(2.12) (6. the independent variables in the above equations are the velocity v.S ξ+ ρ. ξ. (6. the deviations from equilibrium values of entropy s.17) The wave vector K is written K = k + iγ. (6.S ∂A ∂S ∂p ∂S ∂T ∂S S ξ.64).S ∂ξ ∂υ A.S ξ+ ρ.ρ (6.11)–(6.16) yield the following expressions for k and γ ω2 G ω γ= (6.S ∂p ∂ξ v. After some additional simpliﬁcations [194].12).ρ (6.13) ∂A ∂ξ ∂p ∂ξ ∂T ∂ξ ξ+ ρ. Kinetic coeﬃcients are taking into account only in the NavierStokes equation (6.18) k= . density ρ and extent of reaction ξ.ρ (6. ω) = d3 r dtx (r. Eqs.114 Chemistry versus Physics: Chemical Reactions near Critical Points ∂ρ = −ρ0 div (v) ∂t ρ0 ∂v = −grad (p) + η∇2 v ∂t η + ζ grad (div (v)) + ρ0 L1 grad (A) + 3 ∂ξ = −L0 A − L1 div (v) ∂t A= ∂A ∂ρ ∂p ∂ρ ∂T ∂ρ ρ+ ξ. (2.S S ξ. where k is the propagation vector and γ is the attenuation coeﬃcient.11) (6. t) exp (iωt − iKr) .S (6. To derive the dispersion law. C 2ρ0 C 3 where 1 ∂p C ≡− 2 ρ0 ∂υ 2 + ξ. we take the Fourier transform of the hydrodynamic variables ˆ X (K.S 1 + ω2 τ 2 ω2 τ 2 1 + ω2τ 2 + 2ρ0 L1 ∂p ∂ξ (6.S S.S .
This theory contains viscoreactive cross coeﬃcients L1 .s τ ρ0 (1 + ω2 τ 2 ) (6.S (6.A (6. Starting from the same hydrodynamic equations obtained in Chapter 2. However. instead of variables ξ.T kT 2 kp ∇ T + ∇2 p T0 p0 . div (v) and S1 = S − (∂S/∂ξ) ξ. υ.p kp 2 ∇ p p0 (6.91)–(2.p ∂ξ ∂T T− p. which allows us to consider the vicinity of the critical points without the L1 terms [195].21) Some limit cases of these general equations has been considered [194].20) 2τ L1 4η +ζ + 1 + ω2τ 2 3 −1 and the relaxation time τ is given by τ ≡ L0 ∂A ∂ξ . Now let us turn to the slightly diﬀerent hydrodynamic approach.25) ∇2 ξ + p.x T /T0 − αT p/ρ0 .A ∂ξ ∂p p T. which link the chemical reaction to the momentum ﬂow.24) λ ∂S1 kT = ∇2 T + D ∂t ρ0 T 0 T0 +D ∂S ∂ξ ∂A ∂ξ ∇2 ξ + T. we use the set ξ. which is important for a weak electrolyte solution [194]. The hydrodynamic equations (2.94) may then be rearranged to give ∂ρ = −ρ0 div (v) ∂t ρ0 ∂div (v) = −∇2 p + ∂t 4 η + ζ ∇2 div (v) 3 (6. we replace them by the statistically independent variables as it was done there.Sound Propagation and Light Scattering in Chemically Reactive Systems 115 G≡ ∂p ∂ξ + υ. div (v) and φ = Cp.22) (6.23) ∂ξ kT 2 kp = D ∇2 ξ + ∇ T + ∇2 p ∂t T0 p0 + 1 ξ− τ T. p. p.S ∂ξ ∂υ A.
t) = ρ0 It is easy to show [195] that (∂A/∂ξ)T. and ξ (K.p kB T υ d β 1 (K.p = L (∂A/∂ξ)T.p 2 γ T ∂ξ kT D K − . t) = exp − DK 2 + τ −1 t T. (6.28) (6. one obtains β 1 (k.p ρ 0 Numerical estimates indicate that (6.p M = 1/2 ( ∂A )T . 1/2 (6.A − Cp.p is the relaxation time. t) (6.25) in which the pressure is kept constant.ξ .ξ τ T.ξ . t) and β 2 (k. kT D K − T Cp.ξ τ T. t) ξ (−K. t) dt β 2 (K. K > (6.27) where the 2 × 2 matrix M is given by 1/2 ( ∂A )T . t) β 2 (K. t) . Since only the ﬂuctuations of concentration and entropy become anomalously large near the critical point.ξ ρ0Cp.ξ (6. Introducing both normalized and statistically independent spatial Fourier transform variables.p and γT ≡ Cp. t) = ρ0 kB υCp.24)–(6. β 2 (k.116 Chemistry versus Physics: Chemical Reactions near Critical Points −1 where τ T. t) . Cp. t) =M β 1 (K. we may use the simpliﬁed macroscopic equations (6. t) are then eﬀectively decoupled.ξ τ T. The variables β 1 (k.31) The modecoupling method has been used [102] in the analysis of the propagation of sound in a nonreactive binary mixture near the critical .p 2 γ T λK 2 γ ∂ξ + T .ξ 1/2 S1 (k.p τ T. DK 2 + τ −1 T.p T Cp.p ξ (K) ξ (−K) .26) ξ (K.A − Cp.30) is satisﬁed for K > 104 cm −1 .29) The nondiagonal elements of the matrix M are very small and may be neglected if the propagation number K satisﬁes the inequality 1/2 Cp. and all other notation has been explained in Chapter 2.30) λ − DCp.
Sound Propagation and Light Scattering in Chemically Reactive Systems
117
point. The general formula for the complex sound attenuation coeﬃcient α (ω) was obtained in [102], α (ω) ≈ ω 2 ˆ × dK
∞ 0
∂ ln ξ (K) ξ (−K) ∂T
2
dt exp −2 DK 2 + τ −1 t exp (iωt) . T,p
(6.32)
The arguments that lead to Eq. (6.32) remain unchanged when a chemical reaction occurs. The existence of a chemical reaction results in an additional mechanism for the relaxation of the composition back to its equilibrium value, which results in the appearance of the additional τ −1 T,p term in Eq. (6.31). Then, in a reactive system, ω2 α (ω) ≈ ˆ 2 D (K) K 2 + τ −1 (K) − iω T,p ∂ ln < ξ (K) ξ (−K) (6.33) ∂T where the Kdependence of the diﬀusion coeﬃcient, D (K), and of the relaxation time τ −1 (K) = Lχ−1 (K) (which reduces to L (∂A/∂ξ)T,p for T,p K K = 0) have been taken into account. For nonreactive systems, relaxation is associated with diﬀusion. Just as diﬀusion processes show critical slowingdown with time scale τ D ∼ −1 −3ν ∼ [(T − TC ) /TC ] , the sound attenuation will show critical Dk 2 behavior over the same time scale. On the other hand, if the chemical reaction occurs on a faster time scale than diﬀusion and shows critical slowingdown, then the sound attenuation will relax on a time scale × dK ∼ [(T − TC ) /TC ] . Because γ < 3ν, the chemistry τ C ∼ (∂A/∂ξ)T,p is faster and will dominate the diﬀusion which can be seen in sound experiments. However, the diﬀusion coeﬃcient D (k) is wavenumberdependent and, according to Eq. (3.32), at high frequencies, D ∼ k. The characteristic −1 ∼ k −3 . This will occur on diﬀusion time scale will then be τ D ∼ Dk 2 a faster time scale than the chemical reaction, and the crossover will occur by diﬀusioncontrolled sound attenuation. Finally, by plotting the dimensionless sound attenuation coeﬃcient α (ω) /ω2 versus ωτ at various temperatures, one can see whether τ D ∼ −3ν −γ or τ C ∼ [(T − TC ) /TC ] yields a better dynamic scal[(T − TC ) /TC ] ing function. The calculated change in sound velocity ∆C/Cω 2 for ωτ << 1 1/2 in the nonreactive case and to ωτ in the reactive is proportional to (ωτ ) case [195] .
−1 −γ 2
118
Chemistry versus Physics: Chemical Reactions near Critical Points
6.3
Light scattering from reactive systems
Light scattering as a probe of chemical dynamics is able to detect spontaneous ﬂuctuations which are drastically increased near the critical points. Like sound, this is essentially a nonperturbative method which, by using the microscopic polarizability of molecules, provides a complimentary tool to other methods of studying chemical kinetics. Measuring the angular dependence and the shift in the frequency of the light scattered by a system allows one to ﬁnd both the spatial and the time dependence of the ﬂuctuations. The electromagnetic wave impinges upon a molecule inducing the dipole moment proportional to the polarizability. For a macroscopic continuum, the summation over the polarizabilities over individual molecules gives the macroscopic dielectric constant ε = ε0 + δε (r, t), where the latter part comes from the spacetime ﬂuctuations of local thermodynamic variables. Assuming local equilibrium of a reactive binary mixture described by the reduced entropy S1 ≡ S − (∂S/∂ξ)T,p ξ, pressure p, and the extent of reaction ξ, one obtains δε (r, t) = ∂ε ∂S δS1 +
p,ξ
∂ε ∂p
δp +
S1 ,ξ
∂ε ∂ξ
δξ.
p,S1
(6.34)
Light scattering theory assumes [196], [197] that the intensity I of scattered light having wavenumber K0 at point R0 is described by the expression I (r0 , K, ω) =
4 N I0 K0 2 2 sin φ δε (K, ω) δε (−K, 0) 16π 2 R0
(6.35)
where I0 is the intensity of the incident light, N is the number of molecules in the scattered volume, ω is the shift in the frequency of the scattered light, K is the change in the wavenumber in the medium of the scattered light, and φ is the angle between the electric vector of the incident wave and the vector R0 . Inserting the space Fourier transform of Eq. (6.34) into (6.35) gives the light intensity as a function of timedependent correlators of the statistical independent thermodynamic variables S1 (K, t), p (K, t) and ξ (K, t). The hydrodynamic equations express the latter in terms of the static thermodynamic ﬂuctuations of S1 (K) , p (K) and ξ (K). There are two major mechanisms for the relaxation of ﬂuctuations [197]: one arises from transport properties, and the other is due to the local changes induced by chemical reactions. The ﬁrst mechanism depends upon spatial gradients, i.e., the line width of the scattered light is proportional to the square of the wavenumber K 2 , whereas for the second mechanism, the
Sound Propagation and Light Scattering in Chemically Reactive Systems
119
line width is independent on K. Therefore, in principle, one can measure chemical relaxation times by the extrapolation K → 0. However, in a real ﬂuid, the kinetic and chemical modes are coupled. This coupling can be described by the hydrodynamic equations (6.22)–(6.25), which, according to (6.35), have to be written in the ω − K Fourier form, β 1 (K, t) β 1 (K, t) β 2 (K, t) d β 2 (K, t) (6.36) β (K, t) = M β (K, t) dt
3 3
β 4 (K, t)
β 4 (K, t)
where the independent variables β i (K, t) are div (v) and the normalized forms of deviations of pressure, reduced entropy, and the mass fraction of the species with the highest partial speciﬁc enthalpy from their equilibrium values [198], β 1 (k, t) = β 2 (k, t) = β 3 (k, t) = β 4 (k, t) = ρ0 kB υCp,ξ
1/2
S1 (k, t) ,
1/2
ρ0 (∂A/∂ξ)T.p kB T υ (∂v/∂p)S,ξ kB T υ
1/2 1/2
ξ (K, t) , (6.37) p (k, t) ,
ρ0 div (v) kB T K 2 υ and the 4 × 4 matrix M has the following form λ M11 = K 2 + γ T τ −1 , T,p ρ0 Cp,ξ M13 = M31 = C1 K 2 + C2 τ −1 , T,p T Cp,ξ (∂A/∂ξ)T.p
1/2
M12 = M21 = kT D M22 = DK 2 + τ −1 , T,p
K 2 − γ T τ −1 , T,p (6.38)
1/2
M23 = M32 = C3 K 2 + C4 τ −1 , T,p M33 = C5 K 2 + (rM Cp,A /Cp,ξ ) τ −1 , T,p M34 = M43 = Kc∞ , M44 = C6 K 2 , M14 = M24 = M41 = M42 = 0
The four roots of the dispersion equation have been found [199].40) then the dispersion relation reduces in zero approximation to Eqs.p ∂ξ ∂T p. ρ0Cp. (6. 2 c2 0 (6.ξ τ T.p τ −1 << Kc∞ .41). z3.p (T Cp. when the damping terms in the matrix M are much smaller than the frequency of the sound wave Kc∞ .p 2 d2 = DK 2 + τ −1 .p (6.43) while Γ in Eq.2 = 1 2 (d1 + d2 ) ± (d1 − d2 )2 + 4d2 12 1/2 . ρ0 Cp. T. z1.2 are related to the nonpropagating modes and deﬁne the Rayleigh line in the scattered light. 1 ρ0 4 η + ς K 2 << Kc∞ 3 (6.41) and the complex roots z3.p ± iKc∞ .42) The following notation has been introduced in Eq.ξ K2 + .p ∂A ∂ξ d12 = DkT K − T τ T.42) is the same combination of transport coeﬃcients.ξ ) −1 (6. . (6. T. (6. which is the damping factor of the Brillouin lines for a twocomponent nonreactive system.39) and the inverse chemical reaction time τ −1 satisﬁes T.26)– (6. d1 = λ Cp.ξ Cp.A − Cp.120 Chemistry versus Physics: Chemical Reactions near Critical Points where all notations can be found in [198]. (6. However. λ K 2 << Kc∞ . The two real roots z1.4 .ξ DK 2 << Kc∞ . relate to two propagating modes deﬁning two symmetric Brillouin lines.27) with additional ﬁrstorder coupling between the pressure ﬂuctuations and ﬂuctuations in entropy and concentration.A T.4 = ΓK 2 + 1 c2 − c2 −1 ∞ 0 τ T.
contained in a spherical cavity of radius a.Sound Propagation and Light Scattering in Chemically Reactive Systems 121 Let us check the restrictions arising from inequalities (6. or “ﬂuid clustering” [202]. appreciable enhancement up to 2. This phenomenon. The measurement of partial molar volumes of naphthalene in ethylene yields [201] values that are extremely large and negative.39) are always met for suﬃciently small K.5 times the bulk density was seen for supercritical C2 H6. the conditions given in Eq. Conditions (6. (6.40). polarizes the solvent and produces a reaction ﬁeld. A solute with a dipole moment µ. K must be larger than some minimal value which depends on τ T.40 ). clearly show statedependent local density enhancements in dilute supercritical ﬂuid solutions. The magnitude of this enhancement was analyzed in terms of the local CO2 density. Spectroscopic techniques permits us to obtain qualitative information about the local environment around solute molecules.39) are satisﬁed even for the maximum scattering wavenumbers (Kmax ≈ 2 × 105 sm−1 ) encountered in light scattering experiments. ﬂuorescent.p . and its photophysics is well known. Extensive measurements. Using pyrene as a spectroscopic probe.39) and ( 6. CHF3 and CO around pyrene [200]. the inverse of the chemical reaction time must be of order 109 s−1 or smaller. For typical magnitudes of the transport coeﬃcient and the speed of the sound. and dissolved in a solvent with static dielectric constant and refractive index n. To be in the region of wavenumbers used in light scattering experiments. in which the density of solvent around a solute molecule is higher that the bulk density. 6. To satisfy condition (6. called “molecular charisma” [201]. has been studied intensively using spectroscopic measurements. implying that the local environment near the solvent molecules is very diﬀerent from the bulk properties of the ﬂuid. based on the Onsager reaction ﬁeld theory [205]. This shift ∆ν between vapor and solution . or “density augmentation” [200]. using steadystate and timeresolved ﬂuorescence of dissolved pyrene in supercritical CO2 [203] and critical water [204].4 Inhomogeneous structure of nearcritical reactive systems The high compressibility of a dilute solution near the liquidgas critical point gives rise to a local density increase. The use of pyrene as a solute is specially convenient because it is a chromophore. This reaction ﬁeld lowers the groundstate energy of the solvated solute relative to the gas phase value.
6ρloc + 1. n2 − 1 = 6. 6. [203] with permission. as predicted by Eq.45) loc loc n2 + 2 Fig.122 Chemistry versus Physics: Chemical Reactions near Critical Points phase is [206] ∆ν = µ2 − µ2 n2 − 1 g c a3 2n2 + 1 (6. Reproduced from Ref.25ρ2 − 264ρ3 . (6. On the other hand. there is a relationship [207] between the polarizability term n2 − 1 / n2 + 2 and the supercritical CO2 density ρloc . This demonstrates that the local ﬂuid environment surrounding the ground state pyrene molecules is diﬀerent from that predicted by Eq. copyright (1995).44).44). experiment shows that a shift ∆ν between vapor and solution is not linear in n2 − 1 / n2 + 2 .44) 2µc µg − µc − 1 n2 − 1 − + a3 − 2 n2 + 2 where µg and µc are the dipole moments of the ground state and excited states.1 Recovered density enhancement ρloc /ρbulk as a function of the dimensionless ﬂuid density ρ/ρcrit for pyrene in CO2 at temperature T = 45 K. and exceeds the expected value. (6. American Chemical Society. respectively. . However. (6.
45). the local density augmentation ρloc /ρbulk was found as a function of the reduced density ρ/ρcrit of CO2 . connected with the ground state of pyrene molecules.44). which displays a local density augmentation up to 170 at about half the critical density. indicating that pyrene molecules are solvated individually in CO2 . [210]. All these results do not depend on the pyrene concentration.1.Sound Propagation and Light Scattering in Chemically Reactive Systems 123 Equations (6. due to the increased electrostatic interaction between the excited state of pyrene and CO2 .45) were used [203] to ﬁnd the spectral shift ∆ν as a function of bulk density ρbulk . as shown in Fig. Finally.44) and (6. The latter turned out to be 1.5 times greater than in the ground state. which determines [203] the local density ρloc from Eq. In this manner. . The experimental spectral shift ∆ν is used to compute a local polarizability term in Eq.Ndimethylamino) benzonitrile as a probe instead of pyrene. (6. In addition to this local density enhancement. Additional veriﬁcation of the inhomogeneity of a nearcritical reactive system has been obtained by indirect measurements combined with thermodynamic calculations [209]. (6. The same increase of the solvent density around a solute molecule compared with the corresponded bulk density has been obtained [208] using (N. similar experiments have been performed for the pyrene excited state. 6. the same result follows from numerical simulations based on the LennardJones potential [211] and the theory of integral equations [212].
experiment and simulation — we mostly used the ﬁrst in the phenomenological analysis. The theory of critical phenomena. It is not surprising that the methods used in this theory (renormalization group. Of three branches of modern physics — theory. which describe numerical methods of phase equilibria in reactive systems. laser transitions. Although these two phenomena appear to be quite diﬀerent. Indeed.Chapter 7 Conclusions The main body of this book consists of a detailed discussion of the interconnection between “physical” critical phenomena and chemical reactions. both are characterized by the enhancement of ﬂuctuations. longrange order and critical slowingdown. dates from the 1960’s and 1970’s. and the analogy between phase transitions of the ﬁrst and second order and hard and soft transitions of chemical reactions. That is the reason why chemical reactions near the critical point are in such wide use in modern 125 . In addition. modemode coupling. Many references to numerical methods are given in review articles. etc) have been successfully employed in modern chemistry. such as [165]. high compressibility allows the tuning of the density and dissolving power by small changes in pressure. deep analysis shows [22]. which is supported experimentally with a precision of few decimal places. [21] that the instability of diﬀerent nonequilibrium phenomena (hydrodynamic and chemical instabilities. with references to experimental work given in appropriate places. among others) show an intimate connection with equilibrium phase transitions and critical phenomena. A nearcritical ﬂuid has some of the advantages of both a liquid and a gas: high density ensures high dissolving power which is accompanied by high diﬀusivity of solutes and low viscosity facilitates mass transport. Examples of this connection include the relation which has been established [213] between the laser threshold and the critical point of a secondorder phase transition.
sunﬂower and peanuts [221]). This subject still attracts great interest. [165]). metallurgy (the recovery of metals from coal ash [214]. food industry (extraction of caﬀeine from coﬀee beans [220] and oil from corn. solubility of minerals in brine solutions [217]). We may mention applications in chemistry (supercritical extraction [145]). and it is very likely that many new applications will be found in the future. Many applications can be found in review articles ([156]. and the pharmaceutical industry (extraction of vitamin E from soybean oil and its puriﬁcation [222]). phase diagrams in alloy systems [215]).126 Chemistry versus Physics: Chemical Reactions near Critical Points science and technology. mineral equilibria [219]). geology (equilibria between aqueous and mineral phases [218]. . geochemistry (distribution and activity of species in aqueous solutions [216].
Bibliography
[1] H. G. Harris and J. M. Praushnitz, Ind. Eng. Chem. Found. 8, 180, 1969. [2] M. Gitterman, Phys. Stat. Sol. 59, 295, 1980. [3] I. Prigogine and R. Defay, Chemical Thermodynamics, Longmans Green, New York, 1954 [4] L. D. Landau and E. M. Lifshitz, Statistical Physics, Pergamon, London, 1958; Fluid Dynamics, Pergamon, London, 1987. [5] A. A. Galkin and V. V. Lunin, Russian Chem. Rev. 74, 21, 2005. [6] S. N. V. K. Aki and M. A. Abraham, Envir. Prog. 17, 246, 1998. [7] S. N. V. K. Aki and M. A. Abraham, Chem. Eng. Sci. 54, 3533, 1999. [8] A. Kruse and H. Vogel, Chem. Eng. Technol. 31, 23, 2008. [9] A. M. N. da Ponte, J. Supercrit. Fluids 47, 344, 2009. [10] M. Gitterman, Rev. Mod. Phys. 50, 85, 1978. [11] G. Arcovito, C. Falocci, M. Riberty, and L. Mistura, Phys. Rev. Lett. 22, 1040, 1969. [12] H. E. Stanley, Introduction to Phase Transitions and Critical Phenomena, Oxford University Press, 1993. [13] V. L. Ginzburg, Sov. Phys.  Solid State 2, 1824, 1960 [14] D. Yu. Ivanov, Doklady Physics 47, 267, 2002 [15] D. Yu. Ivanov, L. A. Makarevich, and O. N. Sokolova, JETP Letters 20, 121, 1974. [16] W. Wagner, N. Kurzeja, and B. Pieperbeck, Fluid Phase Equilibria 79, 151, 1992. [17] L. A. Makarevich, O. N. Sokolova, and A. M. Rozen, Sov. Phys.  JETP 40, 305, 1974. [18] M. Gitterman, J. Stat. Phys. 58, 707, 1990. [19] J. S. Rowlinson and F. I. Swinton, Liquids and Liquid Mixtures, Butterworth, London, 1982. [20] G. M. Schneider, Adv. Chem. Phys. 17, 1, 1970. [21] A. Nitzan, Phys. Rev. A 17, 1513, 1978. [22] A. Nitzan, P. Ortoleva, J. Deutch, and J. Ross, J. Chem. Phys. 60, 1056, 1974. [23] I. Matheson, D. F. Walls, and C. W. Gardiner, J. Stat. Phys. 12, 21, 1975.
127
128
Chemistry versus Physics: Chemical Reactions near Critical Points
[24] A. S. Mikhailov, Phys. Lett. A 73, 143, 1979; Z. Phys. B 41, 277, 1981. [25] W. Horsthemke and R. Levever, Noiseinduced Transitions, Springer, Berlin, 1984. [26] D. G. Peck, A. J. Mehta, and K. P. Johnston, J. Phys. Chem. 93, 4297, 1998. [27] B. Lin and A. Akgerman, Ind. Eng. Chem. Res. 40, 1113, 2001. [28] H. Li, B. Han, J. Liu,. L. Gao, Z. Hou, T. Jiang, Z. Liu, X. Zhang, and J. He, Chemistry Eur. J. 8, 5593, 2002. [29] J. Quin, M. T. Timko, A. J. Allen, C. J. Russel, B. Winnik, B. Buckley, J. I. Steinfeld, and J. W. Tester, J. Am. Chem. Soc. 126, 5465, 2004. [30] B. Wang, B. Han, T. Jiang, Z. Zhang, Y. Xie, W. Li, and W. Wu, J. Phys. Chem. B 109, 24203, 2005. [31] T. Nishikawa, Y. Inoue, M. Sato, Y. Iwai, and Y. Arai, AIChE J. 44, 1706, 1998. [32] S. Glasstone, K. J. Laidler, and H. Eyring, The Theory of Rate Processes, 1st ed, McGrawHill, New York, 1941. [33] C. A. Eckert, D. H. Ziger, K. P. Johnston, and S. J. Kim, J. Phys. Chem. 90, 2738, 1986. [34] C. Muller, A. Steiger, and F. Becker, Thermochim. Acta 151, 131, 1989. [35] L. Van Hove, Phys. Rev. 95, 1374, 1954. [36] I. Procaccia and M. Gitterman, Phys. Rev. Lett. 46, 1163, 1981; Phys. Rev. A 25, 1137, 1982. [37] A. Z. Patashinskii, V. L. Pokrovskii, and M. V. Feigel’man, Sov. Phys. JETP 55, 851, 1982. [38] R. G. Griﬃths and J. C. Wheeler, Phys. Rev. A 2, 1047, 1970. [39] M. E. Paulaitis and G. C. Alexander, Pure Appl. Chem. 59, 61, 1987. [40] S. T. Milner and P. C. Martin, Phys. Rev. A 33, 1996, 1986. [41] A. Onuki and R. A. Ferrell, Physica A 164, 245, 1989; B. Zappoli, D. Bailly, Y. Garrabos, R. Le Neindre, P. Guenoun, and D. Beysens, Phys. Rev. A 41, 2264, 1990; H. I Boukari, J. N. Shaumeyer, M. E. Briggs, and R. W. Gammon, Phys. Rev. A 41, 2260, 1990. [42] R. P. Behringer, A. Onuki, and H. Meyer, J. Low Temp. Phys. 81, 71, 1990; H. Meyer and F. Zhong, Compt. Rend. Mecanique 332, 327, 2004. [43] B. Zappoli and P. Carles, Acta Astron. 38, 39, 1996. [44] Yu. I. Dakhnovskii, A. A. Ovchinnikov, and V. A. Benderskii, Chem. Phys. 66, 93, 1982. [45] M. Sato, Y. Ikushima, K. Hatekada, and T. Ikeshoji, Int. J. Chem. Reactor Eng. 3, A15, 2005. [46] C. E. Bunker and YaPingSun, J. Am. Chem. Soc. 117, 10865, 1995. [47] I. Raspo, S. Meradji, and B. Zappoli, Chem. Eng. Sci. 62, 4182, 2007. [48] S.T. Chung, H.C. Huang, S.J. Pan, W.T. Tsai, P.Y. Lee, C.H. Yang, and M.B. Wu, Corrosion Science 50, 2614, 2008. [49] M. Gitterman, Physica A 386, 1, 2007. [50] M. Gittermam, Physica A 388, 1046, 2009. [51] M. Green, J. Chem. Phys. 22, 398, 1954. [52] H. Mori, Progr. Theor. Phys. 34, 399, 1965.
Bibliography
129
[53] R. Zwanzig, Lecture in Theoretical Physics, vol. 3, Interscience, New York, 1961. [54] T. Yamamoto, J. Chem. Phys. 33, 281, 1960. [55] B. J. Berne and R. Pecora, Dynamic Light Scattering, Wiley, NewYork, 1976. [56] R. M. Velasco and L. S. GarciaColin, Physica 72, 233, 1973. [57] D. L. Carle, W. G. Laidlaw, and H. N. W. Lekkerkerker, Farad. Soc. Trans. 71, 1448, 1975. [58] R. D. Mountain and J. M. Deutch, J. Chem. Phys. 50, 1103, 1969. [59] L. Letamendia, J. P. Vindoula, C. Vaucamps, and G. Nouchi, Phys. Rev. A 41, 3178, 1990. [60] C. Allain and P. Lallemand, J. de Chimie Physique 77, 881, 1980. [61] A. Munster, Statistical Thermodynamics, Springer, 1974. [62] J. K. Baird and J. C. Clunie, J. Phys. Chem. 102, 6498, 1998. [63] Y. W. Kim and J. K. Baird, Int. J. Thermophys. 22, 1449, 2001. [64] C. D. Specker, J. M. Ellis, and J. K. Baird, Int. J. Thermophys. 28, 846, 2007. [65] Y. W. Kim and J. K. Baird, Int. J. Thermophys. 25, 1025, 2004. [66] I. Procaccia and M. Gitterman, Phys. Rev. A 27, 555, 1983; ibid. 30, 647, 1984. [67] J. C. Wheeler and R. G. Petschek, Phys. Rev. A 28, 2442, 1983. [68] I. R. Krichevskii, Yu. V. Tsekhanskaya, and Z. A. Polyakova, Rus. J. Phys. Chem. 40, 715, 1966; ibid. 43, 1393, 1969. [69] G. Morrison, Phys. Rev. A 30, 644, 1984. [70] M. Gitterman and I. Procaccia, J. Chem. Phys. 78, 2648, 1983; AIChE J. 29, 686, 1983. [71] I. Procaccia and M. Gitterman, J. Chem. Phys. 78, 5275, 1983. [72] T. Toroumi, J. Sakai, T. Kawakami, D. Osawa, and M. Azuma, J. Soc. Ind. Jpn. 49, 1, 1946. [73] S. C. Greer, Phys. Rev. A 31, 3240, 1985. [74] J. L. Tveekrem, R. H. Cohn, and S. C. Greer, J. Chem. Phys. 86, 3602, 1987. [75] R. Narayan and M. J. Antal Jr, J. Am. Chem. Soc. 112, 1927, 1990. [76] R. B. Snyder and C. A. Eckert, AIChE J. 19, 1126, 1973. [77] E. S. Gould, Mechanism and Structure of Organic Chemistry, Holt, Rinehart, and Winston, New York, 1959. [78] Y. W. Kim and J. K. Baird, J. Phys. Chem. A 107, 8435, 2003; ibid. 109, 4750, 2005. [79] J. L. Tveekrem and D. T. Jacobs, Phys. Rev. A 27, 2773, 1983. [80] R. H. Cohn and D. T. Jacobs, J. Chem. Phys. 80, 856, 1984. [81] D. T. Jacobs, J. Chem. Phys. 91, 560, 1989. [82] J. Ke, B. Han, M. W. George, H. Yan, and M. Poliakoﬀ, J. Am. Chem. Soc. 123, 3661, 2001. [83] B. Hu, R. D. Richey, and J. K. Baird, J. Chem. Eng. Data 54, 1537, 2009. [84] M. E. Fisher, Phys. Rev. 176, 257, 1968. [85] M. Gitterman and V. Steinberg, Phys. Rev. A 22, 1287, 1980.
Phys. Sandig. 1984. Ann. [106] R. Anderson and J. Chem. [108] J. 1984. Goldberg. 1989. [100] C. 92. (Leipzig) 28. Phys. [105] Ya. Phys. 76. Phys. 685. Woermann. Langer. [96] G. 1976. Phys. 935. Voronel. Phys. J. C 12. Bunsenges. 1978. 1973. 2606. 76. H. Powers and S. Academic. Phys. Chu. J. 35. V. C. [99] K. 1984. Rev. [110] A. 1968. Ing. 993. 147.Gitterman. and D. Gansen. 1980. S. 5408. 49. and D. Green. S. Macromolecules 21. I. Stein and G. 20. Rev. Lett. 1988. M. Halpern. Greer. 1992. D. Ohbayashi and B. [103] P. Ebeling and R. Phys. C. Chem. 1988 [120] Y. Rev. 1496. Rev. 1973. Scott. Phil. [107] R. 31. Caram and L. Phys. E. Arch. [109] E. U. [112] M. E. Janssen. J. Anal. [118] V. 848. 1979. 31. 1989.JETP 33. Rev. 1983. 2008. Chem. 11. Steinberg. Phys. Khim. C. Dordecht. Gitterman and V. in Phase Transitions and Critical Phenomena. Ber. 84. Neilson. 3403. B. in The Physics and Chemistry of Aqueous Ionic Solutions. Goldburg. A 30. 88. . A. Mech. Moreau. Schonert. 1974. Borgis and M. 59. B. I. vol 5a. Knobler and R. Phys. [115] H. 1236. S. eds. J. [111] M. A 30. 1978. S. Frank. Wheeler. M. 5066. I. 1982. Ration. Kindt. 1980. Chem. S. Mech. 73. Wheeler. [116] G. [101] P. 68. Chem. W. Albeck and M. 69. B 56. Phys. 1971. eds. L. Phys. 1976. J. Tveecom. [117] L. Rev. 65. Phys. A 29. Chem. Kawasaki. Shaw and W. Phys. J. 605. 73. [102] K. 1977. Thoen. [89] A. Rev. K. 1149. BellissentFuel and G. Sci. and J. W. [98] E. [104] G. Phys. 14431. F. Fisher and J. Jacobs. 2002. Fiz. 881. M. M. 27. [114] Y. [91] E. R. Rabin and M . Jasnow. 1978. New York. Zh. Gulari.C. Gitterman and V. [119] D. Chem. J. [93] M. Chem. Reiss and M. Gansen and D. Sci. Phys. 4906. W. Gitterman. Zel’dovich. Chem. Rev. J. 69. Allen. 1965. J. Kogut. and E. L.130 Chemistry versus Physics: Chemical Reactions near Critical Points [86] M. Shugard. and H. Arch. 1938 (in Russian). p. ibid. Lett. 91. Reidel. [95] J. Phys. [88] E. Starikov and B. Norden. Phys. [121] H. [113] W. Phys. J. Rep. 1976. Chem. Phys. 1984. Aris. Greer. Steinberg. A. 665. 75. Chem. C. 2655. Phys. 337. 1976. Mag. 1987. W. Chem. 50. J.. R. [97] J. Physica A 156. P. 358. Wilson and J. M. 2082. R. Ing. Corrales and J. Chu. 1974. 6079. 356. Chem. 1980. Gitterman. Ration. Semura. Phys. C. . Bose. J. 1588. Paulucci. 163. CornishBowden. Phys. [87] J. Woermann. 5778. 2480. Phys. Mod. [92] D. Anderson and S. Hohenberg and B. 1976. 121. Wheeler. 96. Sov. 3129. Rev. Am. Biosciences 27. 19. A 20. 65. Merabet. 1968. 5280. T. Domb and M. 289. B 111. Chem. 81. Stat. J. [94] M. Scriven. 2007. J. Gorodetskii. 355. Chem. Anisimov. T. [90] C. Anal. A 9. Woermann. 1997. Talanquer. L. B. A 28. Phys. Othmer. van Dael. E. 2763. 648. Aris. Phys. and T. 7097. C. J. 1975. Schon. 435.
Lett. E 50. ed. Lett. 875. C. 822. 75. Liu. Principles and Practice. 1994. 1989. Lefever. Lett. Phys. C. 677. Scheﬀer. Carati. He. 155. 578. Phys. 1995. [127] S. Muthukumar. Lett. C. Klein. Nauman. 1994. Rev. [148] L. Chem. E 54. Rev. 6027. 1338. J. K. 1996. Supercritical Fluid Extraction. C. L. Steinberg. E. Glotzer and A. A. 116. 1000. Tong and Y. Frish. in Nonequilibrium Statistical Mechanics. 229. Ohta. T. [132] S. H. C. 1996. Eng. Phys. J. and U. 3820. and G. Hensel. J. Cohen. Taylor. Phys. 1980. Rev. Chem. 258. M. 588. 84. Phys. A. J. [155] B. Phys. S. Lefever. Fogelby. Chem. Jan. J. Wu. Stauﬀer. F. New York. [124] J W. 1674. 45. [134] S. 1994. Palmieri. Gitterman. Christiansen. Chem. Ind. Chem. 75. 1996. Phys. 1988. Motoyama. Elder. 88. Diepen and F. Voronel. 66. Nozieres. . Schneider. and H. Yang. Rev. van Wasen and G. Modern Theory of Critical Phenomena. 1988. Chem. [151] C. and Y. Rev. [139] N. E. D. Q. Rev.Bibliography 131 [122] M. J. Schindewolf. J. H. [154] M. [142] W. Phys. Bunsenges. E 56. 57. 1953. 56. 1997. Boston. Phys. J. Phys. Albeck and M. 1519. Turkevich and M. L. 1982. Mod. Eng. Phys. Yang. D. E. 65. [144] U. [131] S. J. Soc. Tong. 4241. 2001. Jan. 1675. 2003. Chem. C. 48. Stauﬀer. 23. Phys. 1986. Gitterman. Phys. [147] M. Sandler. 1958. [136] C. Res. M. 1996. [153] U. Lett. World Scientiﬁc. Hefner and F. Gitterman. J. 1984. M. J. Phys. p. R2212. [143] V. 70. 1980. Phys. 1968. and N. A. D. Ber. Phys. 88. Phys. 5. 72. Rev. 74. ibid. Chem. Lett. [129] Y. 10091. 40. and S. 4109. Butterworths. Carati and R. and Y. McHugh and V. Phys. Rev. 1980. Mass. 1997. Chem. A. [140] P. I. A. [126] R. E. Japan 65. Diepen. 4085. [135] D. D. 3710. Phys. Phys. FernandezPrini and M. [156] R. [128] S.1994. 2034. Kwak. DiMarzio. Soc. F. Rev. Macromol. Chem. Am. 1984. Glotzer. 2715. Rev. 1995. 1984. D. Scheﬀer. [152] S. Puri and H. 1996. 118. and M. Hillard. [125] M. ibid. Rev. Paulatis. Physica B 117. Rev. Data. 1948. 1894. Phys. [150] M. [123] Y. Chem. 1026. 30. B 105. Linsky. Theory Simul. and N. 88. Van Gunst. A. Schindewolf. Phys. Huo. Chem. T. Wiley. [133] J. X. Chem. 9830. 1983. J. 2002. [141] L. C. 1988. [138] D. Phys. Rev. A 27. Cahn and J. Supercritical Fluid Extraction. Krukonis. Phys. E. 1995. [130] S. Edu. Peter. 1989. J. A254. M. J. Reading. and N. E. Lett. Zhang. A141. McHugh and M. 103. S. A. Japes. M. [137] B. Chem. K. 3127. B. Soc. 28. Am. Hassani. [149] G. J. Mott. C. Yang. 60. C. 801. Ma. Jiang. 3751. 1991. 1976. Conigliuo. Glotzer. 66. [146] M. [145] S. Toxvaerd. E 53. and E. Hernandez. J. Benjamin. Glotzer. Motoyama and T. 326. 16. 25.
Larsen and C. C. D. J. Kwauk. New York. 29. Ishida. Suzuki. 2003. E. D. Phys. Lett. J. 811. Environmental Protection Agency Report No. Krichevskii. J. 1983. Sominskaja. 315. Shirane. Physik Chem. [158] T. Lett. Dowden. 34. Phys. Helling. Krukonis. Ann. 22. Barbosa and M. Chem. Nechitallo. 1980. J. Sov. Doscher and M. Menyhard. 2009. R. Y. 1967. Phys. 311. A. Z. 4439. Squires and Paulaitis. Krichevskii. 80. 1985. J. P. 52. [161] N. eds. Wheeler and B. Schmidt. 83. E. 2723. Sobyanin. Goldstein. D. Cox. Rev. [172] G. Modell. 87. Phys. B. and N. R. Series 329: Washington. Chem. Soc. G. Meier. 47. 113. Widom. and P. 1961. 282. Pergamon. [174] B. . S. J. [165] A. V. Chem. 591. [167] Stortenbekar. 2284. E. 5334. Chem. 1983. Alig. M. Z. Phys. Mather. 1970. Busenges. Ferrell and J. [162] W. A. Collins. K. [177] F. Phys. H. Yeo. C. 79. Doherty. LaPlace. D. 443. K. E. Hutchison and Ross. Jona and G. Szepfalusy. Chem. N. Phys. A. Nechitallo and G. and I. 48. 1988. Lond. Lewis. 1987. Proc.1982. Yonker. Technol. L. ed. 10. 1988. R. Phys. Eisenbach. 1319. R. and S. 1211. 3102. 18. 295. L. [159] R. K. 1892. Phys. S. [173] A. J. A 413. Phys. A. Dunker. Mayer. and M. Rev. De Fillipi. Fluid Phase Equilibria 83. American Chemical Society Symp. 54. P. Rendall.Uspechi. H. Chem. Fluid Phase Equilibria 30. S. McEwan. Rev. E. E 55. W. [188] W. Lindsey. 183. 570. Res. A. K. 4601. Bhattacharjee and R. Soc. and J. 123. L Frye. and S. [183] H. Niemann. R. W. J. Schwabl. Andon and J. Ferroelectric Crystals. and R. [169] V. Rev. Ferrell. [187] Y. G. 1981. 1980. [166] D. ElArabi. 27. S. P. Eckert. X. J. Phys. 1952. Chem. 3. A 31. F. [175] C. Chim. N. Rotenberg and G. Sov. Sorensen. Tester. Widom. A. J. 1974. 1985. Gottsh. 1986. private communication. Ber. M. A 24. Hoﬀman. 343.. McGrawHill. [186] R. Envir. Phys. 703. [164] C. [180] G. in Physical Acoustics.Dokl. Soc. Smith in Supercritical Fluids. W. Wright. A. Strasbourg. 1983. Eng. 1974. R. 1788. and Y. SpringerVerlag. 238. J. N. 1997. Phys. 1962. Einstein. Ferrell. [184] R. N. 1923. Ind. Kuznetsov. 1987. . and K. E. Lukashova. 1643. Chem. Phys.Dokl. Sci. Phys. Jpn. Debenedetti. Saidov. and M. N. 1904. [170] I. AIChE J. Phase Separation in Soft Matter Physics. [176] R. 1816. Sominskaja. 172. 541. 48. [182] A. [168] B. [178] J. Sov. Sundaresan. 1993. Fluid Phase Equilibria 10. Thermodynamics. 1985. Bhattacharjee. EPA600/280054. Tom. [179] R. Rev. 1988. Oil Gas J. Zosel. Bhattacharjee. F. A. [160] H. Woermann. Hudson. Khabibullaev and A. A. Harada. L. 1986. 1987. K. Yang and C.132 Chemistry versus Physics: Chemical Reactions near Critical Points [157] G. 891. [185] P. Rus. 144. Oxford. Soc. M. [163] R. [171] I. [181] P. p. Debenedetti. and J.
A 9. Phys. and C. 1994. 1974. D. J. A. Y. Knutson. F. 1990. Fluid Phase Equilibria 83. Cliﬀord. 26. 1981. Pure Appl. Kautze. 37. Chem. and A. T. Phys. Chem. Kim and K. eds. 346. K. 1966. Rev. E. W. G. N. E. Ser. R. R. 1988. Holloway and J. Chem. D. Johnston and J. 2374. Hentschel and I. Cochran and L. and K. E. C. D. Europhys. Z. Phys. Supercritical Fluid Methods and Protocols. J. 1968. Zalzburg. 1987. Chem. Brennecke. Phys. T. 92. C. Phys. 119. 92. 31. Sanchez and G. 79. E. H. Eng. O. R. and U. Kaatze. G. R. 59. 93. Johnston. Johnston. G. 1988. Chem. 111. Chem. Chem. 1972. S. Phys. U. Garland. Soc. D. Debenedetti. 237. D. Dunbar. Eckert. Chem. Blum and Z. S. Z. P. Liquids 3. Hoytnick. J. 1486. R. P. Chem. L. 2002. Dunbar. Laidlaw. Rev. K. 2000. 2292. A. 1982 R. 117. Rev. H. B. S. Niemeyer. 5832. Am. Physica 75. 48. Onsager. 205. Kaatze. C. W. 1997. (second ed. W. Phys. Mag. Mod. Phys. Procaccia. Geochimica Cosmochimica Acta 44. Geochemical Thermodynamics. 33. and U. Munoz. 31. 2000. Stubbins. Lett. Chem. D. Chem. H. 1682. Liu. P. A 52. F. Weare.Bibliography 133 [189] [190] [191] [192] [193] [194] [195] [196] [197] [198] [199] [200] [201] [202] [203] [204] [205] [206] [207] [208] [209] [210] [211] [212] [213] [214] [215] [216] [217] [218] [219] [220] I. Eng. Bishnoi. R. 225. A. Chem. A. 1974. Kaatze. Mountain. 1980. Svrcek. 1547. Kajimoto. 1987. H. AIChE J. 3100. 225. Chem. V. J. Phys. 221. 7075. Washington. 2004. Geldorf. Phys. 58. L. 1985. Giggenbach. 4. 1347. N. Phys. J. Yamasaki. Spectroscopy 51. B.). Phys. 9828.P. Tomasko. J. J. V. 1985. Helgeson. 1993. Mirzaev. 1989. A. Hordstrom and J. Peshkin. and G. Can. Lekkerkerker and W. Graham and H. Kim and K. Rettschnick. 29. Appl. Chem. Geochimica Cosmochimica Acta 45. Hamilton. I. ACS Symposium. Blackwell Scientiﬁc Publications. 38. . A. Bright. Telgmann. R. 1989. J. Kobayashi. A. de la Selva. Lett. J. R. M. J. and M. J. P. GarciaColin and S. in Supercritical Fluid Science and Technology. 061508. 1983. Lett. 1936. E. G. D. and W. T. Behrends and U. Lekkerkerker and W. Mehrotra. A. S. L. L. Ind. Niemeyer. R. 1206. Phys. Ind. Boston. Behrends. P. 7957. H. G. Chapter 3. K. 1979. L. Soc. 76. 1969. Res. J. L. Eckert and C. J. P. Lee. 1603. and F. 981. 475. 65. Williams. 406. Yonker and R. 393. Chem. M. and F. J. Am. C. Res. Behrends. 117. Phys. J. 666. 1993. and N. Chem. Bright. E 73. W. Sun. K. J. Eng. J. W. Laidlaw. 57. 1970. Humana Press. F. Bunker. 91. Petsche and P. C. Futakami. Rice. Haken. P. Schach. Iwanowski. R. Chem. Smith. 2003. Harvier and J. 210. Phys. 57. 1995. V. M. L. H. B. Penninger. Phil.
357. Phan. [222] G. C.134 Chemistry versus Physics: Chemical Reactions near Critical Points [221] L. Trauber. and P. W. 11. 53. Jessop. G. J. Green Chem. Ann. Burton and M. Brown. . 2009. Hodgson. White. 1990. H. A. Nutr. 10. Rev.
42 . 67 nanocrystalline. 105 ecological theorem. 1 critical slowingdown. 62 kinkantikink solution. 49 consolute point. 5 double critical point. 82 Mott transition. 68 isotope exchange. 89 modiﬁcation of critical indices. 100 carbon dioxide. 29 corrosion eﬀect. 58 DielsAlder reaction. 121 ionization equilibrium. 26 critical dilute solution. 96 critical phenomena. 60 enthalpyenergy compensation. 112 conductance. 7 hard and soft transitions. 87 FokkerPlanck equation. 17 homogeneous nucleation. 77 gas phase. 45 fast heat conductivity. 55 morphologies of phase separation. 12 heat production. 8 light scattering. 118 magniﬁcation of singularities. 4 noncompressible liquid. 50 metalammonia solutions. 3 correlation function. 73 inhomogeneous structure. 95 extent of reaction. 45 fast diﬀusion. 4 ClapeyronClausius equation. 15 diﬀusion coeﬃcient. 74 law of mass action. 89 metallic vapors.Index azeotropic point. 101 Menschutkin reaction. 8 closedloop. 76 degree of dissociation. 22 dielectric constant. 105 complex sound velocity. 72 melting point. 14 electroneutrality. 90 multiple solutions. 60 densityﬁeld variables. 58 meanﬁeld model. 19 DebyeH¨ckel approximation. 51 Ginzburg criterion. 91 u decay of metastable state. 21 fast chemical reaction. 68 equilibrium line. 45 135 FloryHuggins energy.
28 patterns domain. 98 second crossover. 87 threecomponent plasma. 89 transitionstate theory. 34 water. 71 supercritical extraction. 53 statistically independent variables. 16 ultrasound attenuation. 7 viscoreactive cross eﬀects. 67 piston eﬀect. 70 Onsager coeﬃcient. 25 quasisteadystate solutions. 30 reactiondiﬀusion equation. 3 . 83 phase separation.136 Chemistry versus Physics: Chemical Reactions near Critical Points nucleation process. 79 shift of critical parameters. 74 scaling theory. 38 strictly regular solution. 47 ternary polymer blends. 111 rate constants.
This action might not be possible to undo. Are you sure you want to continue?
We've moved you to where you read on your other device.
Get the full title to continue reading from where you left off, or restart the preview.