You are on page 1of 20

This is an open access article published under an ACS AuthorChoice License, which permits

copying and redistribution of the article or any adaptations for non-commercial purposes.

pubs.acs.org/JPCB Feature Article

Gibbsian Surface Thermodynamics


Janet A. W. Elliott*

Cite This: J. Phys. Chem. B 2020, 124, 10859−10878 Read Online

ACCESS Metrics & More Article Recommendations

ABSTRACT: Gibbsian composite-system thermodynamics is the


See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
Downloaded via INST FED EDU CIENCIA E TECH CEARA on July 29, 2023 at 19:21:18 (UTC).

framework governing the equilibrium of composite systems,


including systems that at equilibrium have more than one value
of pressure because of the action of surface tension, semipermeable
membranes, or fields, and thus cannot be treated as simple systems.
J. W. Gibbs’s paper that lays out composite-system thermody-
namics, “On the Equilibrium of Heterogeneous Substances”,
communicated in two parts in 1876 and 1878, is widely regarded
as one of the most important pieces of scientific literature of its
century. Many scientists adopted and stressed the importance of
Gibbsian thermodynamics. In 1960, H. B. Callen wrote a textbook
that made Gibbsian composite-system thermodynamics more
accessible to thermodynamicists. However, Callen’s book left out
Gibbs’s work on curved fluid interfaces and did not treat the complicated nonideal systems of interest to today’s thermodynamicists.
In this Feature Article, I have attempted to convey in a comprehensive manner the framework of Gibbsian composite-system
thermodynamics including in detail the treatment of systems with interface effects and with nonideal, multicomponent phases. This
work lays out the relationships between important equilibrium equations including the following: the Gibbs−Duhem equation, the
Gibbs adsorption equation, the Young−Laplace equation, the Young equation, the Cassie−Baxter equation, the Wenzel equation,
the Kelvin equation, the Gibbs−Thompson equation, and the Ostwald−Freundlich equation, including nonideal and
multicomponent forms. Equations of state that are often useful for Gibbsian composite-system thermodynamics are reviewed
including adsorption isotherms and our own work on two semiempirical equations of state: the Elliott et al. form of the osmotic virial
equation and the Shardt−Elliott−Connors−Wright equation for the temperature and composition dependence of surface tension. I
summarize the work of our group developing Gibbisan composite-system thermodynamics including new equations for such things
as the curvature-induced depression of the eutectic temperature or the removal of azeotropes by nanoscale fluid interface curvature.
Gibbsian composite-system thermodynamics has broad applications in biotechnology, nanostructured materials, surface textures and
coatings, microfluidics, nanoscience, atmospheric and environmental physics, among others.

1. INTRODUCTION tively). Many principles and equations must be added to these


Thermodynamics is the study of mathematical relationships laws: principles and equations that at first glance appear to be
arising from physical laws governing energy and entropy.1 In unrelated, owing to the haphazard way in which they were
different disciplines, the word thermodynamics brings different discovered by a myriad of esteemed scientists starting in the
subsets of the topic to mind. A mechanical engineer may think of 1600s up until 1876. Then, in one of the most important pieces
energy balance and entropic efficiency of devices used in a power of scientific literature of the 19th century, J. Willard Gibbs
generating thermodynamic cycle. A chemical engineer may published in two parts his famous paper, “On the Equilibrium of
think of multicomponent phase equilibrium or chemical Heterogeneous Substances”, that recast thermodynamics as a
reaction equilibrium in liquids and vapors. A materials engineer mathematical study, showing that, from a small number of
may think of thermodynamic phase diagrams of solids. A postulates, the previous thermodynamic results could be
physicist may think of using thermodynamics to unite the obtained by mathematical derivations of multivariable calculus.2
fundamental theories of general relativity and quantum
mechanics or to understand the thermodynamic properties of Received: June 29, 2020
a cosmological object. A physical chemist may think of quantum Revised: September 6, 2020
statistical calculations of molecular energy levels. In the Published: October 22, 2020
historical approach to thermodynamics, the subject is under-
stood in terms of the first and second laws of thermodynamics
(energy balance and the entropy increase principle, respec-

© 2020 American Chemical Society https://dx.doi.org/10.1021/acs.jpcb.0c05946


10859 J. Phys. Chem. B 2020, 124, 10859−10878
The Journal of Physical Chemistry B pubs.acs.org/JPCB Feature Article

In that paper, Gibbs invented many of the cornerstones of system thermodynamics of Gibbs but must be found
physical chemistry such as chemical potential. The 299 page, separately. Equations of state may be found by making
700 equation tome by Gibbs is fairly inaccessible to all but the empirical measurements of some variables and then using
most diligent of well-trained thermodynamicists. Many Gibbsian thermodynamic relationships involving partial
scientists, including Ostwald3 and Guggenheim,4 adopted and derivatives to get the equations in the desired forms.
stressed the importance of Gibbsian thermodynamics. In 1960, Alternatively, equations of state may be derived from
Herbert B. Callen wrote a textbook presenting the Gibbsian statistical thermodynamics.
postulatory approach in a clear and pleasing manner, and his iv. Formally identify which function acts as the thermody-
textbook is used in many thermodynamics graduate courses to namic potential (equivalently free energy). Plot this
this day.5 Although an excellent treatise on thermodynamics in potential versus the variable or variables of interest and
its own right, Callen’s book does not cover one important part of examine the number and nature of extrema to provide
Gibbs’s famous paper: the treatment of systems with curved fluid relative stability of different equilibrium states that satisfy
interfaces. the conditions for equilibrium, thus determining the
Many of the thermodynamic principles with which people are behavior of the system. Maxima and inflection points in
most familiar are only valid for simple systems. A simple system is the free energy correspond to unstable equilibria and
one that at equilibrium has constant values of intensive properties minima correspond to stable equilibria.
throughout. Systems that cannot be treated as simple systems Following each and every one of steps i−iv above prevents the
include systems that have multiple pressures at equilibrium due pitfalls of much surface thermodynamics research that uses
to the action of surface tension of a curved fluid interface, due to equations from different steps derived with inconsistent
the presence of a semipermeable membrane, or due to the action assumptions, or uses a rigorous approach in one step but not
of fields. The important contribution of Gibbs’s famous paper in another. For different geometries, there are specific parts of
was to develop the thermodynamics of composite systems. A the steps above that provide challenge and opportunity for
composite system is a system composed of a number of constituent innovation and fundamental contribution. In specifying the
subsystems, each of which may be treated as a simple system. A system (step i), it is important to use insight to define the system
composite system may or may not itself be a simple system. This to ask the right question. The derivations in step ii may also be
allows the application of rigorous thermodynamics to many challenging; new mathematics may be required. In finding
nonsimple systems including biological systems with semi- solutions to the conditions for equilibrium (iii) and performing
permeable membranes, systems with field-induced pressure the stability analysis (iv), new equations of state may be needed.
gradients, and systems with highly curved fluid interfaces (drops, Even with the equations of state, performing the stability analysis
bubbles, emulsions, foams, aerosols, and multiphase fluids in is not trivial; new analytical and computational techniques are
solid porous media) that occur in almost every manufacturing often required.
and resource industry; are the basis for novel surface, colloidal, 2.1. Finding the Conditions for Equilibrium. The
microfluidic, and nanofluidic technologies; are central to procedure to derive the conditions for equilibrium begins with
environmental processes; and are the focus of much of modern the statement that at equilibrium the entropy of an isolated
science. composite system is an extremum subject to the constraints on
The objectives of this article are to outline briefly the methods the system, including importantly conservation of energy. Thus,
of Gibbsian composite-system thermodynamics, with specific we begin by setting
attention to its application to systems with curved fluid
interfaces, and to review several recent applications. dS C = ∑ dS j = 0
j (1)
2. GIBBSIAN COMPOSITE-SYSTEM
THERMODYNAMICS METHOD where SC is the entropy of the composite system, Sj is the
entropy of simple subsystem j, and the summation is over all
“The single, all-encompassing problem of thermodynamics is subsystems that make up the composite system which may
the determination of the equilibrium state that eventually include as one of the subsystems a surrounding reservoir that
results after the removal of internal constraints in a closed, interacts with the system. An alternative approach, that was the
composite system.”5 original approach of Gibbs,2 is to extremize the energy of an
The steps of Gibbsian composite-system thermodynamics are isolated system subject to entropy being held constant. These
i. Define the system so that the mathematical constraints are two approaches are formally equivalent due to the requirement
clear and the problem is posed to answer the question you that entropy be a monotonically increasing, first-order,
want. homogeneous, differentiable function of the internal energy
ii. Perform a unified derivation of the complete set of that is invertible (see pages 56 and 57 of the Ox Bow Press
equations that must be satisfied for equilibrium by setting version of Gibbs’s paper2 and page 28 of Callen’s book5). The
the differential of the composite-system entropy to zero approach of beginning with eq 1 is more intuitive and
subject to the mathematical constraints. This can be done convenient for many problems, and so, it is the approach
algebraically for some problems and with the calculus of adopted here.
variations and the method of Lagrange multipliers for Next, the entropy of each subsystem in eq 15 is replaced with
others. the appropriate differential form of an Euler relation. For bulk
iii. Analytically or numerically solve the coupled conditions subsystems, the differential Euler relations take the form (eq 12
for equilibrium from step ii using specific equations of of Gibbs’s paper,2 eq 2.6 of ref 5):
state, yielding particular equilibrium states that satisfy the 1 Pj j μi j
conditions for equilibrium. The equations of state do not dS j = j
j dU + dV − ∑ dNij
come from within the postulatory equilibrium composite- T Tj i
Tj (2)

10860 https://dx.doi.org/10.1021/acs.jpcb.0c05946
J. Phys. Chem. B 2020, 124, 10859−10878
The Journal of Physical Chemistry B pubs.acs.org/JPCB Feature Article

where Uj, Vj, and Nji are the internal energy, the volume, and the ∑ dU j = 0
number of moles of component (molecular species) i, j (5)
respectively, in the subsystem j; along with Sj these are the
extensive variables of the bulk subsystem. Tj, Pj, and μji are the where the summation is over all subsystems (bulk phases or
temperature, the pressure, and the chemical potential of interface phases) that can exchange energy with one another. A
component i, respectively, in the subsystem j; these are the conservation of energy constraint is required for each set of such
intensive variables of the bulk subsystem. The summation is over connected subsystems.
all molecular species present in the bulk subsystem. Continuity requires
Interfaces between two bulk subsystems are treated as
additional subsystems (i.e., separate interface phases). In ∑ dV j = 0
Gibbs’s approach, interface phases are modeled by the j (6)
combination of a phase having area but no volume (a dividing where the summation is over all bulk phases that have a freely
surface) with surface excess properties accounting for all effects moving boundary between them. A continuity constraint is
of the interface. By mathematically uniquely placing the dividing required for each set of such connected bulk phases.
surface, all thermodynamic variables of the surface are Conservation of mass requires
accounted for in a way that captures all effects of the interface.
This is an important concept to understand because it is not to ∑ dNij = 0
say that the approach of Gibbs can only be applied to interfaces j (7)
that have no reach into the adjoining bulk phases, but rather that
the excess molecules and energy of that reach must be added to where the summation is over all subsystems that can exchange
the presence of a uniquely defined infinitely thin surface to molecules of component i. Conservation of mass constraints is
completely describe the system. For each interface, the needed for each set of such connected phases. Separate
differential Euler relations take the form (eq 501 of Gibbs’s conservation of mass constraints are needed for each
paper2) component.
There are further geometrical constraints that are imposed by
1 σj j μi j the particular system under consideration. Typically, changes in
dS j = d U j
− dA − ∑ dNij areas of interfaces are related by geometry to changes in bulk
Tj Tj i
Tj (3) phase volumes and areas of other interfaces. For example, for a
spherical drop of liquid of radius r surrounded by vapor, a
where, Aj is the area of the dividing surface, and Sj, Uj, and Nji are change in the area of the liquid−vapor interface ALV is related to
the surface excess entropy, the surface excess internal energy, a change in the volume of the drop VL by
and the surface excess number of moles of component i,
respectively, of interfacial phase j; these are the extensive 2 L
dALV = dV
variables of the interface subsystem. Tj, σj, and μji are the r (8)
temperature, the interfacial tension, and the chemical potential
Finally, there may be other types of constraints imposed, for
of component i, respectively, of the interface j; these are the
example, by complex mechanical connections or stoichiometry
intensive variables of the interface subsystem. The summation in
of chemical reactions. See example 1 on page 51 of ref 5 for a
eq 3 is over all components (molecular species) present in the
worked problem with an interesting constraint imposed by the
interface subsystem. Which components are present in the
mechanical connections of different parts of a composite system
interface subsystem depends on the choice of how to uniquely
by pistons connected by rods, and homework problem 2.9-1 on
place the dividing surface. The Gibbs Dividing Surface approach
page 58 of the same reference for a formally equivalent chemical
is to place the dividing surface such that one of the components
reaction equilibrium with the constraint imposed by stoichiom-
in the bulk phases is not in excess at the interface; this is the usual
etry.
choice for flat surfaces. The Gibbs Surface of Tension approach,
Once the differential Euler equations of the form of eqs 2 and
typically used for curved interfaces, is to place the dividing
3 are substituted into eq 1, along with constraints of the form of
surface such that the property of interfacial tension is
eqs 5−7, along with further geometrical or other constraints of
independent of curvature; in this case, all components will be
the defined composite system, and the like terms collected, the
in excess at the interface. Note that interfacial tension of
extremization of entropy will now be written as an equation of
interfacial phase j is defined

i ∂U j y
the form below

σ j = jjjj j zzzz
k ∂A {S j , Nij (4)
∑ (expression that depends d(extensive variable) = 0
on intensive variables )
(9)
where the subscripts indicate that the derivative is taken with where the differentials of independent variables that remain in
surface excess entropy and surface excess numbers of moles of all eq 9 are all independent because all relationships between
components of the interfacial phase held constant. extensive variables have been applied. Because we are interested
Once the differential Euler equations are substituted into eq 1, in finding the conditions that make entropy an extremum for all
the equation is in terms of differentials of the extensive variables, virtual displacements about equilibrium, eq 9 must be true for all
but not all of these differentials are independent. Next, the possible differentials of the extensive variables; therefore, the
constraints that describe the relationships between changes in coefficients in front of each independent variation must each be
the extensive variables are enumerated. set equal to zero. The equations that result from setting the
Conservation of energy is required. In the absence of fields, coefficients to zero form the complete set of conditions for
this means that equilibrium that the system must satisfy.
10861 https://dx.doi.org/10.1021/acs.jpcb.0c05946
J. Phys. Chem. B 2020, 124, 10859−10878
The Journal of Physical Chemistry B pubs.acs.org/JPCB Feature Article

Fully described worked out examples of applying this where Π is the osmotic pressure which depends on the
procedure to find the conditions for equilibrium for a wide concentration on side A. (See eq 77 of Gibbs’s paper,2 eq 13.38
range of particular systems may be found in refs 5−15. of ref 5, and pages 97−98 of ref 16.)
This procedure yields all the conditions for all three of For a curved fluid interface, the mechanical equilibrium
thermal, chemical, and mechanical equilibrium for a given condition is the Young−Laplace equation (eq 500 of Gibbs’s
paper2):
ij 1 1 yzz
system. Some of the conditions for equilibrium that might be

P j − P k = σ jk jjj z
found from this procedure follow:
jR
k 1 R 2 zz{
thermal equilibrium: +
(15)
Tj = Tk (10)
where P is the pressure of the bulk phase inside the curvature, Pk
j
for all subsystems j and k that can exchange energy, is the pressure of the bulk phase outside the curvature, σjk is the
chemical equilibrium: interfacial tension of the interface between bulk phases j and k,
and (1/R1 + 1/R2) is twice the mean curvature of the interface.
μi j = μik (11) For a sessile drop of liquid, L, on a solid, S, surrounded by a
vapor, V (Figure 1 b), an additional mechanical equilibrium
for all components i that can exchange between subsystems j and condition to the Young−Laplace equation is the Young equation
k, and (eq 672 of Gibbs’s paper2):
∑ νμ
i i = 0 σ LV cos θ = σ SV − σ SL (16)
i (12)
where, θ is the contact angle that the liquid−vapor interface
for each independent reaction with stoichiometric coefficients makes with the solid as measured through the liquid. The
νi, and a wide variety of mechanical equilibrium conditions as contact angle that results from eq 16, or the equilibrium contact
described below. angle on a smooth surface, is often called the Young contact
For two phases j and k separated by a freely moving uncurved angle.
boundary, the mechanical equilibrium condition is For a fluid lens of one liquid, L, on another liquid, L′,
surrounded by a vapor, V (Figure 1 c), there are two Young
P j = Pk (13) equations:9
For two phases A and B separated by a fixed, rigid, σ LV cos β + σ LL′cos α = σ L′V (17)
semipermeable membrane through which can pass the solvent,
but not the solute on side A (Figure 1 a), the chemical− σ LV sin β = γ LL′sin α (18)
mechanical equilibrium condition is
where the angles α and β are as defined in Figure 1 c. Only eq 17
P A − PB = Π (14)
appears in Gibbs’s paper (as eq 561).2 As the number of curved
fluid interfaces intersecting one another in the system increases,
the number of Young equations that are in the set of equilibrium
conditions increases. One independent angle is required to
specify the intersection of one curved interface with a flat
interface; two independent angles are required to specify the
intersection of two curved fluid interfaces with a flat interface,
etc. Also, a Young−Laplace equation (eq 15) applies for each
curved fluid interface.
For a sessile drop sitting on a smooth but chemically
heterogeneous surface, or equivalently for a drop sitting on a
rough surface, where the drop does not penetrate the surface
roughness (the Cassie−Baxter wetting state, Figure 1 d), the
mechanical equilibrium condition (in addition to the Young−
Laplace equation for the liquid−vapor interface) that results
from the Gibbsian composite-system entropy extremization is a
line-fraction form of the Cassie−Baxter equation.14 In the case
of a smooth heterogeneous solid made up of different materials i,
the line-fraction Cassie−Baxter equation is14
cos(θCB) = ∑ λicos(θi)
i (19)
where θCB is the Cassie−Baxter contact angle shown on Figure 1
d, and the line fraction λi is the fraction of the three-phase
contact line that contacts material i with the Young contact angle
of θi. Equation 19 can be adapted to give the contact angle of a
Figure 1. Definition of variables for some of the systems of interest. drop sitting on a flat rough surface, where the drop does not
(a) Osmotic pressure difference across a semipermeable membrane. penetrate the surface roughness:
(b) Sessile drop. (c) Fluid lens. (d) Cassie−Baxter wetting state.
(e) Wenzel wetting state. (f) Phase in a gravitational field. cos(θCB) = λ cos(θY ) − (1 − λ) (20)

10862 https://dx.doi.org/10.1021/acs.jpcb.0c05946
J. Phys. Chem. B 2020, 124, 10859−10878
The Journal of Physical Chemistry B pubs.acs.org/JPCB Feature Article

where line fraction λ is the fraction of the circumference of the mechanical equilibrium must be combined with equations of
spherical cap in Figure 1 d that rests on the flat, but intermittent, state that relate the variables of the system. Equations of state are
solid surface, and θY is the Young contact angle of the solid not found from within Gibbsian composite-system thermody-
surface. namics and must be added separately, typically found from
For a sessile drop sitting on a rough surface, where the drop measurements of the relationships between system variables or
penetrates into the surface roughness (the Wenzel wetting state, from statistical mechanics based on assumptions or calculations
Figure 1 e), the mechanical equilibrium condition (in addition of molecular energies. The equations of state may be in a form
to the Young−Laplace equation for the liquid−vapor interface) relating the intensive properties, i.e., μ−P−T equations for bulk
that results from the Gibbsian composite-system entropy phases or μ−σ−T equations for interface phases, or the
extremization is a new line-roughness form of the Wenzel equations of state may be in other forms depending on the
equation:15 variables naturally appearing in measurements or statistical
ensembles from which the equations of state are found, e.g., P−
cos(θ W ) = ρ cos(θY ) (21) V−T equations, μ−N−T equations, or σ−N−T equations.
where θW is the Wenzel contact angle shown in Figure 1 e, and ρ 2.2.1. Gibbs−Duhem and Gibbs Adsorption Equations.
is the three-phase line roughness with a specific definition that Two important relations of thermodynamics need to be
both accounts for the increased length of the three-phase contact mentioned: the Gibbs−Duhem relation and the Gibbs
line over the length of the spherical cap circumference and takes adsorption equation.
into account the orientation that the liquid−vapor interface From the postulates of thermodynamics about the mathe-
intersects the vertical portions of the solid (see ref 15 for more matical properties of thermodynamic variables, the Euler
description of the definition of ρ). Note that eq 21 is a simplified relations of thermodynamics may be written for each bulk
form of the new Wenzel equation that is valid when the phase j (eq 55 of Gibbs’s paper,2 eq 3.6 of ref 5)
contribution of the liquid in between the surface features is
negligible.15 U j = T jS j − P jV j + ∑ μi j Nij
For a fluid phase in a gravitational field (Figure 1 f), the i (24)
condition for chemical−mechanical equilibrium is (eq 234 of and for each interfacial phase j (eq 502 of Gibbs’s paper2)
Gibbs’s paper2)
U j = T jS j + σ jAj + ∑ μi j Nij
μi j + Migz = const i (22) (25)
i
for each component i and each phase j, where Mi is the molar Taking the derivative of eq 24 and subtracting eq 2 (rearranged
mass of component i, g is the magnitude of acceleration due to for dUj) reveals the Gibbs−Duhem relation (eq 97 of Gibbs’s
gravity, z is the elevation above a reference, and consti is a paper, eq 3.14 of ref 5):
constant.
For a curved fluid interface in a gravitational field, the S j dT j − V j dP j + ∑ Nij dμi j = 0
mechanical equilibrium condition is2,8
ij 1 1 yzz dσ
(26)

PIj − PIk = σ jk jjj


i

z+
jR R 2 zz{
jk

k 1
+ cos ϕ Taking the derivative of eq 25 and subtracting eq 3 (rearranged
dz (23) for dUj) reveals the Gibbs adsorption equation (eq 508 of
Gibbs’s paper2):
where PjI is the pressure of the bulk phase j inside the curvature at
a point on the interface, PkI is the pressure of the bulk phase k S j dT j + A j dσ j + ∑ Nij dμi j = 0
outside the curvature at that point on the interface, σjk is the i (27)
interfacial tension of the j−k interface at that point on the
interface, R1 and R2 are the principal radii of curvature at that Equations 26 and 27 serve to illustrate that changes in the
point on the interface (defined as positive if phase j is inside the intensive properties are not independent and must be related by
curvature), z is the elevation of the point on the interface above a equations of state. Equations 26 and 27 are also useful equations
reference, and ϕ is the angle that the normal to the interface at in converting equations of state from one set of variables to
that point on the interface makes with the vertical direction. another.
Unlike eqs 13−22, the mechanical equilibrium condition in eq 2.2.2. Some Bulk-Phase Equations of State for Use in
23 is not well-known. Equation 23 appears as eq 613 of Gibbs’s Gibbsian Composite-System Thermodynamics. There are
paper,2 but it seems to have gone unnoticed by all but a few more thermodynamic equations of state than there are
researchers17 for 130 years. Gibbs had performed the derivation thermodynamicists. Below are listed only a sample of equations
of the conditions for equilibrium of a sessile drop in a of state that have been useful in applications of Gibbsian
gravitational field in two unconnected parts, one for the curved composite-system thermodynamics.
interface in a gravitational field and one for contact with the The ideal gas law is
solid.2 In 2008, Voitcu and Elliott provided a unified derivation PV = NRT (28)
of all conditions for equilibrium of a sessile drop in a
gravitational field, arriving simultaneously at eqs 10, 11, 16, where R is the universal gas constant. Substituting the ideal gas
22, and 23.8 Although the last term in eq 23 may be negligible in law (eq 28) in the Gibbs−Duhem relation (eq 26) for a single
many practical circumstances, it may become important at larger component and integrating at constant temperature yields the
field magnitudes. dependence of ideal gas chemical potential on pressure:
i P y
μ(T , P) = μref (T , P ref ) + RT lnjjj ref zzz
2.2. Equations of State. In order to make a calculation of an

kP {
equilibrium state of the system in terms of measurable and/or
controllable variables, the conditions for chemical, thermal, and (29)

10863 https://dx.doi.org/10.1021/acs.jpcb.0c05946
J. Phys. Chem. B 2020, 124, 10859−10878
The Journal of Physical Chemistry B pubs.acs.org/JPCB Feature Article

where Pref is a reference pressure and μref is the chemical empirical approaches is to add corrections into the ideal dilute
potential at that reference pressure. solution approaches and then fit to actual data to get a
For an incompressible substance, molar volume v does not parametrization of the corrections that is valid for a specific
depend on pressure. Substituting this into the Gibbs−Duhem solution.
relation (eq 26) for a single component and integrating at In the activity approach,16,20,21 the chemical potentials of
constant temperature yields the dependence of chemical solvent 1 and solute 2 are written as
potential on pressure for an incompressible substance:
μ1(T , P , x1) = μ10 (T , P) + RT ln(γ1x1) (37)
ref ref ref
μ(T , P) = μ (T , P ) + v(P − P ) (30)
and

ÄÅ ÉÑ
A useful chemical potential equation for a slightly

Å Ñ
isothermal compressibility ÅÅÅÅκT = − v ∂P ÑÑÑÑ into the Gibbs−
compressible substance can be obtained by substituting constant μ2 (T , P , x 2) = ψ (T , P) + RT ln(γ2x 2) (38)

ÅÇ TÑ Ö
1 ∂v
( ) where γi is the activity coefficient of species i, and ai = γixi is called
the activity. A variety of models have been developed to capture
Duhem relation (eq 26) for a single component and integrating the dependence of activity coefficients on various variables and
at constant temperature to get18 to facilitate fitting to data.16,20 An important concept in
1 developing such activity coefficient models is thermodynamic
μ(T , P) = μref (T , P ref ) − (v|P − v|P ref )
κT (31) consistency; that is, once the functional form of the activity
coefficient of one species is determined, the functional form of
For a temperature range over which a substance can be the second species is not allowed to be arbitrarily assigned, but
assumed to have a constant molar entropy s, the Gibbs−Duhem rather eqs 37 and 38 along with their activity coefficient models
relation (eq 26) for a single component can be integrated at must satisfy the constant-temperature, constant-pressure form
constant pressure to find the chemical potential dependence on of the Gibbs−Duhem equation (eq 26), i.e.
temperature:19
∑ Nij dμi j = 0
μ(T , P) = μref (T ref , P) − s(T − T ref ) (32) i (39)
For substances at a temperature far enough from the reference 2.2.2.1. Osmotic Virial Equation. In the osmotic virial
temperature that entropy cannot be assumed to be constant, the equation approach,21−29 the correction to the ideal, dilute
chemical potential dependence on temperature can be obtained solution model is in the form of a polynomial expansion of the
by substituting the identity (∂s/∂T)P = CP/T, where CP is the concentration dependence of the solvent chemical potential.
constant-pressure molar heat capacity, into the Gibbs−Duhem This polynomial expansion can be done in different concen-
relation (eq 26) for a single component and integrating at tration units, with each different polynomial expansion giving
constant pressure to get rise to a different solution theory consistent with different
underlying assumptions. In terms of molality, m2 (moles of
μ(T , P) = μref (T ref , P) − s ref (T − T ref ) solute 2 per kg of solvent 1), the chemical potential of the solvent
T T C
is given by
∫ ∫
− ref ref P dT dT
T T T (33)
μ1(T , P , m2) = μ10 (T , P) − RTM1π (40)
For an ideal dilute solution with solvent 1 and solute 2,
considering the entropy of mixing and assuming N2 ≪ N1 yields where M1 is the molar mass of solvent and the osmolality π is
(eqs 13.33 and 13.34 of ref 5) written as a polynomial expansion of molality
μ1(T , P , x 2) = μ10 (T , P) − RTx 2 (34) π = m2 + B2 m22 + C2m23 + ... (41)

where x2 = N2/(N1 + N2) is the mole fraction of solute and where B2 and C2 are, respectively, the second and third osmotic
μ01(T, P) is the chemical potential of pure solvent at the given virial coefficients for use with molality. While there are
temperature and pressure, and thermodynamic approaches for computing values of the osmotic
virial coefficients (see page 132 of ref 16), this is more
μ2 (T , P , x 2) = ψ (T , P) + RT ln(x 2) (35) complicated than for the gaseous virial coefficients, and the
where ψ(T, P) is a concentration-independent function. For a much more practical use of eq 41 is to treat the osmotic virial
solute with a solubility limit, eq 35 can be rewritten in terms of coefficients as empirical constants obtained by fitting to
experimental measurements of osmolality. The osmotic virial

ji x zy
the saturation mole fraction x2∞ as

μ2 (T , P , x 2) = μ20 (T , P) + RT lnjjj 2 zzz


equation was originally developed in terms of molarity by

jx z
MacMillan and Mayer22 and later in terms of molality and mole
k 2∞ {
fraction by Hill.23,24 It is a widely used equation of state for
(36)
aqueous solutions in biology, such that osmometers that
where μ02(T,P) is the chemical potential of pure solute at the measure freezing-point depression, osmotic pressure, or vapor
given temperature and pressure. pressure (phenomena resulting from phase equilibrium equality
There are many equations of state for nonideal multi- of chemical potentials) give readouts in osmolality.
component solutions of which excellent collections and Our group has made several contributions to the osmotic
explanations can be found in refs 16 and 20. Herein, I mention virial equation. First, we proposed a simple extension for
only two empirical approaches that are convenient to use with electrolytes25
Gibbsian composite-system thermodynamics: activity ap-
proaches and the osmotic virial equation. The idea of the π = k 2m2 + B2 (k 2m2)2 + C2(k 2m2)3 + ... (42)

10864 https://dx.doi.org/10.1021/acs.jpcb.0c05946
J. Phys. Chem. B 2020, 124, 10859−10878
The Journal of Physical Chemistry B pubs.acs.org/JPCB Feature Article

where k2 could be referred to as a dissociation constant of


electrolyte solute 2, but it is important to note that k empirically
accounts for more electrolyte effects that just dissociation; so,
the value obtained from fitting to data might not indicate an
actual degree of electrolyte dissociation. Equation 42, truncated
to second order with only two empirical parameters (k2 and B2),
fits sodium chloride solution data as accurately as the Pitzer−
Debye−Huckel equation, a more sophisticated electrolyte
theory requiring 6 empirical parameters.25
Second, we fit experimental data to provide parameters
(dissociation constants and osmotic virial coefficients) for
aqueous solutes of interest for applications in cryobiology
(sodium chloride, potassium chloride, dimethyl sulfoxide,
glycerol, propylene glycol, ethylene glycol, ethanol, methanol,
mannitol, sucrose, dextrose, trehalose, hemoglobin, bovine
serum albumin, ovalbumine,26 and hydroxyethyl starch27).
Third, we contributed combining rules for the osmotic virial
coefficients that when combined with the osmotic virial Figure 2. Predictions (not fits) by the Elliott et al. form of the
equation mixing rule yield the following equations (to third multisolute osmotic virial equation compared with experimental
measurements. The triangles, circles, and squares represent exper-
order) for chemical potential of the solvent in a multisolute imental measurements for each of the multisolute solutions, while the
solution with r − 1 solutes:25,28 colored solid lines represent predictions (not fits) of eq 44 for each
corresponding solution. The triangles (purple) represent data from
μ1(T , P , m2 , m3 , ..., mr ) = μ10 (T , P) − RTM1π
r ÄÅÅ (B + B ) ÉÑ
(43) Hildebrandt’s thesis33 for an aqueous solution of dimethyl sulfoxide

ÅÅ i ÑÑ
π = ∑ kimi + ∑ ∑ ÅÅÅ kimikjmj ÑÑÑÑ
(DMSO) and NaCl, at a solute mass ratio of DMSO:NaCl = 2:1. The

Å ÑÑ
r r

i=2 j=2 Å
circles (blue) represent data from Elliott et al.28 for an aqueous solution

ÅÇ ÑÖ
j
of DMSO and glycerol, at a solute mass ratio of DMSO:glycerol = 2:1.
2
i=2 The squares (green) represent data from Yousef et al.34 for an aqueous
r r r solution of bovine serum albumin (BSA) and ovalbumin, at a solute
+ ∑ ∑ ∑ [(CiCjCk)1/3 kimikjmjkkmk ] mass ratio of BSA:ovalbumin = 3:2. The inset at top left provides a
i=2 j=2 k=2 (44) closer look at the lower end of both the molality and osmolality axes.
Reproduced with permission from ref 29. Copyright 2017 American
Importantly, eq 44 can be used to make predictions of osmolality Chemical Society.
(and therefore freezing-point depression, osmotic pressure,
vapor pressure, etc.) for solutions with one solvent and many
solutes (r components in total) using only tabulated dissociation
constants and osmotic virial coefficients from single-solute yields polynomials, multiple molecular species may be
solutions and not requiring fitting to any multisolute solution represented by a single “grouped solute”, and we show that
data. The combining rules in eq 42, namely using such a grouped-solute approach is mathematically
equivalent to having treated each species individually in the
Bi + Bj multisolute osmotic virial equation.29 This means that the
Bij = multisolute osmotic virial equation can be applied to circum-
2 (45)
stances where the individual molecular solute species cannot be
and enumerated, such as the cytoplasmic solution inside a living
cell.29,35−37
Cijk = (CiCjCk)1/3 (46) Fifth, we derived a solute chemical potential equation29,38 that
are referred to for the gaseous virial equation as “Guggenheim’s is thermodynamically consistent with the solvent chemical
̈ approximation”. However, we point out in Zielinski et al.29
naive potential of the multisolute osmotic virial equation, presented
that because the osmotic virial equation mass balance has one here in molality units:29
more degree of freedom than the gaseous virial equation (due to μs (T , P , m2 , m3 , ..., ms , ..., mr )
the presence of a solvent), the reasons that make these r
combining rules inappropriate for the gaseous virial equation
do not apply to the osmotic virial equation. The multisolute
= ksψs + RTks[ln(M1ms) + ∑ [(Bii + Bss )kimi]
i=2
osmotic virial equation, eq 44, is extremely useful because this r r
single equation can make accurate predictions for combinations 3
of many types of solutes such as organic solvents,25,26,28,30
+ ∑ ∑ [(CiiiCjjjCsss)1/3 kimikjmj]]
2 i=2 j=2 (47)
alcohols,25,26,28,30 sugars,26,30 starches,27 proteins,26,28 electro-
lytes,25,26,30 and ionic31 and nonionic32 micelle-forming where, ψs is a constant that depends only on temperature and
surfactants, whereas most thermodynamic solution theories pressure and is specific to solute s, and M1 is the molar mass of
are only applicable to a small class of molecules and cannot be the solvent.
used for mixtures of molecules from different classes. Three 2.2.2.2. Chemical Potential Equations with Multiple
examples are shown in Figure 2. Dependencies. Finally, note that, through combinations of eq
Fourth, we point out an important feature of the multisolute 34−38, 40, 43, or 47 for chemical potential dependence on
osmotic virial equation. Because the contribution of each solute composition with equations for chemical potential dependence
species is represented by a polynomial, and adding polynomials on pressure (eq 29, 30, or 31) or temperature (eq 32 or 33),
10865 https://dx.doi.org/10.1021/acs.jpcb.0c05946
J. Phys. Chem. B 2020, 124, 10859−10878
The Journal of Physical Chemistry B pubs.acs.org/JPCB Feature Article

chemical potential equations with multiple dependencies can be 2.2.3.1. Adsorption Isotherms. Adsorption isotherms are
found. equations that describe the relationship between the number of
For example, consider the need to compute the colligative adsorbed molecules on a surface (surface excess moles of the
freezing-point depression caused by the presence of a solute in component of interest), Nσ, and the controllable variables of
the absence of any curvature effects of the ice−solution interface temperature, T, and additional independent variables of the
(see Figure 3). adjoining bulk fluid phase such as pressure in an adjoining
single-component gas phase, PG, or composition in the case of an
adjoining multicomponent fluid phase. Because the equilibrium
of a composite system consisting of a surface phase and
adjoining bulk fluid phase requires equality of chemical
potentials for the component adsorbed on the surface and in
the adjoining fluid phase, adsorption isotherms can also be
expressed by the adsorbed component chemical potential
equation with which they are consistent, which is written in
terms of the variables of the adsorbed phase. Adsorption
isotherm equations can have their foundation in empirical
Figure 3. Composite system to consider for freezing-point depression. measurements of adsorption or statistical mechanical deriva-
tions starting with assumptions about properties of the molecule
of interest in the adsorbed and adjoining bulk phases.
A well-known adsorption isotherm is the Langmuir
For clarity, I will assume an aqueous solution, but all that adsorption isotherm, which is the isotherm for noninteracting
follows can be generalized by replacing the word water with the adsorbed molecules that adsorb from a gas phase onto fixed
word solvent. Assuming that a solid composed of pure water is in adsorption sites on a solid that can hold at most one molecule on
equilibrium with an aqueous solution with one solute, the a site. The Langmuir adsorption isotherm40 (eq 7-10 of ref 41)
thermal and mechanical conditions for equilibrium in eqs 10 and and the corresponding chemical potential for the adsorbed
13 indicate that the solid and liquid are at the same temperature molecules (eq 7-8 of ref 41) are given by
and pressure. The chemical potential equilibrium condition for
the pure water solid and the water in the liquid solution can thus MPG
be written from eq 11 as N σ (T , A , PG) =
P + 1/(φGq σ )
G
(52)

ÄÅ ÉÑ
μ1S (Tfp , P) = μ1L (Tfp , P , m2)

ÅÅ ÑÑ
(48) and

μ (T , A , N ) = RT lnÅÅÅ Å ÑÑ
σ σÑ
ÅÅÇ (M − N )q ÑÑÑÖ
where Tfp is the freezing point of the solution. Setting the Nσ
σ σ
freezing point of pure water, T0fp, as the reference temperature,
the chemical potential of the solid pure water ice can be written (53)
from eq 32 as respectively, where M is the number of adsorption sites which is
μ1S (Tfp , P) = μ10 (Tfp0 , P) − s10S(Tfp − Tfp0) proportional to surface area A, φG is a function of temperature
(49)
(the temperature dependence of which depends on the
s0S
where is the molar entropy of pure water solid ice, assumed
1 thermostatistical model used to describe a gas-phase molecule),
for this example to be a constant that is independent of and qσ is the statistical mechanical partition function of a single
temperature. Combining eqs 32 and 40 yields the chemical adsorbed molecule. For the Langmuir adsorption isotherm, qσ is
potential of the water in the liquid solution: a function of temperature. Ward and Elmoselhi42 and Elliott and
Ward43 developed an extension to the Langmuir adsorption
μ1L (Tfp , P , π ) = μ10 (Tfp0 , P) − s10L(Tfp − Tfp0) − RTfpM1π isotherm that allowed for interaction between adsorbed
(50) molecules making qσ a function of both temperature and the
where s0L
1 is the molar entropy of pure water in liquid form, surface density of adsorbed molecules.
assumed to be constant. Substituting eqs 49 and 50 into 48 and The Brunauer−Emmett−Teller (BET) isotherm describes
rearranging yields the freezing-point depression for an aqueous adsorption from a gas onto a solid with fixed adsorption sites
solution with osmolality π:19,26,28,39 where, rather than being limited to one molecule per adsorption
site, any number of molecules can adsorb on a given adsorption
RTfp0M1π
site leading to the amount adsorbed becoming infinite as the gas-
(s10L − s10S)
Tfp0 − Tfp = RM1π
phase saturation pressure is approached and a liquid condenses
1+ on the surface. The BET adsorption isotherm is given by44 (see
(s10L − s10S) (51)
also eqs 7-36 to 7-38 of ref 41)
where the judicious choice of reference temperature as the PG
freezing point of pure water enabled the reference chemical MC P
potentials to cancel out. N σ (T , A , PG) = ∞

PG PG PG
2.2.3. Some Surface Phase Equations of State for Use in
Gibbsian Composite-System Thermodynamics. As with the
(1 − )(1 −
P∞ P∞
+ CP

) (54)
bulk-phase equations of state, I list herein only a few of the where M is the number of adsorption sites which is proportional
surface-phase equations of state that have found use in Gibbsian to surface area A, and P∞ is the saturation pressure (pressure at
composite-system thermodynamics, namely, several adsorption which the gas condenses) which is a function of temperature.
isotherms and one empirical equation for the dependence of While the temperature-dependent constant C can be found from
surface tension on temperature and composition. a rather simplified statistical thermodynamics derivation to be
10866 https://dx.doi.org/10.1021/acs.jpcb.0c05946
J. Phys. Chem. B 2020, 124, 10859−10878
The Journal of Physical Chemistry B pubs.acs.org/JPCB Feature Article

the ratio of the single-molecule partition function of molecules N σ (T , x 2L) Kx 2L


in the first layer of adsorbed molecules to the single-molecule ≡ Γ = Γ∞
A 1 + Kx 2L (58)
partition function of molecules in the second and later layers (eq
7-37 of ref 41), in practice C is used as an empirical fitting where Γ is the surface coverage of adsorbed surfactant molecules
parameter, and the most common usage of the BET isotherm is per unit surface area, Γ∞ = M/A is the maximum possible surface
as a fit to nitrogen adsorption data to measure surface area of coverage of adsorbed surfactant molecules, xL2 is the mole
porous materials (the BET surface area). fraction of surfactant in the liquid phase, and K is a temperature-
Ward and Wu proposed the ζ-isotherm for adsorption from a dependent parameter. Recognizing that changing the units of
gas onto a solid with fixed adsorption sites where a cluster of the solute concentration from mole fraction simply changes the
molecules can adsorb at a given adsorption site and there is a value and units of K, it is common to see eq 58 written in
finite maximum number of molecules per cluster:45 solution concentration units other than mole fraction.
Zhu and Gu48 proposed a general isotherm as an extension of

ÅÄÅ (1 + ζ ) Ñ
ÑÑ
É
N σ (T , A , PG)
PG Å
Mcα P ÅÅÅÅ1 − (1 + ζ ) α P ÑÑ
eq 58

∞Å
ÑÑ
ÑÑÖ
ζ

ÅÇ
PG PG K (c 2L)n
( ) ( )
ÄÅ É
+ ζ αP

Å Ñ
(1 + ζ ) Ñ
Γ = Γ∞

PG Å ÑÑ
∞ ∞

1 − α P ÅÅÅÅ1 + (c − 1)α P − c α P
1 + K (c 2L)n
ÑÑ
= (59)

∞ Å ÑÑ
ÅÇ ÑÖ
PG PG
( ) ∞ ∞
( ) (55)
cL2
where is the concentration of surfactant in the liquid phase.
Originally, n was understood to be the number of surfactant
molecules in a hemimicelle adsorbed on a solid from a liquid
where α and c are temperature-dependent parameters and ζ is (loosely analogous to ζ in the ζ-isotherm) but practically can be
the maximum number of molecules that can adsorb in one left as an empirical fitting parameter to describe a wide variety of
cluster. The ζ-isotherm has been found to provide a good fluid−solid adsorption isotherms. Equation 59 (sometimes
representation of experimental data for adsorption from a gas or referred to as the Langmuir−Freundlich isotherm) is able to
vapor onto a wide variety of solids for which the data is not well- reproduce experimental features such as “S-shaped” adsorption
represented by the BET adsorption isotherm, and to provide a isotherms. Wang, Xu, and Acosta used this isotherm for each
means to obtain values of important material properties from layer of a bilayer to successfully describe adsorption of
adsorption measurements.46,47 surfactants on iron oxide nanoparticles.49
For adsorption at a fluid interface with no fixed adsorption Finally, note that, by substituting any particular adsorption
sites, we take advantage of the chemical potential equilibrium isotherm chemical potential equations into the Gibbs adsorption
between the adsorbed molecules and the solute molecules in the equation, eq 27, noting that isotherms are at constant
adjoining bulk fluid phase. For example, if there is an ideal, dilute temperature, and integrating, important relationships may be
solution of surfactant (surfactant solute 2 in solvent 1) in found between surface tension and adsorption isotherm
equilibrium with surfactant adsorbed at a fluid interface, parameters and controllable variables of the adjoining bulk
equating the chemical potential of surfactant in solution (from phase, such as gas-phase pressure or liquid-phase solute
a combination of eqs 30 and 36) to the chemical potential of the concentration. These relationships are commonly exploited to
surfactant adsorbed at the fluid interface that can contain at most obtain adsorption isotherm parameters from surface tension
M adsorbed molecules (using the chemical potential corre- measurements or to obtain inaccessible surface tension values
sponding to Langmuir adsorption eq 53) yields from knowledge of adsorption isotherm parameters.46,50−52
ij yz
μ20,ref (T , P ref ) + v20(P − P ref ) + RT lnjjj L 2 zz
j x (T , P) zz
2.2.3.2. Shardt−Elliott−Connors−Wright Equation for
xL
k 2∞ {
Surface Tension. The liquid−vapor surface tension depends

ÅÄÅ ÑÉÑ
on temperature and liquid-phase composition. Inserting
Å ÑÑÑ
= RT lnÅÅÅÅ Ñ
temperature-dependent pure species surface tensions into the

ÅÅÇ (M − N σ )q σ ÑÑÑÖ
σ
N Connors−Wright model for the composition dependence of
(56) surface tension at a single temperature gives a useful empirical
equation describing the surface tension dependence on
where qσ is the single-particle partition function for surfactant temperature and composition:53
molecules adsorbed at the fluid interface, assumed to be a
ÄÅ ÉÑ
ÅÅ b(1 − x1) ÑÑÑ
function of temperature. Rearranging eq 56 yields σmix(T , x1) = σ2(T )
Å
− ÅÅÅ1 + ÑÑx1[σ2(T ) − σ1(T )]
Ä ÅÅÇ 1 − a(1 − x1) ÑÑÑÖ
l
o σ ij μ20,ref yz ÅÅÅ v20 ÑÉÑ |
N σ (T , x 2L) =
o L
mq expjj RT zzexpÅÅÅ RT (P − P ref )ÑÑÑÑ/x 2L∞o
o }x 2
k { Å
Ç Ñ
Ö
M n ~
(60)

l Ä ÑÉ L |
o σ ij μ20,ref yz ÅÅÅ v20 ref Ñ Ñ o L
1+m oq expjj RT zzexpÅÅÅÅÇ RT (P − P )ÑÑÑÑÖ/x 2∞}
where a and b are constants that are independent of temperature
ox 2
n k { ~
and composition. The benefit of eq 60 is that, compared with
(57) other approaches, a minimum amount of data is required to
obtain the parameters to make predictions for any temperature
and composition. The temperature-dependent functions σ1(T)
ÅÄÅ v 0 ÑÉÑ
Assuming the pressure effect on chemical potential of the

expÅÅÅ RT2 (P − P ref )ÑÑÑ ≈ 1 (i.e., neglecting the Poynting correc-


surfactant in a liquid solution is negligible, and σ2(T) can be found by fitting data for the temperature

ÅÅÇ ÑÑÖ
dependence of pure component 1 and pure component 2, and
the constants a and b can be found by fitting data for
tion), xL2∞ can be considered to be a function of temperature composition dependence at a single temperature (for example,
only, and eq 57 yields a Langmuir-like adsorption isotherm: room temperature). Equation 60 (and extensions for
10867 https://dx.doi.org/10.1021/acs.jpcb.0c05946
J. Phys. Chem. B 2020, 124, 10859−10878
The Journal of Physical Chemistry B pubs.acs.org/JPCB Feature Article

components above their pure species critical points and for and treating the liquid as compressible.58 Also, for a single-
mixtures of more than two components54) showed excellent component fluid, Wang et al. compared eq 64 to treating both
accuracy across a wide variety of mixtures, including 15 different the liquid and vapor phases with a cubic equation of state and
aqueous mixtures,53 15 different hydrocarbon and organic reviewed other works.59
mixtures,54 and a ternary mixture of methanol/ethanol/water.54 Ward and Levart presented a form of the Kelvin equation for
2.3. Combined Conditions for Equilibrium That Reveal the equilibrium radius of a bubble (critical radius for nucleation)
the Effects of Curved Interfaces. For a state to be a in a liquid with a dilute amount of dissolved gas:60

ÅÄÅ v10L L É
thermodynamic equilibrium state, it must satisfy all three of

nLÑ Ñ
2σ LV
P1∞ÅÅÅ RT (P − P1∞) − 2L ÑÑÑ + P L
thermal, chemical, and mechanical equilibrium. A particular set

ÅÇÅ n1 Ñ
RC =
ÑÖ
of combined conditions for equilibrium are of interest: those n2L
describing phase equilibrium across a curved interface. The ( )−P
n2L∞
L
(65)
combination of eq 10 for thermal equilibrium, eq 11 for chemical
equilibrium, and eq 15 for mechanical equilibrium across a where P1∞ is the flat-interface saturation pressure of the pure
curved interface, along with chosen equations of state for the liquid, v0L L L
1 is the molar volume of the pure liquid at P and T, n2 is
bulk phases, yields a single equation quantifying how interfacial the molar concentration of gas dissolved in the liquid phase, and
curvature and interfacial tension shift thermal or chemical nL2∞ is the saturation molar concentration of dissolved gas (i.e.,
equilibrium. the amount that would be dissolved in the liquid phase were the
2.3.1. Kelvin Equation. The Kelvin equation describes the liquid to be in equilibrium with a gas phase across a flat interface,
change in equilibrium vapor pressure above a curved liquid described for example by Henry’s law). Note that eq 65 has been
interface from that above a flat interface. Consider a single- written with a positive interface curvature defined as curved
component liquid droplet surrounded by its vapor. Noting from toward the vapor (i.e., a vapor bubble) in contrast with eq 64
eq 10 that the temperatures of the liquid and vapor phases are that was written with a positive interface curvature defined as
equal, eq 11 for chemical equilibrium requires curved toward the liquid. Ward’s and Levart’s treatment of
curved interface phase equilibrium in a dilute liquid−gas
μ V (T , P V ) = μ L (T , P L ) (61) solution was used by Zargarzadeh and Elliott for nucleation of
drops and bubbles at solid interfaces with a variety of
Assuming that the vapor is an ideal gas (eq 29) and assuming geometries.61,62

ij PV yz
that the liquid is incompressible (eq 30), eq 61 may be written With the goal of elucidating the effect of interface curvature on

μref (T , P∞) + RT lnjjj zzz = μref (T , P∞) + v L(P L − P∞)


jP z
multicomponent vapor−liquid equilibrium phase diagrams

k ∞{
across the complete composition range of the liquid and the
vapor, Shardt and Elliott incorporated activity coefficients to
(62) capture the thermodynamic nonideality of the liquid phase
Due to the careful choice of the reference pressure as the (while still treating the vapor phase as an ideal gas) and derived
saturation vapor pressure at a flat interface, P∞, the reference forms of the Kelvin equation describing the vapor-phase

Ä ÉÑ|
pressure at the bubble point13,63
l 0L Å
ji P zy
chemical potentials cancel, yielding
o vi ÅÅÅÅ V
o i1
LV j 1 zyz ÑÑo
Ño
RT lnjjj zzz = v L(P L − P∞) m
γi LxiLPi∞expo ÅÅPbubble + σ jj j zz − Pi∞ÑÑÑo }
jP z o Å j R1 z ÑÑÑÖo
V

o RT ÅÅÇ o
k ∞{ n k 2{ ~
V
Pbubble = ∑ +
(63) i
R
(66)
Substituting the mechanical equilibrium condition at a curved
interface, eq 15 into eq 63, yields the complete Kelvin equation and at the dew point13,63
for the difference in the vapor-phase pressure PV above a curved V

ij yz
liquid interface from the saturation pressure of a flat interface Pdew =

jj zz
jj zz
P∞:7,11,55,56

ji P zy ji 1 1 zyz
−1

jj zz
RT lnjjj zzz = v Lσ LV jjj zzz + v L(PV − P∞) jj∑ Ä
Å V ÉÑ zz
jP z jR
V

jj vi0L Å Ñ zz
V yi

k { k { jj i γ LPi∞exp ÅÅP + σ LV − Pi∞ÑÑÑÑ zz


j RT Å ÅÅÇ dew z
+

k
{ ÑÖ
} {
R
∞ 1 2 (64)
i ( 1
R1
+
1
R2 )
where the equation has been written with a positive interface
(67)
curvature defined as curved toward the liquid (i.e., a liquid
droplet). W. Thomson (Baron Kelvin) was the first to note that a where the sums are over all components i, is the activity γLi
curved liquid−vapor interface would alter the vapor-phase coefficient of component i in the liquid phase, xLi is the mole
pressure,57 but the equation he proposed was missing the last fraction of component i in the liquid phase, Pi∞ is the flat-
term, which could not be understood until Gibbs’s later paper.2 interface saturation pressure of pure component i, v0L i is the
For that reason, eq 64 is called the complete Kelvin equation, but molar volume of pure liquid component i, and yVi is the mole
a more apt name might be the Gibbs−Kelvin equation fraction of component i in the vapor phase. Shardt and Elliott
recognizing Gibbs’s insight underpinning the derivation of eq presented numerical methods for solving eqs 66 and 67 along
64. with requisite balance equations to draw isothermal composi-
Equation 64 applies for a single-component liquid−vapor tion phase diagrams63 and isobaric composition phase
system where the liquid is treated as an incompressible diagrams.13 The isobaric phase diagrams are more challenging
substance and the vapor is treated as an ideal gas. Several to draw because of the strong temperature dependence of the
authors have made extensions to eq 64 for nonideal or parameters γLi , Pi∞, v0L
i , and σ which are almost independent of
LV

multicomponent phases.13,18,58−63 For a single-component 13


pressure. The success of eq 67 combined with the Shardt−
fluid, Melrose introduced treating the vapor as a nonideal gas Elliott−Connors−Wright equation for surface tension, eq 60,
10868 https://dx.doi.org/10.1021/acs.jpcb.0c05946
J. Phys. Chem. B 2020, 124, 10859−10878
The Journal of Physical Chemistry B pubs.acs.org/JPCB Feature Article

for predicting dew temperatures of nitrogen/argon mixtures in 2 where the reference state for the chemical potentials has been
nm capillary pores is shown in Figure 4. chosen as the liquid-phase pressure and the corresponding
freezing point at a flat interface, T0fp, so that the reference
chemical potentials cancel out of the equation. Substituting the
mechanical equilibrium condition (eq 15) for the solid−liquid
pressure difference into eq 71 yields the Gibbs−Thomson
equation for interface curvature-induced freezing-point depres-
sion:19,66−69

v Sσ SL ijj 1 1 yzz
S j
jj z
(s − s ) k R1 R 2 zz{
Tfp0 − Tfp = L
+
(72)

where for eq 72 as written, (1/R1 + 1/R2), which is twice the


mean curvature, is defined as positive when the solid is inside the
curvature.
The Gibbs−Thomson effect of curvature-induced freezing-
point depression means that ice can be prevented from growing
into small spaces at temperatures below the normal freezing
Figure 4. Predictions (not fits) by eq 67 of adsorption and desorption point, a phenomenon that is responsible for the action of
dew temperatures as functions of nitrogen mole fraction for antifreeze proteins70,71 and can cause ice-lensing in soils leading
nitrogen(1)/argon(2) mixtures in a capillary with a radius of 2 nm to frost heave.72,73 Karlsson et al. demonstrated that the Gibbs−
compared with experimental measurements of adsorption and Thomson equation was consistent with experimental measure-
desorption at a vapor-phase pressure of 300 kPa from Alam et al.64 ments of the inhibition of ice growth through biological pores
The surface tension of the nitrogen/argon mixtures as a function of and gaps with diameters or gap dimensions ranging from 0.7 to
temperature and pressure was predicted by eq 60. The schematic of the 35 nm.69 The Gibbs−Thomson effect was exploited to build
classical model of adsorption and desorption is adapted from Donohue supercapacitors with aqueous electrolytes that remained func-
and Aranovich.65 Reproduced from with permission from ref 13.
Copyright 2018 American Chemical Society.
tional to −30 °C because a nanostructured polyampholyte
electrolyte could maintain liquid water for ion transport within
its nanopores at low temperatures.74,75
Liu et al. derived a Gibbs−Thomson equation for the
Here, I note that eqs 66 and 67 can be extended for a nonideal curvature-induced freezing-point depression in a nonideal
vapor phase by incorporating fugacity coefficients solution:12

0L Å
Ä ÉÑ|
γi LxiLϕi∞Pi∞ l
V

o vi ÅÅÅÅ V
o i1 1 zyz ÑÑo
Pbubble =

LV j Ño
RTfp0M1π

j
(Ä ) ÉÑ +
zz − Pi∞ÑÑÑ}
1 1
v1Sσ SL
expm ÅPbubble + σ jj ÅÅ
+
o RT ÅÅÅÅÇ ÑÑo ÑÑ
o j R1 z
ÑÖo (s10L − s10S)ÅÅÅÅ1 +
(s10L − s10S)

ÑÑ
R1 R2

o k 2{ o
Tfp0 − Tfp =
n ~ ÑÑÖ

ÅÇ
V
+ RM1π RM1π
ϕi ̂ R 1+
i
(s10L − s10S) (s10L − s10S)
(68) (73)
V

jij zyz
Pdew = The first term on the right-hand side of eq 73 is the capillary
jj zz
jj zz
−1
freezing-point depression (which depends on solution concen-
jj∑ ÄÅ ÉÑ zz
jj zz
tration through osmolality π), and the second term on the right-
V
ϕî yiV
jj i vi0L Å
ÅÅPV ÑÑ zz
jj Ñ zz
RT Å ÅÅ dew i∞Ñ
Ñ
hand side of eq 73 is the solution freezing-point depression. In
k
{ Ç
( )−P ÑÖ
} {
1 1
γi Lϕi∞Pi∞exp + σ LV R1
+ R2
cases where RM1π/(s0L 1 − s1 ) ≪ 1, eq 73 can be linearized with
0S

respect to osmolality to give an equation where the capillary


(69) freezing-point depression does not depend on solution
where ϕi∞ is the fugacity coefficient of pure vapor component i concentration, and the solution freezing-point depression is
and ϕ̂ Vi is the fugacity coefficient of component i in the vapor- proportional to osmolality.19 However, care should be taken
phase mixture. because for concentrated solutions such a linearization may be
2.3.2. Gibbs−Thomson Equation. The Gibbs−Thomson significantly in error.39 Equation 73 reduces to eq 51 in the case
equation describes curvature-induced freezing-point depression. of a flat solid−liquid interface (infinite principal radii of
Thermal and chemical equilibrium (eqs 10 and 11) at a solid− curvature R1 and R2) and reduces to eq 72 in the case of a
liquid interface of a single-component system requires that at the pure liquid (π = 0).
freezing point Tfp Equation 73 can alternatively be written in terms of the
activity coefficient of the solvent (the component that freezes) in
μS (Tfp , P S) = μL (Tfp , P L) (70) the liquid-phase mixture, γL1 , and liquid-phase mole fraction of
Assuming that the liquid and solid are incompressible (eq 30), solvent, xL1 , rather than osmolality12
and that eq 32 applies for the temperature dependence of both RTfp0 ln(γ1Lx1L)

v1Sσ SL
1 1
) ÉÑ −
ÅÅ
solid and liquid phases, eq 70 may be written +
ÑÑ
− s10S)ÅÅÅÅ1 −
(s 0L − s 0S)

ÑÑ
R1 R2
Tfp0 − Tfp = 1 1

ÅÇ ÑÑÖ
ref
μ (Tfp0 , L
P ) − s (Tfp − S
Tfp0) S
+ v (P − P ) L S
R ln(γ1Lx1L) R ln(γ1Lx1L)
(s10L 1 − 0L 0S
(s10L − s10S) (s1 − s1 )
ref
=μ (Tfp0 , L L
P ) − s (Tfp − Tfp0) (71) (74)

10869 https://dx.doi.org/10.1021/acs.jpcb.0c05946
J. Phys. Chem. B 2020, 124, 10859−10878
The Journal of Physical Chemistry B pubs.acs.org/JPCB Feature Article

Figure 5. (a) Fit of the two-parameter Margules activity coefficient equation to the flat-interface experimental solid−liquid equilibrium data80 for the
water/glycerol system at a pressure of 1 atm. (b) Using the activity coefficients obtained from the flat-interface solid−liquid phase diagram, eq 76 was
used to predict the depression of the freezing liquidus and eq 80 was used to predict the depression of the precipitating liquidus for mean curvature
(1/R1 + 1/R2)/2 of 1/(10 nm) curved toward the solid. Predictions are done over a range of possible glycerol solid−liquid interfacial tension values
because of a lack of data for a precise calculation. Adapted (with additional annotations here in panel a) with permission from ref 12. Copyright 2017
American Chemical Society.

and either of eqs 73 or 74 can be rearranged as a single fraction if The Ostwald−Freundlich equation can explain the desirable
desired: increase in solubility of pharmaceutical formulations with
smaller particle sizes.78 Pouralhosseini et al. used the

Tfp0 − Tfp =
(
v1Sσ SL
1
R1
+
1
R2 ) + RT M π
0
fp 1 Ostwald−Freundlich equation to explain the asphaltene solid
size dependence of the phase behavior of asphaltene/toluene/
(s10L − s10S) + RM1π (75) polysterene mixtures.79
Eslami and Elliott77 and Liu et al.12 derived versions of the

Tfp0 − Tfp =
v1Sσ SL ( 1
R1
+
1
R2 ) − RT ln(γ x )
0
fp
L L
1 1
Ostwald−Freundlich equation for nondilute solutions. For
example, Liu et al.’s equation is12
(s10L − s10S) − R ln(γ1Lx1L) (76)
0
Tfp,2 − Tprecipitation
2.3.3. Ostwald−Freundlich Equation. The Ostwald−
Freundlich equation describes the increased solubility that a
highly curved solid precipitate has in a solution. Assuming that a =
(
v2Sσprecipitate
SL 1
R1
+
1
R2 ) − RT 0 L L
fp,2 ln(γ2 x 2 )

pure solid precipitate is in equilibrium with a solution of that (s20L − s20S) − R ln(γ2Lx 2L) (80)
solute requires (eqs 10 and 11)
where xL2 is the osmole fraction of solute in the liquid phase in
μ2S (T , P S) = μ2L (T , P L , x 2L) (77) equilibrium with solid precipitate 2 at the precipitation
0
where xL2 is the mole faction of the solute in the liquid solution. temperature Tprecipitation. Tfp,2 is the equilibrium freezing
Substituting eqs 30 and 36 into eq 77 yields temperature of pure solute.

ij x L yz
μ2S (T , P S) = μ20 (T , P S) − v20L(P S − P L) + RT lnjjj L2 zzz
The equivalence of the nonideal Ostwald−Freundlich

jx z
equation, eq 80, with the nonideal Gibbs−Thomson equation,
k 2∞ {
eq 76, was noted by Liu et al.12 Equations 76 and 80 are the
equilibrium equations for a pure solid with a curved interface in
(78) equilibrium with a solution, the only difference being whether
Substituting eq 15 into eq 78, canceling the reference chemical the solid is made up of frozen solvent (component 1, eq 76) or
potential, and rearranging yields the Ostwald−Freundlich precipitated solute (component 2, eq 80). Equations 76 and 80

ÄÅ 0L SL É
yzÑÑÑÑ
equation:3,12,76,77
ÅÅ v σ i 1
can be used together to describe the effect of solid−liquid

Å j
jj zzÑÑ
x 2 = x 2∞expÅÅÅ
interface curvature (or equivalently solution confinement which

ÅÅ RT jj R1 R 2 zz{ÑÑÑÑÖ
k
1
ÅÇ
L L 2 enforces a solid−liquid interface curvature for solid nucleation)
+
(79)
on a binary solid−liquid phase equilibrium diagram across the
entire composition range.
where xL2∞ is the saturation mole fraction of solute in a liquid in Note that, without deriving actual forms of the Gibbs−
equilibrium with a flat solid−liquid interface, and xL2 is the Thomson or Ostwald−Freundlich equations, with eq 661 of his
enhanced saturation mole fraction in the liquid in equilibrium paper and the discussion that follows, Gibbs provided the setup
with a solid−liquid interface that is curved toward the solid. for solid−fluid equilibrium (in the case of no solid stresses other
10870 https://dx.doi.org/10.1021/acs.jpcb.0c05946
J. Phys. Chem. B 2020, 124, 10859−10878
The Journal of Physical Chemistry B pubs.acs.org/JPCB Feature Article

than isotropic pressure) as the combination of the equality of and they do not indicate the nature of those equilibrium states
chemical potentials with the Young−Laplace equation.2 (stable, metastable, or unstable). For these things, we must find
Figure 5a shows the flat-interface solid−liquid phase diagram which free energy acts as the thermodynamic potential for a
for water(1)/glycerol(2). The freezing liquidus is on the left- given system and examine the extrema of this function.
hand side of the diagram, and the precipitating liquidus is on the The free energy of the system is the energy of the system that
right-hand side of the diagram. If a liquid water/glycerol solution is available to be changed on a path toward equilibrium once
with low solute concentration is cooled, the freezing liquidus is thermodynamic and physical requirements for a given system
reached and pure water ice freezes out of solution. If a liquid have been imposed. The free energy will be a global minimum at
water/glycerol solution with high solute concentration is cooled, stable equilibrium, a local minimum at metastable equilibrium,
the precipitating liquidus is reached and pure solute glycerol and a maximum or inflection (or saddle in multidimensions) at
precipitates out of solution. Figure 5b shows how eqs 76 and 80 unstable equilibrium. When a system that is not in equilibrium
predict that solid−liquid interface curvature will depress both evolves toward equilibrium, the entropy of the system plus the
the freezing liquidus and the precipitating liquidus. surrounding reservoir tends to increase, subject to conservation
2.3.4. Curvature-Induced Shift of Eutectic Temperature. of energy, conservation of mass, continuity of volume, and any
The freezing liquidus and precipitating liquidus meet at the other system-specific constraints. This evolution will occur until
eutectic point. The eutectic point is important because the it is no longer favored, i.e., entropy becomes a maximum, at
eutectic concentration defines what solid will first come out of which point the system will be in a stable equilibrium state and
solution as a liquid solution is cooled and also because no liquid will not change further. Thus, another way of defining free
can exist below the eutectic temperature. Figure 5b shows that energy is the energy that when minimized corresponds to
the Gibbs−Thomson and Ostwald−Freundlich effects will also entropy being maximized subject to constraints. The method of
depress the eutectic temperature and shift the eutectic finding what function acts as the free energy of a given system
concentration. The eutectic concentration may be shifted up follows that described for a single bulk phase on pages 87−91 of
or down from the flat-interface eutectic composition depending the Ox Bow Press version of Gibbs’s paper2 and pages 153−159
on the relative magnitudes of solid−liquid interfacial tensions of of Callen’s book.5 Detailed descriptions of how to find what
the frozen ice and precipitated solute. Noting that, at the eutectic function acts as the thermodynamic potential or free energy of a
mole fraction of solute in the liquid, xE, the freezing temperature composite system that is not a simple system (e.g., a multiphase
and precipitating temperature are equal, eqs 76 and 80 can be system with a curved fluid interface with interfacial tension
rearranged for Tfp and Tprecipitation, respectively, and equated to effects) in a way that is consistent with Gibbsian composite-
yield an equation for the curvature-affected eutectic composi- system thermodynamics have been given by Ward and Levart,60
tion in a binary system:12 Elliott,7 Elliott and Voitcu,81 and Zargarzadeh and Elliott.11
Briefly, as a system evolves toward equilibrium, Planck’s
v1Sσice
SL
( 1
R1
+
1
R2 )−T 0 0L
fp,1(s1 − s10S) statement82,83 of the second law of thermodynamics states that
for macroscopic spontaneous changes
R ln[γ1L(1 − x E)] − (s10L − s10S)
ΔS reservoir + ΔS system ≥ 0 (82)

=
v2Sσprecipitate
SL 1
R1( +
1
R2 ) − 0
Tfp,2(s20L − s20S) when the system plus surrounding reservoir form an isolated
R ln(γ2Lx E) − (s20L − s20S) system. This is subject to conservation of energy,
(81)
ΔU reservoir + ΔU system = 0 (83)
The eutectic composition found from eq 81 can be substituted
into either eq 76 or eq 80 to obtain the curvature-depressed Combining eqs 82 and 83 with any applicable conservation of
eutectic temperature. Depression of the eutectic temperature by mass equations of the form of eq 7, any applicable continuity of
curvature of the solid−liquid interface (or equivalently solution volume equations of the form of eq 6, and any conditions for
confinement which enforces a solid−liquid interface curvature equilibrium that explicitly describe the interaction of the system
for solid nucleation) is the reason that liquid water (amorphous, with the reservoir, one arrives at an equation of the form
mobile water molecules) can exist in a water/sodium chloride Δ(free energy) ≤ 0 (84)
solution confined in nanochannels in a nanostructured
polyampholyte hydrogel at temperatures as low as −49 °C wherein the relevant free energy is identified. Additional
which is far below the water/sodium chloride normal eutectic conditions for equilibrium (i.e., those equilibrium conditions
temperature of −21 °C.75 that describe relationships among constituent subsystems of the
2.4. Procedure for Finding and Examining the System system, rather than relationships with the reservoir) can be
Free Energy. Sections 2.1−2.3 have described the first three substituted in the free energy to write the free energy in other
steps of Gibbsian composite-system thermodynamics: (i) convenient forms. Finally, to make use of the free energy, it is
describe the system so that the mathematical constraints evaluated with respect to a reference state and then plotted with
corresponding to the system description are clear; (ii) derive respect to a system variable, or variables, of interest to identify at
the conditions for thermal, chemical, and mechanical equili- which values the free energy is an extremum (global minimum,
brium by extremizing entropy subject to the constraints; and local minimum, inflection point (saddle), or maximum).
(iii) combine the conditions for equilibrium with equations of Extrema occur at places where the first derivative of the free
state for each phase to arrive at equations that can be solved to energy vanishes. The number of different extrema identifies the
find the properties of interest when the system is in a particular number of different states that satisfy the conditions for
equilibrium state. The first three steps are very useful in that one equilibrium. The second derivative of the free energy identifies
can use the resulting equations to compute many things of the nature of each of those equilibrium states: global minima
interest. However, they do not indicate the number of individual correspond to stable equilibrium states, local minima corre-
states that satisfy the equilibrium conditions for a given system, spond to metastable equilibrium states, and maxima correspond
10871 https://dx.doi.org/10.1021/acs.jpcb.0c05946
J. Phys. Chem. B 2020, 124, 10859−10878
The Journal of Physical Chemistry B pubs.acs.org/JPCB Feature Article

to unstable equilibrium states. Inflection points with respect to because they are imposed by the reservoir and can therefore be
one variable of interest, or saddle points with respect to multiple brought inside the Δ), yields
variables of interest, also correspond to unstable equilibrium
states, albeit where the system is stable with regard to some Δ(U L − T LS L + P LV L) ≤ 0 (93)
changes and unstable with respect to other changes. identifying the free energy (the expression inside the brackets in
As an example, consider the system shown in Figure 6a. A eq 93) as the Gibbs free energy of the liquid. Gibbs free energy is
single-phase liquid system is contained in a piston−cylinder defined as G = U − TS + PV. So, we see that for a single-phase
liquid system with its temperature and pressure fixed by an
external reservoir, maximizing entropy of the system plus the
reservoir subject to constraints is the same as minimizing the
Gibbs free energy of the liquid.
As another example, consider the system shown in Figure 6b.
A liquid−vapor system is contained in a piston−cylinder device
which imposes the reservoir temperature on the system and the
reservoir pressure on the liquid phase. The piston−cylinder
device and its interface with the liquid are assumed to belong to
the reservoir. Conservation of energy, eq 83, implies
ΔU R = −ΔU L − ΔUV − ΔU LV (94)
Continuity of volume implies
Figure 6. (a) Schematic diagram of a single-phase liquid system
contained in a piston−cylinder device which imposes the temperature ΔV R = −ΔV L − ΔV V (95)
and pressure of the surrounding reservoir on the system. (b) Schematic
diagram of a multiphase fluid system with a vapor bubble surrounded by The conditions for equilibrium that explicitly describe the
liquid contained in a piston−cylinder device which imposes the interaction of the system with the reservoir are
temperature of the surrounding reservoir on the system, and the
pressure of the surrounding reservoir on the liquid phase.
T R = T L = T V = T LV (96)
and
PR = PL (97)
R
device which imposes the reservoir temperature, T , and the Equation 82 implies that for any macroscopic spontaneous
reservoir pressure, PR, on the system. The piston−cylinder changes as the system evolves toward equilibrium
device and its interface with the liquid is considered to form part
of the reservoir, and the system is only the liquid. The difference ΔS R ≥ −ΔS L − ΔS V − ΔS LV (98)
form of the fundamental relation for the reservoir can be written
Substituting eqs 87 and 94−98 in eq 85 yields
ΔU R = T R ΔS R − P R ΔV R + ∑ μiR ΔNiR
(85) Δ[(U L − T LS L + P LV L) + (UV − T VS V )
i

Conservation of energy, eq 83, implies + (U LV − T LVS LV ) + P LV V ] ≤ 0 (99)

ΔU R = −ΔU L (86) or
The system does not exchange mass with the reservoir, so for Δ(GL + F V + F LV + P LV V ) ≤ 0 (100)
every species in the reservoir
where F is the Helmholtz free energy, defined as F = U − TS. The
ΔNiR = 0 (87) free energy in the brackets in eq 100, or similar ones, has been
given the symbol B since it corresponds to none of the well-
Continuity of volume implies known free energies.7,11,56,60,81
ΔV R = −ΔV L (88) Equation 99 can be written in another form. Substituting
Euler relations eqs 24 and 25 into the free energies inside the
The conditions for equilibrium that explicitly describe the brackets of eq 99 yields
interaction of the system with the reservoir are
TR = TL (89) B = (P L − PV )V V + σ LVALV + ∑ μiL NiL + ∑ μi V NiV
i i
and
+ ∑ μi LV
NiLV
PR = PL (90) i (101)
Substituting eqs 86−90 in eq 85 yields The Laplace equation, eq 15, is a condition for equilibrium that
L L
0 = ΔU + T ΔS + P ΔV R L L (91) is internal to the system. When eq 15 is substituted into eq 101,
the free energy may be written
Equation 82 implies that for any macroscopic spontaneous
changes as the system evolves toward equilibrium 2σ LV V
B=− V + σ LVALV + ∑ μiL NiL + ∑ μi V NiV
R L R eq
ΔS ≥ −ΔS (92) i i

Substituting eq 92 in eq 91, and rearranging (noting that T and L + ∑ μi LV


NiLV
PL are constant with respect to the evolution toward equilibrium i (102)

10872 https://dx.doi.org/10.1021/acs.jpcb.0c05946
J. Phys. Chem. B 2020, 124, 10859−10878
The Journal of Physical Chemistry B pubs.acs.org/JPCB Feature Article

Figure 7. (a) Nucleation of a liquid drop on a rigid surface. (b) Nucleation of a liquid drop on a fluid surface. (c) Free energy versus water drop volume
for drop nucleation on a fluid (dodecane) substrate or rigid substrate with the same interfacial tensions where T = 22 °C, PV = 2700 Pa, σLV = σwater =
72.43 mN/m, σSV = σ L′V = σdodecae = 25.3 mN/m, and σSL = σ L′L = σwater−dodecae = 53.7 mN/m. Adapted (with additional annotations here) with
permission from ref 9. Copyright 2011 American Chemical Society.

where Req is the radius of the vapor bubble at equilibrium. In Figure 8, the free energy is examined for a bubble growing
So, we see that, for a liquid system with its temperature and out of a water−nitrogen solution where the nucleation happens
pressure fixed by an external reservoir, maximizing entropy inside a finite conical pit in a solid.62 In the case examined, there
subject to constraints is the same as minimizing the Gibbs free are two equilibrium states. There is an unstable equilibrium at a
energy of the liquid. However, for a system that consists of liquid small bubble size indicating the energy barrier that must be
with an internal vapor bubble with the system temperature and overcome for the bubble to nucleate inside the cone, and there is
liquid pressure fixed by an external reservoir, maximizing a stable equilibrium at a larger bubble size that occurs when the
entropy of the system plus reservoir subject to constraints is the bubble is pinned at the cone mouth.62 Depending on the values
same as minimizing the system free energy B. of system parameters, the final stable equilibrium state may
Examples of evaluating the free energy with respect to a occur inside the cone, pinned to the corner of the cone, or
reference state and plotting versus a variable of interest to outside the cone.62
understand the system behavior are shown in Figures 7 and 8. In
Figure 7, the free energy is examined for a water drop nucleating 3. APPLICATIONS OF GIBBSIAN COMPOSITE-SYSTEM
from water vapor on either a rigid or fluid surface in a system
THERMODYNAMICS
surrounded by a reservoir that imposes the reservoir temper-
ature on the system and the reservoir pressure on the vapor While Gibbs created the underlying fabric for composite-system
phase.9 In either case, there is one unstable equilibrium state thermodynamics,2 it was not until various equations of state
(place where the free energy is a maximum). The value of the could be used to solve the conditions for equilibrium and
free energy at the unstable equilibrium quantifies the energy compute the free energy for a particular circumstance that the
barrier that must be overcome in order for a nucleation event to true power of the framework could be wielded. Gibbsian
take place. The difference in energy barriers indicates that composite-system thermodynamics governs the behavior of any
nucleation occurs more readily on a fluid surface than on a rigid system with two or more fluid systems in equilibrium and as new
surface with the same interfacial tension values.9 experimental techniques, technologies, and scientific questions
10873 https://dx.doi.org/10.1021/acs.jpcb.0c05946
J. Phys. Chem. B 2020, 124, 10859−10878
The Journal of Physical Chemistry B pubs.acs.org/JPCB Feature Article

Figure 8. Equilibrium states of a bubble starting with a convex meniscus in a finite cone submerged in a liquid−gas solution. (a) Free energy versus
height of the center of the liquid−vapor interface. (b) Magnification of the energy axis for free energy versus height of the center of the liquid−vapor
interface, to show the maximum and minimum points. The system is set to be at 25 °C and a liquid pressure of 1 atm, with the liquid initially containing
2.5 × 1019 molecules (4.15 × 10−5 moles) of water and 3 × 1014 molecules (4.98 × 10−10 moles) of nitrogen (initial degree of saturation of 1.089). The
cone mouth radius, w, is 50 μm, and the half cone apex angle, β, is 45°. The contact angle, θ, is considered to be 10°. The status of the bubble (inside, or
pinned to the corner, or outside the cone) is shown with different line types. Reproduced with permission from ref 62. Copyright 2019 American
Chemical Society.

continue to be created, the number of applications is ever- surfaces and play a role in interactions between hydrophobic
increasing. solid surfaces in liquids.86 Eslami and Elliott used Gibbsian
Ward and co-workers described the nucleation and stability of composite-system thermodynamics to explain why, under
bubbles from liquid−gas solutions emerging either homoge- conditions where nucleation at surfaces is possible, nucleation
neously from a liquid84 or heterogeneously from a conical solid occurs more readily at a fluid surface (a nucleated lens-shaped
pit,18,60 either at constant liquid-phase pressure60 or at constant drop) than at a solid surface (a nucleated sessile drop) with the
system volume,18,84 with one motivation being the role of bubble same values of interfacial tension.9 Eslami and Elliott described
nucleation in decompression-sickness-caused bone death in the thermodynamics of microfluidic process in which the
humans.18 McGaughey and Ward investigated the stability of concentrating of a solute in an aqueous drop is controlled by the
one or more single-component sessile drops in a closed volume dissolution of water into a surrounding oil phase.10,77,87 They
taking into account the wettability of the volume walls away considered both precipitating and nonprecipitating solutes and
from the sessile drop.85 Zargarzadeh and Elliott investigated the included the roles of both droplet curvature and precipitate
role of conical pit geometry and wettability of the pit walls in the curvature. The complexity of this system can lead to multiple
heterogeneous nucleation of bubbles and drops from either a equilibrium states including equilibrium states in which all water
pure phase56 or a liquid−gas solution.62 Elliott and Voitcu81 and is removed from the drops.
Zargarzadeh and Elliott11 investigated the formation and Ward and Sasges and co-workers used the Gibbsian
stability of liquid capillary bridges between a sphere and a flat composite-system thermodynamics framework to describe the
plate11,56,81 and between two flat plates.56 Zargarzadeh and impact of a gravitational field on the curvature and contact
Elliott included the role of solid wettability and considered both angles of two fluid interfaces located at different heights while in
liquid capillary bridges and bridging bubbles.11,56 Liquid equilibrium with each other6,88−90 and explained several
capillary bridges pull solids together bringing paintbrushes to a thought-provoking experiments at normal gravitational intensity
tip, pulling surface force microscope tips toward surfaces, in a university laboratory88 and at various gravitational
collapsing fine surface texture features, and causing either intensities provided by experiments in a drop shaft89 and on
desired or undesired particle agglomeration in many industrial board the space shuttle.90 Asekomhe and Elliott showed that
settings. Two important results of these thermodynamic studies accounting for gravity-induced deformation of the fluid interface
were that nanoscale confinement between two solid surfaces can changed the line tension inferred from capillary rise in a conical
enhance or suppress phase transitions,11,81 and that nanoscale tube by 50%.91 Eslami and Elliott described the thermodynamics
confinement can lead to disappearance/appearance fluctuations of capillary menisci acted on by a perpendicular gravitational
of the new phase far from the bulk phase critical point of the field across the full range of Bond numbers (capillary
fluid.11,62,81 Zargarzadeh and Elliott provided a thermodynamic diameters).92 Voitcu and Elliott provided a unified derivation
framework for surface nanobubbles,61 microscopic bubbles with of all conditions for equilibrium of sessile drop in a gravitational
nanometer height (nanopancakes) that can form at hydrophobic field.8
10874 https://dx.doi.org/10.1021/acs.jpcb.0c05946
J. Phys. Chem. B 2020, 124, 10859−10878
The Journal of Physical Chemistry B pubs.acs.org/JPCB Feature Article

Milne et al.51 showed that, by combining Gibbs adsorption than 31 nm for isothermal vapor−liquid equilibrium at 298
equations for surfactant adsorption with the equilibrium of a K.13,63 For isobaric vapor−liquid equilibrium at a vapor-phase
sessile drop on a rough surface, analytical expressions for contact pressure of 1 atm, the ethanol/water azeotrope was removed
angle of surfactant-laden sessile drops wetting rough surfaces when the radius of curvature of the liquid−vapor interface
could be found in terms of surfactant concentration and surface (curved toward the liquid) was less than 17 nm.13 Liu et al.12
roughness parameters.50−52 Shardt et al. expanded the approach described the effect of curvature of the ice−solution and
to explain transitions between Cassie−Baxter (drop atop the precipitate−solution interfaces on the solid−liquid phase
roughness trapping air) and Wenzel (drop permeating the diagram across the complete composition range. Interface
roughness) wetting states.52 Shardt and Elliott provided
curvature causes depression of the freezing liquidus and
Gibbsian composite-system derivations of new forms of the
depression of the precipitating liquidus. Importantly, interface
Cassie−Baxter and Wenzel equations that emphasized that it is
roughness at the three-phase line that controls the contact angle curvature depresses the eutectic point (the point where the
of sessile drops on rough surfaces14,15 and showed under what freezing liquidus and precipitating liquidus meet; below the
conditions the volume of liquid penetrating the roughness can eutectic temperature, liquid cannot exist in the system at
be neglected in Wenzel wetting.15 equilibrium). The fact that nanoporosity can suppress freezing
Several of the thermodynamic models developed by our group and allow equilibrium liquid water to exist at low temperatures is
(including equations of state, Gibbsian thermodynamics, and being exploited to design energy storage devices with aqueous
nonequilibrium thermodynamics) have been motivated by electrolytes with excellent low-temperature performance.74,75,97
cryobiology, the study of the response of living cells and tissues Liu et al.97 explored the impact on freezing of nanoscale
to extremely low temperature such as the temperature of liquid confinement and concentrations of additives being used in
nitrogen (−196 °C) used for long-term storage ubiquitously in aqueous electrolytes for zinc−air batteries. Li et al.74 built
cell and tissue banking for transplant medicine, in the supercapacitors with aqueous polyampholyte hydrogel electro-
commercial supply of cells for research, and in managing the lytes with excellent performance down to −30 °C. The
supply of cells within research laboratories.19,25−30,35−39,69,93−96 nanostructured polyampholyte hydrogel maintained liquid
In cryobiology applications, the two fluid phases in equilibrium water in nanopores allowing ionic conductivity to continue at
might be the inside and outside of a cell separated by the plasma low temperatures as low as −49 °C.75
membrane, or pure water ice and the unfrozen fraction more
concentrated in solutes than before freezing.
In 2019 alone, there were 2719 documents published with 4. CONCLUSIONS
“nanofluidics” in the document title, abstract, or keywords Most thermodynamics textbooks cover only simple systems:
(Scopus, June 7, 2020). The equilibrium of two fluid phases systems that at equilibrium have a single constant value of each
inside a solid micro- or nanopore that causes curvature of the intensive property throughout. Systems that have multiple
fluid interface is of intense interest, both because of many pressures at equilibrium due to the action of surface tension of a
practical applications (such as recovery of crude oil from shale) curved fluid interface, due to the presence of a semipermeable
and because of the interest in emerging nanoscience. Zandavi membrane, or due to the action of fields, can be treated as
and Ward47 measured the adsorptive uptake (capillary
composite-systems: systems made up of multiple simple systems
condensation) of three hydrocarbon vapors in nanoporous
silica and showed that pore diameter predicted from the Kelvin in equilibrium with each other but that are not themselves
equation was consistent with an independently measured (with simple systems. Gibbsian composite-system thermodynamics is
transmission electron microscopy) pore diameter of 2.6 nm. a framework to treat these systems consisting of four steps: (i)
They also showed that a more sophisticated analysis where the define the system and constraints, (ii) extremize entropy subject
pressure effects on the adsorption at the solid−liquid interface to constraints (or equivalently extremize free energy) to derive
were described with the zeta isotherm gave the same pore the complete set of conditions for equilibrium, (iii) introduce
diameter in agreement with the measurements. Karlsson et al.69 equations of state to the conditions for equilibrium and solve for
showed that the Gibbs−Thomson equation for the growth of ice equilibrium values of the thermodynamic variables, and (iv)
through biological protein pores led to a single value of the identify which function acts as the thermodynamic potential
protein−water−ice contact angle for three different experi- (equivalently free energy) and examine this function to find the
ments, at three different length scales, with three different number of unstable, metastable, or stable equilibrium states. In
proteins, in three different laboratories, indicating experimental this Feature Article, I have given an overview of Gibbsian
agreement with the Gibbs−Thomson equation for pore composite-system thermodynamics, highlighting many of the
diameters from 0.7 to 35 nm. Shardt and Elliott described the key equations and showing how they are related to one another
role of fluid interface curvature on isothermal63 or isobaric13 in Gibbs’s comprehensive framework. Where possible, I tried to
phase diagrams of multicomponent liquid−vapor phase
give references to the primary literature. Ideal forms of some of
equilibrium. They showed excellent agreement with independ-
the presented equations are more than 100 years old; however,
ent experiments for the isobaric dew temperatures of nitrogen/
argon mixtures condensing in 4 nm diameter pores in Vycor until recently theory had been fully developed only for the least
glass.13 They also showed that the effect of nanoscale fluid complicated of systems (e.g., systems with only a single-
interface curvature on vapor−liquid phase equilibrium in component, or only two ideal phases, or for ideal geometries).
nanopores can be profound. Nonideal mixtures of ethanol/ Here I have highlighted advances made over the past 35 years to
water that exhibit an azeotrope (equal volatility point beyond develop Gibbsian composite-system thermodynamics for
which separation by distillation is not possible) had their application to systems of contemporary interest featuring the
azeotrope completely removed when the radius of curvature of work of my research group. As far as I know, eqs 33, 57, 68, and
the liquid−vapor interface (curved toward the liquid) was less 69 as written are new to this work.
10875 https://dx.doi.org/10.1021/acs.jpcb.0c05946
J. Phys. Chem. B 2020, 124, 10859−10878
The Journal of Physical Chemistry B


pubs.acs.org/JPCB Feature Article

AUTHOR INFORMATION the Canadian Society for Chemical Engineering Syncrude Canada
Corresponding Author Innovation Award (2008), the Natural Sciences and Engineering
Research Council of Canada Doctoral Prize (1998), the Canadian
Janet A. W. Elliott − Department of Chemical and Materials
Engineering, University of Alberta, Edmonton, Alberta T6G 1H9, Council of Professional Engineers Young Engineer Achievement Award
Canada; orcid.org/0000-0002-7883-3243; (2001), the Canadian Institute for Advanced Research Young
Email: janet.elliott@ualberta.ca Explorer’s Prize (2002), and Time Magazine’s Canadians Who Define
the New Frontiers of Science (2002). Dr. Elliott has also received many
Complete contact information is available at: provincial and university awards including the Association of
https://pubs.acs.org/10.1021/acs.jpcb.0c05946 Professional Engineers and Geoscientists of Alberta Summit Excellence
in Education Award (2017). As one student put it, “She could convince
Notes rocks to study thermodynamics.”
The author declares no competing financial interest.
Biography
■ ACKNOWLEDGMENTS
The funding agencies that supported the research in the
reviewed papers are acknowledged in the cited papers. I hold a
Canada Research Chair in Thermodynamics and acknowledge
Discovery Grant RGPIN-2016-05502 from the Natural Sciences
and Engineering Research Council (NSERC) of Canada. I am
grateful to Dr. Nadia Shardt for proofreading this article and
offering helpful suggestions.

■ REFERENCES
(1) Elliott, J. A. W. Introduction to the Special Issue: Thermodynamic
Aspects of Cryobiology. Cryobiology 2010, 60, 1−3.
(2) Gibbs, J. W. On the Equilibrium of Heterogeneous Substances.
Transactions of the Connecticut Academy 1876, 2, 108−248; 1878, 2,
343−524. Republished in The Scientific Papers of J. Willard Gibbs Vol. 1;
Janet A. W. Elliott is a University of Alberta Distinguished Professor and
Ox Bow Press: Woodridge, CT, 1993; Vol. 1, pp 55−353.
Canada Research Chair in Thermodynamics in the Department of (3) Ostwald, W. On the Assumed Isomerism of Red and Yellow
Chemical and Materials Engineering. Dr. Elliott obtained her BASc in Mercury Oxide and the Surface-Tension of Solid Bodies. Z. Phys. Chem.,
the Engineering Physics Option of Engineering Science, and her MASc Stoechiom. Verwandtschaftsl. 1900, 34, 495−503.
and PhD in Mechanical Engineering at the University of Toronto. She (4) Guggenheim, E. A.; Adam, N. K. The Thermodynamics of
has been a Visiting Professor at MIT and at the Oxford Centre for Adsorption at the Surface of Solutions. Proc. R. Soc. London, Ser. A 1933,
Collaborative Applied Mathematics. Dr. Elliott’s research interests 139, 218−236.
include thermodynamics, transport, surfaces, colloids, cryobiology, and (5) Callen, H. B. Thermodyamics and an Introduction to Thermo-
cryopreservation. Broad thermodynamic interests include fundamental statistics, 2nd ed.; John Wiley & Sons Inc.: New York, NY, 1985.
concepts in Gibbsian thermodynamics, mathematics of functions, (6) Ward, C. A.; Sasges, M. R. Effect of Gravity on Contact Angle: A
combining thermodynamics with fluid mechanics, and combining Theoretical Investigation. J. Chem. Phys. 1998, 109, 3651−3660.
thermodynamic insight with experimental data to develop descriptions (7) Elliott, J. A. W. On the Complete Kelvin Equation. Chem. Eng.
of states and processes for a wide range of applications. Colloidal and Educ. 2001, 35, 274−268.
(8) Voitcu, O.; Elliott, J. A. W. Equilibrium of Multi-Phase Systems in
surface thermodynamics interests include drops, bubbles, adsorption,
Gravitational Fields. J. Phys. Chem. B 2008, 112, 11981−11989.
solidification of colloidal suspensions, microfluidic processes, wetting, (9) Eslami, F.; Elliott, J. A. W. Thermodynamic Investigation of the
superhydrophobic surfaces, evaporation, freezing, solidification, Barrier for Heterogeneous Nucleation on a Fluid Surface in
nucleation, phase change in confined geometries, curved fluid Comparison with a Rigid Surface. J. Phys. Chem. B 2011, 115,
interfaces, interfacial and membrane transport, capillarity in gravita- 10646−10653.
tional fields, thermodynamics of solutions and suspensions, and (10) Eslami, F.; Elliott, J. A. W. Design of Microdrop Concentrating
nanoscale science. In addition, Dr. Elliott runs a collaborative, Processes. J. Phys. Chem. B 2013, 117, 2205−2214.
interdisciplinary cryobiology research group with interests in (11) Zargarzadeh, L.; Elliott, J. A. W. Comparative Surface
experimental and computational cryobiology and cryopreservation of Thermodynamic Analysis of New Fluid Phase Formation between a
many cell and tissue types for medical and biotechnology applications. Sphere and a Flat Plate. Langmuir 2013, 29, 3610−3627.
Dr. Elliott currently serves as Associate Editor of the journal (12) Liu, F.; Zargarzadeh, L.; Chung, H. J.; Elliott, J. A. W.
Cryobiology, on the Editorial Advisory Boards of The Journal of Physical Thermodynamic Investigation of the Effect of Interface Curvature on
Chemistry and Langmuir, and on the Editorial Board of Advances in Solid−Liquid Equilibrium and Eutectic Point of Binary Mixture. J. Phys.
Colloid and Interface Science. She has previously served on the Physical Chem. B 2017, 121, 9452−9462.
(13) Shardt, N.; Elliott, J. A. W. Isobaric Vapor−Liquid Phase
Sciences Advisory Committee of the Canadian Space Agency, the
Diagrams for Multicomponent Systems with Nanoscale Radii of
Board of Directors of the Canadian Society for Chemical Engineering, Curvature. J. Phys. Chem. B 2018, 122, 2434−2447.
and the Executive Committee of the American Chemical Society (14) Shardt, N.; Elliott, J. A. W. Gibbsian Thermodynamics of
Division of Colloid and Surface Chemistry. Dr. Elliott’s research has Cassie−Baxter Wetting (Were Cassie and Baxter Wrong? Revisited).
been recognized nationally and internationally in science and Langmuir 2018, 34, 12191−12198.
engineering by Fellowship in the American Institute for Medical and (15) Shardt, N.; Elliott, J. A. W. Gibbsian Thermodynamics of Wenzel
Biological Engineering (2019), Fellowship in the Society for Cryo- Wetting (Was Wenzel Wrong? Revisited). Langmuir 2020, 36, 435−
biology (2018), Fellowship in the Chemical Institute of Canada (2015), 446.

10876 https://dx.doi.org/10.1021/acs.jpcb.0c05946
J. Phys. Chem. B 2020, 124, 10859−10878
The Journal of Physical Chemistry B pubs.acs.org/JPCB Feature Article

(16) Prausnitz, J. M.; Lichtenthaler, R. N.; Gomes de Azevedo, E. (37) Ross-Rodriguez, L. U.; Elliott, J. A. W.; McGann, L. E. Non-ideal
Molecular Thermodynamics of Fluid-Phase Equilibria, 3rd ed.; Prentice- Solution Thermodynamics of Cytoplasm. Biopreserv. Biobanking 2012,
Hall Inc.: Upper Saddle River, NJ, 1999. 10, 462−471.
(17) Rusanov, A. I.; Prokhorov, V. A. Interfacial Tensiometry; Elsevier, (38) Elmoazzen, H. Y.; Elliott, J. A. W.; McGann, L. E. Osmotic
1996. Transport Across Cell Membranes in Nondilute Solutions: A New
(18) Ward, C. A.; Johnson, W. R.; Venter, R. D.; Ho, S.; Forest, T. W.; Nondilute Solute Transport Equation. Biophys. J. 2009, 96, 2559−
Fraser, W. D. Heterogeneous Bubble Nucleation and Conditions for 2571.
Growth in a Liquid−Gas System of Constant Mass and Volume. J. Appl. (39) Shardt, N.; Chen, Z.; Yuan, S. C.; Wu, K.; Laouar, L.; Jomha, N.
Phys. 1983, 54, 1833−1843. M.; Elliott, J. A. W. Using Engineering Models to Shorten
(19) Acker, J. P.; Elliott, J. A. W.; McGann, L. E. Intercellular Ice Cryoprotectant Loading Time for the Vitrification of Articular
Propagation: Experimental Evidence for Ice Growth through Cartilage. Cryobiology 2020, 92, 180−188.
Membrane Pores. Biophys. J. 2001, 81, 1389−1397. (40) Langmuir, I. The Adsorption of Gases on Plane Surfaces of Glass,
(20) Elliott, J. R.; Lira, C. T. Introductory Chemical Engineering Mica, and Platinum. J. Am. Chem. Soc. 1918, 40, 1361−1403.
Thermodynamics, 2nd ed.; Prentice-Hall Inc.: Upper Saddle River, NJ, (41) Hill, T. L. An Introduction to Statistical Thermodynamics; Dover
2012. Publications Inc.: New York, NY, 1986.
(21) Zargarzadeh, l.; Elliott, J. A. W. Comparison of the Osmotic Virial (42) Ward, C. A.; Elmoselhi, M. B. The Coverage Dependence of the
Equation with the Margules Activity Model for Solid−Liquid Heat of Adsorption and the Predicted Wavelength of Surface
Equilibrium. J. Phys. Chem. B 2019, 123, 1099−1107. Luminescence. Surf. Sci. 1988, 203, 463−488.
(22) McMillan, W. G.; Mayer, J. E. The Statistical Thermodynamics of (43) Elliott, J. A. W.; Ward, C. A. Chemical Potential of Adsorbed
Multicomponent Systems. J. Chem. Phys. 1945, 13, 276−305. Molecules from a Quantum Statistical Formulation. Langmuir 1997, 13,
(23) Hill, T. L. Theory of Solutions. J. Chem. Phys. 1957, 26, 955−956. 951−960.
(24) Hill, T. L. Theory of Solutions. II. Osmotic Pressure Virial (44) Brunauer, S.; Emmett, P. H.; Teller. Adsorption of Gases in
Expansion and Light Scattering in Two Component Solutions. J. Chem. Multimolecular Layers. J. Am. Chem. Soc. 1938, 60, 309−319.
Phys. 1959, 30, 93−97. (45) Ward, C. A.; Wu, J. Effect of Adsorption on the Surface Tensions
(25) Prickett, R. C.; Elliott, J. A. W.; McGann, L. E. Application of the of Solid−Fluid Interfaces. J. Phys. Chem. B 2007, 111, 3685−3694.
Multisolute Osmotic Virial Equation to Solutions Containing Electro- (46) Narayanaswamy, N.; Ward, C. A. Specific Surface Area, Wetting,
lytes. J. Phys. Chem. B 2011, 115, 14531−14543. and Surface Tension of Materials from N2 Vapor Adsorption Isotherms.
(26) Zielinski, M. W.; McGann, L. E.; Nychka, J. A.; Elliott, J. A. W.
J. Phys. Chem. C 2019, 123, 18336−18346.
Comparison of Non-Ideal Solution Theories for Multi-Solute Solutions
(47) Zandavi, S. H.; Ward, C. A. Contact Angles and Surface
in Cryobiology and Tabulation of Required Coefficients. Cryobiology
Properties of Nanoporous Materials. J. Colloid Interface Sci. 2013, 407,
2014, 69, 305−317.
255−264.
(27) Cheng, J.; Gier, M.; Ross-Rodriguez, L. U.; Prasad, V.; Elliott, J.
(48) Zhu, B.; Gu, T. J. General Isotherm Equation for Adsorption of
A. W.; Sputtek, A. Osmotic Virial Coefficients of Hydroxyethyl Starch
Surfactants at Solid/Liquid Interfaces. Part 1. Theoretical. J. Chem. Soc.,
from Aqueous Hydroxyethyl Starch−Sodium Chloride Vapor Pressure
Faraday Trans. 1 1989, 85, 3813−3817.
Osmometry. J. Phys. Chem. B 2013, 117, 10231−10240.
(49) Wang, Z.; Xu, S.; Acosta, E. Heat of Adsorption and its Role on
(28) Elliott, J. A. W.; Prickett, R. C.; Elmoazzen, H. Y.; Porter, K. R.;
Nanoparticle Stabilization. J. Chem. Thermodyn. 2015, 91, 256−266.
McGann, L. E. A Multisolute Osmotic Virial Equation for Solutions of
(50) Milne, A. J. B.; Elliott, J. A. W.; Amirfazli, A. Contact Angles of
Interest in Biology. J. Phys. Chem. B 2007, 111, 1775−1785 Included in
Surfactant Solutions on Heterogeneous Surfaces. Phys. Chem. Chem.
a 2017 Virtual Issue of J. Phys. Chem. A, B, C, & Lett. in honor of Marie
Curie’s 150th birthday. Phys. 2015, 17, 5574−5585.
(29) Zielinski, M. W.; McGann, L. E.; Nychka, J. A.; Elliott, J. A. W. A (51) Milne, A. J. B.; Elliott, J. A. W.; Zabeti, P.; Zhou, J.; Amirfazli, A.
Non-ideal Solute Chemical Potential Equation and the Validity of the Model and Experimental Studies for Contact Angles of Surfactant
Grouped Solute Approach for Intracellular Solution Thermodynamics. Solution on Rough and Smooth Hydrophobic Surfaces. Phys. Chem.
J. Phys. Chem. B 2017, 121, 10443−10456. Chem. Phys. 2011, 13, 16208−16219.
(30) Zielinski, M. W.; McGann, L. E.; Nychka, J. A.; Elliott, J. A. W. (52) Shardt, N.; Bigdeli, M.; Elliott, J. A. W.; Tsai, P. A. How
Comment on ‘Determination of the Quaternary Phase Diagram of the Surfactants Affect Droplet Wetting on Hydrophobic Microstructures. J.
Water−Ethylene Glycol−Sucrose−NaCl System and a Comparison Phys. Chem. Lett. 2019, 10, 7510−7515.
Between Two Theoretical Methods for Synthetic Phase Diagrams’ (53) Shardt, N.; Elliott, J. A. W. A Model for the Surface Tension of
Cryobiology 61 (2010) 52−57. Cryobiology 2015, 70, 287−292. Dilute and Concentrated Binary Aqueous Mixtures as a Function of
(31) MacNeil, J. A.; Ray, G. B.; Sharma, P.; Leaist, D. G. Activity Composition and Temperature. Langmuir 2017, 33, 11077−11085.
Coefficients of Aqueous Mixed Ionic Surfactant Solutions from (54) Shardt, N.; Wang, Y.; Jin, Z.; Elliott, J. A. W. Surface Tension as a
Osmometry. J. Solution Chem. 2014, 43, 93−108. Function of Temperature and Composition for a Broad Range of
(32) MacNeil, J. A.; Ray, G. B.; Leaist, D. G. Activity Coefficients and Mixtures. Chem. Eng. Sci. 2021, 230, 116095.
Free Energies of Nonionic Mixed Surfactant Solutions from Vapor- (55) Fisher, L. R.; Israelachvili, J. N. Experimental Studies on the
Pressure and Freezing-Point Osmometry. J. Phys. Chem. B 2011, 115, Applicability of the Kelvin Equation to Highly Curved Concave
5947−5957. Menisci. J. Colloid Interface Sci. 1981, 80, 528−541.
(33) Hildebrandt, W. H. Low Temperature Quantitative Phase (56) Zargarzadeh, L.; Elliott, J. A. W. Surface Thermodynamic
Equilibria and Glass Formation in the Water−Sodium Chloride−Di- Analysis of Fluid Confined in a Cone and Comparison with the
methyl Sulfoxide System. PhD Thesis, Duke University, 1975. Sphere−Plate and Plate−Plate Geometries. Langmuir 2013, 29,
(34) Yousef, M. A.; Datta, R.; Rodgers, V. G. J. Model of Osmotic 12950−12958.
Pressure for High Concentrated Binary Protein Solutions. AIChE J. (57) Thomson, W. On the Equilibrium of Vapor at a Curved Surface
2002, 48, 913−917. of Liquid. Philos. Mag. S. 4 1871, 42, 448−452.
(35) Zielinski, M. W.; McGann, L. E.; Nychka, J. A.; Elliott, J. A. W. (58) Melrose, J. C. Model Calculations for Capillary Condensation.
Measurement of Grouped Intracellular Solute Osmotic Virial AIChE J. 1966, 12, 986−994.
Coefficients. Cryobiology 2020, DOI: 10.1016/j.cryobiol.2019.09.017. (59) Wang, Y.; Shardt, N.; Lu, C.; Li, H.; Elliott, J. A. W.; Jin, Z.
(36) Prickett, R. C.; Elliott, J. A. W.; Hakda, S.; McGann, L. E. A Non- Validity of the Kelvin Equation and the Equation-of-State-with-
ideal Replacement for the Boyle van’t Hoff Equation. Cryobiology 2008, Capillary-Pressure Model for the Phase Behavior of a Pure Component
57, 130−136. under Nanoconfinement. Chem. Eng. Sci. 2020, 226, 115839.

10877 https://dx.doi.org/10.1021/acs.jpcb.0c05946
J. Phys. Chem. B 2020, 124, 10859−10878
The Journal of Physical Chemistry B pubs.acs.org/JPCB Feature Article

(60) Ward, C. A.; Levart, E. Conditions for Stability of Bubble Nuclei (84) Ward, C. A.; Tikuisis, P.; Venter, R. D. Stability of Bubbles in a
in Solid Surfaces Contacting a Liquid-Gas Solution. J. Appl. Phys. 1984, Closed Volume of Liquid-Gas Solution. J. Appl. Phys. 1982, 53, 6076−
56, 491−500. 6084.
(61) Zargarzadeh, L.; Elliott, J. A. W. Thermodynamics of Surface (85) McGaughey, A. J. H.; Ward, C. A. Droplet Stability in a Finite
Nanobubbles. Langmuir 2016, 32, 11309−11320. System: Consideration of the Solid−Vapor Interface. J. Appl. Phys.
(62) Zargarzadeh, L.; Elliott, J. A. W. Bubble Formation in a Finite 2003, 93, 3619−3626.
Cone: More Pieces to the Puzzle. Langmuir 2019, 35, 13216−13232. (86) Lohse, D.; Zhang, X. Surface Nanobubbles and Nanodroplets.
(63) Shardt, N.; Elliott, J. A. W. Thermodynamic Study of the Role of Rev. Mod. Phys. 2015, 87, 981−1035.
Interface Curvature on Multicomponent Vapor−Liquid Phase (87) Eslami, F.; Elliott, J. A. W. Stability Analysis of Microdrops during
Equilibrium. J. Phys. Chem. A 2016, 120, 2194−2200. Concentrating Processes. J. Phys. Chem. B 2014, 118, 3630−3641.
(64) Alam, M. A.; Clarke, A. P.; Duffy, J. A. Capillary Condensation (88) Sasges, M. R.; Ward, C. A. Effect of Gravity on Contact Angle: An
and Desorption of Binary Mixtures of N2−Ar Confined in a Experimental Investigation. J. Chem. Phys. 1998, 109, 3661−3670.
Mesoporous Medium. Langmuir 2000, 16, 7551−7553. (89) Sasges, M. R.; Ward, C. A.; Azuma, H.; Yoshihara, S. Equilibrium
(65) Donohue, M. D.; Aranovich, G. L. Adsorption Hysteresis in Fluid Configurations in Low Gravity. J. Appl. Phys. 1996, 79, 8770−
Porous Solids. J. Colloid Interface Sci. 1998, 205, 121−130. 8782.
(66) Meissner, F. Mitteilungen aus dem Institut für phys. Chemie der (90) Ward, C. A.; Rahimi, P.; Sasges, M. R.; Stanga, D. Contact Angle
Universität Göttingen. Nr. 8. Ü ber den Einfluß der Zerteilung auf die Hysteresis Generated by the Residual Gravitational Field of the Space
Schmelztemperatur. Zeitschrift für anorganische und allgemeine Chemie Shuttle. J. Chem. Phys. 2000, 112, 7195−7202.
1920, 110, 169−186. (91) Asekomhe, S. O.; Elliott, J. A. W. The Effect of Interface
(67) Rie, E. Ü ber den Einfluß der Oberflächenspannung auf Deformation due to Gravity on Line Tension Measurement by the
Capillary Rise in a Conical Tube. Colloids Surf., A 2003, 220, 271−278.
Schmelzen und Gefrieren. Z. Phys. Chem. 1923, 104U, 354−362.
(92) Eslami, F.; Elliott, J. A. W. Gibbsian Thermodynamic Study of
(68) Liu, Z.; Muldrew, K.; Wan, R. G.; Elliott, J. A. W. Measurement of
Capillary Meniscus Depth. Sci. Rep. 2019, 9, 657.
Freezing Point Depression of Water in Glass Capillaries and the
(93) Abazari, A.; Thompson, R. B.; Elliott, J. A. W.; McGann, L. E.
Associated Ice Front Shape. Phys. Rev. E: Stat. Phys., Plasmas, Fluids,
Transport Phenomena in Articular Cartilage Cryopreservation as
Relat. Interdiscip. Top. 2003, 67, 061602. Predicted by the Modified Triphasic Model and the Effect of Natural
(69) Karlsson, J. O. M.; Braslavsky, I.; Elliott, J. A. W. Inhomogeneities. Biophys. J. 2012, 102, 1284−1293.
Protein−Water−Ice Contact Angle. Langmuir 2019, 35, 7383−7387. (94) Ross-Rodriguez, L. U.; Elliott, J. A. W.; McGann, L. E.
(70) Raymond, J. A.; DeVries, A. L. Adsorption Inhibition as a Investigating Cryoinjury Using Simulations and Experiments: 1. TF-1
Mechanism of Freezing Resistance in Polar Fishes. Proc. Natl. Acad. Sci. Cells during Two-step Freezing (Rapid Cooling Interrupted with a
U. S. A. 1977, 74, 2589−2593. Hold Time). Cryobiology 2010, 61, 38−45.
(71) Drori, R.; Davies, P. L.; Braslavsky, I. Experimental Correlation (95) Ross-Rodriguez, L. U.; Elliott, J. A. W.; McGann, L. E.
between Thermal Hysteresis Activity and the Distance between Investigating Cryoinjury Using Simulations and Experiments: 2. TF-1
Antifreeze Proteins on an Ice Surface. RSC Adv. 2015, 5, 7848−7853. Cells during Graded Freezing (Interrupted Slow Cooling without Hold
(72) Dash, J. G.; Rempel, A. W.; Wettlaufer, J. S. The Physics of Time). Cryobiology 2010, 61, 46−51.
Premelted Ice and its Geophysical Consequences. Rev. Mod. Phys. 2006, (96) Elliott, J. A. W.; Elmoazzen, H. Y.; McGann, L. E. A Method
78, 695−741. whereby Onsager Coefficients may be Evaluated. J. Chem. Phys. 2000,
(73) Peppin, S. S. L.; Elliott, J. A. W.; Worster, J. A. W. Solidification of 113, 6573−6578.
Colloidal Suspensions. J. Fluid Mech. 2006, 554, 147−166. (97) Liu, F.; Chung, H.-J.; Elliott, J. A. W. Freezing of Aqueous
(74) Li, X.; Liu, L.; Wang, X.; Ok, Y. S.; Elliott, J. A. W.; Chang, S. X.; Electrolytes in Zinc−Air Batteries: Effect of Composition and
Chung, H.-J. Flexible and Self-Healing Aqueous Supercapacitors for Nanoscale Confinement. ACS Appl. Energy Mater. 2018, 1, 1489−1495.
Low Temperature Applications: Polyampholyte Gel Electrolytes with
Biochar Electrodes. Sci. Rep. 2017, 7, 1685.
(75) Li, X.; Charaya, H.; Bernard, G. M.; Elliott, J. A. W.; Michaelis, V.
K.; Lee, B.; Chung, H.-J. Low-Temperature Ionic Conductivity
Enhanced by Disrupted Ice Formation in Polyampholyte Hydrogels.
Macromolecules 2018, 51, 2723−2731.
(76) Freundlich, H. Kapillarchemie; Akademische Verlagsgessell-
schaft: Leipzig, 1909.
(77) Eslami, F.; Elliott, J. A. W. Role of Precipitating Solute Curvature
on Microdrops and Nanodrops during Concentrating Processes: The
Nonideal Ostwald−Freundlich Equation. J. Phys. Chem. B 2014, 118,
14675−14686.
(78) Johnson, K. C. Comparison of Methods for Predicting
Dissolution and the Theoretical Implications of Particle-Size-Depend-
ent Solubility. J. Pharm. Sci. 2012, 101, 681−689.
(79) Pouralhosseini, S.; Eslami, F.; Elliott, J. A. W.; Shaw, J. M.
Modeling the Phase Behavior of Asphaltene + Toluene + Polystyrene
mixturesA Depletion Flocculation Approach. Energy Fuels 2016, 30,
904−914.
(80) Lane, L. B. Freezing Points of Glycerol and Its Aqueous
Solutions. Ind. Eng. Chem. 1925, 17, 924−924.
(81) Elliott, J. A. W.; Voitcu, O. On the Thermodynamic Stability of
Liquid Capillary Bridges. Can. J. Chem. Eng. 2007, 85, 692−700.
(82) Planck, M. Verdampfen, Schmelzen und Sublirmiren. Ann. Phys.
1882, 251, 446−475.
(83) Deltete, R. J. Planck, Ostwald and the Second Law of
Thermodynamics. HOPOS: The Journal of the International Society for
the History of Philosophy of Science 2012, 2, 121−146.

10878 https://dx.doi.org/10.1021/acs.jpcb.0c05946
J. Phys. Chem. B 2020, 124, 10859−10878

You might also like