9

Future Opportunities in
Thermodynamics
H. Ted Davis
Department of Chemical Engineering & Materials Science
University of Minnesota
Minneapolis, Minnesota

I. Introduction
My job of discussing Keith Gubbins’ paper [ 11 would be easier if I disagreed
with his summary of the current status and future opportunities in the use
of thermodynamics in chemical engineering. Unfortunately for me, his is
a fine paper, and it touches on most of the important areas that are likely
to prosper during the next couple of decades. However, there are some areas
that I believe he underemphasizes or whose importance he underestimates.
I agree with Gubbins that the future advancement in our understanding
will come through the interplay of molecular theory, computer simulation,
and experimentation (Fig. 1). The rapidly increasing speed and memory
capacity of computers enable the simulation of more realistic molecular
systems and the solution of more rigorous molecular theories.
The next two decades will witness a substantial increase in the number
of chemical engineers using computer-aided molecular theory and molecular
dynamics as routine tools for elucidation of thermodynamic mechanisms
and for prediction of properties. Chemical engineers already have a con-
siderable presence in the area. Two of the more significant developments
in computer simulations of the last decade have been accomplished by young

169
Copyright 0 1991 by Audcmic Presi, Inc.
ADVANCE8 IN CHEMICAL ENGINEERING, VOL. 16 AU rightuof reproductionin MY form rcscrvcd.
170 Future Opportunities in Thermodynamics

Figure 1. Developments in thermodynamics.

chemical engineers. One is the Gibbs ensemble Monte Carlo method, cre-
ated by A. Panagiotopoulos [2], which is especially potent in simulating
multiphase equilibria of mixtures. The other is the excluded volume map,
a technique invented by G. Deitrick [3] to reduce by orders of magnitude
the time required to compute chemical potential in molecular dynamics
simulations.
An example drawn from Deitrick’s work (Fig. 2) shows the chemical
potential and the pressure of a Lennard-Jones fluid computed from molecular
dynamics. The variance about the computed mean values is indicated in the
figure by the small dots in the circles, which serve only to locate the dots.
A test of the “thermodynamic goodness” of the molecular dynamics result
is to compute the chemical potential from the simulated pressure by inte-
grating the Gibbs-Duhem equation. The results of the test are also shown
in Fig. 2. The point of the example is that accurate and affordable molecular
simulations of thermodynamic, dynamic, and transport behavior of dense
fluids can now be done. Currently, one can simulate realistic water, elec-
trolytic solutions, and small polyatomic molecular fluids. Even some of the
properties of micellar solutions and liquid crystals can be captured by ide-
alized models [4,51.
H. Ted Davis 171

100

a a
L
10

05 06 07 0 8 09 10 05 06 07 08 09 10
n n
Figure 2. Fugacity a and pressure P, as a function of density n for a 6-12 Lennard-
Jones liquid. Circles denote molecular dynamics results; dashed curves were
derived from integration of the Gibbs-Duhem equation using molecular
dynamics data. Convenient molecular dimensions were used. Reprinted with
permission from Deitrick et al. [3].

In the future, experimentation will play an increasingly important role

in understanding thermodynamic behavior. Current and emerging instrumen-
tation will enable study of structure and dynamics on the molecular or
colloidal scale (angstroms to hundreds of nanometers). With the aid of the
rotating anode and synchrotron radiation sources and sophisticated computer-
aided data analysis, x-ray scattering now provides an accurate noninvasive
probe of molecular-level organization. At national and international facilities,
neutron scattering competes with x-ray scattering in structural analysis and
provides dynamical data on molecular rearrangement not accessible by x-
rays. Another modem tool, the video-enhanced light microscope, can be used
to study structure and transformation kinetics on the scale of tenths to tens
of microns. The transmission electron microscope, extended by fast-freeze
cold-stage technology, enables photography and examination by eye or
computer-aided image analysis of molecular organization on a scale rang-
ing from a few nanometers to tens or even hundreds of nanometers. Scan-
ning electron microscopy spans the gap between transmission electron
microscopy and ordinary light microscopy. The recently invented scanning
172 Future Opportunities in Thermodynamics

tunneling microscopy has begun to yield detailed images of the atomic to-
pography of biological membranes and proteins. The turns in the helical
structure of DNA are clearly visible in several recent publications of scanning
tunneling micrographs [ 6 ] .This remarkable microscope can be operated on
a laboratory bench without the aid of clean room heroics and can be open
to the atmosphere-in fact, it can be operated under liquid. The emerging
atomic force microscope, which does not require proximity of an imaged
object to a conducting substrate, could be even more powerful than the
scanning tunneling microscope.
Modern spectroscopies, such as fast Fourier transform infrared spectros-
copy, picosecond laser spectroscopy, x-ray photoemission spectroscopy,
electron spin resonance, and nuclear magnetic resonance, based on improved
instrumentation and high-speed computer hardware and sophisticated soft-
ware, provide much information concerning molecular structure and dynam-
ics. For example, with the aid of computer-controlled magnetic field pulsing
sequences, nuclear magnetic resonance will give a fingerprint spectrum,
which identifies the phase of the subject system, and magnetic relaxation
rates, from which local molecular rotational behavior can be deduced and
tracer diffusivities of every component of the material can be determined
simultaneously. Furthermore, NMR imaging can be used as a noninvasive
tool to map out the microstructure of fine composites, biological organisms,
and polymeric materials.
The surface force apparatus is now being used routinely to study the
equation of state of solutions confined between opposed, molecularly thin
solid films. The apparatus is also used in one laboratory to study electro-
chemistry of thin films at electrodes a few nanometers thick and in a few
other laboratories to study the behavior of molecularly thin films subjected
to shear and flow [7].
An important point is often overlooked with regard to the role of experi-
mentation. It is not unusual in thermodynamics to think of experiment as
the servant of theory and to consider theory as that which is derived from
the equations of thermodynamics or a statistical mechanics-based hierar-
chy of mechanics (e.g., classical, relativistic, quantum). In my opinion,
experimental thermodynamics has a more proactive role to play in estab-
lishing the conceptual basis on which theoretical models will be built. For
example, it has been observed that certain patterns of phase behavior are
generic, functions of field variables (intensive thermodynamic variables which
are the same in all coexisting phases at equilibrium). Examples include the
law of corresponding states, behavior near a critical point, and the sequence
of phases that appear in scans of temperature, activity of a solute, carbon
number in a homologous series, and the like. The canonization of generic
behavior or patterns shared by a large class of thermodynamic systems is
H. Ted Davis 173

indeed theory. So is the concept that the atomics of a solid sit at three-
dimensionally periodic sites. If the behavior can also be predicted from model
equations of state or statistical mechanics, so much the better.
In his paper, Gubbins identifies the corresponding states theory, pertur-
bation and cluster expansion theory, and density functional theory as those
of particular interest to chemical engineers. I agree that these are impor-
tant theories for future growth in chemical engineering. However, during
the next decade or so, the best one can hope to accomplish with these theories
is handling molecular solutions of not too complex polar, polyatomic
molecules. Microstructural materials such as micellar solutions, liquid crys-
tals, microemulsions, protein solutions, and block copolymer glasses will
require approaches not emphasized in Gubbins’ paper. One of these ap-
proaches is lattice theory, which, contrary to his assertion, is likely to be
very useful in understanding and predicting properties of microstructured
materials. Another approach neglected by Gubbins is what I have chosen
to call the elemental structures model. In what follows, I will present the
phase behavior of oil, water, and surfactant solutions as a case study to
illustrate the special problems posed by microstructured fluids and to in-
dicate how lattice and elemental structures models useful to chemical en-
gineers might be developed.

11. Microstructured Fluids: A Primer

A surfactant molecule is an amphiphile, which means it has a hydrophilic
(water-soluble) moiety and a hydrophobic (water-insoluble) moiety sepa-
rable by a mathematical surface. The hydrophobic “tails” of the most common
surfactants are hydrocarbons. Fluorocarbon and perfluorocarbon tails are,
however, not unusual. Because of the hydrophobic tail, a surfactant resists
forming a molecular solution in water. The molecules will tend to migrate
to any water-vapor interface available or, at sufficiently high concentration,
the surfactant molecules will spontaneously aggregate into association
colloids, i.e., into micelles or liquid crystals. Because of the hydrophific
head, a surfactant (with a hydrocarbon tail) will behave similarly when placed
in oil or when put in solution with oil and water mixtures. Some common
surfactants are sodium or potassium salts of long-chained fatty acids (soaps),
sodium ethyl sulfates and sulfonates (detergents), alkyl polyethoxy alcohols,
alkyl ammonium halides, and lecithins or phospholipids.
The special property of surfactants in solution is that they associate into
a monolayer or sheetlike structure with the water-soluble moieties (hydro-
philic heads) on one side of the sheet and water-insoluble moieties on the
other side [8]. These sheetlike structures provide the building blocks for a
rich variety of fluid microstructures, zhich, depending on thermodynamic
174 Future Opportunities in Thermodynamics

conditions, assemble spontaneously. Several known examples of surfactant

fluid microstructures are illustrated in Fig. 3.
At an air-water interface, a monolayer forms with heads lying down and
tails up (toward air), whereas at an air-hydrocarbon interface the monolayer
lies with tails down. By closing on the tail side, the sheetlike structure can
be dispersed in aqueous solutions as spherical, rodlike, or disklike micelles
(Fig. 3). Closure on the head side forms the corresponding inverted micelles
in oil. Oil added to a micellar solution is incorporated into the interior of
the micelle to form a swollen micellar solution. Thus, surfactant acts to
solubilize substantial amounts of oil into aqueous solution. Similarly, a
swollen inverted micellar solution enables significant solubilization of water
in oil.
Micellar solutions are isotropic microstructured fluids which form under
certain conditions. At other conditions, liquid crystals periodic in at least
one dimension can form. The lamellar liquid crystal phase consists of pe-
riodically stacked bilayers (a pair of opposed monolayers). The sheetlike
surfactant structures can curl into long rods (closing on either the head or
tail side) with parallel axes arrayed in a periodic hexagonal or rectangular
spacing to form a hexagonal or a rectangular liquid crystal. Spherical mi-
celles or inverted micelles whose centers are periodically distributed on a
lattice of cubic symmetry form a cubic liquid crystal.
In addition to the geometrical order associated with the relative arrange-
ments of the surfactant fluid microstructures, the sheetlike assembly of the
surfactant imposes a topological order. The two sides of the sheetlike
monolayer are different, water-rich fluids always lying on the head side and
oil-rich fluids always lying on the tail side. Swollen micellar solutions are
water-continuous and oil-discontinuous; i.e., the water-rich region continu-
ously spans a sample of the solution whereas the oil-rich regions are dis-
joint and lie inside surfaces formed by surfactant monolayers closed on their
tail side. Swollen inverted micellar solution, on the other hand, is oil-con-
tinuous and water-discontinuous. Similarly, the cubic liquid crystals described
above are water-continuous if it is micelles that are periodically arrayed,
and oil-continuous if it is inverted micelles that are periodically arrayed.
As shown in Fig. 3, topologically different microstructures are possible,
namely, bicontinuous microstructures. In these microstructures, the sheet-
like surfactant structure separates a sample-spanning water-rich region from
a sample-spanning oil-rich region. As indicated in Fig. 3, bicontinuous
microstructures occur in liquid crystalline phases (bicontinuous cubic) and
in isotropic phases (some microemulsions having substantial amounts of oil
and water). The existence of bicontinuous microstructures was postulated
by Scriven in 1976 [9] and after several years of controversy has now been
experimentally verified.
 175
H. Ted Davis
176 Future Opportunitiesin Thermodynamics

Proof of the geometric symmetries of the liquid crystals shown in Fig.

3 has been ambiguously established by x-ray scattering. However, unequivo-
cal demonstration of the bicontinuity of some cubic liquid crystals and certain
microemulsions requires more than x-ray scattering. The microemulsions
story is especially interesting since it represents a case of potential engi-
neering application, namely enhanced oil recovery, driving basic scientific
discovery. The triangular phase diagram shown in Fig, 3 is but a single
projection out of multivariable parameter space. By varying a field vari-
able-temperature, pressure, activity of a fourth component, carbon num-
ber of a homologous series of the oil or surfactant, etc.-a triangular prismatic
phase diagram results. The generic pattern of microemulsion phase splits
as a function field variable, which has been identified by the Minnesota group
[ 101 and by Kahlweit, Strey, and co-workers [ 1 1 1 , is illustrated in Fig. 4.
Shown there is temperature-composition diagram indicating the isotropic
phase equilibria that occur in mixtures of pentaethylene glycol dodecyl ether
(C,2E5), water, and octane. At low temperatures two phases coexist, the lower

H20IAJ Oll IS1

-a
Figure 4. Generic phase prism of a nonionic surfactant, water, and oil system. Reprinted
with permission from Kahlweit er al. [ 1 11.
H. Ted Davis 177

phase (in a test tube) being a water-rich microemulsion. As temperature is

decreased, a third phase appears at a critical end point (CEP) at the water-
rich end of the two-phase tie line. With continued increase in temperature,
the middle phase microemulsion increases in oil concentration until the
middle phase disappears at a CEP at the oil-rich end. This pattern of phase
behavior has been designated by Knickerbocker et al. [lo] as a &3,2 se-
quence: denotes a two-phase system in which the microemulsion is the
lower, water-rich phase; 3 denotes a three-phase system in which the
microemulsion is the middle phase and contains appreciable amounts of oil
and water; and 5 denotes a two-phase system in which the microemulsion
is the upper, oil-rich phase.
The special significance of the &3,2 microemulsion sequence to petro-
leum recovery is that, under certain conditions (of temperature, pH, choice
of surfactant, addition of cosurfactant, salt activity, etc.), a middle phase
microemulsion has ultralow tension against both oil-rich and water-rich
phases [12]. In the example illustrated in Fig. 5 , a 2,3,2 sequence of phase
splits can be generated as a function of salt activity in mixtures of oil, brine,
and ionic surfactants. The ultralow-tension microemulsion is obtained by
selecting a particular surfactant and adding a small amount of a low-mo-
lecular-weight alcohol-all of which amount to locating the right region of
field variable parameter space to find the desired pattern of behavior.

I I 1 I

II 13 I .5 I .7 1.9
w t . % NaCl
Figure 5. Microemulsion phase and interfacial tension behavior in a salinity scan.
178 Future Opportunities in Thermodynamics

The sequence shown in Fig. 5 has played a prominent role in resolving

the question of the microstructure of the mid-range microemulsion. At low
salinity, the microemulsion has a low oil concentration and thus is a wa-
ter-continuous swollen micellar solution. At high salinity, the microemulsion
has a low water concentration and is an oil-continuous swollen inverted
micellar solution. Somewhere between low and high salinities, the
microemulsion undergoes a transition from a water-continuous to an oil-
continuous phase. This could happen by sudden inversion of micellar droplets
or through the occurrence of bicontinuous microstructure [ 131 (Fig. 6). The
sequence of structures hypothesized in Fig. 6 would be accompanied by very
different diffusion behavior (Fig. 7). In the droplet transition case, the self-
diffusion coefficient of water would be high until inversion to oil-continuous
microemulsion, after which water and surfactant would diffuse together as
a relatively slow swollen inverted micelle. Oil and surfactant diffusivities
would be low and equal in the water-continuous region and oil diffusivity
would be high in the oil-continuous region. Alternatively, if the transition
is through bicontinuous microemulsions, there will be an intermediate sa-
linity region in which oil and water diffusivities are comparable while the
surfactant diffusivity is lower than either, since surfactant is constrained
to move along the sheetlike surface layers separating oil-rich and water-
rich regions.
Using pulsed field gradient spin echo NMR, Guering and Lindman [ 141
and, independently, Clarkson et aE. [ 151 measured the self-diffusion coef-
ficients of the components of microemulsions of sodium dodecyl sulfate
(SDS), toluene, butanol, and NaCl brine. The results (Fig. 8) establish
unequivocally the existence of bicontinous microemulsion.
It appears that the role of increasing salinity is to change the mean cur-
vature of the surfactant sheetlike structure from a value favoring closure
on the oil-rich regions (swollen inverted micelles). In between, in bicontinous
microemulsion having comparable amounts of oil and water, the preferred
mean curvature must be near zero.
The globular-to-bicontinuous transition of a microemulsion along the &3,2
sequence of phase splits is of special interest because of its relationship to
enhanced oil recovery technology. However, the microstructural transition
is not controlled by phase transitions. If one considers a constant oil/wa-
ter plane in the prismatic phase diagram given in Fig. 4,the resulting tem-
perature-surfactant composition phase diagram [ 151 (Fig. 9) has a one-phase
(10) corridor lying between a liquid crystalline phase (La)to the right and
two- and three-phase regions to the left. The watedoctane weight percent-
ages are fixed at 60/40 in Fig. 9. This corresponds to about equal volumes
of water and octane. Thus, along the corridor, the states A, B, ..., A' rep-
resent a progression of one-phase microemulsion of about equal water and
H. Ted Davis 179

Figure 6. Schematic diagram comparing the droplet inversion transition with the
bicontinuous transition of microemulsions in the 2.3.2 sequence of phase
splits. Reprinted with permission from Kaler and Prager [ 131.
180 Future Opportunities in Thermodynamics

Droplet Inversion microemulsiontransitlon

SALINITY

Blcontinuous mlcroemulslontransition

SALINITY
Figure 7. Schematic diagram comparing the behavior of self-diffusion coefficients of oil
(Do),water ( D J , and surfactant (D,) expected for the droplet inversion trans-
ition and the bicontinuous transition of microemulsiondepicted in Fig. 6 .

oil volumes. Diffusivities (Fig. 10) measured along this corridor by NMR
show plainly that, between G and €1,a transition from globular to bicontinous
microstructure occurs [ 161.
Although NMR results provide perhaps the most convincing evidence of
the bicontinuous structure of some microemulsions, many other techniques
support their existence. These techniques include electrical conductimetry,
x-ray and neutron scattering, quasielastic light scattering, and electron
H. Ted Davis 181

Toluene

10-
Butanol
Butanol
-0

10.'

lo-'
3 4
SALINITY
5
,
( g/
6 7
100,rn3 I
a 9 10

Figure 8. Self-diffusion coefficients of the components of a microemulsion of sodium

dodecyl sulfate (SDS), butanol, toluene, and NaCl brine. Vertical lines denote
2,3 and 3.5 phase transitions. Reprinted with permission from P. Guering and
B. Lindman, Langrnuir 1,464 (1985) [14]. Copyright 1985 American
Chemical Society.

microscopy. For example, a tantalum-tungsten replica of a fracture surface

of frozen microemulsion viewed in the transmission electron microscope
[17] shows a cross section consistent with a structure of intertwined wa-
ter and oil-continuous regions (Fig. l l is a TEM micrograph of C,?E,, water,
octane microemulsion at state J in Fig. 9). In the globular regime, TEM
micrographs reveal disjoint spherical structures quite different from the
structures seen in Fig. 11 [ 161.

111. Microstructured Fluids: Theory and Simulation

At its most satisfying level, a statistical thermodynamic theory would be-
gin by specifying realistic interaction potentials for the molecular compo-
nents of a complex mixture and from these potentials the thermodynamic
functions and phase behavior would be predicted without further approxi-
mation. For the next decade or so, there is little hope to accomplish such
a theory for microstructured fluids. However, predictive theories can be
obtained with the aid of elemental structures models. Also, lattice models
182 Future Opportunitiesin Thermodynamics

I I

40

T
["CI

t -
tEk
30 lF\ 0

20 - a 0

B A

-
10
0 10 20 30

y(wt % C,,E,)
Figure 9. Temperature-compositionphase diagram of the system pentaethylene glycol
ether (C,2E5), water, and n-octane. The weight ratio of water to octane is
fixed at 60 to 40. Diffusivities were measured in samples denoted by filled
circles. Phase diagram determinedby Kahlweit et al. Reprinted with
permission from Kahlweit et al. [l 11.

provide an affordable means to probe, primarily by computer simulation,

the connection between thermodynamic behavior and molecular interactions.
The ideas underlying elemental structures models are to establish micro-
structures experimentally, to compute free energies and chemical potentials
from models based on these structures, and to use the chemical potentials
to construct phase diagrams. Jonsson and Wennerstrom have used this
approach to predict the phase diagrams of water, hydrocarbon, and ionic
surfactant mixtures [ 181. In their model, they assume the surfactant resides
in sheetlike structures with heads on one side and tails on the other side
of the sheet. They consider five structures: spheres, inverted (reversed)
spheres, cylinders, inverted cylinders, and layers (lamellar). These struc-
tures are indicated in Fig. 12. Nonpolar regions (tails and oil) are cross-
hatched. For these elemental structures, Jonsson and Wennerstrom include
in the free energy contributions from the electrical double layer on the water
H. Ted Davis 183

-
10.9

0
Q, 10-10
u)
cu'
Y
E 10'' 1 + D, PURE HzO
0 X D, PURE Cg

10-12

A B C D E F G H I J I'H'G'FE'D'C'B'A

Mixture #
Figure 10. NMR diffusivities for the components of the samples denoted in Fig. 9 by
tilled circles. The samples are microemulsions lying along the one-phase
comdor in Fig. 9. Reprinted with permission from Bodet er af.[16].

side of the surfactant sheets, the entropy of mixing of micellar aggregates

and of the molecular components in aqueous and oleic regions, and the
surface energy of the surfactant sheets.
Figure 13 compares the observed ternary phase diagram of water, octanol,
and potassium caprate with that calculated by Jonsson and Wennerstrom.
The agreement, though not quantitative in every detail, is remarkable given
the complexity of the phase behavior. The agreement is certainly good enough
to encourage future development of theory along these lines for other
mixtures. Wennerstrom and Jonsson have in fact extended their analysis to
a few other systems.
In solutions of water and surfactant, the surfactant monolayers can join,
tail side against tail side, to form bilayers, which form lamellar liquid crystals
whose bilayers are planar and are arrayed periodically in the direction normal
to the bilayer surface. The bilayer thickens upon addition of oil, and the
distance between bilayers can be changed by adding salts or other solutes.
In the oil-free case, the hydrocarbon tails can be fluidlike (La) lamellar liquid
crystal or can be solidlike (Lp) lamellar liquid crystal. There also occurs
another phase, Pp, called the modulated or rippled phase, in which the bi-
layer thickness varies chaotically in place of the lamellae. Assuming la-
mellar liquid crystalline symmetry, Goldstein and Leibler [ 191 have
constructed a Hamiltonian in which (1) the intrabilayer energy is calculated
184 Future Opportunities in Thermodynamics

Figure 11. Freeze-fracture transmission electron micrograph of the microemulsion

denoted as sample J in Fig. 9. Reprinted with permission from W. Jahn and
R. Strey,J. Phys. Chem. 92,2294 (1988) [17]. Copyright 1988 American
Chemical Society.

a. b. C. d. e.
Figure 12. Schematic representation of the five elemental structures used by Jonsson
and Wennerstrom [ 181: (a) spherical, (b) cylindrical, (c) lamellar, (d)
inverted cylindrical, and (e) inverted spherical. Nonpolar regions are
crosshatched.
H. Ted Davis 185

Octanol

10 20 30 40 50 60 70 80 90
Water Potassium caprate

C OH
8*

"20 C,OOO' 'K+

Figure 13. (a) Experimental ternary phase diagrams of water, octanol, and potassium
caprate at 20°C. (b) Phase diagram predicted by Jonsson and Wennerstrom
[18] using an elemental structures model. Reprinted with permission from B.
Jonsson and H. Wennerstrom, J . Phys. Chem. 91,338 (1987) [ 181. Copyright
1987 American Chemical Society.

from a Landau-Ginzburg Hamiltonion accounting for the energy of thick-

ening, stretching, and bending of the bilayer and (2) the interbilayer inter-
actions are approximated by the van der Waals attractive potential plus the
hydration repulsion potential. The hydration potential has been found
empirically to be an exponential function of the distance between adjacent
186 Future Opportunities in Thermodynamics

o,/H
4

0 02 04 06 0 02 04 06
I-+ ’-4
Figure 14. Predicted (Goldstein and Leibler [ 191)temperature-compositionphase
diagrams of lamellar phases in water and phospholipid mixtures. La and Lp
are periodic liquid crystals and P is a modulated or rippled lamellar phase.
P
a2 is a model parameter that sca es as temperature.fis the volume fraction
of phospholipid.Reprinted with permission from Goldstein and Leibler [ 191.

bilayers. Goldstein and Leibler’s elemental structures model yields a phase

diagram (Fig. 14) similar to those observed for mixtures of water and
phospholipids, which are the well-known surfactants in biological mem-
branes.
Many of the special properties of microemulsions have been predicted
with the aid of the following elemental structures model advanced by Talmon
and Prager [20]: (1) oil- and water-rich domains are randomly interspersed
in space, (2) the surfactant lies entirely in monolayers separating the oil-
and water-rich domains, and (3) the free energy arises from the entropy of
interspersion of the domains and from the energy of curvature of the sur-
factant layers. Talmon and Prager used a random subdivision of space, namely
the Voronoi polyhedral tessellation, to provide a means of randomly inter-
spersing oil and water domains. In later variations of the Talmon-Prager
model, de Gennes and Taupin [21], Widom [22], and Safran et al. [23] used
the simpler cubic tessellation of space to model the random interspersion.
Qualitatively, the choice of tessellation does not appear to make much dif-
ference; from the micrograph shown in Fig. 11, it is clear that no polyhe-
dral tessellation will capture the detailed geometry of a microemulsion,
although it might adequately account for the bicontinuous topology. The
transition from globular to bicontinuous microemulsion is predicted by the
Talmon model, and the electrical conductivity observed in a 33,2 salin-
ity scan is quantitatively predicted by that model. The 2,3,2 sequence of
H. Ted Davis 187

W O

Figure 15. Monte Carlo simulation of the phase diagram of the amphiphile H,T,, water,
and oil modeled on a cubic lattice. Reprinted with permission from Larson
WI.

phase splits and the low-tension behavior are qualitatively accounted for
by various versions of the model. Recent versions of the model [22, 231
also focus attention on the importance of the preferred curvature and bending
energy of the surfactant sheets for the patterns of phase behavior of sur-
factant microstructured fluids.
As I stated earlier, lattice models, aided by computer simulations, are likely
to be very useful in predicting patterns of thermodynamic behavior and in
understanding molecular mechanisms in microstructured fluids. Perhaps the
simplest example of such a lattice model is the one employed recently by
Larson [24] to investigate patterns of liquid crystal formation by Monte Carlo
simulation. He assumes that the molecules occupy sites of a cubic lattice.
Water and oil molecules occupy single sites in the lattice. The surfactant
molecules (HmTn)consist of a string of rn head segments and n tail seg-
ments. A surfactant molecule occupies a sequence of adjacent sites, either
nearest neighbor or diagonal nearest neighbor. In terms of pair interaction
energies, water and a head segment are equivalent, and oil and a tail seg-
ment are equivalent. A simulated phase diagram for H4T4 is shown in Fig.
15. For this symmetric surfactant (equal lengths of head and tail moieties),
no cubic-phase liquid crystals are observed. Nevertheless, a surprisingly com-
plex phase diagram is found, with many of the features depicted in Fig. 3.
Larson found the bicontinuous liquid crystal phase with the asymmetric
surfactant H,T,.
188 Future Opportunities in Thermodynamics

(0) (b) (C)

Figure 16. A lattice model accounting for the bending energy of surfactant layers.
Filled circles denote the head group and unfilled circles the tail group of the
surfactant. Reprinted with permission from Widom [25].

Widom has recently formulated a lattice model that takes into account
the amphiphilic nature of surfactant and introduces molecular interactions
that explicitly affect curvature of surfactant sheetlike structures [25].
The three configurations of adjacent surfactant molecules shown in Fig.
16 have different energies of interaction. In addition to these interactions,
the model allows pair interactions among water, oil, surfactant head groups,
and surfactant tail groups. Taking advantage of the relationship between
Widom’s model and an equivalent Ising model, Dawson et al. [26] have
computed the phase diagram for oil, water, surfactant solutions from known
results for the Ising model and from mean field theory. They find lamellar
liquid crystalline phases, micellar solutions, and bicontinuous microemulsion
phases. For the model microemulsion, Dawson [27] has predicted phases
exhibiting ultralow tensions against oil-rich and water-rich phases having
compositions very different from that of the microemulsion. This is in
agreement with experiment. Further mean field theory and computer simu-
lations based on this model should be fruitful.
There are presently several groups around the world conducting molecular
dynamics simulations of micellization and liquid crystallization of more or
less realistic models of water, hydrocarbon, and surfactants. The memory
and speed of a supercomputer required to produce reliably equilibrated
microstructures constitute a challenge not yet met, in my opinion. By tak-
ing advantage of identified or hypothesized elemental structures one can,
however, hope to learn a great deal about the dynamics and stability of the
various identified microstructures.
H. Ted Davis 189

References
1. Gubbins, K. E., in Perspectives in Chemical Engineering: Research and Education (C.
K. Colton, ed.), p. 125. Academic Press, San Diego, Calif., 1991 (Adv. Chem. Eng. 16).
2. Panagiotopoulos, A. Z., Mol. Phys. 61,8 13 (1987).
3. Deitrick, G. L., Scriven, L. E., and Davis, H. T., J. Chem. Phys. 90,2370 (1 989).
4. Jonsson, B., Edholm, O., and Teleman, O., J. Chem. Phys. 85,2259 (1986).
5. Egberts, E., and Berendsen, H. J. C., J. Chem. Phys. 89,3718 (1988).
6. Lee, G., Arscott, P. G., Bloomfield, V. A., and Evans, D. F., Science 244,475 (1989).
7. Israelachvili, J. N., McGuiggan, P. M., and Homola, A. M., Science 241,795 (1988);
Van Alsten, J., and Granick, S., Phys. Rev. Lett. 61, 2570 (1988).
8. Davis, H. T., Bodet, J. F., Scriven, L. E., and Miller, W. G., in Physics of Amphiphilic
Layers (J. Meunier, D. Langevin, and N. Boccara, eds.), p. 3 10. Springer-Verlag, Ber-
lin, 1987.
9. Scriven, L. E., Nature 283, 123 (1976).
10. Knickerbocker, B. M., Pesheck, C. V., Scriven, L. E., and Davis, H. T.,J. Phys. Chem.
83, 1984 (1979); J . Phys. Chem. 86,393 (1982).
11. Kahlweit, M., Strey, R., Haase, D., Kunieda, H., Schemling, T., Faulhaber, B.,
Borkovec, M., Eicke, H.-F., Busse, G., Eggers, F., Funck, Th., Richmann, H., Magid,
L., Soderman, O., Stilbs, P., Winkler, J., Dittrrich, A., and Jahn, W., J. Colloid fnter-
face Sci. 118,436 (1987).
12. Healy, R. N., Reed, R. L., and Stenmark, D. G., Soc. Pet. Eng. J . 16, 147 (1976).
13. Kaler, E. W., and Prager, S., J. Colloid Interface Sci. 86, 359 (1982).
14. Guering, P., and Lindman, B., Langmuir 1,464 (1985).
15. Clarkson, M. T., Beaglehole, D., and Callaghan, P. T., Phys. Rev. Lett. 54, 1722 (1985).
16. Bodet, J. F., Bellare, J. R., Davis, H. T., Scriven, L. E., and Miller, W. G., J. Phys. Chem.
92, 1898 (1988).
17. Jahn, W., and Strey, R.,J. Phys. Chem. 92,2294 (1988).
18. Jonsson, B., and Wennerstrom, H., J. Phys. Chem. 91, 338 (1987);Wennerstrom, H.,
and Jonsson, B., J. Phys. France 49, 1033 (1988).
19. Goldstein, R. E., and Leibler, S., Phys. Rev. Lett. 61,2213 (1988).
20. Talmon, Y., and Prager, S., J. Chem. Phys. 69,2984 (1978).
21. de Gennes, P. G., and Taupin, C., J. Phys. Chem. 86,2294 (1982).
22. Widom, B., J. Chem. Phys. 84,6943 (1986).
23. Andelman, D., Cates, M. E., Roux, D., and Safran, S. A., J. Chem. Phys. 87, 7229
( 1 987).
24. Larson, R. G., J. Chem. Phys. 89, 1642 (1988).
25. Widom, B., J.Chem. Phys. 84,6943 (1986).
26. Dawson, K. A., Lipkin, M. D., and Widom, B., J. Chem. Phys. 88,5149 (1988).
27. Dawson, K. A,, Phys. Rev. A 35, 1766 (1987).