This article was downloaded by: [Adler, S]

On: 3 February 2011

Access details: Access Details: [subscription number 791040946]
Publisher Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-
41 Mortimer Street, London W1T 3JH, UK

Molecular Physics
Publication details, including instructions for authors and subscription information:
http://www.informaworld.com/smpp/title~content=t713395160

Simulations of phase transitions and free energies for ionic systems

Antti-Pekka Hynninena; Athanassios Z. Panagiotopoulosa
a
Department of Chemical Engineering and, Institute for the Science and Technology of Materials,
Princeton University, Princeton, NJ 08540, USA

First published on: 01 December 2010

To cite this Article Hynninen, Antti-Pekka and Panagiotopoulos, Athanassios Z.(2008) 'Simulations of phase transitions
and free energies for ionic systems', Molecular Physics, 106: 16, 2039 — 2051, First published on: 01 December 2010
(iFirst)
To link to this Article: DOI: 10.1080/00268970802112160
URL: http://dx.doi.org/10.1080/00268970802112160

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.informaworld.com/terms-and-conditions-of-access.pdf

This article may be used for research, teaching and private study purposes. Any substantial or
systematic reproduction, re-distribution, re-selling, loan or sub-licensing, systematic supply or
distribution in any form to anyone is expressly forbidden.

The publisher does not give any warranty express or implied or make any representation that the contents
will be complete or accurate or up to date. The accuracy of any instructions, formulae and drug doses
should be independently verified with primary sources. The publisher shall not be liable for any loss,
actions, claims, proceedings, demand or costs or damages whatsoever or howsoever caused arising directly
or indirectly in connection with or arising out of the use of this material.
 Molecular Physics
Vol. 106, Nos. 16–18, 20 August–20 September 2008, 2039–2051

INVITED TOPICAL REVIEW

Simulations of phase transitions and free energies for ionic systems
Antti-Pekka Hynninen and Athanassios Z. Panagiotopoulos*
Department of Chemical Engineering and Institute for the Science and Technology of Materials, Princeton University,
Princeton, NJ 08540, USA
(Received 26 February 2008; final version received 7 April 2008)

A review of simulation studies of phase equilibria and free energies for systems dominated by coulombic
interactions is presented. Phase transitions occur for ionic systems in the strong-coupling limit realized in low-
dielectric constant solvents, at low temperatures, or for high charge valences. The majority of simulation results
to date are for primitive models that treat the solvent as a uniform dielectric continuum. Transitions involving
fluid and solid phases for such models have been studied extensively in the past decade. There is now strong
evidence that the vapour–liquid transition is in the Ising universality class. For highly charged colloids the
vapour–liquid transition becomes metastable with respect to the fluid–solid transition and the behaviour matches
that of charged hard plates. Phase transitions of charged chains illustrate sensitivity of the phase behaviour to the
charge pattern. Studies of salt solubilities using models with explicit solvent suggest that reasonable agreement
with experiment can be achieved with existing force fields, but there is considerable room for improvement. Areas
Downloaded By: [Adler, S] At: 15:56 3 February 2011

of future research needs are briefly discussed.

Keywords: ionic systems; Monte Carlo; phase transitions; primitive model; charged colloids

1. Introduction They are relevant for systems as diverse as colloids,

A detailed understanding of interactions between
charged objects in solution is important for environ-
mental science, biology and applied chemistry.
Technologically relevant systems for which coulombic
interactions are important include molten and
dissolved salts, charge-stabilized colloids and micellar
solutions of ionic surfactants. Proteins and nucleic
acids carry charges along their backbones and are
strongly influenced by the presence of ions in solution.
Multivalent salts and charged proteins are potent
agents for DNA aggregation and precipitation [1] and
exert a strong influence on the structure and dynamics
of highly charged polyelectrolyte solutions [2]. The
classic Poisson–Boltzmann approach provides excel-
lent results at low electrolyte concentrations in aqueous
solutions, but breaks down for higher concentrations,
multivalent ions and low dielectric-constant solvents
[3,4]. Phase and conformational transitions in ionic
systems take place in the ‘strong-interaction’ limit at
which the Poisson–Boltzmann approach is not valid. A
key aspect of strongly interacting ionic systems is the
existence of effective attractions between like-charged
objects in solution. These attractions result from
correlations between ions and counterions and man-
ifest themselves in low-dielectric constant environ-
ments, low temperatures or for highly charged salts.
micelles, red-blood cells, bundling of DNA, and
aggregation of F-actin, microtubules, and viruses.
While originally controversial, this is now an accepted
idea strongly supported by experimental evidence [5].
Significant progress has been made in recent years
in theoretical understanding and atomistic simulation
of the structure, thermodynamics and phase transitions
of strongly interacting ionic systems. Nevertheless,
atomistic simulations in the presence of explicit solvent
still have difficulty reaching the length and time scales
relevant for phase transitions and achieving adequate
sampling of phase space. Much of the work on phase
transitions in ionic systems has been performed on
coarse grained, ‘primitive models’ that contain no
explicit solvent. Even for these simplified models,
simulations of Coulomb-dominated systems can
encounter significant sampling difficulties. For a
simple 1:1 primitive model electrolyte at conditions
near the liquid-vapour critical point, the thermal
energy is only 1/20th of the energy of two ions at
contact. As a result, the systems are strongly associat-
ing and hard to equilibrate. Early simulation work
produced results that (in retrospect) turned out to
suffer from inadequate sampling and equilibration
problems, even though most of the qualitative features
of the observed behaviour were correct. Partly as a

*Corresponding author. Email: azp@princeton.edu

ISSN 0026–8976 print/ISSN 1362–3028 online

ß 2008 Taylor & Francis
DOI: 10.1080/00268970802112160
http://www.informaworld.com
 2040 A.-P. Hynninen and A.Z. Panagiotopoulos
result of the increased availability of consistent
simulation data, improved theories of systems domi-
nated by coulombic interactions have become available
in the past few years [6–10]. Unlike earlier theoretical
approaches [11,12], these predict the correct trends
with respect to size and valence asymmetries. Recent
simulation studies have also contributed towards
clarification of the character of ionic criticality, a
question of great theoretical importance [13–15] and
the subject of many experimental investigations
[e.g. 16–18]. Recent reviews of theoretical tools for
charged systems [19] and ionic criticality [20] are
available. A review of the statics and dynamics of
charged soft matter [21] covers in detail coulomb-
dominated interactions between charged surfaces,
while an earlier review [22] discusses effective interac-
tions between double layers and initial studies of phase
transitions of charged colloidal systems. Older work on
structure and thermodynamics of polyelectrolytes has
been reviewed in [23].
Downloaded By: [Adler, S] At: 15:56 3 February 2011
The objective of the present review, which expands
and updates a shorter article [24], is to summarize the
state-of-the-art on simulation studies of the phase
behaviour of systems dominated by coulombic inter-
actions. In Section 2, we review the current status of
vapour-liquid and solid–fluid equilibria for the
restricted primitive model of ionic solutions. Other
spherically symmetric ionic systems, including charge-
and size- asymmetric primitive models and charged
colloids are the focus of Section 3. Section 4 is devoted
to charged chains and networks and Section 5 to ionic
liquid phase equilibria and the modelling of solubilities
and activity coefficients in salt solutions. Finally,
Section 6 summarizes the current status and touches
upon unresolved questions and research needs.
2. The restricted primitive model
A simple description of systems dominated by
coulombic interactions is provided by the ‘restricted
primitive model’ (RPM) that consists of charged hard
spheres of equal diameter  in a uniform continuum of
dielectric constant ". Half of the spheres carry a
positive and half a negative charge of identical
magnitude, q. The energy of interaction between two
particles is
(
þ1 rij < 
Uij ¼ qi qj ð1Þ
rij  
4""0 rij
where rij is the distance between spheres i and j, qi and
qj their respective charges and "0 the dielectric
permittivity of vacuum. It is customary to choose a
reduced system of units so that temperature is normal-
ized by the energy of interaction of two ions at contact,
4""0 kT
T ¼ ð2Þ
q2
where k is Boltzmann’s constant. Another common
measure of the strength of interactions is the Bjerrum
length,
B ¼ e2 =""0 kB T ¼ =T  ð3Þ
The total number density in the system is
Nþ þ N 3
 ¼  ð4Þ
V
where Nþ ¼ N is the number of positive and negative
particles and V is the system volume.
Because of its simplicity, the RPM has been
extensively studied by both simulations and theoretical
methods. Vorontsov-Velyaminov et al. [25,26] were the
first to propose that the RPM has a liquid–vapour
transition and critical point; using constant-pressure
Monte Carlo simulations; they obtained Tc  0.095
and c  0.17. Stell, Wu and Larsen [27] used equation
of state data from simulations and theoretical approx-
imations of the free energy to obtain Tc  0.085 and
c  0.01. While the estimated values of the critical
parameters differ significantly from ‘modern’ values
given in Table 1, these early studies established firmly
that the system has a low-temperature vapour-liquid
transition.
In the early 1990s, Panagiotopoulos [28] obtained
an estimate of the critical point as Tc ¼ 0.056 and
c ¼ 0.04 using Gibbs ensemble Monte Carlo [29] with
single-ion transfers. Orkoulas and Panagiotopoulos
[30] introduced biased pair transfers used by many
subsequent investigations and obtained Tc ¼ 0.053 and
c ¼ 0.025. It is now known that the Gibbs ensemble is
not well suited to high precision phase coexistence
calculations in the immediate vicinity of critical points
[31]. Clearly, significant uncertainties remained regard-
ing the location of the critical point of the RPM ten
years ago.
Table 1. Critical parameters for the gas–liquid transition of
the RPM.
Ref. Tc c
44 0.04933  0.00005 0.075  0.001
88 0.04917  0.00002 0.080  0.005
39 0.0490  0.0003 0.070  0.005
40 0.0492  0.0003 0.062  0.005
41 0.0489  0.0003 0.076  0.003
 Molecular Physics
A methodological advance in calculations of
critical points of fluids that do not have particle–hole
symmetry was the mixed-field finite-size scaling
approach of Bruce and Wilding [32,33]. The method
is based on matching results for the scaled order
parameter probability distribution to the universal
curve appropriate for the universality class of the
system under study. For bulk fluids, the appropriate
scaling variable is a linear combination of density and
energy, assuming that there is no pressure mixing in the
scaling fields [34]. This method has been used by most
simulation studies of the vapour–liquid critical point of
the RPM in the past decade. The method works well in
combination with the grand canonical histogram
reweighting approach of Ferrenberg and Swendsen
[35], which allows extrapolation of results from a
simulation run to a range of chemical potentials and
temperatures in its vicinity.
The critical parameters of the RPM were obtained
using the Bruce–Wilding approach by Caillol et al.
Downloaded By: [Adler, S] At: 15:56 3 February 2011
[36–38] in hyperspherical boundary conditions and
Orkoulas and Panagiotopoulos [39] in standard cubic
boundary conditions. Yan and de Pablo [40] used it in
combination with hyper-parallel tempering Monte
Carlo that allows for multiple replicas of the simula-
tion at different conditions of temperature and
chemical potential. Panagiotopoulos [41] obtained the
critical parameters of the continuum RPM by extra-
polating results on the finely discretized lattice analog
to the RPM to infinitely fine discretization.
The Bruce–Wilding approach used in these studies
of RPM criticality assumes that the system is in the
three-dimensional Ising universality class. At most, one
may claim consistency with Ising critical behaviour. An
unbiased finite-size extrapolation method using the
grand canonical ensemble [42] was applied to a
discretized version of the RPM by Luijten et al. [43].
Kim and Fisher [44] examined the discretization
dependence of the critical behaviour and extrapolated
to the continuum limit. The unequivocal conclusion
from these two studies was that the RPM is in the
three-dimensional Ising universality class, as also
argued on theoretical grounds [45]. Some earlier
studies of the heat capacity in the canonical (NVT)
ensemble [46,47] did not observe a peak near the
critical point and interpreted this as an indication of
non-Ising behaviour. However, these observations
were later shown to be due to the suppression of
density fluctuations in the canonical ensemble [48].
Table 1 gives a summary of recent results for the
critical parameters of the RPM. There is generally
good agreement between simulations that used differ-
ent boundary conditions and sampling algorithms. The
critical temperature has relatively small uncertainties;
2041
the two most recent estimates [44,38] disagree by more
than their combined error bars but the difference is
only 0.3% of the value of Tc . The other simulations
listed in the Table (with more generous error bars)
overlap these two. This difference may be due to the
use of the Bruce–Wilding approach in [38] versus an
unbiased extrapolation with no assumption of uni-
versality class in [44]. The situation for the critical
density is less satisfactory, with large relative uncer-
tainties and several differences outside the combined
simulation uncertainties. Kim and Fisher [34] sug-
gested that the use of the Bruce–Wilding method with
no pressure mixing could lead to unreliable estimates
of the critical density, but is not likely to result in
major errors in the critical temperature. Clearly, the
precise determination of the critical density for the
RPM remains a topic for future work.
It has already been mentioned that several recent
studies used a discretized version of the RPM. In this
approach, calculations are performed on a simple cubic
lattice of spacing l less than the particle diameter ,
with pre-computed interactions between all lattice sites
for computational efficiency [49,50]. Parameter  ¼ /l
controls how closely the model approximates contin-
uous space. For  ¼ 1, the lattice-discretized RPM has
a tricritical point and an order–disorder transition, as
first observed by Dickman and Stell [51]. The lattice
RPM model with  ¼ 2 was also found [49] to have
phase behaviour qualitatively different from the
continuum model. However, for   3 the phase
behaviour is qualitatively identical to the continuum;
critical point and coexistence curves match continuum
data within a few percent. Lattice discretization effects
have been subsequently analyzed in detail using
theoretical approaches [52–55] and simulations [41,44].
Several systems closely related to the RPM have
also been studied with respect to their vapour–liquid
phase behaviour. The first is charged hard dumbbells
or tethered dimers, approximating the RPM
because there is strong ion pairing in the (untethered)
RPM at its critical temperature. For the contact
tethered dimer of equal-size opposite charges, the
critical temperature is almost the same as for
the unconstrained RPM, but the critical density
is significantly higher: Tc ¼ 0:04911  0:00003,
c ¼ 0:101  0:003 [56]. Tethered-ion systems of vary-
ing tether length were studied in [57]. Interestingly, as
the tether length approaches infinity, care must be
exercised to prevent an ‘entropy catastrophe’ and
obtain a proper thermodynamic limit. The dependence
of the critical parameters on tether length was found to
be smooth but non-monotonic. The other related
system is that of soft charged spheres studied in [58],
for which Tc  0:047, c  0:070.
 2042 A.-P. Hynninen and A.Z. Panagiotopoulos
Downloaded By: [Adler, S] At: 15:56 3 February 2011

Figure 1. (color online). Unit cells of (a) CsCl or bcc, (b) CuAu, and (c) tetragonal fcc structures. The dark (red) and light
(yellow) colors denote opposite charges. Reprinted with permission from [62]. Copyright ß (2006) American Physical Society.

The RPM has several transitions involving solid 0.4

phases, in addition to the liquid–vapour transition. Fcc
0.35
Early simulations of fluid–solid and solid–solid disordered
equilibria for the RPM by Smit et al. [59] and Vega 0.3
et al. [60], identified a liquid–bcc transition at low 0.25
temperatures and a liquid–fcc transition at high Tetragonal
T*

temperatures. The bcc crystal is made out of two 0.2 Fluid

species of oppositely charged spheres, and therefore 0.15 CuAu
corresponds to the atomic structure of CsCl, see
Figure 1(a). Bresme et al. [61] amended the phase 0.1
CsCl
diagram by identifying an additional tetragonal fcc 0.05
solid (see Figure 1c) that becomes stable at high
0
colloid densities. Recently, Hynninen et al. [62] found 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
a novel stable crystal phase that has the symmetry of Figure 2. Phase diagram of the RPM in the packing fraction-
the CuAu binary alloy, shown in Figure 1(b). reduced temperature T* plane. Reprinted with permission
Interestingly, this crystal structure was first suggested from [62]. Copyright ß (2006) American Physical Society.
by experimental observations on a binary mixture of
oppositely charged colloidal spheres. Indeed, accord-
ing to calculations by Hynninen et al., the phase to a liquid–bcc phase transition [63]. The phase
behaviour of the RPM is qualitatively similar to that diagram of the RPM is shown in Figure 2 in the
of oppositely charged colloids. Such a correspondence packing fraction , reduced temperature T* repre-
is expected since the pair potential in the system of sentation. The gray areas denote coexistence regions
oppositely charged colloids is given by a Coulomb of the first-order phase transitions where tie lines are
potential multiplied
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi by a screening factor er, where horizontal. The CsCl–CuAu transition is a weak first-
 ¼ 8B salt is the Debye screening length and salt order martensitic phase transition, i.e. the transition
is the salt concentration. As  ! 0, the two systems occurs by a continuous deformation of the crystal
become identical. The main qualitative difference unit cell.
between the two systems is that at   4.5/, the The RPM phase diagram was determined using
gas–liquid critical point is metastable with respect thermodynamic integration, a method that calculates
 Molecular Physics
the Helmholtz free energy relative to the reference state
whose free energy is known, enabling the calculation of
the full free energy. This approach was first proposed
by Kirkwood in 1935 [64] and was later extended to
include solids by Frenkel and Ladd [65]. In the fluid
phase, the reference state is the hard-sphere fluid. In
the case of solid phases the reference state is the
Einstein crystal where each particle is tied to its lattice
position by a harmonic spring. The phase boundaries
are drawn using the Helmholtz free energies and a
common tangent construction, which guarantees ther-
modynamic stability, i.e. equality of temperature,
pressure, and chemical potential. A similar computa-
tional approach has been used for some of the work on
solid phases of charged colloids described in the
following section.
3. Generalized primitive models and charged colloids
In the RPM model, all ions have the same size and
Downloaded By: [Adler, S] At: 15:56 3 February 2011
absolute charge and only interact through coulombic
forces and hard-sphere repulsions. This section sum-
marizes simulation studies of ionic systems for which
these constraints have been relaxed, resulting in
generalized primitive models.
Phase diagrams of size-asymmetric electrolytes
were obtained by Romero-Enrique et al. [66] using a
fine-lattice discretization approach and by Yan and
de Pablo [67,68] using a multidimensional parallel
tempering algorithm. The key finding from these
studies was that the critical temperature and critical
density are at a maximum for the size-symmetric case
and then drop off as the size asymmetry increases. This
is in contrast to predictions of commonly used theories
for ionic fluids [11,12]. Newer theories have been
developed [6–10] that predict the correct trends of the
critical parameters with size asymmetry.
Multivalent electrolytes have been studied much
less than monovalent ones, despite their importance in
micellar, colloidal and biological systems. Detailed
studies of 2:1 and 3:1 electrolytes for a broad range of
size ratios were performed by Panagiotopoulos and
Fisher [69] and Yan and de Pablo [70]. The main
finding from these studies is that the critical tempera-
ture and density show a maximum for multivalent ions
of increasing size as the valency is increased.
Panagiotopoulos and Fisher [69] suggest that for
small multivalent ions with large monovalent counter-
ions the vapour–liquid phase transition is likely to
disappear because of the formation of an open
network.
Reščič and Linse [71] have investigated the critical
parameters for a 10:1 electrolyte at the point
2043
counterion limit using thermodynamic scaling Monte
Carlo. Linse [72,73] proposed a scaling law for the
critical temperature of systems with charge ratios up to
80:1, also at the point counterion limit. A study of
charge-asymmetric electrolytes with charge ratios up to
10:1 was performed in [74], using reservoir grand
canonical simulations. Some discrepancies were
observed between the results of this study and the
scaling law of Linse that remain to be resolved.
However, it is clear from these studies that salt-free
colloidal systems phase separate through effective like-
ion attractions at sufficiently high couplings.
The existence of like-charge attraction at low
coupling (e.g. micrometre colloids in aqueous solvent
with monovalent ions) has been a controversial topic
since the experiments that among other things showed
a broad gas-solid coexistence [75,76], gas–liquid con-
densation [77] (see, however, [78]), large stable voids
[79,80], and anomalously long-lived dense clusters [81].
These experiments, which have been proven difficult to
replicate, provided the motivation for numerous
theoretical [82–84] and simulation studies. According
to the well-established DLVO theory, no attractions
are possible at low coupling. However, the theoretical
descriptions propose a situation where so called
‘volume terms’, that include all non-pairwise additive
many-body terms, can give rise to a gas-liquid or gas-
solid phase coexistence. Many of the volume theories
have been criticized for their approximations, most
important of which is the linearization of the Poisson–
Boltzmann equation. In fact, as Lu and Denton [85]
recently showed, the other major approximation, the
use of the first-order thermodynamic perturbation
theory, is of lesser importance. In a recent theoretical
treatment [86], the linearization approximation was
partly avoided by solving the nonlinear Poisson–
Boltzmann equation numerically close to the colloid
surface, where the linearization is most problematic,
and applying the linear Poisson–Boltzmann for dis-
tances further from the surface. This volume theory
also predicted broad coexistence regions at low
coupling, however, consequently the parameter space
where phase separation occurred was considerably
smaller than the one obtained using the purely linear
Poisson–Boltzmann theory in [82].
Realistic intermolecular potentials for ionic systems
include soft repulsions and dispersion forces in
addition to the coulombic interactions [87].
Simulation studies of phase transitions for realistic
salt models of salts have been performed by Guissani
and Guillot for NaCl [88] and NH4Cl [89].
Thermodynamic integration methods were used to
obtain the melting point for the Tosi-Fumi-Tosi model
of NaCl [90] in good agreement to experimental data.
 2044 A.-P. Hynninen and A.Z. Panagiotopoulos
Studies of charged Yukawa systems are also available
[91–93]. In addition to the melting point, a calculation
of the homogeneous nucleation rate for the Tosi-Fumi
model of NaCl [94] was performed at deeply subcooled
temperatures using forward flux [95] and transition
interface [96] sampling methods. The nucleation rate
extrapolated to temperatures closer to melting was
several orders of magnitude lower than experimental
measurements, possibly because of model deficiencies
or the limitations of the extrapolation procedure.
In charged nanoparticle systems, where there are
competing interactions to the electrostatic ones
operating at shorter distances, several unusual
morphologies are possible. For example, when attrac-
tive short-range interactions exist between like-charge
ions, fascinating aggregation patterns have been
observed in two dimensions [97]. Simple systems in
three dimensions in which attractions and repulsions
operate among all ions have been studied [98–100]. For
the case of nearest-neighbor exclusions [101] a charge-
Downloaded By: [Adler, S] At: 15:56 3 February 2011
ordered ‘striped’ phase was found to be stable.
Real charged chain or colloidal systems almost
always have varying amount of salt or other ionic
species. Simulations of phase transitions in charged
colloidal (macroion) systems with added salt have been
performed by Hynninen et al. [102]. The critical
parameters were determined for macroion to counter-
ion charge-asymmetries of 2:1, 3:1 and 10:1. This study
found that binary electrolyte mixtures are type-I
mixtures, with the two-components mixing
continuously.
In recent work [103], we have examined the global
phase diagram of charged colloids with charge
asymmetries up to 2000:1. The full phase diagrams
were calculated for asymmetries 2:1, 10:1, 20:1, and
50:1, and are shown in Figure 3. These phase diagrams
were calculated using the Helmholtz free energy
methods mentioned near the end of Section 2. An
important observation was that the gas–liquid phase
separation becomes metastable with respect to gas-
solid phase separation at charge-asymmetry 20:1 and
beyond. In order to have meaningful critical para-
meters for cases where the gas-liquid critical point is
metastable, we defined a ‘critical temperature’ as the
inflection point of the gas–solid phase separation line.
The advantage of this definition is that the inflection
point coincides with the gas–liquid critical point
continuously as charge-asymmetry is lowered below
20:1. In Figure 3(c) and (d), the inflection points are
denoted by the squares. From Figure 3 we see that the
dominant part of the phase diagram for charge
asymmetries greater than 20:1 is a broad gas–fcc
coexistence region, where the gas phase is very
dilute and the fcc solid phase is almost maximally
(0.74) packed. Since the gas phase is dilute, the
particles rarely interact and one can justify a simula-
tion with only few colloids. Similarly, since the solid
phase is very dense, the colloids are almost fixed to
their ideal lattice sites and one can imagine approx-
imating them with fixed colloids. Hence, we proposed a
minimal system approximation where the gas phase is
replaced by a single colloid fixed at the center of
the box and the fcc phase is replaced by four colloids
fixed at their ideal fcc lattice positions. In both cases,
the counterions are mobile and the free energy is
calculated in the same manner as in the full system.
Since periodic boundary conditions are used, the
colloids interact with the periodic images of themselves
and of the counterions even in the case of a single
colloid. We tested the validity of the gas phase
approximation by using four mobile colloids and
found that the results became closer to the full
system, but the difference was small. In Figure 3(c)
and (d), the results of the minimal system approxima-
tion are denoted by the filled symbols. As can be seen,
the approximate critical temperature is pretty close to
the true value and the approximation becomes better
with increasing charge asymmetry. Using the minimal
system approximation, we calculated the critical point
for charge-asymmetries 200:1, 1000:1, and 2000:1. In
Figure 4, we plot the critical temperature as a function
of the charge-asymmetry Q:1. The solid line is a power-
law fit to the data given by
Tc ¼ 0:27Q0:48 ð5Þ
The dashed line is the scaling proposed by Linse et al.,
Tc  Q0:57 . While our data is based on (minimal
system) free energy calculations, Linse et al. based their
results on the calculation of the structure factor for
10  Q  80. The difference in the two results could
stem from the fact that in the structure factor analysis,
one cannot rigorously tell if the chosen state points are
thermodynamically stable or in phase coexistence,
while in our method thermodynamic stability is
guaranteed. Interestingly, the approximately
Tc  Q1=2 scaling behaviour in Equation (5) can be
derived analytically using known results for planar
charged plates. For two infinite parallel plates with
constant (and similar) surface charge and point-like
counterions, in the low-coupling regime, the plates
repel each other as is expected based on the Poisson–
Boltzmann theory. In the high coupling regime, the
assumptions of the Poisson–Boltzmann theory break
down and plates start attracting each other. In the
asymptotic high coupling limit, this phenomenon is
captured by the so-called strong coupling theory by
Moreira and Netz [104]. The strength of
 Molecular Physics 2045
Downloaded By: [Adler, S] At: 15:56 3 February 2011

Figure 3. (color online). Phase diagrams for charged colloidal suspension with charge ratios: (a) 2:1, (b) 10:1, (c) 20:1, and (d)
50:1. Open symbols represent a full calculation of the free energy and closed symbols an approximation based on a minimal
system. The stars are critical points and the triangles are triple points. Reprinted with permission from [103]. Copyright ß (2007)
American Physical Society.

the electrostatic correlations is described by a coupling
parameter
q2  B
¼ ¼ 2Q=ðT  Þ2 , ð6Þ
where ¼ 1/2qBs is the Gouy–Chapman length and
s ¼ Qq/ 2 is the surface charge density of the colloid.
Figure 5 shows the phase diagram of charged planar
plates as a function of reduced plate separation
d~ ¼ d= and coupling strength  as obtained by
Moreira and Netz [104]. At   c  17 the system
undergoes a first-order transition, where the stable
state is coexistence between densely packed plates and
plates at infinite separation. Respectively, these two
states correspond to our densely packed fcc solid phase
and low-density gas phase. Solving T* from equation
above and using  ¼ c, we obtain
 1=2
2

Tc,plate ¼ Q1=2 ¼ 0:343Q1=2 ð7Þ
c
The fact that the spherical particles and plates have the
same critical scaling law nicely illustrates the common
origin of like-charge attraction in these two systems.
The advanced volume-term theory presented in [86]
shows metastable gas–liquid phase coexistence and a
stable gas–solid coexistence at Q ¼ 2000, T* ¼ 20 and no
added salt (a ¼ 0), in the region where the
Poisson–Boltzmann equation is valid. However, our
simulation results at Q ¼ 2000 show that no broad
coexistence is expected for T*  10. Furthermore,
separate (minimal system) free energy calculations
performed at T* ¼ 20 did not show any signs of
instabilities. Direct comparison is somewhat hampered
by the fact that the theory relies on small amount of
added salt to be present in the system to produce optimal
results. However, this first comparison between simula-
tions and theory shows that there is room for improve-
ment in the accuracy of the volume-term theories.
4. Charged chains and ionic gels
Polyelectrolytes and polyampholytes are polymers that
contain charged groups along their backbones or in
side chains. They can be found naturally in the form of
 2046 A.-P. Hynninen and A.Z. Panagiotopoulos
10
8
6
* There have been relatively few studies of phase
Tc
4 studied fully charged lattice chains of chain lengths up
2 critical density was found to be approximately inde-
0
10 100 1000
Q critical parameters in salt-free polyelectrolyte solutions
Figure 4. (color online). Critical temperature as a function of
colloid charge-asymmetry. The squares are the critical
temperatures. Reprinted with permission from [103].
Copyright ß (2007) American Physical Society.
Downloaded By: [Adler, S] At: 15:56 3 February 2011
Figure 5. MC phase diagram of charged planar plates as a
function of reduced plate separation d~ and coupling strength
. Reprinted with permission from [104]. Copyright ß (2001)
American Physical Society.
proteins and nucleic acids, or they can be made
synthetically. They are involved in industrial processes
such as wastewater treatment or oil recovery and are
found in many consumer products. An important
natural polyelectrolyte is DNA, the main carrier of
genetic information. It is well known that multivalent
ions induce attractions between DNA strands
mediated by strong correlation effects [1].
Experiments have shown precipitation and subsequent
redissolution of DNA with addition of polyamine salts
[105]. Properties of polyelectrolytes in solution have
been extensively studied by simulation [e.g. 106–108]
and theoretical methods [e.g. 109], with generally good
agreement for single-chain conformational properties
at low coupling strengths. Collective (multichain)
properties such as the osmotic pressure are harder to
obtain accurately [110]. Phase transitions and critical
points are especially challenging because they occur at
conditions for which charge correlation effects are
important.
transitions in charged chains. Orkoulas et al. [111]
to 24 beads, along with monomeric counterions. The
pendent of chain length, in sharp contrast to homo-
polymers for which the critical density approaches zero
as the chain length increases. Theoretical predictions for
[112] are in only qualitative agreement with simulations.
Cheong and Panagiotopoulos [113] studied polyam-
pholyte lattice chains, both fully charged and end-
charged. For fully charged chains, they determined that
the critical parameters are sensitive functions of the
charge sequence. End-charged chains with intervening
neutral ‘spacer’ beads have decreasing critical tempera-
tures and densities with increasing the spacer length.
These features are qualitatively reproduced by the
charged hard-sphere chain model of Jiang et al. [114],
but significant quantitative discrepancies remain
between simulation and theoretical results.
Polyelectrolyte networks are of interest as possible
drug delivery systems and pH- or concentration-
sensitive actuators. Of particular interest to the present
review are studies of discontinuous volumetric transi-
tions in polyelectrolyte gels due to coulombic interac-
tions, rather than transitions due to polymer-solvent
interactions. Linse and coworkers [115–117] have used
Monte Carlo methods to investigate such phenomena
in perfect primitive polyelectrolyte networks of dia-
mond topology with explicit counterions. Instabilities
indicative of first-order transitions between collapsed
and expanded gels are found as the electrostatic
coupling or additional solvent-mediated short-range
attractions are increased. The presence of salt [117]
reduced the volume of the swollen state but did not
fully suppress the transition. The salt content of the gel
was found to be greater than the prediction of ideal
Donnan equilibrium. A related model has been studied
using molecular dynamics simulations by de Pablo and
coworkers [118,119], who provide a detailed discussion
of the contributions to the network osmotic pressure
due to chain elasticity and electrostatic contributions.
5. Ionic liquids and aqueous salt solutions
Ionic liquids are organic salts that have melting points
near or below room temperature [120,121]. They have
 Molecular Physics
received considerable attention in recent years as
potential ‘green’ solvents for separations [122] and
chemical reactions. Their primary advantage is their
low to negligible volatility and their ability to dissolve
both polar and non-polar species. The properties of
ionic liquids are strongly influenced by electrostatic
interactions, even though specific short-range ion-ion
and ion-solvent forces also play a role [123].
Many simulation studies of ionic liquids aim to
obtain single-phase thermodynamic and transport
properties such as their density or viscosity [124].
Phase equilibrium properties of ionic liquids that have
been studied by simulation fall under two main
categories. The first is the determination of solid-
fluid equilibria for the pure components. The melting
(or freezing) temperature is one of the key parameters
for many practical applications and its accurate
calculation is impossible in direct simulations because
of strong hysteresis effects. Accurate freezing point
calculations for ionic liquids were first performed only
Downloaded By: [Adler, S] At: 15:56 3 February 2011
recently [125], using a specially developed thermo-
dynamic integration approach [126]. The second
category of simulations of phase equilibria involving
ionic liquids is the determination of small-molecule
solubilities in ionic liquids. This can be done more
easily, as it does not involve the free energy of the ionic
liquid itself. Direct particle insertions [127], thermo-
dynamic integration and Gibbs ensemble Monte Carlo
methods have been used to study the solubility of
water and volatile gases such as CO2 in ionic liquids
[128–131].
Aqueous solutions of electrolytes are of central
importance in geochemical, electrochemical, many
separation processes and for biological systems.
Atomistic explicit-solvent simulations of aqueous salt
solutions with explicit solvent have been used to obtain
structural and dynamic properties [e.g.132–134] and
potentials-of-mean-force (PMFs) [135] in salt solu-
tions. Once again, obtaining free energies (or activity
coefficients) and solubilities from simulations of
explicit-solvent models is cumbersome and computa-
tionally expensive, but is becoming more common as
both computational capabilities and simulation meth-
odologies advance. The calculation of solubilities of
ionic salts in water requires the calculation of the
chemical potential in solution and in the solid phase.
The solubility of KF in water was studied by Ferrario
et al. using thermodynamic integration for the
individual ions in solution and integration to the
Einstein crystal for the solid [136]. Using previously
available non-polarizable force fields, the saturation
concentration at 320 K was obtained from the simula-
tions to be 26 M, in reasonable agreement to
the experimental value of 17 M. The KF/H2O and
2047
NaCl/H2O binary systems were more recently studied
by Sanz and Vega [137] for a range of temperatures.
For the KF/H2O system, discrepancies between model
and experiments were shown to increase rapidly at
lower temperatures, but for NaCl good agreement was
seen. An expanded-ensemble fractional particle
approach for getting electrolyte solution free energies
was recently proposed and applied to the NaCl/H2O
system [138].
The presence of salts modifies the solubility of
other components in water. A recent study for the
solubility of methane in aqueous NaCl solutions used
the Widom test particle approach to obtain the CH4
chemical potential at infinite dilution [139]. The
CH4–H2O interaction parameters were obtained in an
earlier study [140] by matching experimental solubility
data in binary system. For the aqueous salt system, a
salting out effect is correctly predicted, but its
magnitude is about twice as large as that seen
experimentally, a difference attributed in [139] to the
neglect of polarization effects of the ions in solution.
Interestingly, earlier studies on CO2 solubility in NaCl
solutions [141] and mixed solvent electrolyte systems
[142] using Gibbs ensemble simulations [29] also
observed an overprediction of salt effects on the
solubilities of other components.
Deviations of the free energy of mixing from ideal
behaviour are often described in liquid solutions in
terms of activity coefficients. For salt solutions, activity
coefficients are well described at low concentrations
using the Debye–Hückel limiting law, but deviations
become important above 0.1 M for 1:1 electrolytes and
at even lower concentrations for multivalent salts [143].
Many models have been proposed to describe activity
coefficients in concentrated solutions of electrolytes
[144], but these generally rely on a number of
phenomenological parameters that need to be fitted to
experimental data. Several investigators have used
PMFs derived from atomistic simulations to obtain
activity coefficients for use in implicit-solvent calcula-
tions [145,146], but quantitative agreement for solutions
with salt concentrations greater than 1.0 M is generally
not attained.
Recent work in our group [147] has combined
dielectric saturation and hydration forces to obtain
effective potentials that describe quantitatively activity
coefficients over a broad range of concentrations. An
example for the effective potential obtained for NaCl
at ambient conditions is shown in Figure 6, with the
corresponding activity coefficients shown in Figure 7.
Interestingly, the activity coefficients for NaCl
obtained from free energy calculations of atomistic
models in [137] were in only qualitative agreement to
experimental data at intermediate concentrations.
 2048 A.-P. Hynninen and A.Z. Panagiotopoulos
Figure 6. Effective interaction potential between Naþ and
Cl ions and comparison to fixed dielectric constant model
(" ¼ 78). Reprinted with permission from [147]. Copyright ß
(2007) American Institute of Physics.
Downloaded By: [Adler, S] At: 15:56 3 February 2011
Figure 7. Mean ionic activity coefficient for NaCl as a
function of molality at 298 K. Simulations with an implicit
model from [147] are shown as circles. Experimental data are
shows as continuous lines; dashed lines are the Debye-Hückel
limiting law (DH) and a fixed dielectric constant model
(" ¼ 78). Reprinted with permission from [147]. Copyright ß
(2007) American Institute of Physics.
6. Summary and future outlook
Phase transitions of strongly coupled ionic systems
have been extensively studied over the past decade. For
primitive models that describe the solvent as a uniform
dielectric continuum, there are now quantitative phase
diagrams including fluid and solid phases for arbitrary
charge and size asymmetries. The liquid-vapour phase
transition has been confirmed to belong to the Ising
universality class. For highly charge asymmetric
colloids, the phase behaviour scales in a fashion
identical to that of attractions between charged flat
plates.
While significant progress has been made in
simulations of phase equilibria for primitive models,
relatively few studies of realistic potential models for
pure ionic liquids and aqueous salt solutions are
available. Much work remains to be done in obtaining
accurate force fields for ionic liquids and dissolved
salts. It is not clear at the present time if ionic
polarizability needs to be included in these force fields
explicitly to attain quantitative agreement to experi-
ment for broad sets of conditions. An accurate
polarizable force field for water has been proposed
recently [148] but its performance for ionic solution
calculations has not been explored. Effective potentials
for aqueous salt solutions have been obtained to model
solution nonidealities over broad concentration ranges,
but their transferability with respect to salt valence and
temperature remains an open issue.
An important consideration for calculating inter-
actions in ionic systems relates to the surface boundary
conditions. Common assumptions are constant-charge,
constant potential, or full charge regulation. The first
case corresponds to the surfaces having a fixed charge,
independent of local environment and surface separa-
tion, which is physically unrealistic. The second case
can be imposed for a conducting (electrode) surface.
For insulating surfaces or colloidal particles, a more
realistic assumption is often that there is chemical
equilibrium of ionizable groups on the surface but
most simulation studies to date correspond to the fixed
charge condition. The incorporation of ionizable
groups in Monte Carlo simulations can be done
through a ‘Reactive Canonical Monte Carlo’ approach
[149,150] that has also been used for the study of pH
effects in proteins [151].
Acknowledgements
The authors acknowledge financial support for this work by
the National Science Foundation, through the Princeton
Center for Complex Materials (NSF-MRSEC Program,
grant DMR- 0213706) and by the Department of Energy,
Office of Basic Energy Sciences (grant DE-FG201ER15121).
References
[1] E. Allahyarov, G. Gompper, and H. Löwen, Phys. Rev. E
69, 041904 (2004).
[2] Y. Zhang, J.F. Douglas, B.D. Ermi, et al., J. Chem. Phys.
114, 3299 (2001).
 Molecular Physics
[3] L. Guldbrand, B. Jonsson, H. Wennerstrom, et al.,
J. Chem. Phys. 80, 2221 (1984).
[4] M. Boström, D.R.M. Williams, and B.W. Ninham,
Phys. Rev. Lett. 87, 168103 (2001).
[5] O. Zohar, I. Leizerson, and U. Sivan, Phys. Rev. Lett
96, 177802 (2006).
[6] Y.V. Kalyuzhnyi, M. Holovko, and V. Vlachy, J. Stat.
Phys. 100, 243 (2000).
[7] D.M. Zuckerman, M.E. Fisher, and S. Bekiranov, Phys.
Rev. E 64, 011206 (2001).
[8] M.N. Artyomov, V. Kobelev, and A.B. Kolomeisky,
J. Chem. Phys. 118, 6394 (2003).
[9] O. Patsahan, I. Mryglod, and T. Patsahan, J. Phys.
Condens. Matter 18, 10223 (2006).
[10] A. Ciach, W.T. Gozdz, and G. Stell, Phys. Rev. E 75,
051505 (2007).
[11] E. González-Tovar, Mol. Phys. 97, 1203 (1999).
[12] F.O. Raineri, J.P. Routh, and G. Stell, J. Phys. IV
France 10, 99 (2000).
[13] M.E. Fisher, J. Stat. Phys. 75, 1 (1994).
[14] J.M.H. Levelt Sengers and J.A. Given, Mol. Phys. 80,
899 (1993).
Downloaded By: [Adler, S] At: 15:56 3 February 2011
[15] G. Stell, J. Stat. Phys. 78, 197 (1995).
[16] R.R. Singh and K. S., Pitzer, J. Chem. Phys. 92, 6775
(1990).
[17] S. Wiegand, J.M.H. Levelt Sengers, K.J. Zhang, et al.,
J. Chem. Phys. 106, 2777 (1997).
[18] M. Kleemeier, S. Wiegand, W. Schröer, et al., J. Chem.
Phys. 110, 3085 (1999).
[19] Y. Levin, Brazilian J. Phys. 34, 1158 (2004).
[20] H. Weingärtner and W. Schröer, Adv. Chem. Phys. 116,
1 (2001).
[21] H. Boroudjerdi, Y.W. Kim, A. Naji, et al., Phys. Rep.
416, 129 (2005).
[22] J.P. Hansen and H. Löwen, Ann. Rev. Phys. Chem. 51,
209 (2000).
[23] J.L. Barrat and J.F. Joanny, Adv. Chem. Phys. 94, 1
(1996).
[24] A.Z. Panagiotopoulos, J. Phys. Condens. Matter 17,
S3205 (2005).
[25] P.N. Vorontsov-Velyaminov, A.M. Elyashevich,
L.A. Morgenshtern, et al., High Temp. (USSR) 8, 261
(1970).
[26] P.N. Vorontsov-Velyaminov and V.P. Chasovskikh,
High Temp. (USSR) 13, 1071 (1975).
[27] G. Stell, K.C. Wu, and B. Larsen, Phys. Rev. Lett. 37,
1369 (1976).
[28] A.Z. Panagiotopoulos, Fluid Phase Equil. 76, 97 (1992);
erratum in Fluid Phase Equil. 92, 313 (1994).
[29] A.Z. Panagiotopoulos, Mol. Phys. 61, 813 (1987).
[30] G. Orkoulas and A.Z. Panagiotopoulos, J. Chem. Phys.
101, 1452 (1994).
[31] A.Z. Panagiotopoulos, J. Phys.: Condens. Matter 12,
R25 (2000).
[32] A.D. Bruce and N.B. Wilding, Phys. Rev. Lett. 68, 193
(1992).
[33] N.B. Wilding and A.D. Bruce, J. Phys.: Condens.
Matter 4, 3087 (1992).
2049
[34] Y.C. Kim and M.E. Fisher, J. Phys. Chem. B 108, 6750
(2004).
[35] A.M. Ferrenberg and R.H. Swendsen, Phys. Rev. Lett.
61, 2635 (1988).
[36] J.M. Caillol, D. Levesque, and J.J. Weis, Phys. Rev.
Lett. 77, 4039 (1996).
[37] J.M. Caillol, D. Levesque, and J.J. Weis, J. Chem. Phys.
107, 1565 (1997).
[38] J.M. Caillol, D. Levesque, and J.J. Weis, J. Chem. Phys.
119, 10794 (2002).
[39] G. Orkoulas and A.Z. Panagiotopoulos, J. Chem. Phys.
110, 1581 (1999).
[40] Q.L. Yan and J.J. de Pablo, J. Chem. Phys. 111, 9509
(1999).
[41] A.Z. Panagiotopoulos, J. Chem. Phys. 116, 3007 (2002).
[42] G. Orkoulas, M.E. Fisher, and A.Z. Panagiotopoulos,
Phys. Rev. E 63, 051507 (2001).
[43] E. Luijten, M.E. Fisher, and A.Z. Panagiotopoulos,
Phys. Rev. Lett. 88, 185701 (2002).
[44] Y.C. Kim and M.E. Fisher, Phys. Rev. Lett. 92, 185703
(2004).
[45] G. Stell, Phys., Rev. A 45, 7628 (1992).
[46] J. Valleau and G. Torrie, J. Chem. Phys. 108, 5169
(1998).
[47] J. Valleau and G. Torrie, J. Chem. Phys. 117, 3305
(2002).
[48] C.D. Daub, P.J. Camp, and G.N. Patey, J. Chem. Phys.
121, 8956 (2004).
[49] A.Z. Panagiotopoulos and S.K. Kumar, Phys. Rev.
Lett. 83, 2981 (1999).
[50] A.Z. Panagiotopoulos, J. Chem. Phys. 112, 7132 (2000).
[51] G. Stell, in New Approaches to Problems in Liquid State
Theory, edited by C. Caccamo, J.P. Hansen and G. Stell,
NATO ASI Series (Kluwer Academic Publishers,
Dordrecht, 1999).
[52] S. Moghaddam, Y.C. Kim, and M.E. Fisher, J. Phys.
Chem. B 109, 6824 (2005).
[53] A. Ciach and G. Stell, Phys. Rev. E 70, 016114 (2004).
[54] A. Ciach, Phys., Rev. E 70, 046103 (2004).
[55] M.N. Artyomov and A.B. Kolomeisky, Mol. Phys. 103,
2863 (2005).
[56] C.D. Daub, G.N. Patey, and P.J. Camp, J. Chem. Phys.
119, 7952 (2003).
[57] J.M. Romero-Enrique, L.F. Rull, and
A.Z. Panagiotopoulos, Phys. Rev. E 59, 041204 (2002).
[58] G. Ganzenmüller and P.J. Camp, J. Chem. Phys. 126,
191104 (2007).
[59] B. Smit, K. Esselink, and D. Frenkel, Mol. Phys. 87, 159
(1996).
[60] C. Vega, F. Bresme, and J.L.F. Abascal, Phys. Rev. E
54, 2746 (1996).
[61] F. Bresme, C. Vega, and J.L.F. Abascal, Phys. Rev.
Lett. 85, 3217 (2000).
[62] A.-P. Hynninen, M.E. Leunissen, A. van Blaarderen,
et al., Phys. Rev. Lett. 96, 018303 (2006).
[63] A. Fortini, A.-P. Hynninen, and M. Dijkstra, J. Chem.
Phys. 125, 094502 (2006).
[64] J.G. Kirkwood, J. Chem. Phys. 3, 300 (1935).
 2050 A.-P. Hynninen and A.Z. Panagiotopoulos

[65] D. Frenkel and A.J.C. Ladd, J. Chem. Phys. 81, 3188 [96] T.S. van Erp, D. Moroni, and P.G. Bolhuis, J. Chem.
(1984). Phys. 118, 7762 (2003).
[66] J.M. Romero-Enrique, G. Orkoulas, A.Z. Panagiotopoulos, [97] S.M. Loverde, Y.S. Velichko, and M.O. de la Cruz, J.
et al., Phys. Rev. Lett. 85, 4558 (2000). Chem. Phys. 124, 144702 (2006).
[67] Q.L. Yan and J.J. de Pablo, J. Chem. Phys. 114, 1727 [98] V. Kobelev, A.B. Kolomeisky, and
(2001). A.Z. Panagiotopoulos, Phys. Rev. E 68, 066110 (2003).
[68] Q.L. Yan and J.J. de Pablo, Phys. Rev. Lett. 86, 2054 [99] A. Diehl and A.Z. Panagiotopoulos, J. Chem. Phys.
(2001). 118, 4993 (2003).
[69] A.Z. Panagiotopoulos and M.E. Fisher, Phys. Rev. Lett. [100] A. Diehl and A.Z. Panagiotopoulos, Phys. Rev. E 71,
88, 045701 (2002). 046118 (2005).
[70] Q.L. Yan and J.J. de Pablo, Phys. Rev. Lett. 88, 095504 [101] A. Diehl and A.Z. Panagiotopoulos, J. Chem. Phys.
(2002). 124, 194509 (2006).
[71] J. Reščič and P. Linse, J. Chem. Phys. 114, 10131 [102] A.-P. Hynninen, M. Dijkstra, and
(2001). A.Z. Panagiotopoulos, J. Chem. Phys. 123, 084903
[72] P. Linse, J. Chem. Phys. 113, 4359 (2000). (2005).
[73] P. Linse, Trans. Phil. Soc. London A 359, 853 [103] A.-P. Hynninen and A.Z. Panagiotopoulos, Phys. Rev.
(2001). Lett. 98, 198301 (2007).
[74] D.W. Cheong and A.Z. Panagiotopoulos, J. Chem. [104] A.G. Moreira and R.R. Netz, Phys. Rev. Lett. 87,
Phys. 119, 8526 (2003). 078301 (2001).
[75] N. Ise, M. Sugimura, K. Ito, et al., J. Chem. Phys. 73, [105] J. Pelta, D. Durand, J. Doucet, et al., Biophys. J. 71, 48
536 (1983). (1996).
[76] N. Ise and M.V. Smalley, Phys. Rev. B 50, 16722 (1994). [106] M. Deserno, C. Holm, and S. May, Macromolecules
Downloaded By: [Adler, S] At: 15:56 3 February 2011

[77] B.V.R. Tata, M. Rajalakshmi, and A.K. Arora, Phys.
Rev. Lett. 69, 3778 (1992).
[78] T. Palberg and M. Würth, Phys. Rev. Lett. 72, 786
(1994).
[79] K. Ito, H. Yoshida, and N. Ise, Science 263, 66
(1994).
[80] B.V.R. Tata, E. Yamahara, P.V. Rajamani, et al., Phys.
Rev. Lett. 78, 2660 (1997).
[81] A.E. Larsen and D.G. Grier, Nature (London) 385, 230
(1997).
[82] R. Van Roij and J.P. Hansen, Phys. Rev. Lett. 79, 3082
(1997).
[83] P.B. Warren, J. Chem. Phys. 112, 4683 (2000).
[84] Y. Levin, E. Trizac, and L. Bocquet, J. Phys.: Condens.
Matter 15, S3523 (2003).
[85] B. Lu and A.R. Denton, Phys. Rev. E 75, 061403 (2007).
[86] B. Zoetekouw and R. van Roij, Phys. Rev. Lett. 97,
258302 (2006).
[87] F.G. Fumi and M.P. Tosi, Phys. Chem. Solids 25, 31
(1964).
[88] Y. Guissani and B. Guillot, J. Chem. Phys. 101, 490
(1994).
[89] Y. Guissani and B. Guillot, J. Chem. Phys. 116, 2058
(2002).
[90] J. Anwar, D. Frenkel, and M.G. Noro, J. Chem. Phys.
118, 728 (2003).
[91] T. Kristóf, D. Boda, J. Liszi, et al., Mol. Phys. 101, 1611
(2003).
[92] J.B. Caballero, A.M. Puertas, A. Fernandez-Barbero,
et al., J. Chem. Phys. 121, 2428 (2004).
[93] T. Kristóf, D. Boda, and D. Henderson, J. Chem. Phys.
120, 2846 (2004).
[94] C. Valeriani, E. Sanz, and D. Frenkel, J. Chem. Phys.
122, 194501 (2005).
[95] R.J. Allen, P.B. Warren, and P.R. ten Wolde, Phys.
Rev. Lett. 94, 018104 (2005).
33, 199 (2000).
[107] S. Liu and M. Muthukumar, J. Chem. Phys. 116, 9975
(2002).
[108] A. Jusufi, C.N. Likos, and H. Lowen, J. Chem. Phys.
116, 11011 (2002).
[109] A. Bizjak, et al., J. Chem. Phys. 125, 214907
(2006).
[110] R. Chang and A. Yethiraj, Macromolecules 38, 607
(2005).
[111] G. Orkoulas, S.K. Kumar, and A.Z. Panagiotopoulos,
Phys. Rev. Lett. 90, 048303 (2003).
[112] J.W. Jiang, L. Blum, O. Bernard, et al., Mol. Phys. 99,
1121 (2001).
[113] D.W. Cheong and A.Z. Panagiotopoulos, Mol. Phys.
103, 3031 (2005).
[114] J.W. Jiang, J. Feng, H. Liu, et al., J. Chem. Phys.
124, 144908 (2006).
[115] S. Schneider and P. Linse, J. Phys. Chem. B 107, 8030
(2003).
[116] S. Schneider and P. Linse, Macromolecules 37, 3850
(2004).
[117] S. Edgecombe, S. Schneider, and P. Linse,
Macromolecules 37, 10089 (2004).
[118] Q.L. Yan and J.J. de Pablo, Phys. Rev. Lett. 91,
018301 (2003).
[119] D.W. Yin, Q.L. Yan, and J.J. de Pablo, J. Chem. Phys.
123, 174909 (2005).
[120] J.F. Brennecke and E.J. Maginn, Amer. Inst. Chem.
Eng. J. 47, 2384 (2001).
[121] R.D. Rogers and K.R. Seddon, Science 302, 792
(2003).
[122] X. Han and D.W. Armstrong, Acc. Chem. Res. 40,
1079 (2007).
[123] R.M. Lynden-Bell, M.G. Del Pópolo, T.G.A. Youngs,
et al., Acct. Chem. Res. 40, 1138 (2007).
[124] E.J. Maginn, Acct. Chem. Res. 40, 1200 (2007).
 Molecular Physics 2051

[125] S. Jayaraman and E.J. Maginn, J. Chem. Phys. 127, [139] H. Docherty, et al., J. Phys. Chem. B 111, 8993
214504 (2007). (2007).
[126] D.M. Eike, J.F. Brennecke, and E.J. Maginn, J. Chem. [140] H. Docherty, et al., J. Chem. Phys. 125, 074510
Phys. 122, 014115 (2005). (2006).
[127] B. Widom, J. Chem. Phys. 39, 2808 (1963). [141] J. Vorholz, et al., Fluid Phase Equilibria 226, 237
[128] R.M. Lynden-Bell, N.A. Atamas, A. Vasilyuk, et al., (2004).
Molec. Phys. 100, 3225 (2002). [142] H.J. Strauch and P.T. Cummings, Fluid Phase
[129] J.K. Shah and E.J. Maginn, Fluid Phase Equilibria Equilibria 83, 213 (1993).
222, 195 (2004). [143] P.W. Atkins, Physical Chemistry, 4th ed. (W.H.
[130] J.K. Shah and E.J. Maginn, J. Phys. Chem. B 109, Freeman, New York, 1990), x10.2.
10395 (2005). [144] J.W. Tester and M. Modell, Thermodynamics and its
[131] I. Urukova, J. Vorholz, and G. Maurer, J. Phys. Chem. Applications, 3rd ed. (Prentice-Hall, Upper Saddle
B 109, 12154 (2005). River, NJ, 1996).
[132] D.E. Smith and L.X. Dang, J. Chem. Phys. 100, 3757 [145] B. Hess, C. Holm, and N. van der Vegt, Phys. Rev.
(1994). Lett. 96, 147801 (2006).
[133] H. Du, J.C. Rasaiah, and J.D. Miller, J. Phys. Chem. B [146] S. Gavryushov and P. Linse, J. Phys. Chem. B 110,
111, 209 (2007). 10878 (2006).
[134] J. Alejandre and J.P. Hansen, Phys. Rev. E 76, 061505 [147] P.J. Lenart, A. Jusufi, and A.Z. Panagiotopoulos,
(2007). J. Chem. Phys. 126, 044509 (2007).
[135] A.P. Lyubartsev and S. Marcelja, Phys. Rev. E 65, [148] P. Paricaud, et al., J. Chem. Phys. 122, 244511
041202 (2002). (2005).
[136] M. Ferrario, G. Ciccotti, E. Spohr, et al., J. Chem. [149] J. K.Johnson, A.Z. Panagiotopoulos, and
Downloaded By: [Adler, S] At: 15:56 3 February 2011

Phys. 117, 4947 (2002). K.E. Gubbins, Mol. Phys. 81, 717 (1994).
[137] E. Sanz and C. Vega, J. Chem. Phys. 126, 014507 [150] W.R. Smith and B. Triska, J. Chem. Phys. 100, 3019
(2007). (1994).
[138] M. Lisal, W.R. Smith, and J. Kolafa, J. Phys. Chem. B [151] J. Mongan, D.A. Case, and J.A. McCammon,
109, 12956 (2005). J. Comput. Chem. 25, 2038 (2004).