THE JOURNAL OF CHEMICAL PHYSICS 129, 134710 共2008兲

Phase diagrams for model heteronuclear dimers confined

in narrow slitlike pores
W. Rżyskoa兲
Department for the Modelling of Physico-Chemical Processes, Maria Curie-Skłodowska University,
20031 Lublin, Poland
共Received 27 May 2008; accepted 27 August 2008; published online 7 October 2008兲

Monte Carlo simulation is used to study the phase behavior of a model of heteronuclear dimers on
a lattice. The calculations are carried out for two- and three-dimensional bulk systems and for
dimers confined in slitlike pores of different widths and for different segment-wall interactions. The
calculations recover a rich variety of phase behaviors that includes transitions between gaslike and
liquidlike phases, the latter being ordered or disordered. The topology of the obtained diagrams
depends on the energy of interactions between segments forming a dimer, on the segment-wall
interaction, and on the pore width. © 2008 American Institute of Physics.
关DOI: 10.1063/1.2985607兴

I. INTRODUCTION An important question is how physical behavior of fluids

is affected by the microscopic structure of molecules. From a
In narrow pores, with pore widths of a few molecular
theoretical point of view adsorption of dimeric fluids is par-
diameters, fluids exhibit significantly different physical be-
ticularly interesting. A dimer is a molecule composed of two
haviors than that observed for bulk systems. The competition
between fluid–pore wall and fluid-fluid interactions leads, segments which can be identical 共homogeneous dimer兲 or
under certain thermodynamic conditions, to surface-driven unlike 共heterogeneous dimer兲. The model can mimic di-
phase transitions, such as layering, wetting, and capillary atomic molecules as, for example, O2, N2, CO, and NO, or
condensation. The phase transitions depend on the geometry more complex compounds such as alcohols, amines, or even
and size of the pore. The confinement leads to a shift in the amphiphiles. For many years systems involving dimers have
coexistence curve. The critical temperature of capillary con- been studied in the framework of both lattice and continuum
densation was found to decrease with decreasing pore width. models.23–37 The thermodynamics of adsorbed dimers has
The shift ⌬Tc is different for narrow and wide pores.1,2 been a subject of numerous works.23–37 Despite the number
Moreover, the competition between different transitions can of contributions to the understanding of dimer adsorption
result in a quite complex phase diagram. To understand these there are many aspects which are not yet solved. One such
phenomena, computer simulation techniques, as well as the- difficult problem is the phase behavior of dimeric fluids. The
oretical methods, including density functional theory and in- structural ordering of interacting dimers has been analyzed
tegral equations, are commonly used. The studies are carried by Phares et al.26 The authors considered the structural or-
out for both off-lattice and lattice models. In many cases dering of interacting dimers on a semi-infinite square lattice
simulations of lattice models are advantageous because they using the transfer matrix method. They found that there is a
not only allow for much faster calculations but also permit to finite number of ordered structures for repulsive nearest-
study much larger systems. neighbor interactions between dimers.
In general, a relation between a model of the system and Recently, we have studied phase transitions in mono-
its phase behavior is one of the fundamental and appealing layer films formed by heteronuclear dimers. The results have
problems in modern statistical thermodynamics. However, revealed a great variety of scenarios leading to the vapor-
even for bulk systems evaluation of complete phase dia- liquid coexistence. These scenarios depend on the relative
grams corresponding to a particular model is a demanding strengths of interactions between dimer segments.33 In par-
task. Of course, this task becomes much more complex and ticular, we found phase diagrams which include a tricritical
laborious when one considers confined systems. Usually, de- point and a critical end point.34,35 The aim of this work,
velopment of analytical theories for confined fluids is diffi- however, is to consider phase transitions for model hetero-
cult and introduction of several approximations is nuclear dimers confined in slitlike pores. Because the bulk
necessary.1–11 Therefore, an attractive alternative is to em- properties of the employed model were not investigated be-
ploy computer simulation techniques for that purpose. In fore, the calculations for the confined systems are preceded
fact, numerous simulations carried out for various confined by the relevant studies of bulk, two-dimensional 共2D兲, as
fluids12–21 discovered interesting changes in their structure well as three dimensional 共3D兲 fluid. The theoretical tool
during phase transitions. The confinement effects have been used for these purposes is the Monte Carlo 共MC兲 method and
studied both for simple and molecular fluids.22 the analysis of multiple histograms,38–40 together with finite-
size scaling technique.41,42
a兲
Electronic mail: wojtekrzysko@gmail.com The paper is organized as follows: In Secs. II and III, the

0021-9606/2008/129共13兲/134710/8/$23.00 129, 134710-1 © 2008 American Institute of Physics

134710-2 W. Rżysko J. Chem. Phys. 129, 134710 共2008兲

model is described and important features of simulation and ␮* = ␮ / 兩uBB兩. Moreover, all lengths are measured in units
methodology are outlined. Section IV gives the results. The of the lattice constant and the density is ␳ = 2N / V, where V is
paper is summarized in Sec. V. the volume 共equal to L3 and L2 for bulk 3D and 2D systems
and DL2 for confined systems兲.
II. MODEL

The phase behavior of a model heteronuclear diatomic III. MC PROCEDURES AND FINITE-SIZE SCALING
bulk fluid in three and two dimensions and the same fluid
confined in slitlike pores is studied. A simple lattice model is MC simulation in the grand canonical ensemble has been
employed. In the case of pores the simulation box is a frag- performed using hyperparallel tempering. Because all tech-
ment of a cubic lattice consisting of L2 ⫻ D sites, where L nical details are well known, see, e.g., Ref. 40, the descrip-
denotes a linear size of the system in X and Y directions and tion of the methodology is confined to its most significant
D is the pore width. A dimer molecule is modeled as two aspects.
interaction centers, A and B, at a fixed separation, i.e., the The simulations were carried out for both bulk 共3D and
bond between the segments cannot be broken. The bond 2D兲 and confined systems. In the case of bulk systems the
length is equal to the lattice constant. The energies of inter- cubic simulation cell was used, with the standard boundary
action between segments belonging to different dimers are conditions applied in three directions. Obviously, for bulk 2D
denoted as uAA, uAB, and uBB, respectively. systems the periodic boundary conditions in directions X and
The relations between the energy parameters determine Y were imposed. The linear size of the systems, L, ranged
the fundamental properties of the fluid. This work is devoted from 10– 24 for 3D systems and from 40– 80 for 2D systems.
to dimers for which interactions B-B and A-A are fixed and In the case of slitlike pores the boundary conditions ap-
equal to uAA = uBB = −1, while interaction A-B is weaker and plied only in the X and Y directions because the size of the
equals uAB / 兩uBB兩 = −0.25. system in the Z direction, D, was delimited by real planar
In the case of slitlike pores the walls are modeled as a walls. Three pore widths were studied, namely, D = 4, 5, and
square lattice strictly compatible with the fluid lattice. The 6. For porous systems the values of L ranged from 20– 64. In
surfaces forming the pore are homogeneous. The potential some cases L was as large as 100.
energy exerted by the walls on a given center of a dimer, A single MC step consisted of an attempt to insert a
located within layer i, is dimer molecule at a randomly chosen position or an attempt
to remove the existing particle. These attempts were selected
Vx,i = Vx共1/i3 + 1/共D − i + 1兲3兲, 共1兲
with equal probability. The equilibrium states were usually
where x denotes the kind of the segment 共A or B兲. Two types reached after 105 − 107 MC steps per lattice site. The aver-
of interactions between a molecule and the surface are con- ages were calculated over the next 2 ⫻ 107 MC steps. To
sidered. In the first case the energy of interaction between a avoid correlations between subsequently generated states, the
given surface site depends on the kind of the segment, sampling of averages was carried out at the intervals of 200
namely, VA = −2 and VB = 0. In the second case the interaction MC steps. As a standard, the average density ␳ and the av-
with the walls is independent of the segment label and VA erage energy u were calculated.
= VB = −2. To explore the phase behavior of the systems the
Let us introduce the occupation variables cxi which can multiple-histogram reweighting and finite-size techniques
take the value cxi = 1 when site i is occupied by atom x and were applied. The 2D probability distributions P共␳ , u兲 were
cxi = 0 in all other cases. Each lattice site is characterized by obtained for the systems of different sizes L and the phase
a pair 共cAi , cBi兲. Then, in the grand canonical ensemble the diagrams were estimated according to the “two-state ap-
Hamiltonian is given by proximation” from the density distribution P共␳兲,

H = uAA 兺 cAicA j + uBB 兺 cBicB j + uAB 兺 cAicB j P共␳兲 = 兺 P共␳,u兲. 共3兲

具i,j典 具i,j典 具i,j典
u
D D
+ 兺 VAi 兺 cAi + 兺 VBi 兺 cBi − NuAB − N␮ , 共2兲 In the case of first-order transitions the density distribution
i=1 i i=1 i P共␳兲 exhibits a double peaked structure. The location of co-
where N is the number of dimers and ␮ is the chemical existence points have been achieved by tuning the chemical
potential. The first three sums are over all pairs of nearest- potential at a given temperature until the areas under both
neighbor lattice sites. The last two sums are over all lattice peaks became identical.40
sites. Because we have taken into account the interaction To study the gas-liquid coexistence the usual order pa-
energies between all possible pairs of segments, we subtract rameter is used,
the interactions between atoms belonging to the same mol-
m = ␳ − 具␳典. 共4兲
ecule, NuAB. Obviously, in the case of bulk 3D and 2D sys-
tems the contribution to the Hamiltonian due to the presence Apart from condensation transitions one can expect
of walls is zero. order-disorder transitions in the studied systems. The knowl-
The dimensionless units, i.e., the reduced temperature T* edge of the possible ordering in dense phases can help elu-
and the chemical potential ␮*, are defined as T* = kT / 兩uBB兩 cidate the nature of phase transitions. To detect a given or-
134710-3 Phase diagrams for dimers in pores J. Chem. Phys. 129, 134710 共2008兲

␾2D,SAF,Lz = 冑␾1,x
2
+ ␾2,x
2
+ ␾21,y + ␾22,y , 共8兲
and for all lattice layers parallel to the pore walls we calcu-
late
D
1
␾2D,SAF = 兺 ␾2D,SAF,Lz .
D Lz=1
共9兲

Finally, we introduce the order parameter for the SAF struc-

ture in the 3D system,
␾SAF = 冑␾1,x
2
+ ␾2,x
2
+ ␾21,y + ␾22,y + ␾1,z
2
+ ␾2,z
2
. 共10兲
FIG. 1. The decomposition of a cubic lattice into four equivalent sublattices
a , b , c , d used to calculate the order parameters. In perfectly ordered phases the order parameters ␾2D,SAF,Lz,
␾2D,SAF, and ␾SAF achieve unity. For disordered phases they
are close to zero.
dered structure one should define a suitable order parameter.
The above-discussed order parameters are comple-
The order parameters for different ordered phases are dis-
mented by the corresponding susceptibilities,
cussed elsewhere.43
It follows from an analysis of the ground state of the 1
␹␸,L = 关具␸2典 − 具兩␸兩典2兴, 共11兲
bulk 3D system with considered parameters 共uAA = uBB = −1, kT
uAB = −0.25兲 that in this case the quasilamellar structure is
formed by dimers lying parallel to the X or Y or Z axis. The and the fourth cumulants,
AABB structure is replicated to make alternating strips of A U␸,L = 1 − 具␸L4 典L/3具␸L2 典2 , 共12兲
and B segments 关the superantiferromagnetic 共SAF兲 phase
共4 ⫻ 1 ⫻ 1兲兴. In lattice layers parallel to the chosen axis the where ␾ is any of the previously defined order parameters
2D SAF 共4 ⫻ 1兲 structures are formed. However, lattice lay- and ␾L denotes the average for the system of the size L. It is
ers perpendicular to this axis consist of the same segments, known that the cumulants U␾,L have a nontrivial universal
and the following sequence is observed: two successive lay- value at a critical temperature 共U␾* 兲.40–42 Thus, the plots of
ers are covered by segments A and the two next layers con- U␾,L共T*兲 for different L intersect, and the crossing point cor-
tain only segments B. It is interesting that there is the order- responds to the critical temperature of an infinite system.
ing of segments as well as the ordering of bonds. This method 共with ␾ = m兲 was used to estimate critical tem-
To describe the above structure one can split the lattice peratures of layering transition and critical temperatures of
into four interpenetrating and fully equivalent sublattices, as capillary condensation in the studied systems.
shown in the Fig. 1. The abcd strip that consisted of sites
belonging to these sublattices is replicated in three dimen- IV. RESULTS AND DISCUSSION
sions. Then, we can consider four sublattice densities. How-
ever, the use of spin variables is more convenient and sim- A. Bulk systems
plifies the notation, so we define the average magnetization In our previous work33 we have reported the results of
of each sublattice as simulations for 2D heteronuclear dimers, whose pair interac-
tions between like segments were identical, uAA = uBB = −1,
1
mi = 兺 Si ,
L2 i苸k
共5兲 while the interaction between unlike segments was weakened
and equal to uAB = −0.375, −0.35, and 0. It was proved that
the topology of the phase diagram obtained for uAB = 0 was
and where k = a, b, c, d. The spin variable Si = cAi − cBi and it
different than in the two remaining cases and that this change
is equal to 1 共−1兲 when the ith site is occupied by segment A
was a direct consequence of the uAB interaction energy. For
共B兲 and 0 when the site is empty. The above sum is extended
uAB = −0.375 the phase diagram showed two first-order phase
over all sites belonging to a given sublattice k. For example,
transitions that met at the triple point. At some range of
if all sites of sublattice a are occupied by segments A 共B兲, the
temperatures we observed the first transition from a vapor-
sublattice magnetization is equal to ma = 41 共− 41 兲.
like phase to a disordered liquid, while the second transition
Using the above sublattice magnetizations we define the
was connected with ordering of the fluid. For uAB = −0.35 the
following order parameters:
triple point temperature moved toward the critical tempera-
␾1,␣ = ma + mb − mc − md , 共6兲 ture of the gas-disordered fluid transition, but the topology of
the diagram remained the same. However, for uAB = 0 the
␾2,␣ = ma − mb − mc + md , 共7兲 triple point vanished and the coexistence envelope in the
temperature-density plane assumed a highly nonsymmetric
where ␣ = x , y , z. swan-neck shape.
These order parameters represents two components of Similar investigations carried out for 3D systems36 also
the order parameter suitable to detect the presence of the indicated that the topology of phase diagrams for hetero-
SAF structures. For the lattice layer Lz we can define the nuclear dimers strongly depends on the interaction param-
order parameter ␾2D,SAF,Lz as eters uAA, uBB, and uAB. It was found that depending on the
134710-4 W. Rżysko J. Chem. Phys. 129, 134710 共2008兲

FIG. 3. Phase diagrams in the density-temperature plane for 共a兲 2D and 共b兲
3D systems. Black circle in 共b兲 marks the critical point. The lattice sizes
were L = 60 for 2D and L = 12 for 3D systems.

branch that meets the first one at the triple point 共Tt*
⬇ 0.79兲 and ends at the critical point 共Tc* ⬇ 0.84兲 describes
the transition between disordered and SAF-type ordered liq-
uids.

FIG. 2. Density histograms for the 2D system with the system size of L B. Confined systems
= 80 for ␮* = −1.8452 at T* = 0.545 共dashed-dotted line兲, ␮* = −1.8362 at
T* = 0.5475 共dashed line兲, ␮* = −1.82 at T* = 0.55 共dotted line兲, and The next step is the presentation of the diagrams for
␮* = −1.775 at T* = 0.555 共solid line兲. confined fluids. The study was concentrated mainly on evalu-
ation of the diagrams for capillary condensation, connected
with the filling of the most inner part of the pore. However,
values of these parameters two ordered structures may be in the majority of the cases layering transitions within the
formed, i.e., that resembling the antiferromagnetic phase and layers adjacent to the pore walls were also investigated.
that corresponding to the SAF phase. Depending on the We begin with the model systems for which the energy
model, the estimated phase diagrams involved the vapor- of interaction of segments A and B with the pore walls is
liquid coexistence, triple points, tricritical points, and the different 共VA = −2 and VB = 0兲. One can expect that the above
critical end points.36 Therefore, in this work the studies on energy difference should prefer orientation of the first-layer
confined fluids were preceded by the relevant simulations of dimers perpendicular to the pore walls with segments A
2D and 3D bulk systems with periodic boundary conditions pinned at the surfaces.
in all dimensions. The model parameters of this work were First, a rather narrow pore, D = 4, is considered. The
selected as to fill the gap in the space of the energy param- phase diagram in the density-temperature plane is shown in
eters investigated so far. Fig. 4共a兲. To evaluate precisely the critical point temperature
Figure 2 shows examples of the histograms of the den- a finite-size analysis of the fourth-order cumulant of Binder,
sity P共␳兲 collected for the 2D systems. The histograms Um,L 关Eq. 共12兲兴, was carried out. Figure 4共b兲 displays the
clearly illustrate the coexistence between two phases. At the Um,L curves, obtained for the system sizes L = 16, 24, 48, and
lowest displayed temperature the densities of the rarefied 64. The critical temperature resulting from that analysis is
phase is quite high; the maximum of the first peak is at ␳ * ⬇ 1.102. The estimated fixed point value U* = 0.602 is
Tc,c m
⬇ 0.29, and the maximum of the second peak is at ␳ ⬇ 0.98. very close to that characteristic for the 2D Ising model. The
When the temperature is raised, the first maximum of the crossing point moves with the system size and the results
histograms shifts toward higher densities, while the position presented here indicate that a precise evaluation of the criti-
of the second peak is always close to 1, suggesting a swan-
neck topology of the phase diagram. Indeed, the coexistence
envelope in the density-temperature plane, presented in Fig.
3共a兲, confirms the above prediction. The existence of the
concavity on the high-density branch of the phase diagram at
T 苸 关0.544, 0.548兴 was confirmed by additional simulations
for much larger systems of L = 80 and 100 within that tem-
perature range. The structure of the dense phase is such as
displayed in Fig. 2 of Ref. 33.
When a 3D system is considered, the topology of the
phase diagram changes, see Fig. 3共b兲. Here the coexistence
envelope consists of two branches. The first one ends at the FIG. 4. 共a兲 Phase diagram in the density-temperature plane; 共b兲 values of
Um,L at different temperatures for different system sizes: L = 64 共solid line兲,
critical point at Tc* = 0.9985 and ␳c = 0.458 and characterizes 48 共dotted line兲, 24 共dashed line兲, and 16 共dashed-dotted line兲. All calcula-
the transition between gas and disordered liquid. The second tions are for D = 4, VA = −2, and VB = 0.
134710-5 Phase diagrams for dimers in pores J. Chem. Phys. 129, 134710 共2008兲

FIG. 5. 共a兲 Adsorption isotherm 共solid line兲 and the ␾SAF order parameter
共dotted line兲 vs the chemical potential at the temperature T* = 1.3. 共b兲 Sus-
ceptibility of ␾SAF order parameter vs the chemical potential at T* = 1.3 for
different system sizes. There are four curves, obtained for L = 20, 40, 60, and
80, that flow together. All calculations are for D = 4, VA = −2, and VB = 0.

cal temperature requires the simulation of very large sys-

tems. The conclusion that capillary condensation belongs to
the universality class of the 2D Ising model is consistent
with results of other studies.44 FIG. 6. 共a兲 Layering phase diagram in the temperature-density plane. Part
To determine the structure of the dense, ordered phase 共b兲 Unweighted histograms, evaluated for ␮* = −4.285 at T* = 0.54 共solid
the ␾SAF order parameters were investigated. It was found line兲, ␮* = −4.255 at T* = 0.55 共dotted line兲, ␮* = −4.240 15 at T* = 0.56
共dashed line兲, and ␮* = −4.155 at T* = 0.56 共dashed-dotted line兲. The inset
that the dense coexisting phase is highly ordered with two shows the capillary condensation phase diagram in the density-temperature
layers of dimers pinned at their A ends at the surfaces. Such plane. The calculations are for L = 40. 共c兲 The ␾2D,SAF,3 order parameter at
an ordering is enforced by the pore walls and additionally T* = 0.5 for L = 40. All calculations are for D = 5, VA = −2, and VB = 0.
stabilized by interactions between segments. The formation
of two layers on the walls means simultaneous filling of the
whole pore. There is no empty room in the pore middle. As a have two phase transitions in the system: the layering tran-
consequence, the ordering is almost perfect. There are 共al- sition 关Fig. 6共a兲兴 and the capillary condensation 关Fig. 6共b兲兴.
most兲 no A-B contacts between segments belonging to dif- The layering transition leads to a vertically oriented layer
ferent dimers and the internal energy of the system is very with A segments being attached to the pore walls. This struc-
low. It is worth noting that the critical temperature for the ture is almost the same as for the pore with D = 4. However,
confined system is now significantly higher than the critical the existence of one empty layer in the pore center partially
temperature of the bulk 3D system. This observation is rather destroys the ordering and causes an increase in the internal
unusual but it becomes clear when we compare the nature of energy. Nevertheless, there is the transition between a disor-
both phase transitions. In the bulk 3D system condensation dered gas and the ordered adsorbed layer. The topology of
corresponds to the transition between a disordered gas and a the phase diagram of the layering is analogous to that ob-
disordered liquid. However, in the case of this special pore served for the capillary condensation in the pore of the width
we have the transition between a disordered gas and the per- D = 4. The critical temperature of the layering transition is
* ⬇ 0.96. It is only slightly lower than the critical tem-
now Tc,l
fectly ordered liquid.
A very interesting phenomenon is that the ordering exists perature of capillary condensation for the pore with D = 4.
at the temperatures well above the critical temperature. This Moreover, due to perpendicular orientation of the first-layer
point is illustrated in Fig. 5, which displays the changes in dimers, the scenario of the layering transition for the system
the order parameter ␾SAF 共dotted line兲 along the adsorption under study is, obviously, quite different than that for the
isotherm 共solid line兲 at T* = 1.3 关Fig. 5共a兲兴 and its fluctuations condensation in the 2D 共bulk兲 system 关see Fig. 3共a兲兴.
关susceptibilities, Fig. 5共b兲兴. In order to check the role of the The filling of the inner layer is a more complex phenom-
finite-size effects, the relevant calculations for several values enon, which is illustrated in Fig. 6共b兲. Here, histograms ob-
of L were carried out. The obtained curves 关Fig. 5共b兲兴 almost tained from grand canonical simulations at low temperatures
flow together, indicating that the influence of the finite-size are shown. At T* = 0.54 and ␮* = −4.285 the histogram con-
effects on the observed behavior is negligible. The shapes of sists of two well-separated peaks with the first maximum
these curves 共Fig. 5兲 prove that above the critical temperature around ␳ ⬇ 0.83 and the second at ␳ ⬇ 0.995. However, for
of condensation there is no second-order phase transition be- T* = 0.55 and ␮* = −4.255 the histogram exhibits three peaks,
tween disordered and ordered fluids in the system. This which indicates that this temperature is close 共slightly above
means that neither tricritical nor critical end points exist for this system size兲 to the triple point temperature. At T*
here. = 0.56 the situation is otherwise. For both lower and higher
We now report the results obtained for a slightly wider values of the chemical potential, ␮* = −4.24 and
pore, namely, for D = 5, with all the parameters characteriz- ␮* = −4.155, two peaks appear, but their locations are differ-
ing the interactions in the systems unchanged. In this case we ent. At the former value of ␮* the maxima are at ␳ ⬇ 0.84
134710-6 W. Rżysko J. Chem. Phys. 129, 134710 共2008兲

and 0.93; the second is smeared out. At the latter value the
maxima occur at 0.98 and 0.99. This result undoubtedly
shows that two phase transitions appear in the system. In-
deed, the approximate capillary condensation part of the
phase diagram 关see the inset to Fig. 6共b兲兴 divides into two
branches. The first branch 共i.e., that at lower densities兲 is the
transition between one-layer gaslike phase and a disordered
one-layer liquid, while the second depicts the transition from
disordered one-layer liquid 共located at the pore center兲 to
one-layer ordered phase of the SAF type, oriented in the X or
Y direction. The two branches meet at the triple point. How-
ever, the triple point temperature is close to the critical tem-
perature of the order-disorder phase transition, outlined
above.
In light of the phase diagram for a strictly 2D system
关Fig. 3共a兲兴, the shape of the capillary phase diagram is unex-
pected. At low temperatures almost perfectly oriented layers
of dimers occupying two layers at each wall cause that the
middle layer could be considered as a 2D system. This is not
the case. Therefore, the only explanation is perhaps that
small nonidealities of the first-layer dimer structure 共the ex-
istence of “holes” and “elevated” dimers was incidentally
observed兲 lead to the change in the phase diagram topology.
Figure 6共c兲 shows ␾2D,SAF,3 the order parameter, but
only for the molecules from the middle layer, calculated
along the adsorption isotherm at T* = 0.5. The adsorption iso-
therm shows two jumps, corresponding to the layering tran-
sition 共the first jump兲 and to the capillary condensation 共the
second jump兲. However, the order parameter curve exhibits
only one sharp step, clearly indicating the point at which the FIG. 7. 共a兲 Layering phase diagram in the temperature-density plane. 共b兲
Unweighted histograms, evaluated to characterize the capillary condensation
ordered middle layer appears. The critical temperature of
and for the system with L = 40. The consecutive curves are for ␮* = −3.554 at
capillary condensation is considerably lower than the critical T* = 0.705 共solid line兲, ␮* = −3.5852 at T* = 0.705 共dotted line兲, ␮* =
temperature of the layering transition 关compare Figs. 6共a兲 −3.5563 at T* = 0.725 共dashed line兲, and ␮* = −3.5370 at T* = 0.74 共dashed-
and 6共b兲兴. dotted line兲. All the calculations are for D = 6, VA = −2, and VB = 0.
Similar calculations as for the pore of D = 5 were also
carried out for D = 6, see Fig. 7. In the latter case the phase
diagrams for the layering and capillary condensation transi- dimers, exhibiting the structure shown in Fig. 2 of Ref. 33.
tions are very similar to those estimated for D = 5. However, The direction of dimer orientation at both walls may be iden-
the critical temperature of the capillary condensation is much tical 关i.e., they are oriented in the X共Y兲 direction兴 or different
higher than for D = 5 关see Fig. 6共a兲兴. The capillary condensa- 共i.e., the dimers at opposing walls are oriented along differ-
tion obeys now two inner layers and the dimers located ent axes兲. Obviously, the free energy at T = 0 is identical for
within these layers can assume orientation either perpendicu- both cases.
lar or parallel to the walls. Such situations were observed Let us concentrate on the description of the capillary
during the simulations 共in fact, relevant order parameters condensation 共i.e., the transition connected with an instanta-
were also calculated兲, but these results are not shown for the neous filling of three inner layers兲. Figure 8共a兲 shows the
sake of brevity. coexistence envelope whereas in Fig. 8共b兲 some selected his-
Finally, let us report the results for the case when the tograms P共␳兲 used to its construction are presented. The
energy of interaction of a segment with the wall is indepen- phase diagram consists of two branches. The first one char-
dent of its label 共VA = VB = −2兲. We begin our discussion with acterizes the transition between the rarefied 共within three in-
the pore of D = 5. In contrast to the cases discussed so far, the ner layers兲 phase and the disordered fluid; the second transi-
layering transition leads to a dense fluid with dimers oriented tion leads to the ordered fluid phase. In this case the critical
parallel to the walls. Thus the capillary condensation obeys temperature of the layering transition is considerably lower
three inner layers. Because of the parallel orientation of the than the critical temperature of capillary condensation. Let us
adsorbed dimers, the distance between the walls is large analyze the phase behavior of the system at temperatures
enough to assure statistical independence of the layering higher than the critical temperature of the layering transition
transitions and each wall. Consequently, the phase diagram but lower than the critical temperature of the transition be-
for the layering transition is close to that observed for a tween the disordered and ordered liquids 关the second branch
strictly 2D system. It should be recalled that a 2D dense of the phase diagram of capillary condensation in Fig. 8共a兲兴.
fluid, coexisting with rarefied phase, is composed of ordered In such conditions during the capillary condensation the or-
134710-7 Phase diagrams for dimers in pores J. Chem. Phys. 129, 134710 共2008兲

FIG. 9. Histograms, evaluated to characterize the capillary condensation and

for the system with L = 40 and D = 4. The consecutive curves are for
␮* = −4.335 at T* = 0.780 共dashed line兲, ␮* = −4.367 at T* = 0.775 共dotted
line兲, and ␮* = −4.395 at T* = 0.770 共solid line兲.

smeared out. This suggests that the gas-disordered liquid

FIG. 8. 共a兲 The capillary condensation phase diagram in the temperature-
density plane for the system with L = 20. 共b兲 Unweighted histograms, evalu-
transition does not exist in the system. All these data point to
ated to characterize the capillary condensation and for the system with L a swan-neck topology of the phase diagram.
= 20. The consecutive curves are for ␮* = −3.91 at T* = 0.755 共solid line兲,
␮* = −3.93 at T* = 0.755 共dotted line兲, ␮* = −3.80 at T* = 0.760 共dashed line兲,
V. CONCLUSIONS
and ␮* = −3.91 at T* = 0.760 共dashed-dotted line兲. 共c兲 The adsorption iso-
therm and the SAF order parameter for the system of L = 20 at T* = 0.7. All Summing up the obtained results one should stress a rich
the calculations in 共a兲–共c兲 are for D = 5 and VA = VB = −2. 共d兲 The P共␳兲 his-
tograms for D = 6 and L = 20 and all remaining model parameters are the variety of phase diagram topologies observed for a very
same as in 共a兲–共c兲. Solid line is for ␮* = −3.35 at T* = 0.79, dotted line is for simple system of heteronuclear dimers in slitlike pores. Mod-
␮* = −3.61 at T* = 0.79, and dashed line is for ␮* = −3.595 at T* = 0.80. erate changes in system parameters may result in remarkable
changes in the phase behavior.
Regarding the case when the segments interact differ-
dering appears instantaneously in the whole pore. This is ently with the walls the following were found:
illustrated in Fig. 8共c兲 where the adsorption isotherm and the
␾2D,SAF order parameter at T* = 0.7 are presented. The first 共1兲 For the narrowest, D = 4, pore the critical temperature
共continuous兲 step on the isotherm corresponds to the adsorp- of the capillary condensation is much higher than the
critical temperature of the bulk, 3D system. Also, the
tion within the layers adjacent to the walls and the second
topologies of the phase diagrams are different in these
共discontinuous兲 to the capillary condensation. Within the first
two cases.
step the order parameter is very close to zero and its jump to
共2兲 For D = 5 the estimated critical temperature of the lay-
共almost兲 unity coincides with the adsorption isotherm jump.
ering transition is much higher than the critical tem-
The orientation occurs layer by layer with all the dimers perature of a 2D bulk system. The topology of the
parallel to the walls. phase diagrams for layering transition and for conden-
When one proceeds to the pore D = 6, the scenario de- sation in the 2D fluid is different. Capillary condensa-
scribed above remains unchanged. Therefore, only the histo- tion that obeys only one layer 共located at the pore cen-
grams P共␳兲 关Fig. 8共d兲兴 are shown here. However, when the ter兲 also exhibits different topology than bulk 2D
pore becomes narrower, D = 4, the topology of the capillary system, but the capillary condensation critical tempera-
condensation phase diagram changes completely. We have ture is close to that for a 2D system.
encountered extremely large difficulties during the simula- 共3兲 The phase behavior of the fluid confined in the pore of
tions aiming at the evaluation of the phase diagram. How- D = 6 is similar to that observed for D = 5. The only
ever, the obtained results undoubtedly point to a swan-neck remarkable difference is that the critical temperature is
topology of the capillary condensation phase diagram. None higher than for D = 5 because the condensation obeys
of the evaluated histograms 共see Fig. 9兲 exhibits the exis- now two layers.
tence of more than two peaks. Simultaneously, with the tem-
perature increase the first maximum of the histograms moves When the interactions of both segments with the walls
are identical, the following were observed:
toward higher densities, while the position of the second
peak is always close to 1. Moreover, the first peaks are 共1兲 For D = 4 the layering transition is similar to that ob-
134710-8 W. Rżysko J. Chem. Phys. 129, 134710 共2008兲

18
served for a strictly 2D system. Capillary condensation E. A. Müller, L. F. Rull, L. F. Vega, and K. E. Gubbins, J. Phys. Chem.
100, 1189 共1996兲.
envelope has the swan-neck shape. The estimated criti- 19
* ⬇ 0.80. S. Sokołowski and J. Fisher, Mol. Phys. 71, 393 共1990兲.
cal temperature is Tc,c 20
Z. Tan and K. E. Gubbins, J. Phys. Chem. 96, 845 共1992兲.
共2兲 The phase behavior changes when the pore width in- 21
V. Babin, A. Ciach, and M. Tasinkevych, J. Chem. Phys. 114, 9585
creases. For D = 5 共or 6兲 the capillary condensation part 22
共2001兲.
of the phase diagram exhibits two phase transitions. L. D. Geln, K. E. Gubbins, R. Radhakrishnan, and M. Śliwińska-
Bartkowiak, Rep. Prog. Phys. 62, 1573 共1999兲.
The topology of the phase diagram of capillary conden- 23
R. H. Fowler and G. S. Rushbrooke, Trans. Faraday Soc. 33, 1272
sation is similar to that estimated for the 3D bulk sys- 共1937兲.
24
tem. P. J. Flory, J. Chem. Phys. 10, 51 共1942兲.
25
P. W. Kasteleyn, Physica 共Utrecht兲 27, 1209 共1961兲.
26
A. J. Phares, F. J. Wunderlich, J. D. Curley, and D. W. Grumbine, J. Phys.
ACKNOWLEDGMENTS A 26, 6847 共1993兲.
27
K. W. Wojciechowski, D. Frenkel, and A. C. Brańka, Phys. Rev. Lett. 66,
This research was supported from the Alexander von 3168 共1991兲.
28
Humboldt Foundation. K. W. Wojciechowski, Phys. Rev. B 46, 26 共1992兲.
29
A. J. Ramirez-Pastor, J. L. Riccardo, and V. D. Pereyra, Surf. Sci. 411,
1 294 共1998兲.
H. Nakanishi and M. E. Fisher, J. Chem. Phys. 78, 3279 共1983兲. 30
J. E. Gonzalez, A. J. Ramirez-Pastor, and V. D. Pereyra, Langmuir 17,
2
R. Evans, U. M. B. Marconi, and P. Tarazona, J. Chem. Phys. 84, 2376
6974 共2001兲.
共1986兲. 31
3 F. Romá, A. J. Ramirez-Pastor, and J. L. Riccardo, J. Chem. Phys. 114,
M. E. Fisher and H. Nakanishi, J. Chem. Phys. 75, 5857 共1981兲.
4 10932 共2001兲.
R. Evans and P. Tarazona, Phys. Rev. Lett. 52, 557 共1984兲. 32
5 M. Borówko and W. Rżysko, J. Colloid Interface Sci. 244, 1 共2001兲.
R. Evans and U. M. B. Marconi, Chem. Phys. Lett. 114, 415 共1985兲. 33
6 W. Rżysko and M. Borówko, J. Chem. Phys. 117, 4526 共2002兲.
R. Evans, U. M. B. Marconi, and P. Tarazona, J. Chem. Phys. 84, 2376 34
共1986兲. W. Rżysko and M. Borówko, Surf. Sci. 520, 151 共2002兲.
35
7
P. Tarazona, U. M. B. Marconi, and R. Evans, Mol. Phys. 60, 573 共1987兲. W. Rżysko and M. Borówko, Surf. Sci. 600, 890 共2006兲.
36
8
A. O. Parry and R. Evans, Phys. Rev. Lett. 64, 439 共1990兲. W. Rżysko and K. Binder, J. Phys.: Condens. Matter 20, 415101 共2008兲.
37
9
T. L. Hill, J. Chem. Phys. 15, 767 共1947兲. D.-W. Wu, G. L. Aranovich, and M. D. Donohue, J. Colloid Interface Sci.
10
D. Nicholson, J. Chem. Soc., Faraday Trans. 1 71, 238 共1975兲. 212, 301 共1999兲; 230, 281 共2000兲.
38
11
F. Brochard-Wyart and P. G. de Gennes, C. R. Seances Acad. Sci., Ser. 2 A. M. Ferrenberg and R. H. Swendsen, Phys. Rev. Lett. 61, 2635 共1988兲.
39
297, 223 共1983兲. A. M. Ferrenberg and R. H. Swendsen, Phys. Rev. Lett. 63, 1195 共1989兲.
12 40
T. Wolfram, R. E. DeWames, W. F. Hall, and P. W. Palmberg, Surf. Sci. J. J. de Pablo, Q. L. Yan, and F. A. Escobedo, Annu. Rev. Phys. Chem.
28, 45 共1971兲. 50, 377 共1999兲.
41
13
D. Nicolaides and R. Evans, Phys. Rev. B 39, 9336 共1989兲. K. Kaski, K. Binder, and J. D. Gunton, Phys. Rev. B 29, 3996 共1984兲.
42
14
E. Kierlik, Y. Fan, P. A. Monson, and M. L. Rosinberg, J. Chem. Phys. K. Binder, Phys. Rev. Lett. 47, 693 共1981兲.
43
102, 3712 共1995兲. A. Patrykiejew, S. Sokołowski, and K. Binder, Surf. Sci. Rep. 37, 207
15
J. P. Donley and A. J. Liu, Phys. Rev. E 55, 539 共1997兲. 共2000兲.
16 44
K. S. Page and P. A. Monson, Phys. Rev. E 54, 6557 共1996兲. O. Dillmann, W. Janke, M. Müller, and K. Binder, J. Chem. Phys. 114,
17
Z. Zhang and A. Chakrabarti, Phys. Rev. E 52, 2736 共1995兲. 5853 共2001兲.