7170 J. Phys. Chem.

C 2009, 113, 7170–7180

Vapor-Liquid Phase Coexistence, Critical Properties, and Surface Tension of Confined

Alkanes

Sudhir K. Singh, Ankit Sinha, Goutam Deo, and Jayant K. Singh*

Department of Chemical Engineering, Indian Institute of Technology Kanpur, Kanpur 208016, India
ReceiVed: August 18, 2008; ReVised Manuscript ReceiVed: March 17, 2009

Configurational-bias grand-canonical transition-matrix Monte Carlo simulations are conducted to investigate

various thermophysical properties, such as phase coexistence, critical properties, density and orientation profiles
of liquid and vapor phases, and vapor-liquid surface tension of methane, ethane, propane, n-butane, and
n-octane in bulk and slit pores of graphite and mica surfaces. An exponential-6 (exp-6) model is used for the
investigation of normal alkanes with a cutoff radius, 15 Å. It is found that the surface tension of the bulk
n-alkane, C1-C4, based on a truncated exp-6 model, agrees reasonably well with the experimental data. Critical
See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

properties are reported by means of the rectilinear diameter approach and least-squares technique. The shift
in the critical temperature under confinements follows more than two linear regimes with an inverse in the
Downloaded via HERIOT-WATT UNIV on February 27, 2024 at 13:58:58 (UTC).

slit width, as the slit width approaches the two-dimensional limit. This is contrary to what has been previously
reported in the literature. The behavior of the critical temperature shift is sensitive to the nature of the surface.
The critical density, on the other hand, fluctuates with a decrease in the slit width. The shift in the critical
vapor pressure continuously increases with a decrease in the slit width toward the two-dimensional value and
becomes constant for pore sizes less than 5 Å for the fluid studied in this work. Corresponding state plots
suggest that the deviation of the saturation vapor pressure from the bulk saturation pressure under confinement
is positive for large pores and negative for smaller pores. Vapor-liquid surface tension values for n-alkanes,
in both types of slit pores, are computed via the finite-size scaling method of Binder and compared with their
bulk values. Our investigation reveals that under slit-pore confinement the vapor-liquid surface tension
decreases many fold.

1. Introduction catalyst supports for chemical reactions.2-13 The design of such

processes is largely empirical with an underdeveloped scientific
Alkanes are commonly used in various industries. They are basis because the theory-based understanding of confined fluids
present in lubricants, fuels, polymers, surfactants, and liquid is still not completely understood.14-18 However, with a rapid
crystals as well as many biological molecules such as lipids. increase of computer power and advanced algorithms, molecular
Understanding the behavior of alkanes under nanoconfinements simulation can help to understand the underlying physics of
is of great interest from fundamental and practical perspectives.
confinement effects on various thermophysical and interfacial
In general, fluids under any type of confinement within pores
properties, which could lead to significant improvements in these
of nanometer size exhibit significant deviation from bulk
industrial processes.
thermophysical and structural properties.1 Such systems are in
between bulk matter and single atoms or molecules, and as a Molecular simulations have been used to study the adsorption
consequence, finite-size and surface effects become imperative properties and isotherms of single-component and binary
for the investigation of such systems. Any real system interfaces mixtures of normal and branched alkanes in various porous
either with walls of a container or with different phases of the materials such as zeolites.19-23 Capillary phase transitions and
substance under consideration. An extra free energy contribution adsorption and separation properties of mixtures of n-alkanes
arises because of these interfaces, which plays an important role in a carbon nanotube were studied using gauge-cell Monte Carlo
in nanoconfined systems and affects the state of the substance. simulation.24,25 It was observed by the investigators that confine-
Interplay of fluid-wall and fluid-fluid forces can lead to ment of n-alkanes in a carbon nanotube decreases its critical
fascinating surface-driven phase changes and various other temperature and increases its critical density. Recently, molec-
thermophysical and interfacial phenomena. The fundamental ular simulation has been applied as a screening tool to identify
interest in these phase changes and other related properties is the new materials for adsorption applications. Such an approach
to understand the new physics that results from finite-size effects, was used for metal-organic frameworks26,27 to study the
varying dimensionality, and surface forces. In addition, a adsorption characteristics of novel metal-organic materials for
fundamental understanding of the various thermophysical and methane and hydrogen storage applications. Zhang28 used a
interfacial phenomena, in the presence of surfaces, is necessary molecular dynamics simulation technique to study the density
for many industrial and geophysical operations. For example, profiles, structural properties, and average diffusion coefficients
micro- and mesoporous materials are widely used in chemical, of n-decane in a carbon nanotube. The author observed well-
oil and gas, food, and pharmaceutical industries for pollution defined peaks in the density profiles, which indicated a strong
control, mixture separation, and purification and as catalysts and tendency of layer formation in the nanotube. Further, the
structural properties were found to vary strongly with the radial
* Corresponding author. E-mail: [email protected]. position of the chain’s center of mass. In some recent work,
10.1021/jp8073915 CCC: $40.75  2009 American Chemical Society
Published on Web 04/07/2009
Properties of Confined Alkanes J. Phys. Chem. C, Vol. 113, No. 17, 2009 7171

single-component adsorption of methane, 29,30 ethane,31-33 and V1 V2 V3

some other higher n-alkanes34,35 along with its adsorption utor(φ) ) V0 + (1 + cos φ) + (1 - cos 2φ) + (1 +
2 2 2
isotherms and density and orientation profiles was investigated
in carbon slit pores using Monte Carlo simulations. Furthermore, cos 3φ) (4)
there have been numerous molecular dynamics simulations for where V0 ) 0, V1 ) 355.03 K, V2 ) -68.19 K, and V3 )
chain molecules in slit pores. These molecular dynamics 791.32 K.
simulations were used to examine the density profiles and chain In this work, pore is of slit geometry with smooth and
conformations of linear and branched alkanes under shear-flow structureless surfaces. Wall-fluid interaction is described by
conditions.4,36-40 These investigations indicate that there is a the 9-3 Steele potential:45
profound influence of the slit surfaces on the fluid structure.
However, in all of the above-mentioned work, authors did not
investigate rigorously changes in the critical and vapor-liquid
2
φwf(z) ) πFwεwfσwf3
3 15 z { ( ) ( )}
2 σwf 9
-
σwf
z
3
(5)
whereZ is the distance of the fluid particle from the wall and
interfacial properties under confinement with changes in the pore
Fw, εwf, and σwf are the parameters of the Steele potential. In eq
width, particularly those approaching the two-dimensional limit.
5, σwf ) (σw + σii)/2 where σw denotes the “diameter” of a
This is the major objective of the current work. In this work,
wall atom and σii refers to the molecular diameter of corre-
we have investigated thermophysical properties of n-alkanes
sponding CH2-CH2 or CH3-CH3- interactions. Potential
confined in graphite and mica slit pores. In particular, we present
parameters, Fw, εwf, and σw for graphite and mica surfaces are
the phase coexistence properties (saturated densities and vapor
0.033 Å -3, 84 K, and 3.92 Å and 0.097 Å-3, 100 K, and 2.22
pressures), critical properties (critical temperature, pressure, and
Å, respectively, whereas for methane σwf for graphite and mica
density), local density and orientation profiles of liquid and vapor
surfaces are 3.7995 and 2.9805 Å, respectively.46
phases, and interfacial properties (vapor-liquid surface tension)
2.2. Simulation Details. In this work, we have used a
of n-alkanes of chain length C1-C8, with changes in the slit-
configurational-bias47-49 grand-canonical transition-matrix Monte
width, temperature, and surface characteristics. The paper is
Carlo (GC-TMMC)50,51 simulation scheme to explore the
organized as follows. Section 2 outlines the potential models
configuration space more efficiently. A configurational-bias
and simulation details. In section 3, we present the results of
Monte Carlo technique is incorporated within the GC-TMMC
our investigation, and we conclude in section 4.
framework along with a multicanonical sampling scheme.52 In
2. Simulation Methods this approach, Monte Carlo simulations are conducted in a
standard grand-canonical ensemble, where the volume (V),
2.1. Potential Model. A united-atom approach41 is used to chemical potential (µ), and temperature (T) are held constant
model the n-alkane molecules. Nonbonded site-site interactions and the particle number (N, density) and energy (U) fluctuates.
are described with the modified Buckingham exponential During a simulation, attempted transitions between states of
intermolecular potential of Errington and Panagiotopoulos,42 for different densities are monitored.53,54 At regular intervals during
which the pair interaction energy, U, is represented as

{ ]) ( ) ]}
a simulation, this information is used to obtain an estimate of

[ ([ rm 6 the density probability distribution, which is subsequently used

ε 6 r
U(r) ) exp R 1 - - for r > rmax to bias the sampling to low probability densities. Over time, all
6 R rm r
1- densities of interest are sampled adequately. The result is an
R efficient self-adaptive method for determining the density
probability distribution over a specified range of densities
)∞ for r < rmax (1) (typically a range that corresponds to the densities of two
potentially coexisting phases). Once a probability distribution
where ε, rm, and R are adjustable parameters. The variable rm has been collected at a given value of the chemical potential
is the radial distance at which U(r) reaches a minimum, and (µ0), histogram reweighting55 is used to shift the probability
the cutoff distance rmax represents the smallest radial distance distribution to other values of the chemical potential using the
for which d[U(r)]/d(r) ) 0. The radial distance for which U(r) following relationship:
) 0 is denoted by σ. The parameters ε, σ, and R are 129.63 K,
3.679 Å, and 16, respectively, for the methyl group (-CH3), ln Π(N, µ))ln Π(N, µ0) + β(µ - µ0)N (6)
73.5 K, 4.00 Å, and 22, respectively, for the methylene group To determine the coexistence chemical potential, we apply the
(-CH2-), and 160.3 K, 3.73 Å, and 15, respectively, for CH4. above relationship to estimate the chemical potential that
The following combining rules are used to determine the cross produces a coexistence probability distribution. Saturated densi-
parameters: ties are related to the first moment of the vapor and liquid peaks
of the coexistence probability distribution, Πc(N). To calculate
1
σij ) (σi + σj) the saturation pressure, we use the following expression:
2
εij ) (εiεj)1⁄2
βPV ) ln[ ∑ Πc(N) ⁄ Πc(0)] - ln(2) (7)
In addition to the calculation of the saturation pressure, densities,
Rij ) (RiRj)1⁄2 (2) and energies, GC-TMMC simulation can also yield the inter-
The bond lengths CH3-CH3, CH3-CH2, and CH2-CH2 are facial free energy, βFL, for a finite system size with a cell length
1.839, 1.687, and 1.535 Å, respectively. Bond-bending angles L, which is determined from the maximum likelihood in the
l v
are generated according to the bending potential43 liquid (Πmax ) and vapor (Πmax ) regions and the minimum
likelihood in the interface region (Πmin):
Kθ 1
ubend(θ) ) (θ - θeq)2 (3) βFL ) (ln Πlmax + ln Πvmax) - ln Πmin (8)
2 2
where Kθ ) 62 500 K/rad2 and θeq ) 114°. Torsion angles are This method does not require the establishment and maintain-
generated according to the potential44 ance of an interface, as is necessary in slab-based methods.56
7172 J. Phys. Chem. C, Vol. 113, No. 17, 2009 Singh et al.

From the formalism of Binder,57 the infinite-size interfacial free aligned perpendicularly to the wall, a value of -0.5 if they are
energy of a two-dimensional surface can be evaluated using all parallel to the walls, and a value of 0 if they are randomly
the following relationship: oriented.
To perform the simulations in an efficient manner, we take
βFL 1 ln L advantage of the fact that the TMMC algorithm enables one to
βγL ) ) C1 + C2 + βγ∞ (9)
2S S S fill the overall collection matrix through a series of independent
where γL is the apparent system size-dependent surface tension simulations, each restricted to a limited range of macrostates.
and γ∞ is the infinite-system size (L f ∞) surface tension, S is In this work, we used a scheme in which a series of semi-
the interface area (LH), L and H are the simulation cell size independent simultaneous simulations sample overlapping win-
and pore width, respectively, and C1 and C2 are constants. The dows in the macrostates with periodic swapping of configura-
expression (9) suggests that the group βFL/2S becomes linear tions between processors. Further details can be found in ref
with the scaling variable (ln L)/S as the system size approaches 65. In our simulations, the Monte Carlo move distribution for
infinity. The method enables one to evaluate the infinite-system
size interfacial tension by extrapolating a series of finite-system
size interfacial free energies. This approach has now been
applied to obtain the surface tension of various fluids58-61 but
mainly in the bulk phase. Recently, we have applied it for the
evaluation of the vapor-liquid surface tension of a confined
square-well fluid.62
In this work, we have estimated the vapor-liquid critical
parameter by fitting the coexistence densities to the law of the
rectilinear diameter63 and the scaling law for the density.64

Fl - Fv ) B 1 - ( T
Tc ) β
(10)

(Fl + Fv) ⁄ 2 ) Fc + A 1 - ( T
Tc ) (11)

where Fl, Fv, Fc, Tc, and β are the liquid-phase density, vapor-
phase density, critical density, critical temperature, and critical
exponent, respectively.
The critical pressure, Pc, is calculated using the least-squares
fitting of the saturation pressure data obtained from the GC-
TMMC simulations to the following expression, which has a
form similar to that of the Antoine equation:
B
ln Pc ) A - (12)
Tc
where A and B are adjustable parameters. The above empirical
form is also utilized to obtain the critical pressure for confined
fluids, as shown earlier for a confined square-well fluid.62
The saturated z density profiles (profiles perpendicular to the
slit surface), Fz, are obtained by recording F(N,z) for each particle
number sampled during GC-TMMC simulations. Coexistence
density profiles are finally obtained using the following expres-
sion:

∑ Πc(i) F(i, z)
i∈vapor/liquid
F(z)vapor/liquid ) (13)
∑ Πc(i)
i∈vapor/liquid

where i is the particle number and its range varies for liquid
and vapor phases.
The orientation of the molecules in the slit pore is evaluated
using the following orientation order parameter (Sz), which is
defined as

Sz ) 〈 3 cos2R - 1 〉 (14)

where R is defined as the angle between the head-to-tail vector Figure 1. (a) Vapor-liquid coexistence envelope, (b) saturation
pressure, and (c) vapor-liquid surface tension of bulk methane,
of the alkane molecule and the z axis. Sz is evaluated for n-butane, and n-octane. Solid lines and filled and open symbols
coexistence liquid and vapor phases akin to the calculation of represent the data of the experiments,68 the exp-6 model with full range
coexistence density profiles, as described above. The orientation interaction, and the exp-6 model with rc ) 15 Å, respectively. The
order parameter takes a value of 1 if all of the molecules are error bars are smaller than the symbol size.
Properties of Confined Alkanes J. Phys. Chem. C, Vol. 113, No. 17, 2009 7173

Figure 3. Local z-density profiles of the saturated vapor phase, in the

Figure 2. Vapor-liquid coexistence envelopes of alkanes in graphite inset, and the saturated liquid phase of methane in (a) graphite and (b)
and mica slit pores: (a) methane; (b) n-butane. Solid curves represent mica slit pores, respectively, at a reduced temperature, TR ) 0.90.
bulk simulation data with pentagons as the bulk critical points. Filled
symbols represent graphite (G) slit pores and open symbols mica (M)
slit pores. In part a, symbols square, inverted-triangle, star, circle, properties.66 Earlier, Goujon and co-workers67 have investigated
triangle, and diamond represent slit widths of 40, 15, 10, 8, 6, and 5 the effect of the cutoff radius, rc, on the liquid-vapor surface
Å, respectively. In part b, filled diamond and triangle symbols represent tension of n-pentane using the TraPPE model. The authors found
6 and 5.8 Å, respectively, and open diamond and triangle symbols the TraPPE model with rc ) 17 Å, a reasonably good model
represent 5.7 and 5.5 Å slit widths, respectively; other symbols have for the prediction of the bulk coexistence density and surface
the same meanings as those in part a. The error bars are smaller than
the symbol size. tension of n-pentane. In our earlier work, some of us compared
the bulk and interfacial properties of the exp-6 model of normal
butane, hexane, and octane with the experimental values. The
evaluating phase coexistence properties is as follows: 15% exp-6 model though predicts the bulk coexistence property
particle displacement, 50% particle insertion/deletion, 15% exceedingly well; however, it overpredicts the surface tension
particle rotation, and 20% particle regrowth. In the finite-size by 26%. In this work, we first investigate the effect of the cutoff
scaling approach, the maximum number of particles taken in radius with rc ) 15 Å on the bulk and interfacial properties of
this work varied from around 2100 particles for methane to 1400 the exp-6 model and compare it with the literature values. The
particles for octane. The simulations are performed with results for the surface tension of bulk methane, ethane, and
truncated nonbonded interaction without long-range correction propane using the full exp-6 model (nontruncated form) are new
(LRC) unless otherwise stated. Three independent runs are and were not reported earlier. Figure 1a shows the bulk phase
conducted to obtain the statistical error in the coexistence coexistence curve of C1, C4, and C8 based on the truncated exp-6
density, pressure, and vapor-liquid surface tension. The errors model with rc ) 15 Å and the full exp-6 model along with
in the coexistence density, vapor pressure, and surface tension experimental data.68 It is known that the full exp-6 model
of bulk n-alkanes are less than 1%, 0.4%, and 4%, respectively. predicts the phase equilibrium data of n-alkane accurately, which
On the other hand, statistical errors in the saturated density, also is reflected from Figure 1. Phase coexistence data calculated
pressure, and vapor-liquid surface tension of confined n-alkanes with rc ) 15 Å without LRC, for methane, also appear to be in
are less than 3%, 1.5%, and 5%, respectively. Errors in Tc, Fc, good agreement with the experimental data. However, for the
and Pc of bulk n-alkanes are less than 0.3%, 0.7%, and 3.5%, case of butane and octane, saturated vapor densities are
respectively, whereas errors in Tc, Fc, and Pc of confined overestimated whereas saturated liquid densities are underes-
n-alkanes are less than 0.4%, 1.5%, and 4.2%, respectively. timated slightly with the implementation of the cutoff radius,
rc ) 15 Å without LRC. Because of the shrinkage of the
3. Results and Discussion
coexistence envelope, the critical temperature is underestimated
It has been observed that the cutoff radius of the intermo- by 3-4% for butane and n-octane, respectively, whereas the
lecular potential influences the phase equilibrium and interfacial critical density is overestimated by 2% for both fluids. Figure
7174 J. Phys. Chem. C, Vol. 113, No. 17, 2009 Singh et al.

Figure 4. Shift in the critical temperature ∆T* vs inverse of the slit Figure 5. Local z-density profiles of (a) the saturated vapor phase
width (H) for n-alkanes in (a) graphite and (b) mica slit pores. and (b) the saturated liquid phase of methane in graphite slit pores of
three different slit widths at a reduced temperature, TR ) 0.90.
1b presents the saturation vapor pressure of the bulk n-alkanes.
The behavior of the vapor pressure for the truncated exp-6 model graphite and mica slit pores of various slit widths. Typical phase
is similar to that seen for the orthobaric densities. The critical coexistence envelopes of methane and butane are shown in parts
pressure for the truncated exp-6 model, with rc ) 15 Å, is a and b of Figure 2, respectively, which present the transforma-
underestimated by 1.8%, 7.6%, and 8.6% for methane, butane, tion of the behavior from nearly three-dimensionally to two-
and n-octane, respectively. On the other hand, the full exp-6 dimensionally induced by the confinement. A large fraction of
model (nontruncated form) overestimates the critical pressure the confined molecules experience a reduction in the number
by 1.25%, 6.3%, and 6.5% for methane, butane, and n-octane, of nearest-neighbor molecules; hence, under confinement, a
respectively. lowering of the critical temperature is observed. The critical
The vapor-liquid surface tension is evaluated using a finite- temperature continues to decrease with a decrease in the slit
size scaling approach in a way similar to that reported in our width up to a certain extent, which for all of the cases is close
earlier work.69 The surface tension based on n-alkane with the to two to three molecular diameters. A subsequent decrease in
truncated exp-6 model, with rc ) 15 Å, is in good agreement the slit width has no effect on the critical temperature, and it
with the experimental data for methane and ethane; however, appears to become constant. The critical density, on the contrary,
for the larger alkanes, the truncated potential underestimates oscillates as we approach from bulk to two-dimensional
the surface tension, as shown in Figure 1c. The deviation from geometry, as observed in Figure 2a,b. In general, the critical
the experimental data gradually increases from propane to density at a large pore is higher than the bulk critical density.
octane. Our current investigation along with earlier work69 shows This is due to the layering of molecules in the vapor phase near
that the exp-6 model with full range interaction overestimates the surface, which increases the density of the vapor phase. The
the surface tension value of C1-C8 by 26%. On the other hand, stronger the surface, the higher would be the vapor density,
with a fixed rc ) 15 Å without LRC, the average absolute leading to higher critical density. However, such behavior is
deviation from the experimental values for methane, ethane, absent in smaller pores because of the decrease in saturated
propane, butane, and octane are 9, 2, 5, 12, and 16%, liquid densities. This trend is observed for all of the studied
respectively. To study the various coexistence and interfacial cases of alkanes, C1, C4, and C8, except for methane in a graphite
properties under different slit-pore confinements, we have used slit pore, which has shown lower critical density even for higher
a cutoff radius rc ) 15 Å without LRC in this work. This choice slit width, which is probably due to the relatively less effect of
of the cutoff radius seems to be reasonable based on the analysis the graphite surface on the vapor-phase density. Recently, Jiang
that most properties deviate slightly from the experimental and Sandler observed similar behavior of higher critical density
values for C1-C8. and a lowering of the critical temperature for C1-C8 in a carbon
Phase coexistence properties and critical properties of n- nanotube.25 In addition, we observed that, as the system
alkanes of carbon chains C1, C4, and C8 are investigated in approaches the two-dimensional regime, the critical density
Properties of Confined Alkanes J. Phys. Chem. C, Vol. 113, No. 17, 2009 7175

Figure 6. Corresponding state plot of vapor-liquid coexistence

Figure 7. (a) Reduced local z-density and (b) orientation profiles of
envelopes of n-butane and n-octane in graphite (G) and mica (M) pores
saturated vapor and liquid phases of n-octane in graphite (G) and mica
for two different slit widths.
(M) slit pores at a fixed slit width, 30 Å, and at a reduced temperature,
TR ) 0.93.
increases toward the two-dimensional value even though the
critical temperature is almost constant. To understand this against the inverse of the slit width is observed to change at
behavior, we studied the coexistence vapor and liquid density smaller pore size and, in particular, first increases for the studied
profiles at extremely small pores. Parts a and b of Figure 3 system and subsequently approaches zero near the two-
present the z-density profiles of methane at a reduced temper- dimensional limit. For example, in the case of a graphite-methane
ature TR ) T/Tc ) 0.90 in graphite and mica slit pores, system, as shown in Figure 4a, the rate of the critical temperature
respectively, at two different slit widths. Figure 3a depicts that shift is found to increase with a decrease in the slit width from
in a graphite slit pore of slit width 5.4 Å, coexistence vapor 10 to 9 Å, which is relatively higher compared to that seen for
and liquid densities are more than that in a 8 Å slit pore; a pore size higher than 10 Å. This interesting behavior is due
however, the relative increase in the liquid-phase density is to changes in the coexisting liquid and vapor densities with a
larger than that in the vapor-phase density as the slit width is reduction in the slit width. Equation 10 suggests that the
reduced from 8 to 5.4 Å. This is also visible in the coexistence estimation of Tc depends on the difference between coexisting
envelope of methane in Figure 2a. This clearly explains the liquid and vapor densities. Higher differences in coexisting
increase in the critical density even though the critical temper- densities will result in comparatively higher Tc values. For
ature remains insensitive to the decrease in the slit pore with a example, at a reduced temperature of 0.90, the average vapor
slit width of less than 8 Å. Similar behaviors of methane in density decreases by 60% and the liquid density increases by
mica slit pores with slit widths of 5 and 7 Å are observed, as 3.5%, as the slit width is decreased from 15 to 10 Å. On the
seen in Figure 3b. other hand, the average vapor density decreases by 33% and
Bruno et al.70 and Vishnyakov et al.71 reported that the shift the liquid density decreases by 12.5% as the slit width decreases
in the critical temperature, ∆T * ) (Tcb - Tcp)/Tcb, where Tcb is from 10 to 9 Å. This remarkable change in coexisting densities
the critical temperature of the bulk fluid and Tcp is the critical is reflected in a sudden relative decrease in the Tc value, with
temperature of the confined fluid, for small pores, is roughly a change in the slit width from 10 to 9 Å. This behavior is also
inversely proportional to the slit width. Parts a and b of Figure reflected in local z-density profiles, as shown in Figure 5a,b, of
4 present the shift in the critical temperature against the inverse coexisting vapor and liquid phases of methane in graphite slit
of the slit width for the two different types of slit pore pores at these three different slit widths, namely, 9, 10, and 15
confinements, graphite (G) and mica (M), which depict more Å at a fixed reduced temperature of 0.90. This transition in the
than two linear regimes for the critical temperature. Our slope for different linear regimes for smaller slit widths is
observations are in good agreement with the investigation of indicative of the fact that, in the case of a confinement decrease,
Bruno et al. and Vishnyakov et al. but for pore sizes larger than the critical temperature does not obey a single linear dependency
approximately three molecular diameters is not extensible for throughout the range of pore widths. This was also observed
smaller pore size. The slope of the critical temperature shift recently for a square-well fluid in a hard slit pore.72
7176 J. Phys. Chem. C, Vol. 113, No. 17, 2009 Singh et al.

Figure 8. Temperature dependence on the density and orientational profiles of n-octane in a mica slit pore for a 30 Å slit width. Plots a and b are
saturated liquid and vapor density profiles, respectively, and the corresponding orientation profiles are in given in parts c and d, respectively.

Parts a and b of Figure 6 present corresponding state plots shown in Figure 8. Parts a and b of Figure 8 indicate that, with
of coexistence envelopes of n-butane and n-octane in terms of an increase in the temperature, the density of the liquid phase
reduced temperature (T/Tc) and density (F/Fc) for pores with in each layer decreases, whereas the density of the vapor
two different slit widths, 40 and 8 Å. Figure 6a shows that, at phase increases, as expected. This is also evident from the phase
larger slit width, the coexistence density of the vapor phase in coexistence plots of Figures 2 and 6. On the other hand, the
a mica slit pore is more than that in a graphite slit pore, whereas orientation of the molecules in the liquid and vapor phases does
the reverse is the case for the liquid-phase density. However, not change appreciably with a change in the temperature,
in a smaller slit-width pore, for example, 8 Å, surface especially near the slit surfaces, as shown in Figure 8c,d. This
characteristics appear to be noninfluential, as seen in Figure 6b, is in agreement with the experimental evidence with long-chain
which shows an indistinguishable behavior of coexistence n-alkanes in the slit-pore confinement.73 The experimental
densities for fluids in graphite and mica pores at two to three evidence of ref 73 along with our study indicates that a change
molecular diameter slit pores. To gain insight into this behavior, in the temperature has insignificant influence on the orientation
local density and orientational profiles are being studied. Parts of the molecules near the surface.
a and b of Figure 7 present z-density profiles and corresponding The vapor-pressure data of a pure substance are fundamentally
orientation profiles of n-octane in a 30 Å slit width and at a important for phase equilibrium calculations and engineering
reduced temperature of 0.93. Figure 7a shows that, in spite of applications, such as distillation, evaporation, and two-phase
having greater density in the first layer of the liquid-phase reactors. We have investigated the effect of the extent of
profile, which is attributed to the stronger attractive nature of confinement in two different types of slit pores on the saturation
mica as compared to the graphite surface, the overall average vapor pressure for a range of temperatures. Parts a and b of
liquid density in a mica slit pore is less than that in a graphite Figure 9 show comparisons of the saturation vapor pressure in
slit pore. In contrast, the average density in the vapor phase in confinement with respect to bulk, in a corresponding state plot.
a mica pore is more than that in a graphite pore. Similar behavior Saturation vapor pressures under either type of confinement
is observed for other alkanes in higher slit-width pores. A (graphite and mica) show positive and negative deviation with
comparison of the orientation profiles of n-octane in Figure 7b respect to the bulk value at a give reduced temperature. The
shows that the molecules in each layer are ordered and lie saturation vapor pressure at larger slit width is more than that
parallel to the walls (Sz < 0). As the distance from the wall of bulk at a given reduced temperature; for n-octane, this value
increases, the orientation order fluctuates and eventually ter- is approximately 20 Å in mica and graphite. A subsequent
minates at zero, indicative of the random nature of the system. decrease in the slit width (as seen for 10 and 8 Å) reduces the
We have studied the temperature dependence of the density and reduced vapor pressure, which is substantially smaller than the
corresponding orientation profiles of liquid and vapor phases corresponding bulk value. A deviation of the reduced vapor
of a typical case of n-alkane (n-octane) in a mica slit pore, as pressure from the corresponding bulk value is seen as larger
Properties of Confined Alkanes J. Phys. Chem. C, Vol. 113, No. 17, 2009 7177

Figure 9. Variation of reduced saturation vapor pressures vs reduced temperature of the bulk and confined n-octane in (a) graphite and (b) mica
slit pores of varying slit widths. Solid lines represent bulk values; filled and open symbols show positive and negative deviations, respectively. Part
c shows the shift in the critical pressure ∆P* of butane and octane with the inverse of the slit width (H) in graphite (G) and mica (M) slit pores.

for a mica pore because of its stronger attraction. Chen et al.74 critical pressure shift is completely in contrast with the non-
earlier predicted that the nanocapillary pressure of n-hexadecane monotonic behavior seen for the critical temperature and density.
in terms of ln(P/Ps) vs 1/r oscillates when the capillary size, r, Figure 9c also reflects that the nature of the surface is more
approaches a few molecular diameter, which suggests that the dominant on the critical pressure, particularly at larger slit
behavior of fluids in a nanosize capillary deviates from that of widths, which is not seen for the critical temperature. Such an
Kelvin’s equation. In this work, specific calculations to check effect diminishes and disappears as the slit width is reduced
the behavior of Kelvin’s equation in small pores have not been toward the two-dimensional value.
performed; however, a certain indication of the deviation from One of the advantageous aspects of the GC-TMMC method
Kelvin’s equation is probably reflected in our saturation vapor is that it can be used in conjunction with the finite-size-scaling
pressures (pressure component parallel to the slit surfaces and (FSS) technique of Binder57 to evaluate surface tension values.
perpendicular to the vapor-liquid interface), under confinement, The effectiveness of this technique over other methods is
as shown in Figure 9a,b. Nevertheless, a more rigorous study demonstrated by Singh and Kofke.60 This technique also benefits
is required to test the correctness of Kelvin’s equation and a from the ease with which calculations can be performed using
modified version of it74 for n-alkane in nanocapillaries, which multiple processors. The combined GC-TMMC + FSS method
is out of the scope of the current work and is reserved for a has been used to determine surface tension values for Lennard-
future study. Jones,51 square-well,60 model associating fluids61 and chain
We have also evaluated vapor-liquid critical pressures under molecules.69 In this work, we have investigated the effect of
various confinement using eq 12, in the same way as was the surface characteristics and the extent of confinement on the
reported for bulk n-alkanes69 and confined square-well fluids62 vapor-liquid surface tension of n-alkanes. Our investigation
in earlier investigations. Equation 12 is used as a fitting tool to indicates that the bulk vapor-liquid surface tension of n-alkanes
calculate the critical pressure once the critical temperature is decreases under confinement of graphite (G) and mica (M) slit
obtained. Such usage of eq 12 is based on the observation that pores. This reduction of the surface tension is attributed to the
the logarithm of the saturation pressure under confinement, reduction in the vapor-liquid surface free energy due to
ln(P), scales linearly with 1/T, in a way similar to that for the confinement. Parts a-e of Figure 10 present a comparison of
bulk vapor pressure.69 In Figure 9c, we have shown the shift in the vapor-liquid surface tension of bulk n-alkane evaluated with
the critical pressure, ∆P* ) (Pcb - Pcp)/Pcb, where Pcb is the the truncated exp-6 potential with rc ) 15 Å and the same under
bulk critical pressure and Pcp is the confined critical pressure, different slit-pore confinement. The surface tension diminishes
with the inverse of the slit width. The behavior seen for the with a decrease in the slit width because for a given temperature
7178 J. Phys. Chem. C, Vol. 113, No. 17, 2009 Singh et al.

Figure 10. Vapor-liquid surface tensions of bulk and confined (a) methane, (b) ethane, (c) propane, (d) butane, and (e) octane in graphite (G) and
mica (M) pores for two slit widths, 20 and 30 Å. Solid lines represent the bulk surface tension value (data points are the same as those reported
in Figure 1c, with open symbols) obtained with rc ) 15 Å without LRC. In part f, the surface tensions of bulk and confined methane, ethane,
butane, and octane are shown versus the reduced temperature (T/Tc). Surface tensions of confined n-alkanes are shown only with 30 Å slit width
of graphite and mica surfaces, with dotted and solid lines representing fits to the data of bulk and confined n-alkanes, respectively. Filled and open
symbols represent the surface tension values of fluids in graphite and mica slit pores, respectively, and have the same meaning as those in parts
a-e. Half-filled symbols (half-left triangle for methane, half-square for ethane, half-circle for butane, and half-pentagon for octane), in part f,
represent bulk phase simulated data points of n-alkanes. The error bars are smaller than the symbol size.

the free energy difference with respect to the interface decreases observed for the smaller pore width of 20 Å, which perhaps is
with a reduction in the slit width. However, for a constant slit the signature of insensitiveness of the thermophysical properties
width, the interfacial free energy decreases with an increase of toward the surface characteristic at such a scale of confinement.
the temperature; hence, the vapor-liquid surface tension This is also reflected from the corresponding state plot of
decreases with an increase of the temperature as expected. We orthobaric densities, as seen in Figure 6b. Figure 10f presents
also observed that, in general, at the same absolute temperature, the vapor-liquid surface tension of bulk and confined n-alkanes
for a 30 Å slit width, the reduction in the vapor-liquid surface as a function of reduced temperatures (T/Tc). Interestingly, the
tension from the bulk is more for the mica (M) slit pore than surface tension values of all typical bulk n-alkanes, except for
for the graphite (G) slit pore, especially for comparatively longer methane, studied in this work fall reasonably around one linear
chain molecules (C3 or higher). Conversely, this trend is not curve, as shown by the dotted line for the range of studied
Properties of Confined Alkanes J. Phys. Chem. C, Vol. 113, No. 17, 2009 7179

properties of alkanes in nanopores. Our investigations indicate

that the surface characteristics and the extent of confinement
affect dramatically various properties of n-alkanes. The critical
temperature decreases continuously up to a certain minimum
slit width, and then it becomes constant as the system approaches
the two-dimensional regime. We observed that a shift in the
critical temperature with respect to the bulk value follows more
than two linear regimes as a function of the inverse slit width
from nearly three-dimensional to two-dimensional geometry.
The critical density, on the other hand, fluctuates with a decrease
in the slit width. On the other hand, a shift in the critical pressure
monotonically increases with a decrease in the slit width and
approaches to a constant value at a smaller pore. Corresponding
state plots suggest that the deviation of the saturation vapor
pressure from the bulk saturation pressure, for a range of
temperatures, under confinement is positive for large pores and
Figure 11. Vapor-liquid surface tension, γ, of n-butane in a mica negative for smaller pores. As the temperature approaches Tc,
slit pore vs (1 - T)/Tc. Symbols represent the simulation data obtained this deviation slowly disappears. The corresponding state plot
with the FSS technique, and the solid line is a fit of the data. reveals that at higher slit width vapor-liquid coexistence
properties are sensitive to the nature of the interacting surfaces.
On the other hand, these properties are seemingly insensitive
temperatures. Surprisingly, the behavior remains the same for
to the surface characteristics at the smaller pore width.
the confined fluid; however, the slope of the master curve for
Liquid-vapor surface tension values for n-alkanes in graphite
the confined fluid, as shown by the solid line, is less than that
and mica pores are studied, for the first time, and found to be
of the bulk fluid. Figure 10f contains surface tension data for
suppressed by many fold under confinement.
n-alkanes in a slit pore of 30 Å slit width. Corresponding data
in a 20 Å slit width (data not shown for the sake of clarity) fall
Acknowledgment. This work was supported by the Depart-
also on the same linear curve, except for methane. The probable
ment of Science and Technology, Government of India (Grants
reason for deviation of methane from the master curve is its
SR/S3/CE10/2006 and IR/S3/EU/005/2007). We acknowledge
spherical nature, as compared to the other chain molecules. It
further computational support from the Computer Center, Indian
is apparent from Figure 10f that bulk phase surface tension
Institute of Technology Kanpur.
values and corresponding confined values are different at the
same reduced temperature, specifically at lower temperatures.
In particular, confinement diminishes the surface tension value References and Notes
at a given reduced temperature. As the temperature increases, (1) Gelb, L. D.; Gubbins, K. E.; Radhakrishnan, R.; Sliwinska-
such that T/Tc > 0.9, both bulk and confined surface tension Bartkowiak, M. Rep. Prog. Phys. 1999, 62, 1573.
values appear to collapse on each other, which indicates the (2) Jackson, C. L.; McKenn, G. B. J. Chem. Phys. 1990, 93, 9002.
(3) Lee, S. C.; Cheng, H. S. Tribol. Trans. 1991, 113, 327.
insignificant role of confinement and surface characteristics on
(4) Gupta, S. A.; Cochran, H. D.; Cummings, P. T. J. Chem. Phys.
the surface tension at high temperature. 1997, 107, 10327.
It is known that critical property calculation from the (5) Israelachvili, J. N. Intermolecular and Surface Forces; Academic:
rectilinear approach is an approximate technique compared to New York, 2003.
(6) Baljon, A. R. C.; Robbins, M. O. Mater. Res. Bull. 1997, 22, 22.
the more rigorous FSS technique.75 Hence, we have verified (7) Matranga, K. R.; Myers, A. L.; Glandt, E. D. Chem. Eng. Sci. 1992,
the correctness of the critical temperature, estimated in this work, 47, 1569.
with the value extrapolated from the surface tension values at (8) Bhatia, S. Langmuir 1998, 14, 6231.
higher temperatures. We have evaluated some of the vapor-liquid (9) Altwasser, S.; Welker, C.; Traa, Y.; Weitkamp, J. Microporous
Mesoporous Mater. 2005, 83, 345.
surface tension values using the GC-TMMC + FSS technique, (10) Maciver, D. S.; Emmett, P. H.; Frank, H. S. J. Phys. Chem. 1958,
at some higher temperatures, namely, 335, 337, 338, 339, and 62, 935.
340 K for a typical case of alkane, n-butane. Surface tension, (11) Gregg, S. J.; Sing, K. S. W. Adsorption, Surface Area, and Porosity,
γ, data calculated at the above temperatures are used for 2nd ed.; Academic: New York, 1982.
(12) Liapis, A. J., Ed. Fundamentals of Adsorption; Engineering
extrapolation to obtain the critical temperature as γ approaches Foundation: New York, 1987.
zero. Figure 11 presents the simulation data and the fitted curve. (13) Rouquerol, F.; Rouquerol, J.; Sing, K. S. W. Adsorption by Powders
The critical temperature obtained through the above method is and Porous Solids: Principles, Methodology, and Applications; Academic:
extremely close to our calculation based on the rectilinear San Diego, 1999.
(14) Rowlinson, J. S.; Widom, B. Molecular Theory of Capillarity;
diameter approach. For example, in the case of butane in a mica Oxford University Press: London, 1982.
slit pore of 20 Å slit width, the Tc value estimated from the (15) Croxton, C. A., Ed. Fluid Interfacial Phenomena; Wiley: New York,
rectilinear approach is 347.31 K and the extrapolated surface 1985.
tension value at 347.31 K is around 2.24 × 10-24, which is (16) Charvolin, J., Joanny, J.-F., Zinn-Justin, J., Eds. Liquids at
Interfaces; North-Holland: Amsterdam, The Netherlands, 1990.
practically zero. This suggests that the technique employed for (17) Henderson, D., Ed. Fundamentals of Inhomogeneous Fluids;
calculation of the critical temperature in this work gives Dekker: New York, 1992.
reasonably accurate results. (18) Binder, K.; Landau, D.; Müller, M. J. Stat. Phys. 2003, 110, 1411.
(19) Fuchs, A. H.; Cheetham, A. K. J. Phys. Chem. B 2001, 105, 7375.
(20) Smit, B.; Siepmann, J. I. Science 1994, 264, 1118.
4. Conclusions (21) Fox, J. P.; Roy, V.; Bates, S. P. Microporous Mesoporous Mater.
2004, 69, 9.
The configurational-bias Monte Carlo technique combined (22) Gupta, A.; Clark, L. A.; Snurr, R. Q. Langmuir 2000, 16, 3910.
with the original GC-TMMC formalism produces an efficient (23) Chempath, S.; Denayer, J. F. M.; De Meyer, K. M. A.; Baron, G. V.;
and self-adaptive method for evaluating the thermophysical Snurr, R. Q. Langmuir 2004, 20, 150.
7180 J. Phys. Chem. C, Vol. 113, No. 17, 2009 Singh et al.

(24) Jiang, J. W.; Sandler, S. I.; Schenk, M.; Smit, B. Phys. ReV. B (50) Errington, J. R. Phys. ReV. E 2003, 67, 012102.
2005, 72, 045447. (51) Errington, J. R. J. Chem. Phys. 2003, 118, 9915.
(25) Jiang, J. W.; Sandler, S. I. Fluid Phase Equilib. 2005, 228, 189. (52) Berg, B. A.; Nneuhaus, T. Phys. ReV. Lett. 1992, 61, 9.
(26) Duren, T.; Sarkisov, L.; Yaghi, O. M.; Snurr, R. Q. Langmuir 2004, (53) Fitzgerald, M.; Picard, R. R.; Silver, R. N. Europhys. Lett. 1999,
20, 2683. 46, 282.
(27) Frost, H.; Duren, T.; Snurr, R. Q. J. Phys. Chem. B 2006, 110, (54) Fitzgerald, M.; Picard, R. R.; Silver, R. N. J. Stat. Phys. 2000, 98,
9565. 321.
(28) Zhang, F. J. Chem. Phys. 1999, 111, 9082. (55) Ferrenberg, A. M.; Swendsen, R. H. Phys. ReV. Lett. 1988, 61,
(29) Do, D. D.; Do, H. D. J. Phys. Chem. B 2005, 109, 19288. 2635.
(30) Kowalczyk, P.; Tanaka, H.; Kaneko, K.; Terzyk, A. P.; Do, D. D. (56) Goujon, F. M.; Boutin, A.; Fuchs, A. H. J. Chem. Phys. 2002, 116,
Langmuir 2005, 21, 5639. 8106.
(31) Vishnyakov, A.; Piotrovskaya, E. M.; Brodskaya, E. N. Langmuir (57) Binder, K. Phys. ReV. A 1982, 25, 1699.
1996, 12, 3643. (58) Potoff, J. J.; Panagiotopoulos, A. Z. J. Chem. Phys. 2000, 112,
(32) Klochko, V. A.; Brodskaya, E. N.; Piotrovskaya, E. M. Langmuir 6411.
1999, 15, 545. (59) Singh, J. K.; Kofke, D. A.; Errington, J. R. J. Chem. Phys. 2003,
(33) Calleja, G.; Coto, B.; Pinar, A.; Morales-Cas, A. M. Adsorption 119, 3405.
2006, 12, 45. (60) Singh, J. K.; Kofke, D. A. Mol. Simul. 2004, 30, 343.
(34) Do, D. D.; Do, H. D. Chem. Eng. Sci. 2005, 60, 1977. (61) Singh, J. K.; Kofke, D. A. J. Chem. Phys. 2004, 121, 9574.
(35) Severson, B. L.; Snurr, R. Q. J. Chem. Phys. 2007, 126, 134708. (62) Singh, J. K.; Kwak, S. K. J. Chem. Phys. 2007, 126, 24702.
(36) Balasundaram, R.; Jiang, S. Y.; Belak, J. Chem. Eng. J. 1999, 74, (63) Rowlinson, J. S.; Swinton, F. L. Liquids and Liquid Mixtures, 3rd
117. ed.; Butterworth: London, 1982.
(37) Wang, J. C.; Fichthorn, K. A. Colloids Surf. A 2002, 206, 267. (64) Rowlinson, J. S.; Widom. B. Molecular Theory of Capillarity;
(38) Cui, S. T.; Gupta, S. A.; Cummings, P. T.; Cochran, H. D. J. Chem. Clarendon: Oxford, U.K., 1982.
Phys. 1996, 105, 1214. (65) Errington, J. R. Langmuir 2004, 20, 3798.
(39) Cui, S. T.; Cummings, P. T.; Cochran, H. D.; Moore, J. D.; Gupta, (66) Baidakov, V. G.; Chernykh, G. G.; Protsenko, S. P. Chem. Phys.
S. A. Int. J. Thermophys. 1998, 19, 449. Lett. 2000, 321, 315.
(40) Martini, A.; Liu, Y.; Snurr, R. Q.; Wang, Q. J. Tribol. Lett. 2006, (67) Goujon, F.; Malfreyt, P.; Boutin, A.; Fuchs, A. H. J. Chem. Phys.
21, 217. 2002, 116, 8106.
(41) Ryckaert, J. P.; Bellemans, A. Chem. Phys. Lett. 1975, 30, 123. (68) NIST Chemistry Web Book. http://webbook.nist.gov/chemistry/
(42) Errington, J. R.; Panagiotopoulos, A. Z. J. Phys. Chem. B 1999, fluid.
103, 6314. (69) Singh, J. K.; Errington, J. R. J. Phys. Chem. B 2006, 110, 1369.
(43) Van der Ploeg, P.; Berendsen, H. J. C. J. Chem. Phys. 1982, 76, (70) Bruno, E.; Marini Bettolo Marconi, U.; Evans, R. Physica A 1987,
3271. 141, 187.
(44) Smit, B.; Karaborni, S.; Siepmann, J. I. J. Chem. Phys. 1995, 102, (71) Vishnyakov, A.; Piotrovskaya, E. M.; Brodskaya, E. N.; Votyakov,
2126. E. V.; Tovbin, Yu. K. Langmuir 2001, 17, 4451.
(45) Steele, W. A. Surf. Sci. 1972, 36, 317. (72) Subimal, J.; Singh, J. K.; Kwak, S. K. J. Chem. Phys. 2008,
(46) Porcheron, F.; Rousseau, B.; Fuchs, A. H.; Schoen, M. Phys. Chem. submitted for publication.
Chem. Phys. 1999, 1, 4083. (73) Maeda, N.; Kohonen, M. M. J. Phys. Chem. B 2001, 105, 5906.
(47) Siepmann, J. I.; Frenkel, D. Mol. Phys. 1992, 75, 59. (74) Chen, Y.; Wetzel, T.; Aranovich, G. L.; Donohue, M. D. J. Colloid
(48) De Pablo, J. J.; Laso, M.; Suter, U. W. J. Chem. Phys. 1992, 96, Interface Sci. 2006, 300, 45.
2395. (75) Wilding, N. B. Phys. ReV. E 1995, 52, 602.
(49) Rosenbluth, M. N.; Rosenbluth, A. W. J. Chem. Phys. 1956, 23,
356. JP8073915