Adsorption of Water in Finite Length Carbon Slit Pore: Comparison

For describing the interaction of water, we used the potential model proposed by Muller et al., and the simulated isotherms of water in slit pores are...
0 downloads 0 Views 982KB Size
J. Phys. Chem. B 2007, 111, 13949-13956

13949

Adsorption of Water in Finite Length Carbon Slit Pore: Comparison between Computer Simulation and Experiment Atichat Wongkoblap and Duong D. Do* Department of Chemical Engineering, UniVersity of Queensland, St Lucia, Queensland 4072, Australia ReceiVed: June 18, 2007; In Final Form: September 6, 2007

The effects of surface dimensions and topology on the adsorption of water on a graphite surface at 298 K were investigated using the grand canonical Monte Carlo (GCMC) simulation. Regarding the surface topology, we specifically considered the functional group and its position on the surface. The hydroxyl group (OH) is used as a model for the functional group. For describing the interaction of water, we used the potential model proposed by Muller et al., and the simulated isotherms of water in slit pores are found to depend on the position and concentration of the functional group. The onset of adsorption shifts to lower pressure when the concentration of functional group increases or when the functional group is positioned at the center of the graphene surface. The configuration of a group of functional groups also affects the adsorption isotherm. In all cases investigated, we have found that the hysteresis loop always exists, and the loop size depends on the concentration of the functional group and its position. Finally, we tested the molecular model of water adsorption on a functional graphite pore against the experimental data of a commercial activated carbon. The agreement is found to be satisfactory when the model porous solid is composed of pores having width in the range between 10 and 20 Å and functional groups positioned at the center of the graphitic wall.

1. Introduction Adsorption phenomena are increasingly studied by molecular simulation tools, among which molecular dynamics (MD) and Monte Carlo simulation (MC)1,2 are two popular methods applied to solve numerous adsorption problems involving solid adsorbents. Activated carbon represents an important group of adsorbents because of its large micropore and mesopore volumes and high surface area. Models of porous carbon are usually assumed to be composed of pores whose surfaces are of infinite extent.3-9 Such a model is far from reality because pores are neither infinite nor uniform, but rather the pore surface is finite in lateral directions parallel to the surface,10 and it contains functional groups, chemical impurities, and morphological defects on the basal graphene layers.11 We have recently presented adsorption isotherms of argon in homogeneous and heterogeneous finite length carbon slit pores.12,13 However, the adsorption behavior of simple fluids, such as argon, on carbon is different from that of water.5,6 This is due to the greater interaction energy between fluid particles (hydrogen bonding) than between a fluid particle and a carbon surface. The choice of molecular model for water is also important; two types of water models have been put forward in the literature. One is the point charge models such as SPC,14 SPC/ E,15 and ST216 models, while the other is the square-well site models such as the primitive model (PM)17-20 and the Muller and co-workers5,6 models. The simulated adsorption isotherms of water on activated carbon show a distinctly vertical pore filling, while experimental studies21-24 show a slight adsorption at low relative pressures (P/P0, where P0 is the saturation pressure), a steep increasing (but not vertical) of adsorption isotherm at moderate P/P0, and a saturation at high P/P0.25 The common reason used to explain the difference between the * To whom correspondence should be addressed. Phone: +61-7-33654154; fax: +61-7-3365-2789; e-mail: [email protected].

simulation results and the experimental studies is that porous carbon contains functional groups; however, other factors such as pore width and length could also contribute to this difference. To develop a reliable solid model and a water model to describe the adsorption behavior of water on activated carbon is still a challenging issue. Therefore, the aim of this work is to use homogeneous and heterogeneous finite pores together with the Muller et al. water model to study the adsorption behavior of water in activated carbon. The Muller et al. water potential model is used because the point charge model involves longrange Coulomb forces which require a large simulation system size to minimize the cutoff errors induced because of the effect of simulating a finite system5 and the greatly increased computation time,26 and the Muller et al. model can mimic the association to avoid the large simulation system and computation time.25 The more accurate molecular potential model for water should account for not only the usual dispersion and repulsive forces and hydrogen bonding but also the long-range electrostatic forces, polarizability, quantum effects, bond flexibility, and multibody effects. The Muller et al. water potential model, however, uses the off-center square-well interactions to describe bonding sites27 and increases the energy well depth for unlike sites to form the hydrogen bonding which accounts for the electrostatic and attractive forces.5 The quantum and bond flexibility effects are not taken into account; this is because the influence of these effects is small enough to be neglected.27 The molecular parameters are evaluated by using Wertheim’s TPT1 theory and are fitted to thermodynamic properties such as the gas-liquid coexistence properties of bulk water at 298 K and vaporization energy.5,6 This water model cannot be used as a general water potential in other applications because the optimized molecular parameters are calculated at a single temperature of 298 K5. However, it is a simple model which can describe some of the most important adsorption behavior

10.1021/jp0747297 CCC: $37.00 © 2007 American Chemical Society Published on Web 11/29/2007

13950 J. Phys. Chem. B, Vol. 111, No. 50, 2007

Wongkoblap and Do

TABLE 1: Molecular Parameters Used in This Studya fluid

σOO, Å

OO/k, K

rOH, Å

σHB, Å

HB/k, K

H2O

3.06

90.0

1.2852

0.612

3800

a

O: oxygen atom; H: hydrogen atom; HB: hydrogen bonding.

of water in a finite length pore in this study. Furthermore, this model can describe adsorption behavior equally well when compared with the behaviors described by models with partial charges such as SPC/E.25,26 In this paper, the adsorption isotherms of water in the finite length carbon slit pore both in the presence and in the absence of functional groups at 298.15 K are obtained for pore widths of 10, 16, and 20 Å by using a grand canonical Monte Carlo simulation (GCMC). The functional group used in this study is the hydroxyl (OH) form. Two different locations of the functional group, either at the center or at the corner of the upper pore wall, are used to study the effects of the position and the concentration of the functional group on the adsorption isotherm. The details of solid surface and functional group will be described in sections 2.2 and 2.3, respectively. 2. Methodology 2.1. Water Model. The water model used in this study is taken from the work of Muller and co-workers.5,6 In this model, water is treated as a spherical Lennard-Jones (LJ) molecule with one dispersive site (oxygen atom) at the center of a tetrahedron and four square-well (SW) associating sites positioned at the vertices of the tetrahedron. These four sites represent two hydrogen atoms and two lone pairs of electrons. The molecular parameters used in this study are listed in Table 1,5,6 where OO and σOO are the well depth of the interaction potential and the collision diameter of oxygen, respectively, and k is the Boltzmann’s constant, rOH is the distance between an SW site and the oxygen atom, HB is the energy well depth for unlike associating sites (a hydrogen atom of one water molecule and a lone pair of electrons of another water molecule) to form a hydrogen bond while that for the like sites interaction is zero, and σHB is the collision diameter of two unlike associating sites. A cutoff radius in the calculation of interaction energy of 5 times the collision diameter (5σOO) is used in this study. The interaction energy between two dispersive sites (φOO) is calculated using the Lennard-Jones 12-6 equation.3,5

φOO(r) ) 4OO

[( ) ( ) ] σOO r

12

-

σOO r

6

(1)

where r is the separation distance. If A represents a hydrogen atom on one water molecule and B is a lone pair of electrons on another water molecule, the interaction potential energy between A and B to form a hydrogen bond can be calculated by using eq 2.5,6

φHB(rAB) )

{

-HB rAB < σHB 0 otherwise

(2)

where rAB is the distance between A and B. 2.2. Solid Model. The solid model used in this study is the carbon-based adsorbents whose pores have a slit-shaped geometry.12,13 A simple slit pore of finite length is modeled as a parallel pair of finite length walls which consist of graphene layers. Each wall consists of three graphene layers, and these layers are stacked on top of each other with an interlayer spacing (∆) of 3.354 Å. The width H of this slit pore model is defined as the distance between a plane passing through all carbon atom

Figure 1. The solid configuration for the first (a) and second (b) models. White spheres represent carbon atoms while black spheres and gray spheres represent oxygen atom and hydrogen atom of the OH groups, respectively.

centers of the outermost layer of one wall and the corresponding plane of the other wall. The carbon-carbon bond length in a graphene layer is 1.42 Å.11 The LJ parameters for a carbon atom of the graphene layer, σss and ss/k, are 3.4 Å and 28 K, respectively.3 The interaction energy between a water molecule and a carbon atom is calculated by the Lennard-Jones 12-6 equation in the case of a pore of finite dimensions and the 10-4-3 potential in the case of a pore of infinite lateral dimensions. The cross molecular parameters are calculated from the Lorentz-Berthelot rule.3,5-6 Assuming that the pairwise additivity holds, the total potential energy of the simulation box is calculated by summing the pairwise interactions between two dispersive sites, between hydrogen atoms and lone pair electrons, and between individual carbon atoms and dispersive sites.5-8

U)

φik(|ri - rk|) + ∑ φHB(i,j) ∑ φij(|ri - rj|) + ∑ i,k m,n

i, j>i

(3)

where ri and rj are the positions of the dispersive sites i and j, respectively, rk is that of a carbon atom, φij is the pair interaction potential between two dispersive sites, φik is that between the dispersive site of water molecule i and a carbon atom k or a dispersive site of functional group k, and φHB is that between a hydrogen atom m on water molecule i or functional group i and a lone pair of electrons n of water molecule j. 2.3. Functional Group Model. The functional groups on activated carbon surfaces have been identified as hydroxyl, carbonyl, carboxylic, phenolic, lactonic, and pyrone;6 however, to simplify the model, we assume that the representative functional group is the hydroxyl (OH) group as done in McCallum et al.6 The functional group is modeled as an LJ dispersive site at the center of oxygen atom and a single squarewell (SW) site at the hydrogen atom, and their molecular parameters are the same as those of the water model except for the hydrogen bond strength. The energy well depth to form hydrogen bond between the SW site on a hydroxyl group and the lone pair of electrons on a water molecule is 5000 K6 compared to 3800 K for water-water interactions. It is reported in the literature28 that the heteroatoms chemically combined with carbon atoms on the graphene layers are mainly located at the edges and corners. In this paper, we choose two models to investigate the effects of active site location on the adsorption behavior of water. In the first model, the OH groups are positioned at only one corner of the upper wall (Figure 1a); for clarity, we show only one graphene layer for each wall. The other model assumes that the active sites are placed at the center of the upper wall (Figure 1b). The LJ dispersive site of the hydroxyl group is located at a distance of 1.364 Å below the graphene layer while the SW site is placed at 1.2852 Å from the LJ center6 and is perpendicular to the pore wall. Hereafter,

Adsorption of Water in Finite Length Pore

J. Phys. Chem. B, Vol. 111, No. 50, 2007 13951

Figure 2. Adsorption isotherms of water at 298.15 K for infinite pore (triangle symbols; filled symbols for adsorption and unfilled symbols for desorption) and finite length slit pore (circle symbols) of various widths: H ) 10 (a), 16 (b), and 20 Å (c).

we will use the terms corner and center topologies to describe the first and second models, respectively. We also study the effects of the concentration of functional groups and their configuration on the adsorption isotherm. The effects of concentration are studied with zero, two, and six functional groups on the graphene layer, while the effects of configuration are investigated via the separation distance between functional groups. We choose the following distances in our investigation: 1.42, 4.26, and 7.10 Å. 2.4. Simulation Method. The simulation box of a slit pore is used in this study. For a finite pore model, all graphene layers are rectangular with linear dimensions in x and y of 37.6 and 39.4 Å, respectively. The top and the bottom of the simulation box are two walls of the slit pore, and each wall consists of three graphene layers. In the case of an infinite pore, a linear dimension of 30 Å is used. The pore widths used in this investigation are 10, 16, and 20 Å to represent the micropore in activated carbon. The grand canonical Monte Carlo method (GCMC) with the Metropolis algorithm1,2 is used to obtain the adsorption isotherm of water in a slit pore. One GCMC cycle consists of 1000 displacement moves and attempts of either insertion or deletion with equal probability. In each displacement move, the particle is also rotated randomly around x, y, or z-axis with equal probability. For the adsorption branch, we use an empty box as the initial configuration, and the simulation is carried out until the number of particles in the box does not change (in statistical sense). Once this is achieved, the chemical potential is increased to a new value, and the final configuration of the previous chemical potential is used as the initial configuration for the MC simulation. The initial orientation of the newly inserted water molecule is the horizontal tetrahedron configuration. After the insertion, the water molecule is rotated randomly around x, y, and z-axis in sequence. For each point on the adsorption branch, 700 million configurations are used to reach equilibrium and to obtain ensemble averages. On the other hand, desorption branch of the isotherm is started with the highest chemical potential, and then the chemical potential is decreased to a new value. The equilibrium configuration of the previous chemical potential is used as the starting point for the new chemical potential, and the process is repeated until the simulation box is empty. The displacement step length is initially chosen as 0.5 times the collision diameter of oxygen atom, and it is increased 5% when the acceptance ratio is greater than 0.5 and is decreased 5% when this ratio is less than 0.5. All calculations are carried out at 298.15 K. The virial equation29 is used to determine the reduced pressure of the bulk gaseous phase for a given chemical potential. In the case of infinite pore, periodic boundary conditions are applied in x and y directions.1,2 In the case of

finite pore, the particle move is rejected if the particle is displaced to a position outside the simulation box. 3. Results and Discussions We start our discussion by presenting the adsorption isotherms of water using the Muller et al. model for homogeneous (i.e., no functional group) pores of 10, 16, and 20 Å widths. Next, the adsorption isotherms obtained for heterogeneous pores of corner and center topologies for functional group will be compared with those obtained for the homogeneous pores to show the effects of the functional group and its location on the adsorption behavior. 3.1. Homogeneous Pores: Effects of Pore Length. The simulated isotherms versus relative pressure for water adsorption in graphitic pores at 298.15 K are shown in Figure 2. The curves with triangle symbols are those for infinite pore, and circle symbols are those for finite pore. The adsorption isotherms for infinite and finite length pores show the type V isotherm classified by Brunauer et al.30 and a very large hysteresis loop. The adsorption branch shows a sharp pore filling (less sharp is observed with finite pore), and the onset of this pore filling occurs at a pressure higher than the saturation pressure of the water potential model. The onset pressure increases with pore width, and such a behavior is also observed in the study of Liu and Monson25 for the infinite pore with the square-well sites model. The sharp change in density, the large hysteresis loop, and the onset of pore filling occurring at a pressure greater than the saturation pressure are not observed experimentally. This suggests that real surfaces are not homogeneous. Let us now explore the heterogeneity in the form of functional group and see whether we can reconcile the simulation results and the experimental data. 3.2. Heterogeneous Pores: The Role of Functional Group. We now study the effect of functional group and its topology on the water adsorption in finite length pores having widths 10, 16, and 20 Å. The two solid models mentioned in section 2.3, corner and center topologies, are considered, and two hydroxyl functional groups are assumed to be located on the graphene layer with a separation distance (δ) of 4.26 Å. First, we present in Figure 3 the adsorption isotherms obtained for heterogeneous finite pores with the corner topology (triangle symbols) and the center topology (circle symbols). For comparison, we also present the isotherm of homogeneous finite pore (dashed line) in the same figure. Figure 4 shows the snapshots of water molecules in heterogeneous pores having widths of 10 and 16 Å at low relative pressures to investigate how water adsorption is initiated. The following differences between the homogeneous and heterogeneous pores are observed.

13952 J. Phys. Chem. B, Vol. 111, No. 50, 2007

Wongkoblap and Do

Figure 3. Adsorption isotherms at 298.15 K for the Muller et al. water model in finite length pores of different widths of (a) 10, (b) 16, and (c) 20 Å for the corner (triangle symbols; filled symbols for adsorption and unfilled symbols for desorption), center topologies (circle symbols), and homogeneous pore (dashed line).

Figure 4. (a) Snapshots of water molecules in heterogeneous finite length slit pores for width of 10 Å. In these figures, big white spheres represent carbon atoms, big and small black spheres represent oxygen and hydrogen atoms of hydroxyl group, respectively, and big dark gray spheres represent oxygen atoms of water while small gray spheres and small white spheres represent hydrogen atoms and lone pair electrons of water molecules, respectively. (b) Snapshots of water molecules in heterogeneous finite length slit pores for width of 16 Å. Symbols are the same as those in Figure 4a.

1. An early onset of adsorption is observed for heterogeneous pores, and this is due to the favorable adsorption of water molecules around functional groups via hydrogen bonding. 2. The isotherms of heterogeneous pores are greater than those of homogeneous pores. 3. The relative pressure at which the pore filling occurs in heterogeneous pores is lower, and it occurs at a pressure below the saturation pressure (P < P0). This is consistent with what is observed experimentally.21-24 This is due to the greater interaction energy between fluid and functional group that leads to the favorable nucleation of water clusters. 4. The hysteresis loop is smaller for heterogeneous pores, and this is due to the early onset of adsorption as mentioned in point 3 above. Next, we discuss the effects of the position of functional group on adsorption isotherm. As seen in Figure 3, the adsorption isotherm for the center topology model is greater than that for the corner topology model at low relative pressures. The reason

is that water molecules which directly formed hydrogen bonds to the functional group and become nucleation centers for other water molecules to form aggregates are surrounded by more carbon atoms of graphene layers in the case of center topology. For example, in the case of 20 Å pore, the interaction energy between a water molecule and carbon atoms of the surface (φsf ) -4.7 kJ/mol) contributes a fair fraction to the total interaction potential (φ ) -62 kJ/mol), while in the case of corner topology, this interaction is only -2.4 kJ/mol. After the water nucleation around the functional group has been completed, the total energy of the system depends mainly on the fluid-fluid interaction energy. The effect of the location of active sites on the adsorption of water is also observed in the literature.26,31-32 The adsorption isotherms for corner topology finite pores exhibit vertical pore filling; however, those for center topology model show a gradual increase in the adsorbed amount that is quite similar to the experimental results.21-24

Adsorption of Water in Finite Length Pore

J. Phys. Chem. B, Vol. 111, No. 50, 2007 13953

Figure 5. Snapshots of water molecules in heterogeneous finite pore for width of 16 Å obtained by using canonical ensemble (NVT). Symbols are the same as those in Figure 4a.

Figure 6. The adsorption isotherms of water in 16 Å pores with two functional groups (square symbols; filled symbols for adsorption and unfilled symbols for desorption) and six functional groups (triangle symbols). (a) Corner topology and (b) center topology.

Figure 4 shows the snapshots of water molecules in heterogeneous finite pores (10 Å pore in Figure 4a and 16 Å in Figure 4b). For clarity, we show only one graphene layer of each wall. We observe that the adsorption behavior of water is not the same as that for nonpolar molecules such as argon for which the adsorbed phase is started by forming the two contact layers adjacent to the two walls at low pressure.12,13 In the case of water adsorption, a few water molecules are nucleated around the functional group. Once nucleated, water cluster grows in all directions but predominantly in the direction perpendicular to the pore wall. This is due to the tetrahedral structure of the water molecule of which one lone pair of electrons forms a hydrogen bond with the functional group. This makes two hydrogen atoms and the remaining lone pair of electrons of that water molecule pointing downward, and they are available for the other water molecules to form a hydrogen bond with.32 This argument is supported by the snapshots obtained from the canonical ensemble (NVT) simulation of adsorption in 16 Å pore with the corner topology, and the results are shown in Figure 5. 3.3. Effects of the Concentration of the Functional Group. We have shown the effects of the functional group and its topology on the adsorption isotherm. Now, we turn to the investigation of the effects of its concentration on the isotherm. The concentration of the functional group is either two or six molecules on one graphene layer of the upper wall, and the separation distance (δ) between two functional groups is 4.26 Å. In the case of six functional groups, they are located in a circular arrangement with one group at the center of a circle of radius 4.26 Å, and the other five are placed on the periphery of the circle. We show the adsorption isotherms of water in heterogeneous 16 Å pores having the corner and center

topologies in Figure 6a and 6b, respectively, and those for 20 Å pores in Figure 7a and 7b. The common characteristics of these isotherms are observed with different concentrations of functional group. 1. In the low relative pressure region, the amount adsorbed increases with an increase in the concentration of the functional group. This is due to the more available strong energy sites for water molecules to interact with. This simulation result agrees with the experimental observations of Morimoto and Miura33-36 which show that the reduced amount of surface groups leads to a drastic decrease of the amount of adsorbed water. 2. An increase in the functional group concentration leads to a marginally early onset of pore filling. This could be because the pore filling is governed dominantly by the fluid-fluid interaction. 3.4. Effects of the Separation Distance of the Functional Group. Next, we present the effects of the separation distance of functional group on the adsorption isotherms. A series of simulations was carried out for both corner and center topologies with various separation distances of 1.42, 4.26, and 7.10 Å. The number of functional groups is kept constant at six. The results are shown in Figures 8 and 9 for 16 Å and 20 Å pores, respectively. The following conclusions can be drawn. 1. The adsorption isotherm increases with increasing separation distance in the region prior to the pore filling. In the case of separation distance of 1.42 Å, this distance is shorter than the collision diameter of the dispersive site (oxygen atom), and therefore one water molecule adsorbs and covers more than one hydroxyl group. In the case of greater separation distances, each functional group can be adsorbed by one water molecule. However, for the separation distance of 7.1 Å which is about

13954 J. Phys. Chem. B, Vol. 111, No. 50, 2007

Wongkoblap and Do

Figure 7. The adsorption isotherms of water in 20 Å pore for the heterogeneous pores. (a) Corner topology and (b) center topology. Symbols are the same as those in Figure 6.

Figure 8. The adsorption isotherms of water in pore width of 16 Å for the heterogeneous pore with six functional groups and the separated distance of 1.42 Å (triangle symbols; filled symbols for adsorption and unfilled symbols for desorption), that of 4.26 Å (square symbols), and that of 7.10 Å (diamond symbols). (a) Corner topology and (b) center topology.

Figure 9. The adsorption isotherms of water in pore width of 20 Å. Symbols are the same as Figure 8. (a) Corner topology and (b) center topology.

twice the collision diameter of oxygen atom, the distance between the two hydroxyl groups is such that additional water molecules can form hydrogen bonds with those water molecules that bind directly on these functional groups. 2. The pore filling pressure is lower for larger separation distance. This is due to the same reason that we mentioned in part 1 above. 3.5. Comparison between Data and Simulation Results for Water Adsorption in Activated Carbon. The experimental data for water adsorption in Ajax activated carbon type 976 extruded (Norit RB; Norit, Amersfoort, The Netherlands) was

obtained at 298 K (shown in Figure 10) by using an intelligent gravimetric analyzer (IGA) model IGA-002 supplied by Hiden Analytical Ltd., United Kingdom. Prior to adsorption experiments, the carbon sample (0.12 g) was outgassed at 200 °C for 10 h and then was allowed to cool to the adsorption temperature. The experimental isotherm shows a gradual increase in slope at relative pressures (P/P0) lower than 0.5, after which the isotherm increases dramatically, and then it increases with a decreasing rate. The behavior of the isotherm near P/P0 ) 0.5 is also observed in Kaneko et al.37 for water adsorption in pitchbased activated carbon fiber (ACF) which is considered to be

Adsorption of Water in Finite Length Pore

J. Phys. Chem. B, Vol. 111, No. 50, 2007 13955 mental data is quite satisfactory, although there are some underestimations at low and high relative pressure regions. The lower theoretical isotherm in these regions could be because the simulated isotherms for pores having widths smaller than 10 Å and greater than 20 Å are not included in this study. 4. Conclusions

Figure 10. The experimental isotherm of water in activated carbon at 298 K.

Figure 11. The experimental isotherm of water in activated carbon at 298 K (solid line) and the combined isotherm at the same temperature (dotted line with circle symbols).

uniform in pore structure. Four samples of ACF were used in Kaneko et al. having widths varying from 6.9 to 14.5 Å, which are typical micropore sizes for activated carbons. This range of pore size is in accord with that determined for Ajax activated carbon type 976 extruded which has broad bimodal peaks at 10 and 19 Å,38 and the physical properties of this activated carbon were reported by Gray39 with a micropore volume of 0.40 cm3 micropore/cm3 particle or 0.61 cm3/g (pore width H e 50 Å), a macropore volume of 0.31 cm3 macropore/cm3 particle (H > 50 Å), Brunauer-Emmett-Teller (BET) (nitrogen) surface area of 1200 m2/g, and particle bulk density of 0.66 g/cm3. The surface functional groups were determined using Boehm40 titration, and it is reported in the literature6 that the surface site density for Norit activated carbon is 0.675 sites/ nm2. In this study, we choose the combination of simulation results for three pore widths of 10, 16, and 20 Å with center topology to fit the water adsorption in Ajax activated carbon. These widths represent the micropore commonly found in activated carbons. The choice of center topology in preference to the corner topology is because it shows a gradual increase in adsorption isotherm which is similar to the experimental data. Six hydroxyl groups are placed on one pore wall with a separation distance of 4.26 Å, and this amount represents the surface site density of 0.405 sites/nm2. The combined adsorption isotherm of these pores (10, 16, and 20 Å) is shown in Figure 11 as circle symbols with dashed line. Also shown in the same figure is the experimental data as solid line. The matching between the simulation results and the data yields the specific pore volume of 10, 16, and 20 Å pores as 0.11, 0.05, and 0.27 cm3/g, respectively. The total specific volume of 0.43 cm3/g is close to the micropore volume of 0.472 cm3/g found in the literature.38 The agreement between the combined isotherm for the heterogeneous pores with center topology and the experi-

In this paper, we have presented the adsorption of water in carbon slit pores of finite and infinite lengths and have considered the effects of the functional group on the adsorption isotherm using a grand canonical Monte Carlo (GCMC) simulation. In this study, the Muller et al. water model is used while the functional group is assumed to be a hydroxyl (OH) group. Two solid models, the corner and the center topologies, are used to study the effects of the position of functional group on the adsorption of water. We have observed the isotherm shape of type V and the large hysteresis for both homogeneous and heterogeneous finite pores. In comparing the isotherms obtained for heterogeneous pores with those for homogeneous pores, the functional group shows significant effects on the adsorption isotherm. We observe a greater adsorbed amount at low relative pressures and a smaller hysteresis loop. The adsorption isotherm obtained for the heterogeneous pore with center topology shows similar behavior to the experimental data. Agreement between the theoretical isotherm and the experimental data of water adsorption on Ajax activated carbon at 298 K is found to be satisfactory. Acknowledgment. This research is supported by the Australian Research Council. We also acknowledge the Royal Thai Government for financial support in the form of scholarship to A.W. We thank Associated Professor Tangsathitkulchai C., Junpirom S., and Suranaree University of Technology for supporting the experimental work for IGA in Thailand. References and Notes (1) Frenkel, D.; Smit, B. Understanding Molecular Simulation; Academic Press: New York, 2002. (2) Allen, M. P.; Tildesley, D. J. Computer Simulation of Liquids; Oxford Science Publications: New York, 1987. (3) Do, D. D.; Do, H. D. Langmuir 2003, 19, 8302. (4) Nguyen, C.; Do, D. D. Langmuir 1999, 15, 3608. (5) Muller, E. A.; Rull, L. F.; Vega, L. F.; Gubbins, K. E. J. Phys. Chem. 1996, 100, 1189. (6) McCallum, C. L.; Bandosz, T. J.; McGrother, S. C.; Muller, E. A.; Gubbins, K. E. Langmuir 1999, 15, 533. (7) Striolo, A.; Chialvo, A. A.; Cummings, P. T.; Gubbins, K. E. Langmuir 2003, 19, 8583. (8) Striolo, A.; Gubbins, K. E.; Chialvo, A. A.; Cummings, P. T. Mol. Phys. 2004, 102, 243. (9) Maddox, M.; Ulberg, D.; Gubbins, K. E. Fluid Phase Equilib. 1995, 104, 145. (10) Franklin, R. E. R. Soc. London 1951, 209, 196. (11) Do, D. D. Adsorption Analysis: Equilibria and Kinetics; Imperial College Press: New Jersey, 1998. (12) Wongkoblap, A.; Junpirom, S.; Do, D. D. Adsorpt. Sci. Technol. J. 2005, 23, 1. (13) Wongkoblap, A.; Do, D. D. J. Colloid Interface Sci. 2006, 297, 1. (14) Berendsen, H. J. C.; Postma, J. P. M.; van Gunsteren, W. F.; Hermans, J. In Intermolecular Forces; Pullman, B., Ed.; Reidel: Dordrecht, Netherlands, 1981; p 331. (15) Berendsen, H. J. C.; Grigera, J. R.; Straatsma, T. P. J. Phys. Chem. 1987, 79, 926. (16) Stillinger, F. H.; Rahman, A. J. Chem. Phys. 1974, 60, 1545. (17) Nezbeda, I.; Slovak, J. Mol. Phys. 1997, 90, 353. (18) Nezbeda, I. J. Mol. Liq. 1997, 73-74, 317. (19) Predota, M.; Nezbeda, I. Mol. Phys. 1999, 96, 1237. (20) Predota, M.; Nezbeda, I.; Cummings, P. T. Mol. Phys. 2002, 100, 2189. (21) Mahle, J. J.; Friday, D. K. Carbon 1989, 27, 835. (22) Kimura, T.; Kanoh, H.; Kanda, T.; Ohkubo, T.; Hattori, Y.; Higaonna, Y.; Denoyel, R.; Kaneko, K. J. Phys. Chem. B 2004, 108, 14043.

13956 J. Phys. Chem. B, Vol. 111, No. 50, 2007 (23) Ohba, T.; Kanoh, H.; Kaneko, K. J. Phys. Chem. B 2004, 108, 14964. (24) Vartapetyan, R. Sh.; Voloshchuk, A. M.; Buryak, A. K.; Artarnonova, C. D.; Belford, R. L.; Ceroke, P. J.; Kholine, D. V.; Clarkson, R. B.; Odintsov, B. M. Carbon 2005, 43, 2152. (25) Liu, J. C.; Monson, P. A. Langmuir 2005, 21, 10219. (26) Birkett, G. R.; Do, D. D. Mol. Phys. 2006, 104, 623. (27) Muller, E. A.; Gubbins, K. E. Ind. Eng. Chem. Res. 1995, 34, 3662. (28) Skalny, J.; Bodor, E. E.; Brunauer, S. J. Colloid Interface Sci. 1971, 37, 476. (29) Muller, E. A.; Hung, F. R.; Gubbins, K. E. Langmuir 2000, 16, 5418. (30) Brunauer, S.; Deming, L. S.; Deming, W. E.; Teller, E. J. J. Am. Chem. Soc. 1940, 62, 1723. (31) Picaud, S.; Hoang, P. N. M.; Hamad, S.; Mejias, J. A.; Lago, S. J. Phys. Chem. B 2004, 108, 5410.

Wongkoblap and Do (32) Wongkoblap, A.; Do, D. D. International Conference on Modeling in Chemical and Biological Engineering Sciences Proceeding [CD-ROM]; Bangkok, 2006. (33) Morimoto, T.; Miura, K. Langmuir 1986, 2, 43. (34) Miura, K.; Morimoto, T. Langmuir 1986, 2, 824. (35) Miura, K.; Morimoto, T. Langmuir 1991, 7, 374. (36) Miura, K.; Morimoto, T. Langmuir 1994, 10, 807. (37) Kaneko, K.; Hanzawa, Y.; Iiyama, T.; Kanda, T.; Suzuki, T. Adsorption 1999, 5, 7. (38) Birkett, G. R.; Do, D. D. Langmuir 2006, 22, 7622. (39) Gray, P. G. Fundamental Studies of Sorption Dynamics of SO2, NO2 and CO2 on Activated Carbon. Ph.D. Thesis, University of Queensland, Australia, 1991. (40) Boehm, H. P. In AdVances in Catalysis; Academic Press: New York, 1966; Vol. 16.