Vous êtes sur la page 1sur 9

MOLECULAR SIMULATION OF GAS ADSORPTION IN

REALISTIC MODELS OF SILICA NANOPORES

Benoit COASNEa,b, Francisco R. HUNGa, Flor R. SIPERSTEINc, Keith E. GUBBINSa

a
Department of Chemical and Biomolecular Engineering, North Carolina State University, Raleigh,
NC 27695-7905, USA.
b
Laboratoire de Physicochimie de la Matière Condensée (UMR CNRS 5617), Université de
Montpellier II, Place Eugène Bataillon, 34095 Montpellier, Cedex 5, France.
c
Departament d'Enginyeria Química, Universitat Rovira i Virgili, Campus Sescelades Av. dels Països
Catalans, 26, 43007 Tarragona, Spain.

Abstract – This paper presents a molecular simulation study of Xe adsorption at 195 K in two
atomistic models of silica MCM-41 pores. Model A consists of a regular cylindrical pore having a
constant section. Model B has an important surface disorder that reproduces the morphological features
of an on-lattice simulation mimicking the synthesis process of MCM-41 pores. The adsorption isotherm
for model A exhibits a large hysteresis loop that is typical of capillary condensation in regular
nanopores. In contrast, the adsorption/desorption process for model B is a quasi-continuous and quasi-
reversible mechanism. Due to the important surface roughness for model B, the isosteric heat of
adsorption for this sample is much larger than that for the regular cylindrical pore. Where possible,
comparison with experimental data is made.

Résumé – Simulation moléculaire de l’adsorption de gaz dans des modèles réalistes de nanopores
de silice. Cet article présente une étude par simulation moléculaire de l’adsorption de Xe à 195 K dans
deux modèles atomiques des pores de silice MCM-41. Le modèle A consiste en un pore cylindrique
régulier à section constante. Le modèle B présente un important désordre de surface qui reproduit les
aspects morphologiques d’un modèle obtenu par simulation sur réseau du processus de synthèse des
pores MCM-41. L’isotherme d’adsorption pour le modèle A comporte une large boucle d’hystérésis,
typique de la condensation capillaire dans des nanopores réguliers. Au contraire, le processus
d’adsorption/désorption pour le modèle B a une allure quasi continue et quasi réversible. A cause de
l’importante rugosité de surface du modèle B, la chaleur d’adsorption isostérique pour cet échantillon
est beaucoup plus grande que celle obtenue pour le pore cylindrique régulier. Dans la mesure du
possible, les résultats sont comparés à des données expérimentales.
______________________________________________________________________________________________
Reprints: Benoit Coasne, Laboratoire de Physiocochimie de la Matière Condensée, Université Montpellier II, place
Eugène Bataillon, 34095 Montpellier cedex 5, France.
1. INTRODUCTION

Mesoporous silica MCM-41 and SBA-15 are made up of a hexagonal array of cylindrical pores
having a narrow pore size distribution that can be varied from 1.5 up to 20 nm [1-3]. Adsorption of gas
in these porous solids has been extensively studied using both experiments and simulations (for a
review see Ref. [4]). This effort is relevant (i) to the promising uses of these materials as adsorbents or
catalytic supports and (ii) to the fundamental understanding of phase transitions in confined geometry.
Most simulation and theoretical works were performed for pores modeled as structureless pores having
a regular cylindrical geometry. Recent simulation studies have also investigated the behavior of pure
fluids and mixtures confined in atomistic pore models of MCM-41 or SBA-15 materials [5-11]. The
effect of surface roughness and morphological defects, such as constrictions, on gas adsorption and
capillary condensation was also recently addressed [10,11].

It is not clear whether the atomistic silica pore models used in the simulation studies mentioned above
are realistic representations of the morphology of MCM-41 pores. Recently, Siperstein and Gubbins
[12,13] reported lattice Monte Carlo simulations of surfactant – solvent – silica systems that were able
to reproduce the formation of hexagonal structures, resembling the arrangement of silica MCM-41
pores. We were able to prepare an atomistic silica mesopore that keeps the morphological features of
this on-lattice model. In this paper, we report Grand Canonical Monte Carlo (GCMC) simulations of Xe
adsorption at 195 K in this on-lattice model of MCM-41 pores. We discuss the effect of the shape of the
pore by comparing the adsorption isotherm and the isosteric heat curve with those obtained for an
atomistic, regular cylindrical nanopore having a similar pore diameter.

2. COMPUTATIONAL DETAILS

In this section, we briefly present (i) the method used to generate atomistic silica mesopores, and (ii)
the details of the GCMC simulation of Xe adsorption in the porous sample. Full details of this study can
be found elsewhere [14].

Siperstein and Gubbins [12,13] studied the phase behavior of surfactant – solvent – silica containing
systems using lattice Monte Carlo simulations. For a concentration ratio of surfactant/silica equal to
0.2, the authors found that the system forms a hexagonal phase mimicking the MCM-41 porous
structure. Assuming a lattice unit of 0.5 nm, the porous material was found to have an average pore
radius of about 3-4 nm and a pore length of 28 nm [12]. In order to generate an atomistic pore that
keeps the morphological features of the material obtained by Siperstein and Gubbins, we carved out of
an atomistic silica block the porous matrix that corresponds to the on-lattice structure. We mimic the
pore surface in a realistic way by removing the silicon atoms that are in an incomplete tetrahedral
environment and by saturating oxygen dangling bonds with hydrogen atoms (this procedure ensures the
electroneutrality of the sample). Following the previous work by Pellenq and Levitz [15], we relaxed
the porous structure by performing Monte Carlo simulations in the canonical NVT ensemble.
Interactions between atoms of the sample were calculated using the two-body BKS potential [16,17],
except for the OH group that was modeled using a Morse potential [18]. Figure 1 shows transversal and
plane views of the atomistic porous MCM-41 obtained using the above procedure. The pore size
distribution of the sample was estimated as the distribution of the distances between hydrogen atoms at
the pore surface and the center of the pore. The material generated using this procedure (model B) has
an average pore radius of 3.2 nm with a dispersion of +/- 1 nm. Pore model A was generated following
a similar procedure, but has a regular cylindrical shape with a constant section (radius R = 3.2 nm).
The GCMC technique is a stochastic method that simulates a system having a constant volume V (the
pore with the adsorbed phase), in equilibrium with an infinite fictitious reservoir of particles imposing
its chemical potential µ and its temperature T [19-21]. The absolute adsorption isotherm is given by the
ensemble average of the number of adsorbed atoms versus the pressure of the gas reservoir P (the latter
is obtained from the chemical potential according to the bulk equation of state for an ideal gas).
Interactions between Xe atoms and substrate atoms were calculated using the PN-TraZ potential as
reported for rare gas adsorption in zeolite [22] or in porous silica glass [15]. Values of the interaction
parameters as well as details of the intermolecular potential functions can be found in the previous work
by Pellenq and Levitz [15]. The Xe/Xe interactions were calculated using a Lennard-Jones potential
with εXe = 211 K, σXe = 0.41 nm, that correctly accounts for the bulk liquid properties [15].

Figure 1. Plane (left) and transversal (right) section views of the atomistic MCM-41
nanopore (model B). This configuration was obtained by carving out of an atomistic
silica block the on-lattice pore structure generated by Siperstein and Gubbins [12] (see
text). White and gray spheres are oxygen and silicon atoms, respectively. Black spheres
are hydrogen atoms that delimitate the pore surface.
3. RESULTS AND DISCUSSION

The Xe adsorption isotherm at 195 K for the atomistic MCM-41 pore model B is shown in Figure 2.
We also report the adsorption curve obtained for the cylindrical pore model A having the same radius
(R = 3.2 nm). Adsorbed amounts have been normalized to the total number of Xe atoms, N0, when the
pores are filled. The adsorption isotherm for the pore model A conforms to the typical behavior
observed for adsorption/condensation of gas in regular cylindrical pores [4,11,23]; the adsorption
isotherm increases continuously (multilayer adsorption) until P = 0.70 P0, where a jump in the adsorbed
amount occurs (capillary condensation). The evaporation process occurs at a pressure, P = 0.55 P0,
lower than the capillary condensation pressure, so that a large hysteresis loop is observed. In contrast,
the adsorption/desorption isotherm for model B exhibits different features; the condensation and
evaporation processes are quasi-continuous mechanisms that are composed of reversible paths and two
irreversible small jumps in the adsorbed amount (P ~ 0.47 P0 and P ~ 0.56 P0). We note that the
hysteresis loops that accompany these jumps are much smaller than that observed in the case of the
regular cylindrical pore. The experimental observation of such two-step filling mechanisms is very
difficult as the hysteresis loops are very narrow and the jumps in the adsorbed amounts very small. The
filling and emptying processes for model B are located at much smaller pressures than that for model A.
It is also found that, at a given pressure, the adsorbed amount prior to capillary condensation is much
larger for model B than for model A. This result is in agreement with our previous work, where it was
shown that the surface roughness leads to an increase of the adsorbed amount in the multilayer
adsorption regime [10,24].

1.0

0.8

0.6
N/N0

0.4

0.2

0.0
0.0 0.2 0.4 0.6 0.8 1.0

P/P0

Figure 2. Xe adsorption isotherms at 195 K in pore model A (circles), R = 3.2 nm, and in
pore model B (squares), R = 3.2 nm. Open and closed symbols are the adsorption and
desorption data, respectively. P0 is the bulk saturation pressure of the gas. Adsorbed
amounts have been normalized to the number of Xe atoms, N0, when the pore is
completely filled.

In order to interpret the filling of the MCM-41 pore model B, we show in Figure 3 density profiles
ρ(Z*) along the pore axis at pressures located slightly below the jumps observed in the adsorption
isotherm (P ~ 0.47 P0 and P ~ 0.56 P0). The density ρ(Z*) of the confined fluid has been reduced with
respect to the density when the pore is filled ρ0(Z*). For the same pressures, we also report in Figure 4
configurations of Xe atoms adsorbed in the pore. In both Figure 3 and Figure 4, the extremity located
at Z* = 30 is connected to the extremity Z* = -30, due to the use of periodic boundary condition along
the pore axis. We first discuss the results obtained at the lowest pressure P ~ 0.47 P0. The confined fluid
exhibits two regions of low density, ρ/ρ0 ~ 0.8. These two regions are located between Z* = -15 and 10
(middle of the x-axis) and between Z* = 20 and -25, respectively (left and right extremities of the x-
axis).

1.0

0.8

0.6

ρ/ρ0 0.4

0.2

0.0
-30 -20 -10 0 10 20 30
Z*

Figure 3. Density profiles along the pore axis of Xe adsorbed at 195 K in the MCM-41
pore model B (R = 3.2 nm). The density is reduced with respect to the density when the
pore is completely filled, ρ0 (P ~ P0). The grey and black thick lines correspond to P =
0.47 P0 and P = 0.56 P0, respectively. The thin straight line ρ = ρ0 is a guide for the eye
that indicates the density profile when the pore is filled.

Figure 4. Transversal section views of Xe adsorbed at 195 K in the MCM-41 pore model
(R = 3.2 nm) at P = 0.47 P0 (top) and P = 0.56 P0 (bottom). Black spheres are hydrogen atom
at the pore surface and grey spheres are Xe atoms. The black lines delimitate the regions
low density that are not filled (nano-bubbles).
Inspection of the snapshots shown in Figure 4 reveals that these regions correspond to nano-bubbles
that coexist with the liquid-like regions in the pore. Such a coexistence between low and high density
regions has been already reported for adsorption in chemically heterogeneous slit pores [25,26],
cylindrical pores with constrictions [11,23,27-29] and disordered porous materials [15,30-32]. When
the pressure increases slightly above P ~ 0.47 P0, the fluid condenses in the center part of the pore and
only the bubble located between Z* = 20 and -25 exists (see Figure 3 and Figure 4 for P ~ 0.56 P0).
The latter remains stable until its condensation at a pressure P ~ 0.56 P0 (see adsorption isotherm shown
in Figure 2).

To compare the results for both pore models in terms of isosteric heat of adsorption, Qst, we show in
Figure 5 the Qst curves versus the pore filling fraction N/N0. Both curves are typical of adsorption on
heterogeneous surfaces as they decrease down to a plateau value that is close to the heat of liquefaction
of bulk Xe, ~ 13 kJ/mol. We also report in Figure 5 the experimental data obtained by Burgess et al.
[33] for Xe adsorption at 195 K in a silica sample (Vycor porous glass) having an average pore
diameter of 4.5 nm. It is found that the isosteric heat of adsorption for model A matches better the
experiments than that for model B. At low pore filling fractions (N/N0 < 0.3), Qst for model B is
overestimated. In agreement with a previous work on adsorption in smooth and rough pores [15], this
result explains why the adsorbed amount at a given pressure prior to capillary condensation is larger for
model B than for model A (Figure 2).

40

30
Qst(kJ/mol)

20

10

0
0.0 0.2 0.4 0.6 0.8 1.0
N/N0

Figure 5. Isosteric heat of adsorption of Xe at 195 K as a function of the number of


adsorbed atoms N/N0: (circles) pore model A, R = 3.2 nm; (squares) pore model B, R =
3.2 nm. The black diamonds show the experimental data obtained by Burgess et al. for
silica glass Vycor having an average pore diameter of 4.5 nm (From Ref. [33]).

In order to understand the differences between the isosteric heats for the two atomistic pores, we
report in Figure 6 the wall – fluid and fluid – fluid contributions to the total isosteric heat of adsorption.
The fluid – fluid contribution is very similar for both pores over the entire range of filling fractions. In
contrast, it is found that the wall – fluid contribution for model B is significantly larger than that for
model A. This wall – fluid contribution, which explains the large isosteric heat of adsorption Qst found
in Figure 5, is due to the important surface roughness of the MCM-41 model B, compared to that of
model A. The comparison with the experimental data by Burgess et al. [33] suggests that the degree of
surface disorder for model B is somewhat too great (assuming that MCM-41 and Vycor samples have
similar surface roughness). On the other hand, the isosteric heat of adsorption for model A reproduces
reasonably the experimental curve. Further work including a direct comparison with experiments for
MCM-41 will allow us to discuss the ability of the silica nanopores presented in this paper to capture
the features of gas adsorption in these materials [14,34].

40

30

Qst(kJ/mol) 20

10

0
0.0 0.2 0.4 0.6 0.8 1.0

N/N0

Figure 6. Fluid – wall (open symbols) and fluid – fluid (closed symbols) contributions to
the isosteric heat of adsorption of Xe at 195 K in silica pores. Circles and squares are data
for MCM-41 models A and B, respectively.

4. CONCLUSION

In this paper, we report grand canonical Monte Carlo simulations of Xe adsorption at 195 K in
atomistic silica nanopores. Two models are considered: a regular cylindrical pore having a constant
section (model A) and a sample with an important surface disorder that reproduces the morphological
features of an on-lattice simulation mimicking the synthesis process of MCM-41 materials (model B)
[12,13]. The adsorption isotherm for model A exhibits a large hysteresis loop, as expected for capillary
condensation in regular cylindrical nanopores [4,11]. The surface roughness for model B has a strong
impact on gas adsorption in nanopores; the adsorption/desorption process is a quasi-continuous
mechanism made up of reversible and weakly irreversible paths. The isosteric heat of adsorption for
model B is significantly larger than for model A; the latter is in good agreement with the experimental
data obtained for a silica porous glass. This difference between the isosteric heat curves is due to the
wall – fluid interaction, which is more attractive for pore model B than that pore model A.

Acknowledgments - It is a pleasure to thank Dr. Roland J. –M. Pellenq for helpful discussions. We are
grateful to the National Science Foundation (grants no. CTS - 0211792 and INT – 0089696) and the
Petroleum Research Fund of the American Chemical Society for funding of this work. FRS thanks the
Ramón y Cajal program from the Ministerio de Ciencia y Tecnología and the support of the project
CTQ2004-03346/PPQ. This research was performed using supercomputing resources from the National
Partnership for Advanced Computational Infrastructure (NSF/NRAC grant MCA93SO11).
5. REFERENCES

[1] A. Corma, From microporous to mesoporous molecular sieve materials and their use in
catalysis, Chem. Rev. 97 (1997) 2373 – 2419.
[2] U. Ciesla and F. Schüth, Ordered mesoporous materials, Microp. Mesop. Mater. 27 (1999) 131
– 149.
[3] G. J. de A. A. Soler-Illia, C. Sanchez, B. Lebeau, and J. Patarin, Chemical strategies to design
textured materials: from microporous and mesoporous oxides to nanonetworks and hierarchical
structures, Chem. Rev. 102 (2002) 4093 – 4138.
[4] L. D. Gelb, K. E. Gubbins, R. Radhakrishnan, and M. Sliwinska-Bartkowiak, Phase separation
in confined systems, Rep. Prog. Phys. 62 (1999) 1573 – 1659.
[5] M. J. Bojan, A. V. Vernov, and W. A. Steele, Simulations studies of adsorption in rough-walled
cylindrical pores, Langmuir 8, 901 – 908 (1992).
[6] M. W. Maddox, J. P. Olivier, and K. E. Gubbins, Characterization of MCM-41 using molecular
simulation: heterogeneity effects, Langmuir 13 (1997) 1737 – 1745.
[7] L. D. Gelb, The ins and outs of capillary condensation in cylindrical pores, Mol. Phys. 100 (13)
(2002) 2049 – 2057.
[8] B. Coasne, A. Grosman, C. Ortega, and R. J. –M. Pellenq, Physisorption in nanopores of various
sizes and shapes: a grand canonical Monte Carlo simulation study, in Studies in Surface Science
and Catalysis 144; Rodriguez-Reinoso, F.; McEnaney, B.; Rouquerol, J.; Unger K., Eds.; Elsevier
Science (2002) 35 –42.
[9] Y. He and N. A. Seaton, Experimental and computer simulation studies of the adsorption of
ethane, carbon dioxide, and their binary mixtures in MCM-41, Langmuir 19 (2003) 10132 – 10138.
[10] B. Coasne and R. J. –M. Pellenq, Grand canonical Monte Carlo simulation of argon adsorption
at the surface of silica nanopores: effect of pore size, pore morphology, and surface roughness, J.
Chem. Phys. 120 (2004) 2913 – 2922.
[11] B. Coasne and R. J. –M. Pellenq, A grand canonical Monte Carlo study of capillary
condensation in mesoporous media: effect of the pore morphology and topology, J. Chem. Phys.
121 (2004) 3767 – 3774.
[12] F. R. Siperstein and K. E. Gubbins, Synthesis and characterization of templated mesoporous
materials using molecular simulation, Mol. Sim. 27 (5-6) (2001) 339 – 352.
[13] F. R. Siperstein and K. E. Gubbins, Phase separation and liquid crystal self-assembly in
surfactant-inorganic-solvent systems, Langmuir 19 (6) (2003) 2049 – 2057.
[14] B. Coasne, F. R. Hung, R. J. –M. Pellenq, F. R. Siperstein, and K. E. Gubbins, Gas adsorption
in realistic models of silica mesopores, Langmuir, to be submitted (2005).
[15] R. J. –M. Pellenq and P. E. Levitz, Capillary condensation in a disordered mesoporous
medium: a grand canonical Monte Carlo study, Mol. Phys. 100 (13) (2002) 2059 – 2077.
[16] B. W. H. van Beest, G. J. Kramer, and R. A. van Santen, Force fields for silicas and
aluminophosphates based on ab-initio calculations, Phys. Rev. Lett. 64 (16) (1990) 1955 – 1958.
[17] J. D. Gale, Empirical potential derivation for ionic materials, for Phil. Mag. B 93 (1) (1996) 3 –
19.
[18] N. H. de Leeuw, F. Manon Higgins, and S. C. Parker, Modeling the surface structure and
stability of α-quartz, J. Phys. Chem. B 103 (1999) 1270 – 1277.
[19] D. Nicholson and N. G. Parsonage, Computer Simulation and the Statistical Mechanics of
Adsorption; Academic Press: New York (1982).
[20] M. P. Allen and D. J. Tildesley, Computer Simulation of Liquids; Oxford: Clarendon (1987).
[21] D. Frenkel and B. Smit, Understanding Molecular Simulation: From Algorithms to
Applications, 2nd Ed.; Academic Press: London (2002).
[22] R. J. –M. Pellenq and D. Nicholson, Intermolecular potential function for the physical
adsorption of rare gases in silicalite, J. Phys. Chem. 98 (1994) 13339 – 13349.
[23] B. Coasne, K. E. Gubbins, and R. J. –M. Pellenq, A grand canonical Monte Carlo study of
adsorption and capillary phenomena in nanopores of various morphologies and topologies:
Testing the BET and BJH characterization methods, Part. Part. Syst. Char. 21 (2004) 149 – 160.
[24] R. J. M. Pellenq, B. Coasne, and P. E. Levitz, Adsorption/condensation of xenon in mesopores
having a microporous texture or a surface roughness, in Adsorption and transport at the nanoscale,
N. Quirke and D. Nicholson, Eds.; Taylor and Francis, London (2004).
[25] H. Bock and M. Schoen, Phase behavior of a simple fluid confined between chemically
corrugated substrates, Phys. Rev. E 59 (1999) 4122 – 4136.
[26] H. Bock, D. J. Diestler, and M. Schoen, Phase behaviour of fluids confined between
chemically decorated substrates, J. Phys.: Condens. Matter 13 (2001) 4697 – 4714.
[27] L. Sarkisov and P. A. Monson, Modeling of adsorption and desorption in pores of simple
geometry using molecular dynamics, Langmuir 17 (2001) 7600 – 7604.
[28] A. Vishnyakov and A. V. Neimark, Monte Carlo simulation test of pore blocking effects,
Langmuir 19 (2003) 3240 – 3247.
[29] B. Libby and P. A. Monson, Adsorption/desorption hysteresis in inkbottle pores: a density
functional theory and Monte Carlo simulation study, Langmuir 20 (2004) 4289 – 4294.
[30] L. Gelb and K. E. Gubbins, Correlation functions of adsorbed fluids in porous glass: a
computer simulation study, Mol. Phys. 96 (1999) 1795 – 1804.
[31] F. Detcheverry, E. Kierlik, M. L. Rosinberg, and G. Tarjus, Local mean-field study of capillary
condensation in silica aerogels, Phys. Rev. E 68 (2003) 061504.
[32] F. Detcheverry, E. Kierlik, M. L. Rosinberg, and G. Tarjus, Mechanisms for gas adsorption and
desorption in silica aerogels: the effect of temperature, Langmuir 20 (2004) 8006 – 8014.
[33] C. G. V. Burgess, D. H. Everett, and S. Nuttall, Adsorption of CO2 and xenon by porous glass
over a wide range of temperature and pressure: applicability of the Langmuir case VI equation,
Langmuir 6 (1990) 1734 – 1738.
[34] F. R. Hung, B. Coasne, K. E. Gubbins, F. R. Siperstein, and M. Sliwinska-Bartkowiak, in
Studies in Surface Science and Catalysis, to be submitted (2005).

Vous aimerez peut-être aussi