You are on page 1of 4

JOURNAL OF CHEMICAL PHYSICS

VOLUME 109, NUMBER 12

22 SEPTEMBER 1998

Monte Carlo simulations of hydrogen adsorption in single-walled carbon nanotubes


Farida Darkrima)
LIMHP Universite Paris XIII, Avenue Jean Baptiste Clement, 93430 Villetaneuse, France

Dominique Levesqueb)
LPTHE Universite Paris XI, Batiment 211 91405 Orsay, France

Received 17 October 1997; accepted 16 June 1998 Within the framework of a study on the properties of carbon nanotubes, a promising new material, we performed numerical simulation of hydrogen adsorption at room temperature in single-walled nanotubes. The structure of this material is favorable to the adsorption phenomenon because of the narrow size distribution of the nanotube diameters, which have dimensions on the order of the range of the carbon attractive interaction. We discuss the inuence of the single-walled carbon nanotube diameters on the relative arrangement of carbon atoms and hydrogen molecules within an array of parallel single-walled carbon nanotubes. We also studied the inuence on adsorption of the distance between the nearest-neighbor nanotubes. 1998 American Institute of Physics. S0021-9606 98 50636-8
I. INTRODUCTION

The carbon nanotubes discovered by Iijima in 1991 and synthesized by arc discharge evaporation can be obtained with a wall made up of a single rolled graphite basal plane and a denite diameter.2,3 Ge et al.4 described the synthesis of these giant fullerenes. Nanotubes are studied primarily because of their high conductivity and their optical and specic properties sensitive to their particular structures.59 Furthermore, under dened thermodynamic conditions, nanotubes can occasionally draw up liquids by capillarity.10,11 Taking into account the microporous nature of these carbon macromolecules, Gubbins et al.1214 performed Monte Carlo simulations of gas adsorption in cylindrical pores which represent the nanotube structure. They concluded that nanotube materials have an advantageous adsorptive capacity, which can be used for separation of gases. Concerning hydrogen adsorption on nanotubes at room temperature, Dillon et al.15 predicted from their measurements that the possible number of hydrogen molecules inside a nanotube would be maximal for a diameter equal to 2 nm. In the present study, we calculated, by numerical Monte Carlo simulations, the hydrogen adsorption on an array of singlewalled nanotubes SWNTs of different diameters separated by various distances Sec. II . The adsorption takes place both inside and outside the nanotubes. Nanotubes have a very narrow diameter distribution, and this characteristic, which might be favorable to hydrogen adsorption if such a distribution is adequately chosen with respect to the hydrogen molecule diameter and the carbon attractive force range, is discussed in Sec. III.
II. ADSORPTION SIMULATIONS ON SWNTs

We computed hydrogen adsorption by grand canonical Monte Carlo simulations. This standard Monte Carlo method
a b

Electronic mail: darkrim@limhp6.univ-paris13.fr Electronic mail: dl@stat.th.u-psud.fr 4981

of simulation and its use for the study of adsorption are well documented in the literature and are described in Refs. 16 and 17. In the simulations, interactions between hydrogen molecules result from a Lennard-Jones potential located at the center of mass of the molecules and a quadrupolar interaction. The hydrogen quadrupole is described by three charges: two charges q located on the protons, distant from 0.0741 nm, and one charge 2q located at the center of mass (q 0.615 10 26 esu . This intermolecular interaction model has been used successfully in previous simulations.18 The cross interaction between hydrogen molecules and carbon atoms of the nanotube wall is a Lennard-Jones potential obtained by using Berthelot rules which presume that carbon atoms interact by a ctitious Lennard-Jones potential following a procedure already widely used in simulations.1719 The parameters of the carbon Lennard-Jones interactions are 0.34 and 28.2 K, respectively. The parameters of the Lennard0.2958 nm and 36.7 K are Jones hydrogen interaction used as units of length and energy to dene reduced units of 3 . In the simulatemperature T * kT/ and density * tions realized in this work, the chemical potential expressed in reduced units was kept equal to 2.7, and was chosen to give, at T 293 K (T * 7.988), a reduced bulk density equal to * 0.06 when simulation was performed for a simulation cell without adsorbent material. This density corresponded to a pressure of 10 MPa. In this thermodynamic state, the thermal wavelength of hydrogen molecules h/ 2 mkT (h is the Planck constant, m the mass of the hydrogen molecule, and k the Boltzmann constant was equal to 0.07 nm. The ratios between and or between a typical distance between hydrogen molecules in the bulk phase and were in the range of values 0.070.2. These values were such that the quantum effects could modify our estimates by classical simulations of the thermodynamic properties, such as adsorption, by a few percent, since the magnitude of quantum corrections, for instance on free energy, was ( / ) 2 . It did not seem useful to
1998 American Institute of Physics

0021-9606/98/109(12)/4981/4/$15.00

Downloaded 07 Jun 2011 to 203.90.91.225. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions

4982

J. Chem. Phys., Vol. 109, No. 12, 22 September 1998


N conf i N ads i 1

F. Darkrim and D. Levesque

N eff

N conf

* V* , bulk a

FIG. 1. Typical arrangement of nanotubes in the simulation cell. The axes of the nanotubes are parallel to the y direction and their centers are placed at sites of a square lattice in the x, z plane. The distance between the walls of nanotubes is equal to d 0.334 nm. This latter distance corresponds to the minimal approach distance between two graphite planes. The length of the cell sides is equal to 4.01613, 3.4076, and 4.0613 nm along the x, y, and z directions, respectively.

where N iads is the number of hydrogen molecules present in the simulation cell, which contains SWNTs at the ith conguration generated by the numerical simulation. The average value of N iads is computed from the N conf congurations generated by the simulation. * V * is the number of hydrobulk a gen molecules present at the reduced bulk density * in the bulk empty volume of the simulation cell V * . Then, N eff is the a difference between the average number of hydrogenadsorbed molecules in V * lled with SWNTs and the numa ber of hydrogen molecules in the same volume lled by the gas at the bulk density * . bulk In the simulations, the Lennard-Jones interactions were truncated at a distance r C 5.2 . Beyond this distance, the Lennard-Jones potential was smaller than 2 10 4 . The length of the sides of the parallelepipedic volume V a was 11 and larger than 2r C . Hence, all interactions between hydrogen molecules and carbon atoms in the range of the interactions were considered and no appreciable nite-size effect on our simulation result was expected by increasing the size of the V a .
III. RESULTS

attempt to take into account these corrections in the present work, the main purpose of which was to give an estimate of hydrogen adsorption in a simple model of materials made by SWNTs. The adsorbent material was formed by an array of nanotubes which was inside the parallelepipedic volume of the simulation cell having periodic boundary conditions in the x, y, and z directions. The typical arrangement of these nanotubes is given in Fig. 1. The axes of the nanotubes were parallel to the y direction and their centers were located at the sites of a square lattice in the x, z plane. The nanotubes were constructed by rolling up a basal plane of graphite in order to form a cylindrical wall where parallel bonds between carbon atoms were parallel to the cylinder axis cf. Fig. 1 . In this conguration without chirality, we presumed that the Lennard-Jones potential previously designed for taking into account the interaction between hydrogen and graphite basal planes remained valid, considering that the curvature of the carbon hexagonal unit has a negligible effect on the effective solidgas potential. In this work, both the diameters of SWNTs and the distances between their walls could vary. We considered the cases of nanotubes with diameters equal to 0.704, 1.018, 1.174, 1.331, 1.644, and 1.957 nm arranged in an order such that the smallest distance d between the walls was equal to 0.334 nm. We next considered cases of nanotubes with a diameter equal to 1.174 nm in an arrangement such that the values of the minimal distances between the walls varied from 0.5 to 0.9 nm. We considered as adsorbed the excess average number N eff of hydrogen molecules due to the presence of the absorbent in the simulation cell.20 N eff is dened by

Our theoretical results are presented in Table I. We report i the value of N eff for different SWNT diameters (D) at constant distance between the walls d 0.334 nm and ii the values of N eff on SWNTs with denite diameter (D 1.174 nm and different distances between the walls d. We stress that the value of N eff takes into account the adsorption of hydrogen molecules inside and outside the nanotube walls, because in the grand canonical sampling of the congurations, hydrogen molecules are inserted or deleted at random in the whole volume of the simulation cell. Each grand canonical Monte Carlo simulation corresponded to a rst run of 1.5 106 elementary Monte Carlo moves displacement, insertion, and deletion of molecules to reach an equilibrium conguration. Then, the average value of N eff was computed by runs of 8 106 Monte Carlo moves. In these runs, because each type of elementary move was performed with equal probability 1/3, 2.6 106 attempts to insert or to delete one molecule were made and about 10%50% following the SWNT congurations were successful. The statistical uncertainty on N eff could be evaluated as equal to 1% by the safe estimate that the number of uncorrelated values of N eff obtained during one Monte Carlo run was 104 . Adsorption varied with the diameter nanotubes and the distance between the tube walls. In our simulations, the nanotube diameter corresponding to the maximal adsorbed hydrogen amount was equal to 1.174 nm for a minimal distance between nanotube walls of 0.7 nm. The distribution of hydrogen molecules in this case is illustrated in Fig. 2. In Fig. 2 are drawn the isodensity lines of hydrogen molecules in a zx plane perpendicular to the axis of the nanotubes. Clearly, the molecules are packed near the internal and external surfaces of the nanotube walls. We notice the low

Downloaded 07 Jun 2011 to 203.90.91.225. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions

J. Chem. Phys., Vol. 109, No. 12, 22 September 1998

F. Darkrim and D. Levesque

4983

TABLE I. Theoretical results obtained with single-walled nanotubes of diameters (D) placed in the volume of the simulation cell (V a ). The distance between the carbon walls is equal to (d). The reduced temperature is 7.988 and the value of the reduced chemical potential is equal to 2.7, corresponding to a reduced bulk density * bulk 0.06. N carbon is the number of carbon atoms on the circumference of the nanotubes. N bulk is the number of hydrogen molecules present in the simulation cell without the adsorbent at a reduced bulk density of * * bulk 0.06, and is given by the relation N bulk V a bulk . The total number of hydrogen molecules present in the simulation cell which contains the adsorbent is N ads . The efciency of adsorption E ff is given by (N eff /N bulk)100. The error on E ff is about 1%2%. D nm 0.704 1.018 1.174 1.331 1.644 1.957 1.174 1.174 d nm 0.334 0.334 0.334 0.334 0.334 0.334 0.5 0.7

N carbon 9 13 15 17 21 25 15 15

V* a 1293.142 2184.686 2717.621 3308.663 4665.072 6253.913 3322.232 4163.312

* bulk
0.06 0.06 0.06 0.06 0.06 0.06 0.06 0.06

N ads 90 152 194 234 329 437 245 342

N bulk 77.6 131.1 163.1 198.5 279.9 375.2 199.3 249.8

E ff 15% 15% 19% 16% 16% 16% 23% 37%

density of the molecules near the axis of the nanotubes and in the region most distant from the external surfaces of the nanotubes. The comparison between the range of the crossinteraction carbonhydrogen, the hydrogen molecule size, and the SWNT diameter could qualitatively explain these results by considering that repulsive interactions on the two sides of the nanotube wall prevented hydrogen molecules from approaching the carbon wall within a distance of less than 0.3 nm. Due to the repulsive interactions between hydrogen molecules, their smallest approach distance was about 0.3 nm. From these steric effects, we concluded that inside a nanotube of 1 nm along a diameter perpendicular to the axis, it was possible to place two hydrogen molecules. When the diameter of the tube was smaller than 1 nm, it was only possible to place one hydrogen molecule; when the SWNT diameter was larger than 1 nm for instance 1.957 nm , the central space along the axis of the cylinder would

be lled by a gas having a density close to the bulk density * 0.06 corresponding to a pressure of 10 MPa at T 293 K . The approximate value of the diameter of the SWNT for which adsorption was maximal was in relative agreement with the diameter of 2 nm predicted by Dillon et al.15 Finally, we found that, for a SWNT of 1.174 nm diameter and a value of the distance between the walls of the tubes d equal to 0.7 nm, the adsorption effect on the SWNT increased by 37% the number of hydrogen molecules present in a given volume. Our observations on the adsorption of the hydrogen molecules in the conned volume inside the nanotubes were supported by the results of simulations for a periodic array of parallel basal graphite planes performed for an identical chemical potential and temperature. In the slits formed by the array of the graphite planes, we computed the density proles of the hydrogen molecules since the distance between the planes varied. These proles are presented in Fig. 3. Clearly, for the thermodynamic state of hydrogen considered in this work, they show that, in the central part of the volume between two graphite planes separated by a distance larger than 3 0.9 nm , the value of the density of hydrogen molecules is equal to that of the bulk density.
IV. CONCLUSION

FIG. 2. Isodensity lines of the density of hydrogen molecules inside a simulation cell containing an array of single-walled carbon nanotubes of diameter equal to 1.174 nm. The distance between the carbon walls is equal to d 0.7 nm. The bulk density is equal to 0.06. Dimensions of the simulation cell along x and z axes are given in reduced units.

By considering Monte Carlo simulation, and under a current thermodynamic state 10 MPa and 293 K , SWNTs seemed to be good adsorbents for hydrogen. This is mainly due to the favorable potential inside and outside the SWNT, which attracts gas molecules on each side of the nanotube with a minimal loss of volume. The decrease in adsorption as the SWNT diameter increases corresponds to the fact that a large part of the volume inside or outside the nanotube is out of range of the attractive forces of the solidgas interaction. For the largest diameters, the central part of the nanotube was lled with gas at the bulk density. We stress that our simulation results depended on our choice of intermolecular potentials between hydrogen molecules and carbon atoms, but such potentials seem to be a reasonable estimate of these

Downloaded 07 Jun 2011 to 203.90.91.225. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions

4984

J. Chem. Phys., Vol. 109, No. 12, 22 September 1998

F. Darkrim and D. Levesque

Our simulation results should be useful, through quantitative meaningful comparison with adsorption experiments of hydrogen on nanotubes, in showing this favorable change in the interaction between hydrogen molecules and nanotube walls.

FIG. 3. Density proles of hydrogen molecules inside a simulation cell, which contains an array of graphite planes, disposed perpendicularly to the z axis. The simulation cell has periodic boundary conditions in the three space directions. The length of the cell along z is 3.3479 nm. From bottom to top, the density proles * (z) correspond to the case of graphite planes spaced by 3.3479, 1.116, 0.6696, and 0.5580 nm. Taking into account the periodic boundary conditions, the proles are drawn only for a half cell. The simulations were performed at T * 7.988 and a pressure of 10 MPa corresponding to a bulk density of * 0.06. For this adsorbent model, the maximal efciency of adsorption cf. Table I was 80% for planes spaced at 0.6696 nm.

molecular interactions. In the near future, the possible synthesis of nanotubes with solidgas potential more favorable to adsorption can be envisaged. But, it is noteworthy that synthesis of the SWNT with dened diameters and distances between the walls is difcult to perform.

S. Iijima, Nature London 354, 56 1991 . E. Dujardin, T. W. Ebbesen, H. Hiura, and K. Tanigaki, Science 265, 1850 1994 . 3 P. M. Ajayan, O. Stephan, C. Colliex, and D. Trauth, Science 265, 1212 1994 . 4 M. Ge and K. Sattler, Science 260, 515 1993 . 5 S. Lijima and T. Ichihashi, Nature London 363, 603 1993 . 6 D. S. Bethune, C. H. Kiang, M. S. de Vries, G. Gorman, R. Savoy, J. Vasquez, and R. Beyers, Nature London 363, 605 1993 . 7 J. W. Mintmire, B. I. Dunlap, and C. T. White, Phys. Rev. Lett. 68, 631 1992 . 8 N. Hamada, S. Sawada, and A. Oshiyama, Phys. Rev. Lett. 68, 1579 1992 . 9 B. I. Yacabson and R. E. Smalley, Am. Sci. 85, 323 1997 . 10 M. R. Pederson and J. Q. Broughton, Phys. Rev. Lett. 69, 2689 1992 . 11 P. M. Ajayan and S. Lijima, Nature London 361, 333 1993 . 12 B. K. Peterson and K. E. Gubbins, Mol. Phys. 62, 215 1987 . 13 M. W. Maddox, S. L. Sowers, and K. E. Gubbins, Adsorption 2, 23 1996 . 14 M. W. Maddox and K. E. Gubbins, Langmuir 11, 3988 1995 . 15 A. C. Dillon, K. M. Jones, T. A. Bekkedahl, C. H. Kiang, D. S. Bethune, and M. J. Heben, Nature London 386, 377 1997 . 16 J. Vermesse and D. Levesque, Mol. Phys. 77, 837 1992 . 17 J. Vermesse and D. Levesque, J. Chem. Phys. 101, 9063 1994 . 18 D. Marx, O. Optiz, P. Nielaba, and K. Binder, Phys. Rev. Lett. 70, 2908 1993 ; D. Marx and P. Nielaba, Phys. Rev. A 45, 8968 1994 . 19 W. A. Steele, Surf. Sci. 36, 317 1973 . 20 A. S. Coolidge, J. Am. Chem. Soc. 56, 554 1934 .
2

Downloaded 07 Jun 2011 to 203.90.91.225. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions

You might also like