5598
J. Phys. Chem. C 2008, 112, 5598-5604
Silicon Nanotube as a Promising Candidate for Hydrogen Storage: From the First Principle Calculations to Grand Canonical Monte Carlo Simulations Jianhui Lan, Daojian Cheng, Dapeng Cao,* and Wenchuan Wang DiVision of Molecular and Materials Simulation, Key Lab for Nanomaterials, Ministry of Education, Beijing UniVersity of Chemical Technology, People’s Republic of China ReceiVed: December 14, 2007; In Final Form: January 21, 2008
We employ a multiscale theoretical method, which combines the first-principle calculation and a grand canonical Monte Carlo (GCMC) simulation, to predict the adsorption capacity of hydrogen in silicon nanotube (SiNT) arrays at T ) 298 K in the pressure range from 1 to 10 MPa. In the multiscale method, the binding energy obtained from the first-principle calculation is used as an input in the GCMC simulation. It is found from the first-principle calculation that the SiNT arrays exhibit much stronger attraction to hydrogen both inside and outside SiNTs, compared to the isodiameter carbon nanotubes (CNTs). The subsequent GCMC simulations indicate that the SiNT arrays present distinct improvements of 106%, 65%, and 52% in the gravimetric adsorption capacity of hydrogen at P ) 2, 6, and 10 MPa, respectively, compared to the isodiameter CNTs. This suggests that the SiNT array is a promising candidate for hydrogen storage.
1. Introduction In the past 10 years, many efforts have been focused on the hydrogen adsorption capacity of carbon nanotubes (CNTs) due to potential application of hydrogen for fuel cell vehicles.1-6 However, most of these efforts failed to reach the gravimetric density (6.5 wt %) proposed by the U.S. Department of Energy (DOE) for the hydrogen plan of fuel cell vehicles. Although some reports7,8 claimed that their investigations reached the DOE target, their results cannot be applied in practice or could be reproduced until now. For example, Chen et al.8 found that lithium- and potassium-doped CNTs can store 14-20 wt % hydrogen. However, the stored hydrogen can be released only at high temperatures, which is not applicable in fuel cell vehicles. Schlapbach et al.9 believed that the carbon materials are difficult for the high adsorption capacity of hydrogen unless the carbon materials hold an extremely high specific surface. Evidently, the design of novel materials and the modification of porous materials significantly increase the adsorption capacity of hydrogen.10-12 Cao et al.11 designed a novel class of carbonaceous materials called graphitic carbon inverse opal (GCIO) that can load the hydrogen delivery capacity of 5.9 wt % at room temperature and P ) 30 MPa, where the delivery amount is defined as the adsorption amount at high pressure minus that at the discharge pressure, because the GCIO materials possess an extremely high specific surface area of 4200 m2/g. Mpourmpakis et al.12 proposed a kind of SiC nanotube in which the gravimetric adsorption capacity of hydrogen gains an improvement of 20%, compared with that of a pure carbon nanotube. Recently, Bhatia et al.13 investigated adsorption of hydrogen and methane in porous adsorbents of activated carbon and CNTs systematically from the thermodynamic point of view. They believed that the thermodynamic requirement of an adsorbent being capable of efficiently storing hydrogen at room temperature is the adsorption enthalpy change of hydrogen, equal to * To whom correspondence should be addressed. E-mail: caodp@ mail.buct.edu.cn.
15.1 kJ/mol, if hydrogen gas is charged at 3 MPa and released at 0.15 MPa. With the successful synthesis of silicon nanotubes (SiNTs) by the chemical vapor deposition (CVD) method in 2002,14 a lot of other methods were also developed to fabricate the SiNTs15-18and well-aligned SiNT arrays.19 Compared to carbon, silicon has more electrons in the outer shells, which leads to higher polarizability and a stronger dispersion force. Accordingly, the SiNT may exhibit a stronger van der Waals (VDW) attraction to hydrogen than CNTs. Therefore, we report a multiscale theoretical method to investigate adsorption of hydrogen in the SiNT arrays. The multiscale theoretical method combines the first-principle calculations to obtain the binding energy between hydrogen and the SiNT and a grand canonical Monte Carlo (GCMC) simulation20 to evaluate the hydrogen adsorption capacity in the SiNT arrays, where the calculated binding energy is provided as an input in the GCMC simulation. The rest of this paper is organized as follows. First, we give the detailed descriptions of the first-principle calculations, the fitting of the force field, and the GCMC simulations. Then, we present the comparison of the gravimetric adsorption capacities of hydrogen in the SiNT and CNT arrays and explore the effects of the geometrical arrangement and size of the SiNT on the adsorption amount of hydrogen. Finally, some discussion is addressed. 2. Models and Methods 2.1. First-Principle Calculation. The first-principle calculations used for the parametrization of the interaction potentials of H2 with SiNTs were performed in the framework of density functional theory (DFT) with the Gaussian 03 program package.21 The cluster model method was used in the first-principle calculation, and a large curved graphite-like sheet was adopted to represent the SiNT adsorbent. It should be mentioned that similar cluster models for nanotubes were also used by Gubbins and co-workers22 and Mpourmpakis et al.,12,23 on the basis of the fact that the adsorbed hydrogen mainly interacts with the
10.1021/jp711754h CCC: $40.75 © 2008 American Chemical Society Published on Web 03/19/2008
SiNT as a Promising Candidate for Hydrogen Storage
J. Phys. Chem. C, Vol. 112, No. 14, 2008 5599
Figure 1. Schematic representations of the SiNT cluster model, where all the terminals are saturated with H atoms and the brown yellow and gray spheres represent Si and H atoms, respectively: (a) (9, 9) SiNT, (b) (5, 5) SiNT, (C) H2 adsorption mode at a hollow site with the H-H bond vertical to the tube surface.
neighboring carbon atoms. Figure 1a shows a schematic diagram of the cluster model used in the first-principle calculation for the (9, 9) SiNT. This model contains 22 framework silicon atoms, and all the terminals are saturated with H atoms. The analogous cluster model was also adopted to represent the (7, 7) SiNT and (14, 14) CNT. For the (5, 5) SiNT, the cylinder cluster model of Si70H20 was selected due to its relatively small diameter and is shown in Figure 1b. To achieve more accurate results, the H2 molecule and its nearest six silicon atoms were treated with the 6-311++G (d,p) basis set, whereas the other atoms were treated with the commonly used 3-21G basis set. All the calculations were carried out by the PW9124 exchangecorrelation function, which can provide a more accurate dispersion force than B3LYP.25,26 Tsuzuki and Lu¨thi27 also demonstrated that the PW91 functional is accurate for predicting the energetics of neutral complexes where the dispersion interactions play an important role. On nanotube surfaces, there are three typical adsorption sites: on-top, bridge, and hollow. A hydrogen molecule approaches the tube wall with multiple orientations from both sides of the tube wall.28 All these cases have been considered for adsorption of hydrogen on the SiNT surface. Our results show that, among different adsorption orientations, hydrogen adsorbing on the hollow sites with the H-H bond vertical to the tube surface is the most favorable adsorption mode for both inside and outside the SiNT wall (see Figure 1c). This phenomenon has already been observed for adsorption of hydrogen on a CNT surface.12 Figure 2 shows the potential energy curves of H2 interaction with the (9, 9) SiNT at different adsorption sites by the vertically approaching mode. For comparison, the potential energy curves of H2 interaction with the (9, 9) SiNT and the (14, 14) CNT at the optimum orientation (i.e., the H-H bond vertical to the tube surface) are shown in Figure 3. It is found that the H2-SiNT interaction gains an increase of about 20% at the equilibrium adsorption position, compared to the H2-CNT interaction. This phenomenon mainly stems from the dense electron cloud around the SiNT surface, which produces a strong VDW attraction to hydrogen.
Figure 2. Calculated potential energy curves of H2 adsorption in different adsorption sites of the tube wall of the (9, 9) SiNT with the H-H bond vertical to the tube surface outside and inside: (a) adsorption outside the (9, 9) SiNT, (b) adsorption inside the (9, 9) SiNT.
2.2. Fitting of the Force Field. To efficiently implement the multiscale investigation from the first-principle calculation to GCMC simulation, it is necessary to use a function to bridge the first-principle calculation and the GCMC simulation. In this work, the first-principle calculation results of H2 with SiNTs as well as CNTs were fitted to a Morse potential in eq 1 to obtain the interaction between a fluid molecule and the surface of the nanotube. That is to say, the Morse potential acts as the bridge of the first-principle calculation and the GCMC simulation. The parameters in the Morse potential were obtained according to the way that the fitted binding energy of the molecule to the surface is as close as possible to that determined by the first-principle calculations described in the previous section. Although most researchers have used the Lennard-Jones potential to describe the interaction between hydrogen and atoms of the nanotube in previous works,12,23 it is difficult to use the Lennard-Jones potential to model the interaction derived from the first-principle calculations. This is because as an adsorbate molecule deviates from the equilibrium adsorption position, the Lennard-Jones potential rises rapidly, while the potential energy derived from the first-principle calculation rises slowly. In this work, our calculations show that the Morse potential including three parameters is a better selection to describe the binding energies obtained from the first-principle calculations, compared with the Lennard-Jones potential. It should be mentioned that the Morse potential has been used successfully to fit the
5600 J. Phys. Chem. C, Vol. 112, No. 14, 2008
Lan et al. tubes is set to 0.7 nm, because previous studies37,38 indicate that this distance corresponding to 2 times the molecular diameter (here, it is about 0.7 nm) is an optimal spacing for gas adsorption, at which the gas molecule can form an ordered monolayer between two tubes. The periodic boundary conditions were applied in all three dimensions. The cutoff radius was 5 times the collision diameter. For each state point, the GCMC simulation consisted of 1 × 107 steps to guarantee equilibration, and the following 1 × 107 steps were used to sample the desired thermodynamics properties, such as the adsorption amount and isosteric heat of hydrogen. In the GCMC simulations, the diameter and length of the (9, 9) SiNT are 1.93 and 2.14 nm. The corresponding volumes of the simulation boxes for rhombic and square arrangements are 12.82 and 14.80 nm3, respectively. The diameter and length for the (14, 14) CNTs are 1.90 and 2.34 nm. Correspondingly, the volumes of the simulation boxes for rhombic and square arrangements were 13.70 and 15.82 nm3, respectively. 2.3.1. Chemical Potential. To obtain the relationship between the chemical potential and pressure, Widom’s test particle insertion method in the NVT ensemble was used in terms of20
Figure 3. Potential energy curves of H2 adsorption on the hollow site of the tube wall: (a) (9, 9) SiNT, (b) (14, 14) CNT.
interaction potential determined by the first-principle calculations in previous works.29-32 The interaction between a fluid molecule and the surface of the nanotube was fitted to the Morse potential, and the form is expressed as
( ( ))
Ui ) 2D[x2 - 2x], x ) exp -
γ ri -1 2 re
(1)
where ri is the distance between hydrogen and the tube surface (Å). The constants D, γ, and re for hydrogen adsorption both inside and outside the nanotubes are given in Table 1. 2.3. Grand Canonical Monte Carlo Simulations. In this work, it is assumed that the nanotube has a rigid structure and no geometry variation of the adsorbent is considered, since the induced geometric variation of nanotubes by hydrogen can be neglected at room temperature. In our simulations, no phase transition occurred, which is different from the case mentioned by Cole and co-workers.33,34 They believed that the geometry variation of adsorbents should be taken into account, if the simulation was implemented at lower temperatures than the critical point. In the GCMC simulation, the temperature, volume, and chemical potential are specified. In this work, the fluid-fluid interaction was described using the typical 12-6 Lennard-Jones pair potential:
[( ) ( ) ]
σ Ui ) 4 ri
12
σ ri
6
(2)
The parameters of hydrogen adopted here are ff/k ) 42.8 K and σff ) 2.97 Å, which have been reported in the recent literature.36 To explore the effect of the geometrical arrangement of nanotubes on the adsorption of hydrogen, both the rhombic and square arrangements were employed in this work. The crosssection of the simulation box consisted of four lines connecting the centers of the nanotubes in the rhombic and square arrangements (also see Figures 4 and 5). In the two arrangements, the spacing between the tube walls of the neighboring
p)
F
+
β
B(r f ij)‚b r ij ∑i ∑ j>i (3)
3V
where the symbols p, F, β,and V represent the pressure, density, reciprocal temperature, and volume of the fluid, respectively.39 B(r f ij) and b rij denote the force and distance between particle i and particle j, respectively. The excess chemical potential can be given by39
µex ) -
1 ln β
∫dbs N +1 〈exp(-β∆u)〉N
(4)
where 〈...〉N denotes the canonical ensemble average over the configuration space of the N-particle system. Actually, in Widom’s test particle method, the bulk density is used for the establishment of the relationship between the chemical potential and pressure. 2.3.2. Isosteric Heat. A thermodynamic quantity of interest in adsorption studies is the isosteric heat, which is the released heat for each molecule added to the adsorbed phase. By using the fluctuation theory, the isosteric heat is calculated from40
qiso )
〈U〉〈N〉 - 〈UN〉 〈N2〉 - 〈N〉〈N〉
+ kBT
(5)
where 〈...〉 denotes the ensemble average, N is the number of particles, and U is the configuration energy of the system. 3. Results and Discussion 3.1. Comparison of Adsorption of Hydrogen in the SiNT and CNT Arrays. By inputting the Morse potential into the GCMC simulation, we evaluated the gravimetric adsorption capacity of hydrogen in the SiNT arrays at room temperature. The gravimetric adsorption capacity is defined as the mass percentage of the adsorbed hydrogen in nanotube arrays. Parts a and b of Figure 4 show the snapshots of hydrogen adsorption in the rhombic SiNT and CNT arrays at T ) 298 K and P ) 2, 6, and 10 MPa, respectively. The adsorption isotherms are also presented in Figure 4c. Our GCMC simulations indicate that gravimetric adsorption capacities of hydrogen in the (9, 9) SiNT arrays reach up to 1.30, 2.33, and 2.88 wt % at 2, 6, and 10 MPa at T ) 298 K, respectively. These results present
SiNT as a Promising Candidate for Hydrogen Storage
J. Phys. Chem. C, Vol. 112, No. 14, 2008 5601
Figure 4. Snapshots and gravimetric adsorption capacities of hydrogen in the rhombic tube arrays at T ) 298 K and P ) 2, 6, and 10 MPa: (a) (9, 9) SiNT array, (b) (14, 14) CNT array, (c) gravimetric adsorption isotherms of hydrogen. The rhombic area within the dashed lines represents the cross section of the simulation box used in our GCMC simulations.
TABLE 1: Parameters of the Morse Potential Used To Describe the Interaction between a Hydrogen Molecule and SiNTs as Well as CNTs H2-(5, 5) SiNT D (kcal/mol) g re (Å)
H2- (7, 7) SiNT
H2- (9, 9) SiNT
H2- (14, 14) CNT
in
out
in
out
in
out
in
out
1.370 4.651 3.179
1.044 4.658 2.817
1.165 4.736 3.163
0.978 4.644 2.956
1.077 4.887 3.125
0.941 4.830 2.969
0.927 5.635 2.900
0.767 5.828 2.890
corresponding adsorption increases of 106%, 65%, and 52%, compared to the adsorption of the isodiameter (14, 14) CNTs. Furthermore, it can be found from Figure 4c that the gravimetric adsorption amount of hydrogen in the SiNT array is always significantly greater than that in the CNT array, due to the stronger interaction of the SiNT array with hydrogen.
3.2 Effect of the Geometrical Arrangement of the Tubes on Hydrogen Adsorption. To consider the effect of the geometrical arrangement of the nanotubes on hydrogen adsorption, we present the snapshots and the corresponding gravimetric capacities of hydrogen adsorbed in the square array in Figure 5a,b. The corresponding adsorption isotherms are also shown
5602 J. Phys. Chem. C, Vol. 112, No. 14, 2008
Lan et al.
Figure 5. Snapshots and gravimetric adsorption capacities of hydrogen in the square tube arrays at T ) 298 K and P ) 2, 6, and 10 MPa: (a) (9, 9) SiNT array, (b) (14, 14) CNT array, (c) gravimetric adsorption isotherms of hydrogen. The square area within the dashed lines represents the cross section of the simulation box used in our GCMC simulations.
in Figure 5c. Although the gravimetric capacities of hydrogen in the rhombic and square arrays remain basically the same, the rhombic array presents a larger hydrogen density, because the rhombic box has a smaller volume than the square one. In particular, at relatively low pressure, the rhombic SiNT array presents not only a higher gravimetric capacity, but also a larger hydrogen density, compared to the square array. This is because the interstice channels of the SiNT rhombic array are obviously narrower than those of the square array. Therefore, the potential field in the rhombic array is stronger, which leads to a stronger attraction to hydrogen. Mpourmpakis et al.12 reported the hydrogen adsorption capacity increases of the square SiC tube array by 46% and 21% at pressures of 5 and 10 MPa and T ) 175 K, respectively, compared to that of the isodiameter CNTs.
Obviously, our calculation indicates that the SiNTs present a higher hydrogen adsorption capacity than SiC nanotubes. Figure 6 shows the isosteric heats of hydrogen adsorbed in rhombic and square tube arrays at T ) 298 K in the pressure range from 1 to 10 MPa. As the pressure increases, the isosteric heat in the SiNT arrays decreases slightly, while that in the CNT arrays basically remains unchanged. Evidently, the enthalpy change for hydrogen adsorption in SiNT arrays is much larger than that in CNT arrays. This phenomenon is in good agreement with the above results of the first-principle calculations. That is to say, the SiNTs exhibit larger affinity to hydrogen than the isodiameter CNT, which arises from the strong VDW interaction between hydrogen and SiNTs. In the same thermodynamic state, the constant enthalpy change for hydrogen adsorption in the
SiNT as a Promising Candidate for Hydrogen Storage
J. Phys. Chem. C, Vol. 112, No. 14, 2008 5603
Figure 7. Adsorption amount of hydrogen in the (5, 5), (7, 7), and (9, 9) SiNT rhombic arrays at pressures of 4, 6, 8, and 10 MPa.
Figure 6. Isosteric heats for hydrogen adsorption in the rhombic and square tube arrays at T ) 298 K. For comparison, the isosteric heat calculated by using the Lennard-Jones potential from the literature13 for the (14, 14) CNT rhombic array is also presented.
SiNT rhombic array is 1.24 kJ/mol larger than that in the SiNT square array on average, while this value is 0.82 kJ/mol for the CNT arrays. This is due to the relatively compact arrangement of the rhombic array, which causes larger fluid-adsorbent interaction, compared to the square array with the same spacing between neighboring tube walls. Moreover, the adsorption heats calculated by the Morse potential here are higher than those calculated by the LennardJones potential.13,41 To explain this difference, the GCMC simulation for adsorption of hydrogen in the (14, 14) CNT rhombic array by using the Lennard-Jones potential in eq 2 was also implemented. To confirm the validity of our simulation in this calculation, we also used these parameters from the literature (ff/k ) 34.2 K and σff ) 2.96 Å for fluid-fluid interaction, sf/k ) 30.95 K and σsf ) 3.18 Å for fluid-solid interaction)13 to calculate the isosteric heat of hydrogen at T ) 298 K, and the calculated result is also shown in Figure 6. The calculated isosteric heat changes from about 5.75 to 5.9 kJ/mol as the pressure decreases, which is in good agreement with the literature, in which the isosteric heat is 5.85 kJ/mol for hydrogen adsorption on the (9, 9) CNT.41 There are several factors which may contribute to the enhancement of the adsorption heat of hydrogen in this work. The main reason is the distinction between the Morse and Lennard-Jones potentials as mentioned above. As the hydrogen molecule approaches the tube surface, the Morse potential varies smoothly. Thus, many molecules are adsorbed near the bottom of the potential well, leading to a relatively large enthalpy change. However, the Lennard-Jones potential goes up rapidly as the molecule deviates from the equilibrium adsorption position. As a result, the isosteric heat calculated by the Lennard-Jones potential is clearly lower, as shown in Figure 6. The second possible reason is the exchangecorrelation function used in the first-principle calculation. As is well-known, different exchange-correlation functions such as B3LYP25,26 and PW9124 may generate slightly different binding energies,12 and the former is generally considered to underestimate the binding energy to some extent. The DFT methods may also give results somewhat different from those of the MP2 method.42,43 Mpourmpakis et al.12 found that the B3LYP method underestimates the binding energy by about 0.88 and 0.87 kcal/ mol for hydrogen adsorption on SiCNT and CNT, respectively, compared to the PW91 method at the equilibrium position.
Bhatia and Myers13 believed that for the optimum delivery of hydrogen between pressures of 3 and 0.15 MPa at 298 K, an adsorption enthalpy change of 15.1 kJ/mol is most appropriate. On the basis of the opinion of Bhatia et al., it is confirmed that the SiNT array possesses a better adsorption capability for hydrogen than the CNT array, because the SiNT array gives an adsorption heat closer to the enthalpy change of 15.1 kJ/mol recommended by Bhatia and Myers. 3.3. Effect of the Size of the Tube on Hydrogen Adsorption. Not only the geometrical arrangement of the tubes but also the diameter and curvature of the tube affect the adsorption of hydrogen in the SiNT array. Therefore, gravimetric adsorption capacities of hydrogen in (5, 5) and (7, 7) rhombic SiNT arrays were also investigated. The Morse potential parameters were obtained by fitting the first-principle calculations and are listed in Table 1. The diameters of the (5, 5) and (7, 7) SiNTs are 1.08 and 1.50 nm, respectively. The first-principle calculations show that, as the diameter declines, the interaction between hydrogen and SiNTs increases obviously for both inside and outside adsorption. However, the GCMC calculations prove that, among the three kinds of SiNT arrays, the (7, 7) structure gives the highest hydrogen storage capacity in the medium-pressure range from 4 to 10 MPa. Figure 7 shows the gravimetric adsorption amount of hydrogen in the (5, 5), (7, 7), and (9, 9) SiNT arrays at T ) 298 K, where the spacing between the tube walls of the neighboring tubes is still set to 0.7 nm. Clearly, as the diameter increases, the total surface area of the tube array increases, while the interaction between hydrogen and SiNTs declines. Therefore, there exists an optimum diameter for hydrogen storage within different pressure ranges. In this work, the (7, 7) SiNT array rather than the (9, 9) SiNT array exhibits the best hydrogen adsorption capacity within the moderate pressure range from 4 to 10 MPa. 4. Conclusions In summary, the multiscale theoretical investigations indicate that the SiNT array gives a higher hydrogen adsorption capacity than the CNT array at T ) 298 K in the pressure range from 1 to 10 MPa. Excellent performance of the SiNT array in hydrogen storage is attributed to the denser electron cloud of the SiNT surface compared to the CNT surface, which produces much stronger VDW attraction to hydrogen. As a result, the SiNT arrays present distinct improvements of 106%, 65%, and 52% in the gravimetric adsorption capacities of hydrogen at P ) 2, 6, and 10 MPa at T ) 298 K, respectively, compared to the isodiameter CNTs. This suggests that the SiNT array is a promising candidate for hydrogen storage. By comparison of rhombic and square arrangements of the tubes, it is found that the rhombic tube array can adsorb more hydrogen molecules.
5604 J. Phys. Chem. C, Vol. 112, No. 14, 2008 The dependence of hydrogen storage on the diameter and curvature of the tube denotes that, among the (5, 5), (7, 7), and (9, 9) rhombic SiNT arrays, the (7, 7) structure exhibits the largest hydrogen adsorption capacity at T ) 298 K in the pressure range from 4 to 10 MPa. In short, the multiscale theoretical method presented in this work provides a useful bottom-to-top way to investigate properties of novel materials. Since the SiNT is a novel kind of material, there is no useful experimental information on hydrogen storage in the SiNT reported at present. It is expected this work can be verified by experimental measurement in the future. Acknowledgment. This work is supported by the National Natural Science Foundation of China (Grants 20776005 and 20736002), Beijing Novel Program (Grant 2006B17), Program for New Century Excellent Talents (Grant NCET-06-0095) from the Ministry of Education, and “Chemical Grid Project” and Excellent Talents Funding of the Beijing University of Chemical Technology. References and Notes (1) Liu, C.; Fan, Y. Y.; Liu, M.; Cong, H. T.; Cheng, H. M.; Dressethaus, M. S. Science 1999, 286, 1127. (2) Zhang, X. R.; Cao, D. P.; Chen, J. F. J. Phys. Chem. B 2003, 107, 4942. (3) Wang, Q.; Johnson, J. K. J. Phys. Chem. B 1999, 103, 4809. (4) Yin, Y. F.; Mays, T.; McEnaney, B. Langmuir 2000, 16, 10521. (5) Eric, M.; Pierre, B. Langmuir 2004, 20, 7852. (6) Cao, D. P.; Wang, W. C. Int. J. Hydrogen Energy 2007, 32, 1939. (7) Dillon, A. C.; Jones, K. M.; Bekkedahl, T. A.; Kiang, C. H.; Bethune, D. S.; Heben, M. J. Nature 1997, 386, 377. (8) Chen, P.; Wu, X.; Liu, J.; Tan, K. L. Science 1999, 285, 91. (9) Schlapbach, L.; Zuttel, A. Nature 2001, 414, 353. (10) Zhao, Y. F.; Kim, Y. H.; Dillon, A. C.; Heben, M. J.; Zhang, S. B. Phys. ReV. Lett. 2005, 95, 155504. (11) Cao, D. P.; Feng, P. Y.; Wu, J. Z. Nano Lett. 2004, 4, 1489. (12) Mpourmpakis, G.; Froudakis, G. E.; Lithoxoos, G. P.; Samios, J. Nano Lett. 2006, 6, 1581. (13) Bhatia, S. K.; Myers, A. L. Langmuir 2006, 22, 1688. (14) Sha, J.; Niu, J. J.; Ma, X. Y.; Xu, J.; Zhang, X. B.; Yang, Q.; Yang, D. R. AdV. Mater. 2002, 14, 1219. (15) Jeong, S. Y.; Kim, J. Y.; Yang, H. D.; Yoon, B. N.; Choi, S. H.; Kang, H. K.; Yang, C. W.; Lee, Y. H. AdV. Mater. 2003, 15, 1172. (16) Hu, J. Q.; Bando, Y.; Liu, Z. W.; Zhan, J. H.; Golbergl, D.; Sekiguchi, T. Angew. Chem., Int. Ed. 2004, 43, 63. (17) Chen, Y. Q.; Tang, Y. H.; Pei, L. Z.; Guo, C. AdV. Mater. 2005, 17, 564. (18) Balasubramanian, C.; Godbole, V. P.; Rohatgi, V. K.; Das, A. K.; Bhoraskar, S. V. Nanotechnology 2004, 15, 370. (19) Mu, C.; Yu, Y. X.; Liao, W.; Zhao, X. S.; Xu, D. S.; Chen, X. H.; Yu, D. P. Appl. Phys. Lett. 2005, 87, 113104.
Lan et al. (20) Frankel, D.; Smit, B. Understanding Molecular Simulation: From Algorithms to Applications, 2nd ed.; Academic Press: San Diego, 2002. (21) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Montgomery, J. A., Jr.; Vreven, T.; Kudin, K. N.; Burant, J. C.; Millam, J. M.; Iyengar, S. S.; Tomasi, J.; Barone, V.; Mennucci, B.; Cossi, M.; Scalmani, G.; Rega, N.; Petersson, G. A.; Nakatsuji, H.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.; Klene, M.; Li, X.; Knox, J. E.; Hratchian, H. P.; Cross, J. B.; Bakken, V.; Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.; Austin, A. J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.; Ayala, P. Y.; Morokuma, K.; Voth, G. A.; Salvador, P.; Dannenberg, J. J.; Zakrzewski, V. G.; Dapprich, S.; Daniels, A. D.; Strain, M. C.; Farkas, O.; Malick, D. K.; Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Ortiz, J. V.; Cui, Q.; Baboul, A. G.; Clifford, S.; Cioslowski, J.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi, I.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.; Peng, C. Y.; Nanayakkara, A.; Challacombe, M.; Gill, P. M. W.; Johnson, B.; Chen, W.; Wong, M. W.; Gonzalez, C.; Pople, J. A. Gaussian 03, revision B.02; Gaussian, Inc.: Wallingford, CT, 2004. (22) Kostov, M. K.; Santiso, E. E.; George, A. M.; Gubbins, K. E.; Nardelli, M. B. Phys. ReV. Lett. 2005, 95, 136105. (23) Mpourmpakis, G.; Tylianakis, E.; Froudakis, G. E. Nano Lett. 2007, 7, 1893. (24) Perdew, J. P.; Wang, Y. Phys. ReV. B 1992, 45, 13244. (25) Lee, C.; Yang, W.; Parr, R. G. Phys. ReV. B 1988, 37, 785. (26) Becke, A. D. J. Chem. Phys. 1993, 98, 5648. (27) Tsuzuki, S.; Luthi, H. P. J. Chem. Phys. 2001, 114, 3949. (28) Mpourmpakis, G.; Tylianakis, E.; Froudakis, G. E. J. Nanosci. Nanotechnol. 2006, 6, 87. (29) Han, S. S.; Goddard, W. A. J. Am. Chem. Soc. 2007, 129, 8422. (30) Smith, R.; Nock, C.; Kenny, S. D.; Belbruno, J. J.; Vece, M. D.; Palomba, S.; Palmer, R. E. Phys. ReV. B 2006, 73, 125429. (31) Jiang, A. Q.; Awasthi, N.; Kolmogorov, A. N.; Setyawan, W.; Borjesson, A.; Bolton, K.; Harutyunyan, A. R.; Curtarolo, S. Phys. ReV. B 2007, 75, 205426. (32) Vervisch, W.; Mottet, C.; Goniakowski, J. Phys. ReV. B 2002, 65, 245411. (33) Galbi, M. M.; Toigo, F.; Cole, M. W. Phys. ReV. Lett. 2001, 86, 5062. (34) Cole, M. W.; Vincent, H.; Stan, G.; Ebner, C.; Hartman, J. M.; Moroni, S.; Boninsegni, M. Phys. ReV. Lett. 2000, 84, 3883. (35) Wang, B. L.; Yin, S. Y.; Wang, G. H.; Zhao, J. J. J. Phys.: Condens. Matter 2001, 13, L403. (36) Williams, K. A.; Eklund, P. C. Chem. Phys. Lett. 2000, 320, 352. (37) Cao, D. P.; Zhang, X. R.; Chen, J. F.; Wang, W. C.; Yun, J. J. Phys. Chem. B 2003, 107, 13286. (38) Kowalczyk, P.; Tanaka, H.; Holyst, R.; Kaneko, K.; Ohmori, T.; Miyamoto, J. J. Phys. Chem. B 2005, 109, 17174. (39) Cao, D. P.; Wu, J. Z. Langmuir 2004, 20, 3759. (40) Nicholson, D.; Parsonage, N. Computer simulation and statistical mechanics of adsorption; Academic Press: London, 1982. (41) Cheng, H. M.; Cooper, A. C.; Pez, G. P.; Kostov, M. K.; Piotrowski, P.; Stuart, S. J. J. Phys. Chem. B 2005, 109, 3780. (42) Frisch, M. J.; Head-Gordon, M.; Pople, J. A. Chem. Phys. Lett. 1990, 166. (43) Head-Gordon, M.; Pople, J. A.; Frisch, M. J. Chem. Phys. Lett. 1988, 153, 503.
