Physical Chemistry Chemical Physics

Physical Chemistry Chemical Physics Volume 10 | Number 35 | 21 September 2008 | Pages 5321–5476

www.rsc.org/pccp

ISSN 1463-9076

COVER ARTICLE Zhou et al. Hydrogen dissociative chemisorption and desorption on saturated subnano palladium clusters (Pdn, n = 2–9)

PERSPECTIVE Macht and Iglesia Structure and function of oxide nanostructures: catalytic consequences of size and composition

1463-9076(2008)10:35;1-X

www.rsc.org/pccp | Physical Chemistry Chemical Physics

PAPER

Hydrogen dissociative chemisorption and desorption on saturated subnano palladium clusters (Pdn, n = 2–9) Chenggang Zhou,a Shujuan Yao,a Jinping Wu,a Robert C. Forrey,b Liang Chen,c Akitomo Tachibanad and Hansong Chenge Received 25th March 2008, Accepted 29th May 2008 First published as an Advance Article on the web 1st July 2008 DOI: 10.1039/b804877k H2 sequential dissociative chemisorption on small palladium clusters was studied using density functional theory. The chosen clusters Pdn (n = 2–9) are of the lowest energy structures for each n. H2 dissociative chemisorption and subsequent H atom migration on the bare Pd clusters were found to be nearly barrierless. The dissociative chemisorption energy of H2 and the desorption energy of H atom in general decrease with the coverage of H atoms and thus the catalytic efficiency decreases as the H loading increases. These energies at full cluster saturation were identified and found to vary in small energy ranges regardless of cluster size. As H loading increases, the clusters gradually change their bonding from metallic character to covalent character. For the selected Pd clusters, the capacity to adsorb H atoms increases almost proportionally with cluster size; however, it was found that the capacity of Pd clusters to adsorb H atoms is, on average, substantially smaller than that of small Pt clusters, suggesting that the catalytic efficiency of Pt nanoparticles is superior to Pd nanoparticles in catalyzing dissociative chemisorption of H2 molecules.

1. Introduction Transition metal catalyzed chemical reactions are of great industrial importance and have been a subject of extensive studies both experimentally and theoretically for many years.1–4 Palladium (Pd) and platinum (Pt) catalysts, in particular, have been widely utilized in many chemical processes such as hydrogenation, oxidation and reduction process.5–12 Understanding the elementary processes of these chemical reactions is essential for design and development of novel catalysts to achieve high catalytic efficiency. The focus of the present study is to address some of the fundamental issues concerning hydrogen dissociative chemisorption on Pd catalytic particles. Of particular interest are hydrogen-saturated metal catalysts, which would occur at typical experimental conditions due to constant H2 pressure.13,14 In general, the size of the metal catalytic particles ranges from nano-scale to meso-scale.6,15,16 It has been widely recognized that chemical reactions occur at the sharp corners and defect sites of catalysts.17–20 Most theoretical studies have employed single crystalline surface models at low coverage of molecular species to represent catalyst surfaces.21,22 The catalytic performance, however, is known to depend on the surface roughness of the catalyst particles19,20 and is also expected to a

Institute of Theoretical Chemistry and Computational Materials Science, China University of Geosciences, Wuhan, 430074, China b Department of Physics, Penn State University, Berks Campus, Reading, PA 19610-6009, USA c Ningbo Institute of Materials Technology and Engineering, Chinese Academy of Sciences, Ningbo, Zhejiang, 315201, China d Department of Microengineering, Kyoto University, Kyoto 606-8501, Japan e Air Products and Chemicals, Inc. 7201 Hamilton Boulevard, Allentown, PA 18195-1501, USA

This journal is

c

the Owner Societies 2008

be coverage dependent.23 At a low coverage, the dissociative chemisorption energy of H2 on catalyst surfaces is considerably higher than at a high coverage.23,24 Through catalytic activity using low coverage, a single crystalline surface model has been obtained. The adsorption of H on Pd(111) has been studied using density functional theory/generalized gradient approximation (DFT/GGA) method.23–28 The GGA reported adsorption energy ranges from 0.27 to 0.53 eV (the experimental value is 0.45 eV).25 The adsorption and thermally activated diffusion of hydrogen on Pd(111) surface were studied with the variation of adsorption sites at low H coverage by Watson et al.24 The fcc hollow site was found to be the most stable adsorption site with an adsorption energy of 0.53 eV. Nobuhara et al.23,26 investigated the low coverage dependence of hydrogen absorption onto Pd(111) and the corresponding variation in the energy barrier for H absorption. Dong et al.27 also calculated H2 dissociative adsorption process on Pd(111) and found an activation energy of 0.02 eV, nearly barrierless. The reported H desorption energy from a Pd single-crystal surface is approximately 2.7–2.9 eV.23,28 In a typical hydrogenation reaction, a certain pressure of hydrogen is constantly maintained. Consequently, the surface of catalyst particles is expected to be fully covered by either molecular or atomic hydrogen. It is therefore essential to understand the catalytic activity at full catalyst saturation in order to design novel catalysts to achieve best catalytic efficiency. Comprehensive theoretical modeling of realistic particle size of catalysts using first-principles methods is computationally difficult since the calculations would involve a very large number of atoms. In a previous paper on sequential H2 dissociative chemisorption on small Pt clusters,29 we showed that some of the most fundamental properties of catalyst particles may not be too sensitive to the particle Phys. Chem. Chem. Phys., 2008, 10, 5445–5451 | 5445

size at full catalyst saturation of H atoms. Specifically, the dissociative chemisorption energy of H2 and the desorption energy of H atoms at full H coverage only vary in small energy ranges. These two physical quantities are important since catalytic hydrogenation is critically dependent on how easily an H2 molecule dissociates onto and how fast an H atom desorbs from a fully H saturated catalyst. For a realistic size of catalyst particles, we anticipate that both the dissociative chemisorption energies of H2 and the desorption energy of H atom would be lower than for subnano clusters. Studies of these physical quantities on fully saturated small metal clusters, particularly their evolution with the cluster size, provide useful insight into the properties of real size catalysts. In this paper, we explore these fundamental properties for sequential H2 dissociative chemisorption on small Pd clusters using quantum-mechanical density functional theory. The structure of small Pd clusters has been previously reported.30 The lowest energy cluster structures were chosen as substrates for the present study of sequential H2 dissociative chemisorption. The energetics of successive H2 loading is calculated and the dependence of H2 dissociative chemisorption energy and H desorption energy on H coverage is investigated. We show that these energies in general decrease as the coverage increases and that both quantities at full H saturation vary in small ranges, similar to what was found for small Pt clusters.29,31 Our results demonstrate, however, that much fewer H2 molecules would undergo dissociative chemisorption on small Pd clusters than on Pt clusters of similar size, suggesting that Pd clusters are considerably less catalytically efficient in catalyzing dissociative chemisorption of H2 molecules.

2. Computational method All calculations were performed using DFT/GGA with the non-local Perdew and Wang (PW91)32 exchange–correlation functional as implemented in the DMol3 package.33,34 A double precision numerical basis set augmented with polarization functions was used to describe the valence electrons, while the core electrons were described by an effective core pseudopotential, which also accounts for the scalar relativistic effects.35,36 These basis functions are numerically exact atomic orbitals rather than analytical functions (e.g. Gaussian or Slater type orbitals) and molecules can be dissociated exactly to its constituent atoms within the DFT context. It has been shown that this quality of basis set gives rise to very little superposition effects.37 A spin-polarized scheme was employed to deal with the electronically open-shell systems intrinsic to the Pd atoms. Charge transfer was studied using the Hirshfeld population analysis.38 The Mulliken population analysis is inadequate in the present case since the size of Pd atom is much larger than that of H. All structures were fully optimized without symmetry constraints. The energetically most stable structures were found using the conjugated gradient algorithm. The transition state (TS) structure search was performed only for H2 on Pd6 to gain insight into the barrier of H2 dissociative chemisorption using the LST/QST method.39 TS structures were verified by performing normal mode analysis that gives only one imaginary frequency. The average dissociative chemisorption energy per molecule was evaluated 5446 | Phys. Chem. Chem. Phys., 2008, 10, 5445–5451

using the equation: DECE ¼ 2ðEPdn þ m=2EH2 EPdn Hm Þ=m;

ð1Þ

ðn ¼ 2; 3; . . . ; m ¼ 2; 4; 6; . . .Þ; where EPdn represents the energy of the Pdn cluster, EH2 is the energy of the H2 molecule and EPdn Hm is the total energy of m hydrogen atoms adsorbed on Pdn cluster. In catalysis, the key physical quantity is the desorption energy of molecular/atomic species at a full coverage of catalyst surfaces. In the present case, the energy required to desorb a single H atom is not the same as the dissociative chemisorption energy per molecule defined above. Instead, the sequential desorption energy per atom may be defined as DEDE ¼ EH ðEPdn Hm EPdn Hm 2 Þ=2;

ð2Þ

ðn ¼ 2; 3; . . . ; m ¼ 2; 4; 6; . . .Þ; where EH is the energy of H atom. At the full saturation limit, the sequential desorption energy DEDE represents the threshold energy required for an H atom to desorb from the cluster surface. To quantify if a Pd cluster is fully saturated, we performed room temperature ab initio molecular dynamics (MD) simulations for 2 ps with a time step of 1 fs in a NVT canonical ensemble using the Nose´–Hoover thermostat40,41 to ensure all H atoms remain chemisorbed. Excessive H atoms on the surface will recombine into H2 molecules physisorbed on the cluster upon the MD runs.

3. Results and discussion Our focus in this paper is to study cluster chemical properties under full H saturation. The lowest energy bare Pdn cluster structures were taken directly from our previous calculations.30 Sequential H2 dissociative chemisorption on these clusters was then examined. As shown previously, small Pdn clusters bear some similarities to Ptn clusters. Essentially, there are three possible adsorption sites, namely 1-fold on-top, 2fold edge and 3-fold hollow, to accommodate hydrogen atoms. In contrast to small Pt clusters, we found that at a low coverage there is strong preference for H atoms to reside at the edge and hollow sites of the Pd clusters upon H2 dissociative chemisorption and the on-top site is not energetically stable. For convenience, we chose only a Pd6 cluster for detailed analysis of sequential H2 dissociative chemisorption. The cluster adopts an octahedral configuration with adsorption strength of 0.41 and 0.61 eV per H atom at the edge and hollow sites, respectively. Fig. 1a shows the calculated energy diagram of dissociative chemisorption of H2 molecule on the cluster leading to adsorption of H atoms at the most stable hollow sites. Here the binding energy is defined as the difference of the total electronic energies between the cluster and its constituent atoms. Initially, the H2 molecule approaches the cluster from the on-top site (R0), undergoing a physisorption process with an adsorption energy of 0.59 eV. The calculated HOMO of the bare cluster and the LUMO of H2 shown in Fig. 1b indicates that the symmetry of the molecular orbitals matches with each other, allowing electrons to flow from the This journal is

c


Fig. 1 (a) The calculated energy diagram of dissociative chemisorption of an H2 molecule and the subsequent migrations of the H atoms on the Pd6 octahedral cluster. The unit for the bond distance is A˚; (b) the calculated HOMO of Pd6 cluster and the LUMO of H2; (c) the density of states for sequential dissociative chemisorption of H2 molecules on Pd6 cluster.

bare cluster to H2 to form metal hydride. Subsequently, a metastable adsorption state was found with a significant H–H stretch (R). The H2 molecule continues to undergo a dissociative chemisorption process with a small barrier of 0.13 eV to form two neighboring atoms (P1). The H atom can further diffuse to other adsorption sites with a barrier of no more than 0.11 eV (from P1 to TS2), indicating that H2 dissociative chemisorption and H diffusion processes are facile. When the two H atoms are finally settled down (P3), the system gains a stabilization energy of 1.16 eV. The small barrier for H2 dissociative chemisorption coupled with the highly favorable reaction energies indicates that this process is thermodynamically controlled. The calculated small diffusion barriers for H atoms in the cluster are consistent with the experimental fact that the formation of Pd hydrides is facile and H atoms are highly mobile in the hydride complexes. The high mobility of hydrogen gives rise to numerous isomers, for a given H loading. To simplify the presentation, we will only present the results for the lowest energy hydride structures. Fig. 1c displays the calculated density of states spectra of sequential H2 dissociative chemisorption on Pd6. All clusters are of close shell. The bare cluster is of a typical metallic character. Upon successive H loading, the valence band systematically evolves downward and the gap between the valence band and conduction band gradually opens up, giving This journal is

c


rise to a significant covalent character for higher H coverage clusters. The ionization of d-electrons from Pd6 results in a higher contribution to the empty bands above the Fermi level, while contribution to the valence band from the s-orbital of H atoms increases with H coverage. Fig. 2a displays the fully optimized structures of sequentially dissociated H2 on the Pd6 cluster. At a low coverage, the hollow sites are first populated by H atoms followed by the edge sites. As the H loading increases, some of the on-top sites, on which H adsorption was found unstable at low coverage, are also populated. The cluster reaches its full saturation at m = 14, substantially fewer than m = 24 for the Pt6 cluster.31 Ab initio molecular dynamics simulation indicates that additional H loading would result in recombination of H atoms into H2 molecules as shown in the calculated H–H distance distribution (Fig. 2b). The radial distribution of H–H distance of Pd6H14 cluster shown in Fig. 2b indicates that the H atoms are well separated by at least 1.8 A˚. However, adding an additional pair of H atoms, originally separated, results in a peak at approximately 0.8 A˚ (Fig. 2b), indicating the formation of an H2 molecule. The cluster expands slightly upon chemisorption of H atoms as illustrated in Fig. 2c where the average Pd–Pd and Pd–H distances are displayed. In particular, at m = 8, the average Pd–Pd bond distance becomes 2.95 A˚ due to significant structural deformation that results in considerable bond Phys. Chem. Chem. Phys., 2008, 10, 5445–5451 | 5447

Fig. 2 (a) Fully optimized structures of H2 sequential dissociative chemisorption on a Pd6 cluster; (b) the calculated H–H distance distribution g(r) for Pd6H14 and Pd6H16, respectively. g(r) was obtained by tabulating all the H–H distances at each step of the ab initio MD trajectories fit with Gaussian functions; (c) the average Pd–Pd and Pd–H bond length of the structures in (a).

5448 | Phys. Chem. Chem. Phys., 2008, 10, 5445–5451

elongation upon H chemisorption. Correspondingly, the average Pd–H distance becomes smaller. The Pd–Pd distance is decreased slightly upon further loading of H atoms and the Pd6Hm clusters are moderately close-packed. Fig. 3a displays the optimized structures of the fully H saturated Pdn (n = 2–9) clusters. The saturation was ensured by performing ab initio MD runs at 300 K. Higher H loading leads to recombination of H atoms into H2 molecules. It is interesting that the capacity of small Pd clusters to absorb H atoms (H-capacity) is essentially half of what was found for small Pt clusters except for n = 2 and 3.29,31 Most of the adsorbed H atoms reside at the edge and hollow sites. For larger clusters, some of the on-top sites are also populated. To quantify the optimized structures of the fully saturated clusters, we show in Fig. 3b the average nearest neighboring Pd–Pd distances before and after H-saturation and the average Pd–H bond distances. While the Pd–H distances vary in the range between 1.675 and 1.780 A˚, the metal bonding of the clusters loosens up upon hydride formation as the average Pd–Pd distances increase to accommodate H atoms. Fig. 3c shows the number of chemisorbed H atoms vs. cluster size. Essentially, the H-capacity is almost linearly proportional to the cluster size, similar to Pt clusters. Two of the most important properties for hydrogen dissociative chemisorption on transition metal clusters are the dissociative chemisorption energy of H2 (DECE) and the desorption energy of H atom (DEDE) at full cluster saturation.29,31 In Fig. 4a, the calculated sequential dissociative chemisorption energies of H2 molecules on the selected Pd clusters are displayed. In general, the energy decreases as H coverage increases. Except for n = 2 and 3, the calculated threshold DECE fluctuates between 0.6–0.8 eV, slightly lower than that of small Pt clusters. For Pd2H2, the surprisingly large DECE value can be attributed to its relatively stable electronic structure. Pd2 cluster is magnetic with two unpaired electrons.30 Upon H2 dissociative chemisorption, its electronic structure becomes a close-shell and thus stable. Fig. 4b shows the calculated sequential desorption energies of H atoms from the Pd clusters. For n = 2 and 3, there is only one desorption energy data since H saturation is achieved at m = 2 (see eqn (2)). For larger clusters, the calculated threshold DEDE varies in the range of 2.29–2.67 eV, similar to what was found for small Pt clusters,29 which is considerably lower than the value of H desorption on Pd crystalline surface at high coverage.23,28 Although the general trend is that DEDE decreases with cluster size, there appears to be some fluctuation in the calculated desorption energies. This is due to the fact that the sequential desorption energy is evaluated from the difference between EPdn Hm and EPdn Hm 2 and some of the smaller clusters can be less stable than the larger ones. It is important to point out that for the selected size of clusters, both the calculated dissociative chemisorption energy of H2 and the desorption energy of H atoms at full cluster saturation vary in a small range. This is similar to what was found for Pt clusters and suggests a relative independency of particle size with respect to these important physical properties. Fig. 4c shows the calculated sequential Hirshfeld charges transferred from Pd clusters to H atoms. Like small Pt clusters, the Pd clusters lose electrons to H atoms and the charge transfer from Pd clusters This journal is

c


Fig. 3 (a) The optimized structures of fully H saturated Pdn clusters (n = 2–9); (b) the calculated average Pd–H and Pd–Pd distances at full saturation. Open circles are for the bare clusters and solid ones are for saturated clusters; (c) H coverage at full saturation vs. cluster size.

to H atoms increases monotonically as H coverage increases, leading to hydride formation. Finally, the calculated density of states of the fully saturated Pd clusters is shown in Fig. 5. The electronic structure of all clusters is of a close-shell. Similar to the hydrides of small Pt clusters, the low-lying states of PdnHm are all dominated by dorbitals of Pd atoms. The d-orbitals with electrons transferred to H atoms are pushed above the Fermi level. In parallel, the sorbital of H atoms is buried in the valence bands. The calculated DOS spectra show that some of the metal hydrides still maintain a considerable metallic character (e.g. Pd7H16 and Pd9H22).

4. Summary Transition metal catalyzed hydrogenation represents one of the most important chemical reactions with many industrial applications. Understanding the interactions between catalyst particles and H2 molecules is of fundamental importance. We have attempted to address such interactions by examining H2 dissociative chemisorption on subnano Pd clusters as the clusters approach full H saturation. Despite the relatively smaller size of the Pd clusters chosen in the present study compared with the size of catalyst particle size used in practice, This journal is

c


useful insight into the catalytic activity of Pd catalyst toward H2 can be gained. Like small Pt clusters, dissociative chemisorption of molecular hydrogen on subnano Pd clusters is facile with exothermic reaction energies, small activation barriers and favorable orbital overlaps between the HOMO of metal clusters and the LUMO of H2. However, the capacity of Pd clusters toward H2 dissociative chemisorption is considerably smaller, which is perhaps due to the 4d105s0 close-shell electronic configuration of Pd atoms compared with the 5d96s1 configuration of Pt atoms. This result suggests that the catalytic efficiency of Pt nanoparticles is superior to Pd nanoparticles in catalyzing dissociative chemisorption of H2 molecules. Another difference is the preferred sites for H atoms. For Pt clusters, the on-top site is energetically the most stable, whereas the energetically most preferred sites of H atoms on Pd clusters are edge and hollow sites. As H loading increases, some of the on-top sites for Pd clusters can also be populated. Our results indicate that the dissociative chemisorption energy of H2 and the desorption energy of H atom in general decrease with the coverage of H atoms and consequently the catalytic efficiency is reduced as the H loading on the cluster increases. We found that H2 dissociative chemisorption on the Pd clusters is Phys. Chem. Chem. Phys., 2008, 10, 5445–5451 | 5449

Fig. 4 (a) The calculated H2 dissociative chemisorption energy; (b) H desorption energy and (c) the loss of Hirshfeld charges of Pd clusters vs. H coverage.

Fig. 5

The calculated density of states of small Pd clusters saturated with H atoms.

dictated by the charge flow from metal clusters to H atoms and the electron transfer increases with H coverage. As H atoms are successively adsorbed on the clusters, the gap between the conduction band and valence band gradually opens up and the 5450 | Phys. Chem. Chem. Phys., 2008, 10, 5445–5451

chemical bonding evolves from metallic to covalent. We have identified that at the full H-saturation the threshold dissociative chemisorption energy of H2 and the desorption energy of H atom are in the range of 0.6–0.8 eV and 2.29–2.67 eV, respectively, for This journal is

c


n Z 4. These values do not change significantly with the selected cluster size. The results seem to imply that these important physical quantities for larger Pd clusters near full saturation of H atoms on the accessible Pd atoms may be similar to those found here for small clusters.

Acknowledgements The work conducted at CUG was supported by the National Natural Science Foundation of China for Youth (Grant No. 20703040) and Natural Science Foundation of Hubei Province (Grant No. 2006ABA349).

References 1 R. A. Olsen, G. J. Kroes and E. J. Baerends, J. Chem. Phys., 1998, 109, 2450. 2 G. Kresse, Phys. Rev. B, 2000, 62, 8295. 3 S. Horch, H. T. Lorensen, S. Helveg, E. Lægsgaard, I. Stensgaard, K. M. Jacobsen, J. K. Nørskov and F. Besenbacher, Nature, 1999, 398, 134. 4 A. S. Wo¨rz, K. Judai, S. Abbet and U. Heiz, J. Am. Chem. Soc., 2003, 125, 7964. 5 A. M. Doyle, S. K. Shaikhutdinov, S. D. Jackson and H.-J. Freund, Angew. Chem., Int. Ed., 2003, 42, 5240. 6 U. Heiz, A. Sanchez, S. Abbet and W.-D. Schneider, J. Am. Chem. Soc., 1999, 121, 3214. 7 E. Pijper, G. J. Kroes, R. A. Olsen and E. J. Baerends, J. Chem. Phys., 2002, 117, 5885. 8 S. G. Podkolzin, R. Alcala´ and J. A. Dumesic, J. Mol. Catal. A: Chem., 2004, 218, 217. 9 J.-S. Filhol and M. Neurock, Angew. Chem., Int. Ed., 2006, 45, 402. 10 Y. Wang and P. B. Balbuena, J. Phys. Chem. B, 2004, 108, 4376. 11 C. Hartnig and E. Spohr, Chem. Phys., 2005, 319, 185. 12 Y. Ishikawa, J. J. Mateo, D. A. Tryk and C. R. Cabrera, J. Electroanal. Chem., 2007, 607, 37. 13 G. Ertl, Pure Appl. Chem., 1980, 52, 2051. 14 J. Panpranot, G. J. James G. Goodwin and A. Sayari, J. Catal., 2003, 213, 78. 15 T. Visser, T. A. Nijhuis, A. M. J. van der Eerden, K. Jenken, Y. Ji, W. Bras, S. Nikitenko, Y. Ikeda, M. Lepage and B. M. Weckhuysen, J. Phys. Chem. B, 2005, 109, 3822.

This journal is

c


16 J. d. Graaf, A. J. v. Dillen, K. P. d. Jong and D. C. Koningsberger, J. Catal., 2001, 203, 307. 17 X. Liu, H. Dilger, R. A. Eichel, J. Kunstmann and E. Roduner, J. Phys. Chem. B, 2006, 110, 2013. 18 M. K. Oudenhuijzen, J. A. V. Bokhoven, J. T. Miller, D. E. Ramaker and D. C. Koningsberger, J. Am. Chem. Soc., 2005, 127, 1530. 19 A. Chaudhari, C.-C. S. Yan and S.-L. Lee, Catal. Today, 2004, 97, 89. 20 K. Shun’ichi, Y. Hiroshi and Y. Kazuhiro, Proc. Int. Conf., 2006, 2006, 309. 21 G. Papoian, J. K. Nørskov and R. Hoffmann, J. Am. Chem. Soc., 2000, 122, 4129. 22 H. F. Busnengo, M. A. D. Ce´sare, W. Dong and A. Salin, Phys. Rev. B, 2005, 72, 125411. 23 K. Nobuhara, H. Kasai, H. Nakanishi and A. Okiji, J. Appl. Phys., 2002, 92, 5704. 24 G. W. Watson, R. P. K. Wells, D. J. Willock and G. J. Hutchings, J. Phys. Chem. B, 2001, 105, 4889. 25 J.-F. Paul and P. Sautet, Phys. Rev. B, 1996, 53, 8015. 26 K. Nobuhara, H. Kasai, W. Q. Dino and H. Nakanishi, Surf. Sci., 2004, 566–568, 703. 27 W. Dong and J. Hafner, Phys. Rev. B, 1997, 56, 15396. 28 A. L. Cabrera, E. Morales and J. N. Armor, J. Mater. Res., 1995, 10, 779. 29 C. Zhou, J. Wu, A. Nie, R. C. Forrey, A. Tachibana and H. Cheng, J. Phys. Chem. C, 2007, 111, 12773. 30 C. Luo, C. Zhou, J. Wu, T. J. D. Kumar, N. Balakrishnan, R. C. Forrey and H. Cheng, Int. J. Quantum Chem., 2007, 107, 1632. 31 L. Chen, A. C. Cooper, G. P. Pez and H. Cheng, J. Phys. Chem. C, 2007, 111, 5514. 32 J. P. Perdew and Y. Wang, Phys. Rev. B, 1992, 45, 13244. 33 D. Delley, J. Phys. Chem., 1996, 100, 6107. 34 B. Delley, J. Chem. Phys., 2000, 113, 7756. 35 M. Dolg, U. Wedig, H. Stoll and H. Preuss, J. Chem. Phys., 1987, 86, 866. 36 A. Bergner, M. Dolg, W. Kuechle, H. Stoll and H. Preuss, Mol. Phys., 1993, 80, 1431. 37 B. Delley, J. Chem. Phys., 1990, 92, 508. 38 F. L. Hirshfeld, Theor. Chim. Acta B, 1977, 44, 129. 39 T. A. Halgren and W. N. Lipscomb, Chem. Phys. Lett., 1977, 49, 225. 40 S. Nose´, Mol. Phys., 1984, 52, 255. 41 G. J. Martyna, M. E. Tuckerman and M. E. Tobias, Mol. Phys., 1996, 87, 1117.

Phys. Chem. Chem. Phys., 2008, 10, 5445–5451 | 5451