Hydrogen adsorption and desorption on the Pt and Pd subnano clusters – a review

Abstract

In this review, we present our recent first principles studies on the sequential H₂ dissociative chemisorption and H desorption on the Pt_n and Pd_n clusters (n=2-9, 13). Upon full saturation by H atoms, the calculated H₂ dissociative chemisorption energy and H desorption energy on Pt₁₃ and Pd₁₃ clusters are similar to the corresponding values on smaller close-packed clusters. Indeed, the catalytic performances of these subnano clusters do not vary significantly with the particle sizes or shapes. Instead, they are dependent on the surface metal atoms which can be accessed by H atoms. In addition to the coverage dependency of the H₂ dissociative chemisorption and H sequential desorption energies, the phase transition of both Pt₁₃ and Pd₁₃ from the icosahedral to fcc-like structures at certain H coverage was also investigated.

Graphical abstract

Keywords

hydrogen adsorption / density functional theory / metal clusters / catalysis

1 Introduction

Hydrogen dissociative chemisorption on transition metal catalysts is of great industrial interest in many applications such as hydrogenation, hydrogen storage and electrocatalysis [1-6]. In the past few decades, a great deal of experimental and theoretical efforts have been made to understand the mechanisms of the catalytic processes on Pt, Pd and other precious transition metal surfaces in order to develop low-cost and efficient catalysts for hydrogenation [7-12]. For example, some studies using thermal desorption spectroscopy reported that the dissociative chemisorption energy of hydrogen on the Pt(111) surface is between 0.70 and 0.83 eV [13, 14]. Accordingly, many theoretical studies have employed single crystalline surface models at low coverage of H₂ to represent catalyst surfaces [12, 15, 16]. The reported dissociative chemisorption energy of hydrogen on these Pt crystalline surfaces ranges from 0.8 to 1.54 eV. On the other hand, Watson and coworkers have calculated the adsorption and thermally activated diffusion of hydrogen on the Pd(111) surface with the variation of adsorption sites at low H coverage [17]. They found that the fcc hollow site with an adsorption energy of 0.53 eV is the most favorable site. Nobuhara investigated the low coverage dependency of hydrogen absorption into Pd(111) and the corresponding variation in the energy barrier for H absorption [18, 19].

However, in a realistic catalytic system, the Pt and Pd catalysts are usually dispersed on the support materials in the form of particles (or clusters) with different sizes and shapes. The dissociative chemisorption and desorption behaviors of hydrogen on such particles and perfect crystalline surfaces should be significantly different due to the fact that finite-size particles possess sharp corners and edges, which are usually the most active sites for catalytic reactions [20-22]. A recent desorption experiment of hydrogen/deuterium on a Pt₁₃ cluster supported by NaY zeolite yielded an H₂ dissociative chemisorption energy of 1.36 eV [23], which is much higher than on single crystalline surfaces. Using density functional theory (DFT) methods, Okamoto also reported a similar chemisorption energy of 1.4 eV for H₂ dissociative adsorption at the on-top sites of a Pt₁₃ cluster [24, 25]. These results clearly demonstrate that the chemical reactivity of H₂ on clusters is much higher than on crystalline surfaces. Furthermore, several calculations also suggested substantially higher desorption energy for atomic hydrogen adsorbed on clusters than on flat crystalline surfaces. For example, the desorption energy of an H atom on the Pt(111) surface was found to be 2.60-2.65 eV [26-28]. In contrast, Watari reported a H desorption energy of 2.9-4.5 eV on a Pt₁₃ cluster [29].

Recently, Cheng and coworkers reported strong coverage dependency of catalytic activity of Pt and Pd clusters [30-32]. The dissociative chemisorption energy of H₂ on these catalyst surfaces at low coverages is considerably higher than at high coverages. In a typical hydrogenation reaction, a certain pressure of hydrogen is always maintained and the catalyst particles should be fully covered by either molecular or atomic hydrogen, consequently their catalytic performance is also expected to be coverage dependent. Indeed, much insight into the catalytic activity using low H coverage and single crystalline surface models has already been obtained. However, to more adequately address the realistic catalytic activity, it is very critical to understand the chemical behaviors of the Pt and Pd clusters that are saturated by H atoms to simulate the real catalytic environment. Unfortunately, comprehensive modeling of realistic particle size of catalysts using first principles methods would be computationally difficult since the calculations would involve a very large number of atoms.

In this review, we present our recent first principles studies on the sequential hydrogen chemisorption on a series of subnano Pt_n and Pd_n clusters (n=2-9, 13) to address their catalytic activity [30-33]. We show that at full H-saturation the H₂ dissociative chemisorption energy and the H desorption energy on those clusters do not change significantly with the cluster sizes and shapes. This review is aimed at enhancing our understanding of the interplay between hydrogen and Pt/Pd subnano clusters at or near full H coverage. The literatures on hydrogen sorption on Pt and Pd catalysts at low coverages have been reviewed and discussed by many authors [9, 12, 17, 33], which will not be focused on in this review.

2 Models and computational details

Our calculations were carried out using DFT methods with the Perdew–Wang exchange-correlation functional and a spin-polarized scheme [34]. The effective core pseudopotential (ECP) was utilized to describe the core electrons and a double numerical basis set augmented with polarization function (DNP) was employed to describe the valence electrons. The energy and gradient convergence tolerance was chosen to be 2×10^-5 Ha and 4×10^-3 Ha/Å, respectively. All structures are fully optimized using conjugated gradient algorithm without symmetry constrains. We also employed the LST/QST method to identify the transition state of H₂ dissociative chemisorption and diffusion on the Pt and Pd clusters [35]. The NVT molecular dynamics (MD) simulations and bond-distance distribution analysis were performed to investigate the structural rearrangement and identify the hydrogen saturation point. The Hirshfeld population analysis was performed to evaluate the charge transfer from metal clusters to hydrogen [36]. All computational methods were implemented in DMol³ package [37, 38].

The H₂ dissociative chemisorption energy \mathrm{\Delta }{E}_{\mathrm{CE}} and the H sequential desorption energy \mathrm{\Delta }{E}_{\mathrm{DE}} are defined as follows:

\begin{array}{}\text{(1)}& \mathrm{\Delta }{E}_{\mathrm{CE}}=-\frac{2}{x}(E_{{\mathrm{P}}_n{\mathrm{H}}_x}-E_{{\mathrm{P}}_n}-\frac{x}{2}{E}_{{\mathrm{H}}_2})\end{array}

\begin{array}{}\text{(2)}& \mathrm{\Delta }{E}_{\mathrm{DE}}=E_{\mathrm{H}}-\frac{1}{2}(E_{{\mathrm{P}}_n{\mathrm{H}}_x}-E_{{\mathrm{P}}_n{\mathrm{H}}_{x-2}})\end{array}

where n is the number of H atoms. E_{{\mathrm{P}}_n{\mathrm{H}}_x} is the energy of P_nH_x hydrides, E_{{\mathrm{P}}_n} is the energy of bare metal clusters and E_{{\mathrm{H}}_2} is the energy of a H₂ molecule.

For Pt and Pd, the growth path from a few atoms to a bulk may adopt three patterns, i.e., the close-packed triangular pattern, the icosahedral pattern and the fcc-like pattern. In our recent studies, we found that at the subnano scale, both Pt and Pd clusters initially adopt a close-packed triangular growth pattern, followed by an abrupt transition to the icosahedral growth pattern at a certain point [39]. Therefore, we selected close-packed structures for smaller Pt_n and Pd_n (n=2–9) clusters, which are taken directly from our previous reports [39]. For larger Pt₁₃ and Pd₁₃ cluster, we selected the smallest icosahedral cluster as a model.

3 Results and discussion

3.1 Hydrogen chemisorption on close-packed Pt_n clusters (n=2–9)

We first investigated the chemisorption on the small close-packed Pt_n clusters (n=2–9). For convenience, we choose the Pt₄ cluster as an example to show the sequential H₂ dissociative chemisorption process. On the Pt₄ tetrahedron, there are three possible adsorption sites available for H adsorption: one-fold on-top, two-fold edge and three-fold hollow site. The on-top site is identified as energetically the most favorable binding site, followed by the edge site. Figure 1(a) displays the fully optimized Pt₄ structures at varying H coverages. Depending on the coverage, the cluster is distorted to a certain extent. Full saturation is reached when 16 H atoms are adsorbed, while excessive H adsorption would result in recombination of H atoms to form H₂ molecules. In fact, upon loading 18 H atoms on the cluster, the trajectories of MD simulations for up to 2 ps show that two H atoms are readily squeezed out of the cluster, forming a H₂ molecule which subsequently desorbs to the gas phase. Indeed, the calculated H-H bond distance distribution, as shown in Figs. 1, 2(b), clearly illustrates that all H atoms are well separated by at least 2.1 Å on Pt₄H₁₆, while a peak around 0.75 Å corresponding to a H₂ molecule is observed on Pt₄H₁₈.

For all smaller Pt_n (n=2–9) clusters, the calculated H₂ dissociative chemisorption energy E_{\mathrm{CE}}, H sequential desorption energy E_{\mathrm{DE}} and the loss of electrons of metal clusters Q exhibit similar features, as shown in Fig. 2. The general trends of these quantities are that they decline with the H coverage. However, while both E_{\mathrm{CE}} and Q decrease monotonically with the number of H atoms, some fluctuation of E_{\mathrm{DE}} is observed. This is because H atoms first saturate the energetically most favorable sites. As these sites are filled, their stability decreases and the vacant sites begin to fill. Full H saturation can be identified using MD simulations, which can distinguish the H₂ molecules physisorbed on the clusters. Figure 2 also indicates that in all cases, at the full saturation, the H₂ dissociative chemisorption energy fluctuates in the range between 0.9-1.1 eV, which is 0.2-0.3 eV higher than the experimental value for the H₂ dissociative chemisorption on the Pt(111) surface at zero coverage. Similarly, the calculated threshold H sequential desorption energy varies in a narrow range of 2.45-2.60 eV, slightly smaller than the value for an isolated H atom on the Pt(111) surface reported in a previous study by Papoian [12].

3.2 Hydrogen chemisorption on the Pt₁₃ cluster

We next turn to investigate the hydrogen chemisorption on the larger icosahedral Pt₁₃ cluster. Similarly, the adsorption energy on the three possible sites is calculated: on-top, edge and hollow sites. The adsorption energies on these sites are calculated to be 1.37, 1.46 and 1.21 eV, respectively. In contrast to the smaller close-packed clusters, the edge site is identified as the most favorable binding site on the icosahedral Pt₁₃ cluster, which is in agreement with the recent DFT/GGA results [23]. Many studies have reported that H₂ molecules can readily dissociate on flat Pt surfaces or clusters [12, 17, 24, 25]. Furthermore, the dissociated H atoms have high mobility on those surfaces [12, 17]. For comparison, we thus further studied the reaction pathways of H₂ dissociation and H diffusion on the Pt₁₃ cluster. Initially, a H₂ molecule was placed 3.01 Å above the on-top site to ensure no chemical bonding between H and Pt atoms, as shown in Fig. 3. The dissociation of H₂ is found to be nearly barrierless (TS1, 0.06 eV), and the subsequent H diffusion barriers are smaller than 0.34 eV. The final product (P3) has a chemisorption energy of 1.65 eV with the two H atoms well separated on the edge sites. Therefore, the H₂ dissociation is ready to proceed on the on-top sites, which clearly highlights the catalytic activity of the Pt₁₃ cluster. Subsequently, H atoms can readily overcome the moderate activation barriers and migrate from the on-top sites to the more favorable edge sites.

We then sequentially placed more H atoms onto the edge sites of the Pt₁₃ cluster and then fully optimized the structures. All the optimized configurations were examined by MD simulations to determine whether the system has reached full H saturation. A few selected structures are displayed in Fig. 4(a). MD simulations and bond distance distribution analysis indicate that the H saturation point on Pt₁₃ was located at n=44. Excessive H atoms will recombine to H₂ molecules, resulting in a0.74 Å peak in the H–H distance distribution, as shown in Fig. 4(b).

Figure 5(a) shows that the average Pt–Pt bond distance increases rapidly with the hydrogen loading at low coverage (), implying that the sequential adsorption of H atoms leads to expansion of the Pt₁₃ cluster, which is consistent to the previous reports [4]. However, the icosahedral structure of the Pt₁₃ cluster is still maintained and the most stable configurations have symmetric H–H pair distribution until n=8, as shown in Fig. 4(a₂). We thus denote this stage () as the icosahedron stage. However, the Pt₁₃ icosahedron undergoes a transition to an fcc-like structure [Fig. 4(a₃)] when more than 8 hydrogen atoms are chemisorbed. To ensure that it is not an artifact, we  performed  MD  simulations  for  up  to  2  ps for Pt₁₃H₈ and found that the icosahedral structure was stable. We then placed two more H atoms on the vacant edge sites and performed MD simulations for 2 ps again. The energy evolution curve and the average Pt–Pt bond distance change during the MD simulations are shown in Fig. 5(b₁) and (b₂), respectively. At the initial period (500 ps) of the MD simulations, only thermal motion of Pt and H atoms were observed. However, in contrast to Pt₁₃H₈, there are four additional Pt-H bonds formed and the Pt–Pt bond strength was weakened. Consequently, the symmetry of the Pt₁₃ icosahedron is destroyed and the icosahedron starts to evolve to an fcc-like cluster. Indeed, Fig. 5(b) clearly indicates that with an fcc-like structure, the Pt₁₃H_n hydride can have a minimum total energy and shortest average Pt–Pt bonds. As shown in Fig. 4(a₄), the fcc feature is gradually consummated until the edge sites are all occupied (n=24).

The homogeneously absorbed H atoms at the edge sites are all distributed around the fcc Pt₁₃ and formed two types of 4-membered H rectangles. One type of the rectangles has a Pt atom in the plane center (denoted as Pt@H₄), while the other type of rectangles has no central Pt atoms, namely, the Vac-H₄ sites. Our calculations indicate  that  the  hydrogen  adsorption  energy  at  the Pt@H₄ site is around 0.12 eV higher than the Vac-H₄ site, which is calculated to be 1.41 eV. The Pt@H₄ were then sequentially filled by up to 12 H atoms until n=36, as shown in Fig. 4(a₅) and the middle part of Fig. 5(a). All the hydrogen atoms were absorbed on the surface of the Pt₁₃ cluster and the average Pt–Pt bond lengths fluctuate in a small range of 2.83-2.88Å with the increasing H loading. Additional H atoms placed on the 6 Vac-H₄ sites will penetrate into the cluster and interact with the core Pt atom [Fig. 4(a₆)–(a₈)]. Indeed, the endohedrally adsorbed H atoms will significantly weaken the interactions between the core and shell Pt atoms, leading to further expansion of the Pt₁₃ structure, as shown in the third stage in Fig. 5(a). Such adsorption behavior of H on the Pt₁₃ icosahedral cluster is quite different from the smaller close-packed clusters because those smaller clusters have no inner atoms. However, such phenomena have been observed for H adsorption on the Pt crystalline surfaces where H atom could diffuse into the Pt lattice after all the surface sites are filled [9].

Figure 6 displays the H₂ dissociative chemisorption energies (\mathrm{\Delta }{E}_{\mathrm{CE}}), the H sequential desorption energies (\mathrm{\Delta }{E}_{\mathrm{DE}}) and the Hirshfeld charge  transfer from Pt₁₃ to H atoms (\mathrm{\Delta }Q) with respect to the H coverage. Before the edge sites were filled up, both \mathrm{\Delta }{E}_{\mathrm{CE}} and \mathrm{\Delta }Q curve fluctuates in a very small range of 1.55-1.68 eV and 0.05-0.08 electrons, respectively. Upon the phase transition, when H atoms penetrate into the cluster, \mathrm{\Delta }{E}_{\mathrm{CE}} and \mathrm{\Delta }{E}_{\mathrm{DE}} start to drop rapidly to 0.9 eV and 2.02 eV,respectively. Compared to smaller Pt clusters, we can conclude that the phase transition will be beneficial to the catalytic activity of Pt₁₃ because the dissociative chemisorption energy of H₂ on those small clusters drops rapidly with the increasing loading. The cluster itself relaxes [Fig. 5(a)] due to the H adsorption, allowing additional H atoms to reside at Pt@H₄ and Vac-H₄ sites. Compared to the single crystalline surface, the threshold \mathrm{\Delta }{E}_{\mathrm{CE}} is about 0.2 eV higher, which indicates higher reactivity towards H₂ dissociation. In contrast, the threshold \mathrm{\Delta }{E}_{\mathrm{DE}} is 0.6 eV lower than on the single crystalline surface, implying that the desorption of H atoms is much easier. In addition, compared to the close-packed small Pt clusters, \mathrm{\Delta }{E}_{\mathrm{CE}} is almost identical while \mathrm{\Delta }{E}_{\mathrm{DE}} is slightly lower by about 0.2 eV. Apparently, the vast difference of the sizes and shapes will not change the \mathrm{\Delta }{E}_{\mathrm{CE}} and \mathrm{\Delta }{E}_{\mathrm{DE}} significantly.

3.3 Hydrogen chemisorption on close-packed Pd_n clusters (n=2–9)

We now investigate the H₂ dissociative chemisorption, H sequential desorption on the Pd clusters as well as their structural phase transition. In contrast to the Pt clusters, the edge site was identified as the most favorable binding site for H atoms on both close-packed and icosahedral Pd clusters. Figure 7(a) displays the optimized structures of the fully H saturated Pd_n(n = 2–9) clusters. It is interesting to note that the H capacity of small Pd clusters is roughly half of that of small Pt clusters except for n=2 and 3. Most of the adsorbed H atoms reside at the edge and hollow sites, while H atoms tend to fill in the on-top sites of the close-packed Pt clusters first. For larger Pd clusters, some of the on-top sites are also populated.

In Fig. 8(a), the calculated sequential dissociative chemisorption energies E_{\mathrm{CE}} of H₂ molecules on the Pd_n (n=2–9) clusters are displayed. In general, the energy decreases as H coverage increases. Furthermore, at low H coverage, the dissociative H₂ chemisorption energies are reduced with cluster size. Except for n=2 and 3, the calculated threshold E_{\mathrm{CE}} fluctuates between 0.6-0.9 eV, slightly lower than that of small Pt clusters. For Pd₂H₂, the surprisingly large E_{\mathrm{CE}} value can be attributed to its relatively stable electronic structure. Pd₂ cluster is magnetic with two unpaired electrons. Upon H₂ chemisorption, its electronic structure becomes a close-shell and thus stable. Figure 8(b) shows the calculated sequential desorption energies of H atoms from the Pd clusters. For n=2 and 3, the desorption energies are simply the values of their dissociative chemisorption energies. For larger clusters, the calculated threshold E_{\mathrm{DE}} varies in a range of 2.29-2.80 eV, similar to what was found for Pt clusters, which is considerably higher than the value of H desorption on Pd crystalline surface at high coverage. Although the general trend is that E_{\mathrm{DE}} decreases with cluster size, there appears some fluctuation in the calculated desorption energies due to the fact that the sequential desorption energy is evaluated from the difference between E_{{\mathrm{Pd}}_n{\mathrm{H}}_x} and E_{{\mathrm{Pd}}_n{\mathrm{H}}_x-2} and some of the smaller clusters can be less stable than the larger ones. Figure 8(c) shows the calculated sequential Hirshfeld charges transferred from Pd clusters to H atoms. Similar to small Pt clusters, the Pd clusters donate electron to H atoms and the charge transfer from Pd clusters to H atoms increases monotonically as H coverage increases, leading to hydride formation.

3.4 Hydrogen chemisorption on the Pd₁₃ cluster

The calculated H₂ dissociative chemisorption energy on the edge site of the Pd₁₃ cluster is 1.40 eV. The H₂ dissociation and H diffusion activation barriers are calculated to be 0.16 eV and 0.07 eV, respectively. Compared to the Pt₁₃ cluster, the dissociation of H₂ on Pd₁₃ is slightly more difficult, while the H diffusion is much more facile. Sequentially introducing H atoms to the Pd₁₃ cluster also leads to the phase transition from the icosahedral to fcc-like structures. A few selected Pd₁₃H_n hydride structures are displayed in Fig. 9. The phase transition point is located at n=20 and the transformation consummates at n=24 where all the edge sites are filled. Unlike the Pt₁₃ cluster, the two additional H atoms introduced to the Pd₁₃H₂₄ hydride would be pulled into the cluster and interact with the core Pd atom instead of staying on the surface. However, the Pd₁₃ cluster can hold only 2 H atoms inside, while the Pt₁₃ cluster can accommodate 6 H atoms inside. The H desorption energy on Pd₁₃ is lower than on Pt₁₃. Consequently, H atoms would be much easier to desorb from Pd than Pt clusters, implying that Pt clusters may have a much higher hydrogen sorption capacity. Indeed, upon full H saturation, the Pd₁₃ cluster can only hold around 2/3 of the amount of H atoms that the Pt₁₃ cluster can hold. Additional H atoms placed on the Pd₁₃H₂₆ hydride will not penetrate into the cluster but absorb at the Pd@H₄ corner sites instead. The saturation point identified by MD simulation is located at n=30.

The calculated H₂ threshold dissociative chemisorption energy and H desorption energy at full saturation are 0.76 eV and 2.04 eV [Fig. 10(a) and (b)], comparable to those values on smaller close-packed Pd clusters. Therefore, we believe that, these two important properties, \mathrm{\Delta }{E}_{\mathrm{CE}} and \mathrm{\Delta }{E}_{\mathrm{DE}}, do not vary with sizes and shapes for both Pt and Pd clusters with up to 13 atoms. The main factor that controls the catalytic properties towards H₂ dissociative chemisorption is how many Pt or Pd atoms can be accessed by H₂ molecules. As shown in Fig. 10(c), the loss of electrons \mathrm{\Delta }Q from metal cluster to average per H atom drops almost linearly with the increasing H loading. We also note that charge transfer of Pd₁₃H_n hydride is much higher than that of Pt₁₃H_n.

3.5 Electronic properties

Figure 11 displays the calculated density of states (DOS) of both bare and fully saturated Pt₁₃ and Pd₁₃ clusters. Upon H₂ adsorption on Pt₁₃ and Pd₁₃ clusters, hydrogen coverage alters the electron spin density of the cluster surface, which affects the electronic properties of catalysts. Our calculations indicate that the bare clusters exhibit certain magnetic moments. Upon full saturation by H atoms, the Pt and Pd hydrides exhibit close-shell features due to the fact that the unpaired Pd/Pt-d electrons will interact with the H-1s orbital, which annihilates the magnetic moments. Detailed analysis shows that the valence bands are contributed mostly by the d-orbitals of metal atoms and the s-orbitals of H atoms. The conduction bands are mostly contributed by the 5d-and 6s-orbitals of Pt atoms, while 4d-and 5s-orbitals of Pd atoms. Since the interaction between H-1s and Pt-5s is stronger than the interaction between H-1s and Pd-4d, the band gap of Pt₁₃H₄₄ is smaller than that of Pd₁₃H₃₀.

4 Conclusions

Pt and Pd clusters catalyzed hydrogenation reactions play a very important role in many applications and industrial processes and thus have initiated intensive theoretical and experimental studies. In this review, we summarize our recent studies that attempted to quantitatively address the catalytic performance of small Pt and Pd clusters toward H₂ chemisorption and H desorption using first principles methods. We chose several close-packed Pt_n and Pd_n (n=2–9) clusters, as well as icosahedral Pt₁₃ and Pd₁₃ clusters to systematically investigate the size/shape effects on the sequential H₂ dissociative chemisorption and H desorption. For both Pt₁₃ and Pd₁₃, their H₂ dissociative chemisorption energy and H desorption energy at full saturation are comparable to what were calculated for smaller close-packed clusters. This clearly suggests that some important properties related to the catalytic performance of transition metal clusters may not vary significantly with the particle size or shape. However, these properties are dependent on the available surface metal atoms that can be accessed by H atoms. In addition, our study indicates that Pt clusters have a much higher hydrogen adsorption capacity than Pd clusters. We are also aware that our calculations are based on the Pt and Pd clusters with no more than 13 atoms. Hence, our conclusions are probably limited to the subnano clusters. Further studies on larger-scale clusters are required to enhance our understanding of the size-dependency of metal catalytic performance.

References

[1] B. C. H. Steele, Nature (London), 1999, 400: 619

[2] S. F. Parker, C. D. Frost, M. Telling, P. Albers, M. Lopez, and K. Seitz, Catal. Today, 2006, 114: 418

[3] P. S. Cremer, X. Su, Y. R. Shen, and G. A. Somorjai, J. Am. Chem. Soc., 1996, 118: 2942

[4] Y. Li and R. T. Yang, J. Am. Chem. Soc., 2006, 128: 726

[5] Y. Li and R. T. Yang, J. Am. Chem. Soc., 2006, 128: 8136

[6] T. E. Felter, S. M. Foiles, M. S. Daw, and R. H. Stulen, Surf. Sci. Lett., 1986, 171: 379

[7] S. Horch, H. T. Lorensen, S. Helveg, E. Lægsgaard, I. Stensgaard, W. Jacobsen, J. K. Nøskov, and F. Besenbacher, Nature (London), 1999, 398: 134

[8] S. C. Badescu, P. Salo, T. Ala-Nissila, S. C. Ying, K. Jacobi, Y. Wang, K. Bedurftig, and G. Ertl, Phys. Rev. Lett., 2002, 88: 136101

[9] P. Légaré, Surf. Sci., 2004, 559: 169

[10] M. K. Oudenhuijzen, J. A. Bokhoven, J. T. Miller, D. E. Ramaker, and D. C. Koningsberger, J. Am. Chem. Soc., 2005, 127: 1530

[11] X. Liu, H. Dilger, R. A. , R. A. Eichel, J. Kunstmann, and E. Roduner, J. Phys. Chem. B, 2006, 110: 2013

[12] G. Papoian, J. K. Nøskov, and R. Hoffmann, J. Am. Chem. Soc., 2000, 122: 4129

[13] D. Godbey and G. A. Somorjai, Surf. Sci., 1988, 204: 301

[14] K. Christmann, Surf. Sci. Rep., 1988, 9: 1

[15] B. Hammer and J. K. Nøskov, Nature (London), 1995, 376: 238

[16] R. A. Olsen, G. J. Kroes, and E. J. Baerends, J. Chem. Phys., 1999, 111: 11155

[17] G. W. Watson, R. P. K. Wells, D. J. Willock, and G. J. Hutchings, J. Phys. Chem. B, 2001, 105: 4889

[18] K. Nobuhara, H. Kasai, W. Q. Dino, and H. Nakanishi, Surf. Sci., 2004, 566–568: 703

[19] K. Nobuhara, H. Kasai, H. Nakanishi, and A. Okiji, J. Appl. Phys., 2002, 92: 5704

[20] L. Barrio, P. Liu, J. A. Rodriguez, J. M. Campos-Martin, and J. Fierro, J. Chem. Phys., 2006, 125: 164715

[21] Z. P. Liu and P. Hu, J. Am. Chem. Soc., 2003, 125: 1958

[22] X. Q. Gong, A. Selloni, O. Dulub, P. Jacobson, and U. Diebold, J. Am. Chem. Soc., 2008, 130: 370

[23] X. Liu, H. Dilger, R. A. Eichel, J. Kunstmann, and E. Roduner, J. Phys. Chem. B, 2006, 110: 2013

[24] Y. Okamoto, Chem. Phys. Lett., 2006, 429: 209

[25] Y. Okamoto, Chem. Phys. Lett., 2005, 405: 79

[26] G. E. Gdowski, J. A. Fair, and R. J. Madix, Surf. Sci., 1983, 127: 541

[27] L. J. Richter and W. Ho, Phys. Rev. B, 1987, 36: 9797

[28] C. T. Au, T. J. Zhou, and W. J. Lai, Catal. Lett., 1999, 62: 147

[29] N. Watari and S. Ohnishi, J. Chem. Phys., 1997, 106: 1997

[30] L. Chen, A. C. Cooper, G. P. Pez, and H. Cheng, J. Phys. Chem. C, 2007, 111: 5514

[31] C. Zhou, J. Wu, A. Nie, R. C. Forrey, A. Tachibana, and H. Cheng, J. Phys. Chem. C, 2007, 111: 12773

[32] C. Zhou, S. Yao, J. Wu, R. Forrey, L. Chen, A. Tachibana, and H. Cheng, Phys. Chem. Chem. Phys., 2008, 10: 5445

[33] W. Dong and J. Hafner, Phys. Rev. B, 1997, 56: 15396

[34] J. P. Perdew and Y. Wang, Phys. Rev. B, 1992, 45: 13244

[35] T. A. Halgren and W. N. Lipscomb, Chem. Phys. Lett., 1977, 49: 225

[36] F. L. Hirshfeld, Theor. Chim. Acta B, 1977, 44: 129

[37] B. Delley, J. Chem. Phys., 2000, 113: 7756

[38] B. Delley, J. Chem. Phys., 1990, 92: 508

[39] A. Nie, J. Wu, C. Zhou, S. Yao, C. Luo, R. C. Forrey, and H. Cheng, Int. J. Quantum Chem., 2007, 107: 219

Rights & permissions

©Higher Education Press and Springer-Verlag 2009