Density Functional Theory Study of Benzene Adsorption on Small Pd

11068

J. Phys. Chem. C 2007, 111, 11068-11076

Density Functional Theory Study of Benzene Adsorption on Small Pd and Pt Clusters Maurı´cio T. de M. Cruz,† Jose´ Walkimar de M. Carneiro,*,†,‡ Donato A. G. Aranda,§ and Michael Bu1 hl| Po´ s-Graduac¸ a˜ o em Quı´mica Orgaˆ nica, Departamento de Quı´mica Inorgaˆ nica, Instituto de Quı´mica, UniVersidade Federal Fluminense, Outeiro de Sa˜ o Joa˜ o Batista, s/n, 24020-150 Nitero´ i - RJ, Brazil, Laborato´ rio de Quı´mica Verde (Greentec), Escola de Quı´mica, Centro de Tecnologia, UniVersidade Federal do Rio de Janeiro, E-211, 21949-900 Rio de Janeiro - RJ, Brazil, and Max Planck Institut fu¨r Kohlenforschung, Kaiser-Wilhelm-Platz 1, Mu¨lheim an der Ruhr, D-45470, Germany ReceiVed: April 2, 2007; In Final Form: May 21, 2007

Restricted and unrestricted B3LYP/LANL2DZ and B3LYP/6-31G(d) calculations are employed to study the adsorption of benzene on small Pd7, Pd(7+3), Pt7, and Pt(7+3) clusters. Restricted calculations give the bridge30 adsorption site as the most stable for adsorption of benzene on Pd7 cluster, while on Pd(7+3) benzene is strongly tilted, leading to an adsorption mode where two benzene π-bonds more strongly interact with the metal cluster. Unrestricted calculations (triplet spin state) strongly favor the tilted conformations so that on both Pd7 and on Pd(7+3) the tilted di-π-hcp adsorption mode is the most stable. On the Pt7 and Pt(7+3) clusters even the restricted calculations lead to the tilted di-π adsorption mode as the most stable, a result that is reinforced by the unrestricted formalism, which also favors the tilted adsorption mode. The differences between the restricted and the unrestricted calculations are much larger for platinum than for palladium, which is related to the fact that atomic palladium has closed shell electronic configuration while atomic platinum has open shell electronic configuration. The adsorption process is dominated by electron donation from the highest occupied molecular orbital of benzene to the lowest unoccupied molecular orbital of the metal clusters.

Introduction The adsorption of organic and inorganic molecules on solid surfaces is of considerable interest to material science. Several reactions occurring on surfaces, such as catalysis, corrosion, coating, etc., involve adsorption as the first step,1 and adsorption of small inorganic molecules on metal surfaces has been quite throughly studied.1,2 The interaction of organic aromatic molecules with a transition metal surface can activate them for subsequent catalytic transformations. In this respect, benzene represents the prototypical aromatic system, not only for experimental studies but for computations as well, mainly due to its small size and high symmetry. The adsorption of benzene on different transition metal surfaces has been the subject of ongoing research for a long time.3-19 Even though several experimental5-9,11,13-19,22-24 and theoretical3,4,10,12,20,21,25 procedures have been employed, the orientation of the aromatic ring over the metal surface is not yet firmly established.4,21,25 Indeed, experimental and theoretical studies lead to contrasting conclusions, mainly because the number of variables that have to be controlled is considerably large. The most commonly studied surface for adsorption of aromatic molecules on platinum and palladium is the closepacked (111) face, not only due to its high symmetry but also because it is thermodynamically the most stable.20,21 Temperature-programmed desorption (TPD) experiments have shown * To whom correspondence should be addressed. † Po ´ s-Graduac¸ a˜o em Quı´mica Orgaˆnica, Instituto de Quı´mica, Universidade Federal Fluminense. ‡ Departamento de Quı´mica Inorga ˆ nica, Instituto de Quı´mica, Universidade Federal Fluminense. § Universidade Federal do Rio de Janeiro. | Max Planck Institut fu ¨ r Kohlenforschung.

that the monolayer of benzene on the Pt(111) surface consists of two chemisorbed states, which are desorbed with activation energies of 19.6-21 kcal mol-1 and 30-31 kcal mol-1, respectively, over the 280-520 K temperature range.16,18 Previous experiments by Gland and Somorjai22 showed evidence for a possible π f σ-bond transition upon adsorption of benzene on Pt surfaces. When adsorbed on Pt/Al2O3 benzene coverage reaches 0.31,23 which was rationalized by assuming that some C-H bonds of the chemisorbed benzene are not oriented parallel to the metal surface,17,24 which is an indication for stronger interaction between benzene and the Pt surface. This would likely result in enhanced surface coverage, because a certain fraction of the adsorbed benzene molecules would become σ-bonded to the Pt surface.26-29 Gland and Somorjai22 reported a di-σ-bonded benzene formed in a multistep mechanism, in contrast to the formation of the π-bonded benzene occurring in a single step. Spectroscopic studies for adsorption of benzene on Pt surfaces, usually done using single-crystal analytical characterization techniques,11,17-19,22 yielded results that are not conclusive. At low coverage, benzene adsorbs through its π-electron system, leading to a flat adsorption form.11 However, at high coverage tilted adsorption forms are also observed.7,30 Recent periodic slab density functional theory (DFT) calculations for adsorption of benzene on the Pt(111) surface appear to converge to a flat adsorption form,3,4,21,25,31 with adsorption energies ranging from -214,31 to -29 kcal mol-1.3,21 However, cluster calculations not only yielded higher adsorption energies (up to -43 kcal mol-1) as also lead to tilted adsorption forms.21 In general, it has been agreed that benzene adsorbs parallel to the metallic surface using its π-system to bind to the platinum atoms, although a stable Pt-C-H σ-complex has been also suggested.32

10.1021/jp072572c CCC: $37.00 © 2007 American Chemical Society Published on Web 07/04/2007

Benzene Adsorption on Small Pd and Pt Clusters The results for adsorption of benzene on palladium surfaces are similar to those for platinum. Spectroscopic investigations at room temperature indicate that benzene adsorbs on a single adsorption site of the Pd(111) surface, forming a π-bonded complex, with the aromatic ring oriented parallel to the metal surface. While angle-resolved ultraviolet photoemission investigations indicate an adsorption site with C6V symmetry,13 highresolution electron energy loss spectroscopy measurements suggest adsorption in Cs symmetry.4,14 With the assumption that effects from deeper substrate layers can be neglected, C6V symmetry is only compatible with benzene adsorption on atop sites, a geometry that was also favored by a theoretical investigation.33 Low-energy electron diffraction,9 near-edge X-ray absorption fine structure,7,15 ultraviolet photoelectron spectroscopy,7 X-ray photoelectron spectroscopy,15 and TPD30 experiments all suggest that the adsorption geometry depends on the degree of coverage, being parallel to the surface for low coverage or tilted in the case of high coverage. The most recent periodic slab DFT calculations suggest that the bridge site is the energetically most favorable site for benzene adsorption on Pd(111),4,12,15 with adsorption energies ranging from -274 to -33 kcal mol-1.12 In the present work, we provide additional data about benzene adsorption on small Pt and Pd clusters. The main goal is to broaden the knowledge about energies and geometries for benzene adsorption on small clusters34 when no extended flat surface is available. Indeed, this may considerably diverge from previous reported calculations, usually done on extended surfaces. Our motivation is related to the mechanism of ester enantioselective hydrogenation35,36 that occurs on supported Pt and Pd catalysts in the presence of chiral modifiers such as the cinchona alkaloids. The enantioselective hydrogenation of pyruvate esters catalyzed by Pt modified by the cinchona alkaloids cinchonidine and cinchonine has received much attention recently, as one of the rare effective heterogeneous enantioselective catalysts.37-39 The most widely accepted reaction model for the ester hydrogenation, the so-called 1:1 interaction model, is based on the hypothesis that the cinchona alkaloids adsorb intact on the Pt catalyst surface via the quinoline π-electron system.38,40,41 It has been generally accepted that the adsorption of such modifiers on a Pt or Pd surface, via the quinoline moiety,42 provides a chiral adjacent site at which selective enantioface adsorption of pyruvate occurs with subsequent hydrogenation affording preferentially one enantiomer. Therefore a crucial step in this process is the interaction of the aromatic quinoline moiety with the supported catalyst. At high coverage, the alkaloid is indicated to bind via the nitrogen lone pair to Pt(111) or Pd(111), but at low concentration, more relevant for catalysis, binding occurs via the π-system of quinoline, which is oriented flat (or nearly so) on the metal surface.43,44 According to DFT computations, both of the fused aromatic rings of quinoline appear to attach to the metal to similar extent.45 Here, we used benzene as archetypical model for this π-interaction, striving to provide reference data for eventual comparison with pyridine and quinoline. Although a very large array of metal atoms would be needed to represent an extended metal surface, a supported metal catalyst may also be described reasonably well with a smaller cluster. Experimental evidence shows that on Al2O3 supported platinum catalyst, the average metal coordination number is rather small, indicating that the number of metal atoms in each cluster is also small (on average 3 to 4 atoms),46 although Pt agglomeration may be observed under sulfur-poisoning conditions. In this respect, cluster models may even represent the

J. Phys. Chem. C, Vol. 111, No. 29, 2007 11069 actual supported systems better than extended, plain surfaces. We have now studied the interaction between small metal cluster fragments consisting of up to 10 atoms with benzene, assuming that these interactions will be more important than long-range effects, for example, those stemming from the oxide support. Computational Methods All calculations were done with the B3LYP hybrid functional as proposed and parametrized by Becke.47 This is a mixture of Hartree-Fock and DFT exchange terms with the gradientcorrected correlation functional of Lee et al.48 Calculations were done using both restricted and unrestricted formalism. In the restricted calculations, the LANL2DZ pseudopotential, as proposed by Hay and Wadt,49 was employed to optimize the geometry of benzene interacting with the fixed clusters. LANL2DZ is an effective core potential (ECP) that includes mass-velocity relativistic effects for core electrons. For palladium and platinum, the ECPs include the innermost 28 and 60 electrons, respectively. The remaining 18 “valence” electrons in each metal atom are described by a set of minimal basis set representing the 4s and 4p orbitals of palladium as well as the 5s and 5p orbitals of platinum, together with a double-ζ basis set representing the 4d, 5s, and 5p orbitals of palladium and the 5d, 6s, and 6p orbitals of platinum, respectively.50 For carbon and hydrogen atoms, the valence double-ζ basis sets of Dunning and Huzinaga (D95V)51 are employed. Single-point energy calculations were performed replacing the D95V basis set on benzene with the 6-31g(d) basis set, while retaining the LANL2DZ pseudoptotential for the metals. The B3LYP/ LANL2DZ combination has recently been shown to be not very accurate in reproducing geometries of first-row TM complexes.52 We are interested in relative trends for closely related systems, however, which should qualitatively be captured already at that level. Only the energies obtained with the larger basis set are discussed. The calculations described above were done using closedshell restricted formalism (singlet wavefunction). Nevertheless, whereas atomic palladium has 4d10 closed-shell electronic configuration, atomic platinum has a 5d96s1 ground state electronic configuration that could lead to instabilities in the restricted wavefunction. Considering this possibility, all calculations were repeated with the unrestricted formalism and testing the stability of the final wavefunction. Several electronic states for different spin multiplicities of the bare palladium and platinum clusters were determined, and the two or three most stable in each case were used to further reoptimize the geometry of benzene on the clusters. For the unrestricted calculations, we used the same combination of method (B3LYP) and basis set (LANL2DZ and 6-31g(d)) as was used for the restricted case and did not use any symmetry restrictions. Optimizations were done starting from the restricted optimized geometry using the larger basis set. One-layer M7 and two-layer M(7+3) hexagonal “benzene-like” clusters were designed in an arrangement that would represent a small slice of the (111) face of the bulk metal. Therefore, a bidimensional expansion of these clusters should reproduce the metal fcc(111) surface, usually the thermodynamically most stable. These are the smallest cluster where all the unique adsorption sites of the metal surface may be reproduced, albeit with highly unsaturated border atoms. While the M7 cluster has the best compromise between size and computational cost, the M(7+3) cluster may give information on the effect of a second metal layer and increased saturation on the optimized geometries and adsorption energies in addition to allowing for differentiation

11070 J. Phys. Chem. C, Vol. 111, No. 29, 2007

Cruz et al.

Figure 1. Possible arrangements for the interaction of benzene with the M7 and M(7+3) metal clusters.

between the so-called hollow hexagonal close-packed (hcp) and hollow face-centered cubic (fcc) sites. In all calculations the nearest neighbor Pt-Pt and Pd-Pd distances were held fixed at their bulk values (Pt-Pt ) 2.77 Å and Pd-Pd ) 2.75 Å).53 All calculations were performed using the Gaussian suite of programs.54 Geometric and electronic parameters for the benzene molecule, as well as adsorption energies, were obtained from each benzene/Mn arrangement. Adsorption energies were computed by subtracting the energies of the isolated benzene molecule and of the bare metal cluster from the energy of the benzene/ Mn system, according to eq 1 below. Calculations were done at 0 K without zero-point vibrational corrections, as these are difficult to compute for partially optimized structures. The adsorption energy corresponds to the energy difference between the benzene/Mn complex and its components in the corresponding spin state at infinite separation.

Ead ) E(benzene/Mn) - [E(benzene) + E(Mn)]

(1)

A negative value for the adsorption energy means that the corresponding adsorbed state is thermodynamically more stable than the unbounded state. Results The several unique sites for interaction of the benzene molecule with the M7 and M(7+3) clusters are shown in Figure 1. In the atop(0) and atop(30) sites the benzene molecule lies flat over the center of the cluster allowing interactions mainly

with the highly coordinated central metal atom. As we restricted optimization by symmetry, the aromatic ring is not allowed to move laterally, therefore the main degree of freedom is the distance between the aromatic ring and the metal plane. In the bridge(30) and hollow(0) sites, interactions between the π-bonds of the aromatic moiety with either two or three metal atoms of the cluster are favored. In the bridge(30) site, the central and one of the peripheral metal atoms are mainly involved in the interaction, while in the hollow(0) site three π-interactions are possible, one involving the central atom and two with peripheral metal atoms. Finally the bridge(0) and hollow(30) sites are those that lead preferentially to σ-type interactions. In the bridge(0) site, this involves the central and one of the peripheral metal atoms, while in the hollow(30) site it involves the central and two peripheral metal atoms. Note that while these sites are equivalent to those that have been previously simulated in slab periodical4,12,25 or cluster calculations,21,34 in the present case the position of the benzene molecule on the clusters is allowed to fully optimize, whereby being free to migrate to the border of the cluster, leading to optimized arrangements where the aromatic ring may be strongly tilted. This is mainly the case for the bridge and hollow sites. Migration from the hollow(0) site to the border of the cluster may result in a structure where two metal atoms interact with two π-bonds of benzene in an arrangement that we call di-π. As will be shown below in the final optimized geometry, this type of interaction is similar to those found on interaction of 1,3-butadiene with transition metal atoms.55 In contrast, benzene migration from the hollow(30) site to the border of the cluster results in a structure with a typical 1,3-di-σ bond. Therefore, the final optimized geometries may differ somewhat from those usually described for adsorption on hollow and bridge sites. It should be added that on the hollow sites of the M(7+3) cluster two adsorption forms, a hcp and a fcc type, are possible. In the hcp site, there is no metallic atom of the second layer immediately below the center of the aromatic ring, as is the case for the fcc site. This produces distinct arrangements, as shown in Figure 1. Benzene Adsorption on Pd7 and Pd(7+3) Clusters. Previous experimental studies on benzene adsorption on Pd(111) revealed that the preferred adsorption form depends on the degree of coverage of the metal surface. At low coverage, benzene adsorbs parallel to the surface, while at higher coverage tilted adsorption forms were also observed.7,15,19,30 In the present study, a clear result emerges first. There is no adsorption of benzene on either of the atop sites. During geometry optimization on these sites, the benzene molecule moves away from the metal surface with the interaction energy approaching zero when the distance between benzene and the metal surface is near to 4 Å. At this large separation distance, the geometry of benzene remains unaltered as compared to that of isolated benzene. As this was a general rule and also for platinum, atop sites are not discussed anymore. Restricted calculations indicate that for benzene adsorption on the Pd7 cluster the preferential site is the bridge(30) (Figure 2, Table 1). On this site, benzene adsorbs parallel to the metal surface with adsorption energy of -29.0 kcal mol-1. While the bridge(0) site is only about 3 kcal mol-1 less stable, both hollow sites are at least 6 kcal mol-1 less stable. This adsorption energy, although fortuitously, agrees well with previous experimental studies for benzene adsorption at low coverage, as measured by TPD (31.1 kcal mol-1),30 and is only 2 kcal mol-1 lower than previous calculations.4,12 All C-C bond lengths increase (1.448 to 1.465 Å, Figure 2) in relation to the C-C bond lengths in free benzene (1.40 Å).

Benzene Adsorption on Small Pd and Pt Clusters

J. Phys. Chem. C, Vol. 111, No. 29, 2007 11071

Figure 2. Preferential adsorption geometries with S ) 0 (singlet) and their respective C-C and Pd-C bond lengths (Å) and C-H bond inclination angle (R) for benzene adsorbed on Pd7 and Pd(7+3): bridge(30)/Pd7, (a) superior view and (b) lateral view; di-π-hcp/Pd(7+3), (c) superior view and (d) lateral view.

TABLE 1: B3LYP/6-31g(d)/LANL2DZ (6-31g(d) for Carbon and Hydrogen, LANL2DZ for the Metal) Adsorption Energies (kcal mol-1) for Benzene Adsorption on Pd7, Pd(7+3), Pt7, and Pt(7+3) Pd7 hollow(0) f di-π hollow(30) bridge(0) bridge(30)

Pt7

restricted

unrestricted (S ) 1)a

-22.91 -22.18 -25.88 -29.01

-26.17 -12.91 -11.87c -18.24c

restricted hollow(0) f di-π hollow(30) bridge(0) bridge(30)

Pd(7+3) restricted hollow(0)-hcp f di-π-hcp hollow(0)-fcc f di-π-fcc hollow(30)-hcp f di-σ-hcp hollow(30)-fcc f di-σ-fcc bridge(0) bridge(30)

Pt(7+3)

unrestricted (S ) 1)e

-29.79 -24.22 -26.16 -22.63 -27.84 -21.44 -24.23 -19.88 converged to di-π-hcp -24.03 -19.23

unrestricted

-68.59 -38.10 (S ) 3)b Converged to hollow(0) f di-π -38.10 converged to di-π -41.96 -9.78 (S ) 2)d

restricted hollow(0)-hcp f di-π-hcp hollow(0)-fcc f di-π-fcc hollow(30)-hcp f di-σ-hcp hollow(30)-fcc f di-σ-fcc bridge(0) bridge(30)

unrestricted (S ) 4)

-50.22 -39.83 -39.44 -31.81 converged to di-π-hcp -36.19 converged to di-π-fcc converged to di-π-hcp -32.51 converged to di-π-hcp

a Quintet (S ) 2) states are at least 5.1 kcal mol-1 less stable. bNo stable wavefunction with S ) 2 (quintet) was found for this adsorption form. The triplet state (S ) 1) has higher energy and strong spin contamination. cOptimization of the quintet state (S ) 2) of both the bridge(0) and the bridge(30) adsorption forms converged to the di-π form. dThe septet state (S ) 3) is 4.2 kcal.mol-1 less stable. The triplet state (S ) 1) has higher energy and strong spin contamination. eThe quintet states (S ) 2) are only about 1 kcal mol-1 less stable.

The C-H bonds bend away from the metal plane, indicating an increased sp3 character of the carbon atoms. The shortest Pd-C distances are 2.02 and 2.19 Å, corresponding to the distance from the carbon atoms to the edge and to the central metal atoms, respectively (Figure 2b). The addition of three palladium atoms in the second layer (Pd(7+3)) forces the aromatic ring to move to the edge of the cluster, leading to a tilted adsorption mode (Figure 2c,d), with adsorption energy of -29.8 kcal mol-1. The other adsorption modes, including the parallel one, are found in a 5.8 kcal mol-1 energy range (Table 1). The final optimized geometry of the most stable adsorption mode is now better described as a diπ-hcp form with the benzene molecule tilted by nearly 60° relative to the metal surface. The structural parameters (Figure

2) suggest the formation of two π-interactions, reminiscent of those found in a metal complexes interaction with 1,3butadiene.55a,b,c The C-C bonds that directly interact with the cluster are elongated (1.429 to 1.448 Å), while the C-C bond farther from the metal surface does not change (1.398 Å). The shortest C-Pd distance is 2.270 Å, corresponding to the distance between a carbon atom participating in the π-interaction and the nearest edge metal atom (Figure 2d). The C-H bonds of the carbon atoms that more strongly interact with the metal surface bend away from the carbon plane by 14.0° (Figure 2d). The di-π-hcp adsorption mode is 3.6 kcal mol-1 more stable than adsorption on the fcc site (-26.2 kcal mol-1). In the last case, the C-C bond lengths, the angles involving the C-H

11072 J. Phys. Chem. C, Vol. 111, No. 29, 2007 bonds, and the benzene tilting in relation to the metal plane are all very similar to those calculated for adsorption on the hcp site. For all the adsorption forms but the atop ones, for both Pd7 and Pd(7+3) clusters, the hydrogen atoms bend away from the metal plane. The highest bending is calculated for the bridge(30) adsorption site on Pd7, where bending of the C-H bonds in relation to the carbon ring is up to 25.4°. This fact points to loss of aromaticity in the benzene molecule as a result of rehybridization during the adsorption process. It may be observed that, as a general rule, elongation of the C-C bonds is less relevant in the tilted adsorption forms than in the parallel ones. By increasing the basis set size on carbons and hydrogens (from D95V to 6-31 g(d)), there is a corresponding increase in the adsorption energies (more negative values), leading to an improved description of the benzene/Pdn system. However, such effect is less pronounced for adsorption forms where benzene is tilted (on average 3 kcal mol-1 more strongly adsorbed with the larger basis set) than for parallel adsorption (on average 5 kcal mol-1 more strongly adsorbed with the larger basis set). For the tilted di-π form on Pd7, an opposed behavior is observed with the adsorption energy decreasing by 1.3 kcal mol-1 when the basis set increases. Nonparallel adsorption forms are more usual for adsorption on Pd(7+3), while parallel adsorption forms are predominant for adsorption on Pd7 clusters. Although the presence of the second layer increases the coordination number of the edge metal atoms, the corresponding increase in the coordination number of the central atom is three times larger therefore increasing the relative “unsaturation” (and, thus, the Lewis acidity) of the edge atoms in relation to the central atom. As a consequence, on the Pd(7+3) cluster the benzene molecule remains parallel to the cluster surface only in the bridge(30) adsorption mode. In the other adsorption forms, there is an unbalance of forces between the central and the edge metal atoms due to a higher number of C-Pd interactions involving the edge metal atoms in relation to the corresponding number of interactions involving the central metal atom. Such fact, added to the higher acidity of the edge metal atoms, could be responsible for the tilting of the benzene molecule in some adsorption forms. The degree of deformation of the benzene molecule due to the adsorption process may be given by its reorganization energy. This may be obtained as a difference in energy between the energy of the isolated fully optimized benzene structure and the energy calculated at the same level for benzene at the adsorbed geometry. The highest reorganization energy is found for adsorption on the bridge(30) site of Pd7, 29.1 kcal mol-1. This is not surprising, because it is also the form that leads to the highest adsorption energy. The reorganization energy for adsorption on the bridge(30) site of Pd7 is almost six times greater than the corresponding value for adsorption on the di-σ site of Pd(7+3) (5.3 kcal mol-1). In the most stable di-π-hcp adsorption form on Pd(7+3), the bond reorganization energy of benzene is only 7.2 kcal mol-1. The calculations described above were done using restricted wavefunctions in a singlet state. Because there could be more stable electronic states of higher multiplicities, we also calculated the relevant adsorption forms (excluding the atop ones) using the B3LYP unrestricted formalism with the same combination of basis set as above. The most stable electronic state of the Pd7 cluster is the triplet state (S ) 1). An unrestricted singlet state (S ) 0) 4.4 kcal

Cruz et al.

Figure 3. Preferential adsorption geometries with S > 0 and their respective C-C and Pd-C bond lengths (Å) and C-H bond inclination angle (R) for benzene adsorbed on Pd7 and Pd(7+3): di-π/Pd7 (S ) 1), (a) superior view and (b) lateral view; di-π(hcp)/Pd(7+3) (S ) 1), (c) superior view and (d) lateral view.

mol-1 higher could be found, however, with high spin contamination (〈Sˆ 2〉 larger than 1.3). A quintet state (S ) 2) was found 8.8 kcal mol-1 less stable than the triplet state. Although the triplet state is considerably more stable than the quintet one, all adsorption forms were calculated in both spin states as well as in the septet state (S ) 3). Following the same behavior observed for the bare cluster, the triplet state was found always more stable than either the quintet or the septet state for the four relevant adsorption forms on Pd7. The unrestricted calculations show a clear tendency to preferentially stabilize the tilted adsorption forms. Indeed, the bridge(0) and hollow(30) forms, when optimized in the quintet state, converged to the di-π geometry. This tendency is indicated by the relative energies of the four adsorption forms in the triplet state. While in the restricted formalism the bridge(30) is the most stable adsorption site and relative energies are clustered close together, in the unrestricted calculations the most stable form is the tilted di-π with adsorption energy of -26.2 kcal mol-1 (Table 1). Parallel adsorption forms are at least 8 kcal mol-1 (bridge(30)) and up to 14 kcal mol-1 (bridge(0)) less stable. The di-π form is somewhat more strongly tilted than in the restricted calculation (compare Figures 2 and 3). The metal and the carbon planes are almost perpendicular to each other, and the final optimized structure typically remembers a diene-like interaction. Calculation of Wiberg bond indices (WBI) using the natural bond orbital approach gives notable covalent contributions for Pd-C binding, with 4 WBIs between Pd/C pairs approaching or exceeding 0.3. Following the same behavior observed in the restricted calculations, the C-C bonds interacting with the cluster elongate and the C-H bonds bend away from the cluster by the same amount found in the restricted calculations. The Pd-C distances in the π-interaction are longer than the corresponding distances in the restricted calculation by about 0.1-0.2 Å. For the bare Pd(7+3) cluster, the most stable electronic state is the septet (S ) 3). However, for the adducts we found the triplet state as the most stable with the quintet state about 1.0 kcal mol-1 higher in energy. Relative energies among the different adsorption forms using the unrestricted calculations do not differ much from the restricted results. The most stable

Benzene Adsorption on Small Pd and Pt Clusters

Figure 4. Preferential adsorption geometries with S ) 0 and their respective C-C and Pt-C bond lengths (Å) and C-H bond inclination angle (R) for benzene adsorbed on Pt7 and Pt(7+3): bridge(30)/Pt7, (a) superior view and (b) lateral view; di-σ/Pt(7+3), (c) superior view and (d) lateral view.

adsorption forms are again the tilted arrangements, although with relative energies very close together. The less stable bridge(30) parallel adsorption form is only 5.0 kcal mol-1 above the most stable di-π adsorption form, whose adsorption energy is -24.2 kcal mol-1 (Table 1). The geometries of the di-π-hcp forms, as calculated with the restricted (Figure 2) and the unrestricted (Figure 3) formalisms, do not significantly diverge from each other. Differences in relevant bond distances are below 0.1 Å. The similarities between the restricted and the unrestricted geometries for the di-π-hcp adsorption forms also reflect in the benzene bond reorganization energies, which are 7.2 and 5.5 kcal mol-1 for the restricted and the unrestricted case, respectively. Benzene Adsorption on Pt7 and Pt(7+3) Clusters. Benzene adsorbed on Pt surfaces has been found to orient parallel to the metal surface.11,17,18,21,31 The adsorption site depends on the specific metal surface and on the presence of coadsorbates.55 On Pt(111) a disordered pure benzene layer is formed at bridge sites.24 Semiempirical atom superposition and electron delocalization molecular orbital calculation (ASED-MO) showed that adsorption on 3-fold hollow sites is the most stable for benzene coordination to platinum clusters.57 More recent58 ab initio calculations on the interaction of 1-3 benzene molecules with one or two Pt atoms showed that the π-interaction of the Pt atom with benzene is so strong that the benzene molecule is deformed from its normal regular hexagonal structure, resulting in low-symmetry structures. Although slab periodic DFT calculations considered benzene to adsorb flat on the Pt(111) surface,3,4,21,25 hybrid DFT calculations32 indicate that the benzene-Pt interaction leads to two stable structures, the traditional di-π complex and a σ-bonded structure, which is more stable than the former by 9.5 kcal mol-1. Restricted calculations for adsorption on both Pt7 and Pt(7+3) indicate the tilted di-π adsorption form (Figure 4) as the most stable with adsorption energies of -68.6 and -50.2 kcal mol-1, respectively (Table 1). Two points deserve attention when comparing theses results to those for palladium. The adsorption energy is more than 2 times larger for adsorption on platinum than that found for adsorption on palladium. Additionally, there is a larger difference in adsorption energy between the Pt7 and the Pt(7+3) clusters. On the larger Pt(7+3) cluster, the adsorption is much less exergonic, probably due to a higher saturation of the metal atoms in this cluster. The adsorption energies of the

J. Phys. Chem. C, Vol. 111, No. 29, 2007 11073 most stable parallel adsorption modes on Pt7 and on Pt(7+3) (bridge(30)) are 26 and 17 kcal mol-1, respectively, lower (less negative) than that of the respective di-π forms. However, adsorption energies above 50.0 kcal mol-1 are considerably higher than either experimental16-18 or theoretical reported values,3,4,31 although other calculations on small clusters also led to high adsorption energies.21 As will be shown below, the unrestricted formalism yields much more reasonable values for adsorption energies on the platinum clusters. As observed before for palladium, increasing the basis set on carbon and hydrogen atoms from D95V to 6-31g(d) results in stronger adsorption energies for the parallel adsorption forms. For platinum, the effect is on the average 8 kcal mol-1 for Pt7 and 7 kcal mol-1 for Pt(7+3). However, for the tilted adsorption mode more strongly bonded forms were also calculated with the larger basis set. Adsorption energies become more negative on average by 5.7 kcal mol-1. On both Pt7 and Pt(7+3), benzene adsorbs through two π-bonds, each involving one of the edge metal atoms and two carbon atoms (Figure 4). The tilt angle is larger on Pt7 (90°, Figure 4b) than on Pt(7+3) (67°, Figure 4d). In the tilted adsorption form on Pt7, the carbon pairs nearest to the metal cluster are slightly closer to the metal atoms than in the case of adsorption on Pt(7+3) (2.15 and 2.21 Å versus 2.19 and 2.23 Å, respectively). As a consequence, the C-C bonds formed by these pairs elongate more for adsorption on Pt7 than it does for adsorption on Pt(7+3) (1.456 Å versus 1.452 Å). As a general rule, C-C bonds closer to the metal cluster undergo larger elongation upon adsorption on platinum than on palladium clusters, indicating a higher sp3 character for adsorption on platinum. The smallest C-metal distance in the preferential adsorption forms on platinum clusters is smaller than it is on palladium clusters for the same adsorption site. Adsorption on the di-π-hcp site of Pt(7+3) is 10.8 kcal mol-1 more stable than adsorption on the di-π-fcc site. Nevertheless, on the fcc site benzene is more tilted than on the hcp site. The reorganization energy for benzene adsorbed in the di-π form on Pt7 is 42.5 kcal mol-1, more than twice as high as that calculated for Pt(7+3) (16.9 kcal mol-1). This is a consequence of the stronger interaction of benzene with Pt7 than with Pt(7+3), entailing a stronger deformation of the benzene geometry. The unrestricted calculations yield adsorption energies on Pt7 and on Pt(7+3) that significantly diverge from the restricted results. The most stable bare Pt7 cluster has a septet electronic state59 with the quintet state 9.6 kcal mol-1 higher. No stable wavefunction in the quintet state could be found for the tilted adsorption forms of benzene on the Pt7 cluster. In the septet state, the tilted di-π adsorption form is the most stable, 35.2 kcal mol-1 more stable than the corresponding parallel bridge(30) adsorption site (30.9 kcal mol-1 more stable than the bridge(30) adsorption site in the quintet state). Triplet wavefunctions always lead to higher energy than either the quintet or the septet wavefunctions. A drastic reduction in the adsorption energy was calculated with the unrestricted wavefunctions. For the tilted adsorption form in the septet state, the adsorption energy is reduced to -38.1 kcal mol-1. A corresponding reduction was calculated for the parallel bridge(30) adsorption form in the quintet state (Table 1). Addition of the second layer (Pt(7+3)) does not change this picture significantly. The most stable bare Pt(7+3) cluster has a septet electronic state; however, triplet and nonet electronic states could also be found that are only marginally less stable (by 0.8 and 0.5 kcal mol-1, respectively). Calculations of the several Pt(7+3)/C6H6 adducts in the different spin states also yield

11074 J. Phys. Chem. C, Vol. 111, No. 29, 2007

Figure 5. Preferential adsorption geometries with S > 0 and their respective C-C and Pt-C bond lengths (Å) and C-H bond inclination angle (R) for benzene adsorbed on Pt7 and Pt(7+3): di-π/Pt7 (S ) 3), (a) superior view and (b) lateral view; di-π(hcp)/Pt(7+3) (S ) 4), (c) superior view and (d) lateral view.

energies that are very close, although with the nonet wavefunctions systematically more stable than the corresponding wavefunctions of lower multiplicity. In the nonet state, the most stable adsorption form is the di-π-hcp with -39.8 kcal mol-1 adsorption energy. The alternative di-π-fcc adsorption form is 8.0 kcal mol-1 less stable (Table 1), while all the parallel adsorption sites converged to the corresponding tilted di-π form upon geometry optimization. On both Pt7 and Pt(7+3) clusters, benzene adsorbs strongly tilted (Figure 5). As observed for palladium and in the restricted calculations, the C-C bonds that directly interact with the metal atoms elongate up to 1.480 Å for Pt7 and 1.463 Å for Pt(7+3). The smallest Pt-C distances are 2.154 Å for Pt7 and 2.200 Å for Pt(7+3), corresponding to the Pt-C distance in the di-π interaction. The C-H bonds closer to the metal surface bend away by 21.2° for adsorption on Pt7 and 17.7° for adsorption on Pt(7+3). These values are essentially the same as those obtained in the restricted calculations, although closer Ptbenzene contact and stronger benzene deformation are indicated when compared to adsorption on palladium. This is reflected in the benzene bond reorganization energy. For adsorption on the Pt7 cluster, the benzene geometry deforms by 13.0 kcal mol-1, while for adsorption on Pt(7+3) the benzene bond reorganization energy is 14.0 kcal mol-1. The natural bond orbital (NBO) analysis for the most stable Pt(7+3) cluster in the nonet state gives 4 WBI of 0.34, indicating notable covalent contributions for Pt-C binding. Discussion The present restricted and unrestricted calculations allow us to derive relevant conclusions about the adsorption of benzene on small clusters of palladium and platinum. The most evident result of our calculations is that for these small clusters tilted conformations are preferred as compared to parallel adsorption forms. Adsorption modes with a di-π type interaction are the most stable in all cases we calculated. The final optimized geometry has an arrangement that resembles that found when metals interact with 1,3-butadiene with each of a pair of π-bonds of benzene more strongly interacting with one metal atom of the cluster. This general trend was observed for both palladium and platinum. Although restricted calculations yielded a parallel adsorption form for Pd7, the unrestricted calculations were

Cruz et al. uniform in giving only tilted adsorption modes as the most stable in all the cases we calculated. As a general trend, the energy differences between the parallel and the tilted conformations are smaller for adsorption on palladium than for adsorption on platinum. Similarly, differences between the restricted and the unrestricted calculations are significantly smaller for adsorption on palladium than for adsorption on platinum. This may be rationalized if we consider the electronic nature of palladium as compared to that of platinum. Atomic palladium has a closed shell d10 electronic configuration. For both the Pd7 and the Pd(7+3) clusters, electronic states of low multiplicity (triplet) were found as the most stable but with small energy differences between the restricted (singlet) and the unrestricted (triplet) calculations. Adsorption energies of tilted adsorption modes become more negative with the unrestricted methodology, but the increase in adsorption energy for the relevant di-π form is no more than 3 kcal mol-1. In contrast, for the parallel adsorption forms on Pd7 and on Pd(7+3) the unrestricted calculations lead to more positive adsorption energies. On the other hand, atomic platinum has an open shell d9s1 electronic configuration. It is thus not surprising that the most stable electronic states for both the Pt7 and the Pt(7+3) clusters are of higher multiplicity. The corresponding most stable Pt(7+3)/C6H6 adducts are also those of high multiplicity. This has as consequence that for platinum there is a considerably larger energy difference between the restricted and the unrestricted calculations. While restricted wavefunctions lead to an adsorption energy of up to -68.6 kcal mol-1, the unrestricted wavefunctions yield much more reasonable results with adsorption energy on the order of -38 to -40 kcal mol-1. If basis set superposition error (BSSE) is additionally considered (counterpoise calculations on selected adsorption forms yielded BSSE on the order of 5-8 kcal mol-1), the final adsorption energy approaches values that are reasonably close to the experimental measurements or previous calculations. However, it should be stressed that tilted adsorption forms are strongly preferred for adsorption on these small clusters. How do our results compare to previous calculations that predict the parallel bridge adsorption site as the most stable? As a first point, it should be emphasized that when considering only the parallel adsorption forms, our results also predict the bridge(30) adsorption site as the most stable, which is in complete agreement with the previous results.3,4,21,25,31 As a second point, experimental single-crystal characterization techniques as well as slab periodical calculations are performed under conditions where tilted, lateral adsorption forms, such as those found in the present case, are not allowed. On a flat, extended surface, parallel adsorption modes are doubtlessly the preferred adsorption form. In that case, the bridge(30) adsorption site has been predicted to be preferential, a result with which we agree, because from the possible parallel adsorption modes we also predict the bridge(30) site to be the most stable. However, when adsorbing on small clusters, or on strongly deformed and not flat surfaces, our results indicate that tilted adsorption forms may not be ignored. Tilted adsorption modes were also predicted by others who calculated adsorption on small clusters.21 Additionally, mainly for metals where a closed shell electronic configuration is not the fundamental state, unrestricted wavefunctions are essential to yield reasonable adsorption energies. The total Mulliken charge density on both platinum clusters is negative for all adsorption forms, as was observed for palladium. Therefore, the most strong interaction seems to originate from electron donation from the aromatic ring to the

Benzene Adsorption on Small Pd and Pt Clusters SCHEME 1 : LUMO and HOMO Energies for the Isolated Subunits, Pdn and Ptn (n ) 7 and n ) 7 + 3), and the Benzene Moleculea

J. Phys. Chem. C, Vol. 111, No. 29, 2007 11075 expected, charge transfer is favored when benzene adsorbs parallel to the metal plane, mainly due to better superposition of its π-orbitals with the metal conduction band. This leads, however, to lower adsorption energies. Conclusions

a For the metallic cluster, we take the (open-shell) energy of the highest β occupied orbital as the HOMO energy and the (open-shell) energy of the lowest R unoccupied orbital as the LUMO energy.

metal cluster, although some back-donation may also operate. Considering that donation from a doubly occupied benzene MO into a singly occupied cluster acceptor MO would result in little overall stabilization (the level of one electron would be raised), it is reasonable to accept that interaction between the donor and an empty acceptor MO could well be more favorable, even if that acceptor MO (lowest unoccupied molecular orbital (LUMO)) is higher in energy than any singly occupied orbital. Apparently, this is the case here because there is very low spin density on benzene. By analysis of the most stable open-shell (unrestricted) adduct in each case and taking the LUMO’s energy as the energy of the lowest R unoccupied orbital and the highest occupied molecular orbital’s (HOMO’s) energy as the energy of the highest β-occupied orbital, the picture given in Scheme 1 emerges. It shows that the energetically most favorable electronic interaction involves electron transfer from the HOMO of benzene to the LUMO of the metallic cluster. The cluster’s LUMO energies show a nice correlation with the adsorption energy in each case. The LUMO of Pd7 (-3.58 eV) is only 0.17 eV lower than the LUMO of Pd(7+3) (-3.41 eV). As a consequence, adsorption energies on Pd7 and on Pd(7+3) are similar, being, however, somewhat more negative on Pd7 (-26.2 kcal mol-1) than on Pd(7+3) (-24.2 kcal mol-1). The same behavior is found for platinum. The LUMO of Pt7 (-4.11 eV) is 0.21 eV higher than the LUMO of Pt(7+3) (-4.32 eV). The corresponding adsorption energy on Pt7 (-38.1 kcal mol-1) is less negative than the adsorption energy on Pt(7+3) (-39.8 kcal mol-1). Still more relevant, the LUMO’s energies of the platinum clusters are considerably lower (-0.5 eV for Pt7 and -0.9 eV for Pt(7+3)) than the corresponding LUMO’s energies of the palladium clusters. As a consequence, the adsorption energies on platinum and on palladium follow the same behavior, being considerably more negative for platinum (-11.9 kcal mol-1 for Pt7 and -15.6 kcal mol-1 for Pt(7+3)) than it is for palladium. The difference in LUMO’s energies also reflects the amount of charge transfer between benzene and the clusters when two identical adsorption forms for the same kind of cluster of palladium and platinum are compared. It should be added, however, that the amount of charge transfer between the benzene molecule and the metallic cluster do not show any direct relationship with the magnitude of the adsorption energy. As

This detailed density functional study for adsorption of benzene on Pdn and Ptn small clusters showed that adsorption energies as well as adsorption sites are strongly dependent on the electronic state. While for the palladium systems we found an unrestricted triplet ground state that is only slightly different from the restricted singlet state, for platinum clusters the unrestricted calculation of states with higher multiplicities (up to nonet) changes the adsorption energies by more than 30 kcal mol-1, concomitant with larger changes in the adsorption geometries. With the restricted formalism, we could locate some adsorption forms with benzene parallel to the metal surface. In contrast, the unrestricted formalism strongly favors tilted adsorption forms with the benzene molecule placed on the edge of the cluster almost perpendicular to the metal surface. The final geometry is more akin to a butadiene-like interaction, where two of the benzene π-bonds interact with the cluster. Our final results indicate that benzene adsorption on platinum is more favorable by at least 10 kcal mol-1 than adsorption on palladium. The adsorption process has been shown to be dominated by electron donation from the HOMO of benzene to the LUMO of the metal in the two clusters (M7 and M(7+3)) of both metals. The adsorption energy may be related to the energy level of the metal LUMO with lower LUMOs leading to stronger adsorption energy. Acknowledgment. J.W. de M.C. and D.A.G.A. received a research fellowship from CNPq. M.T. de M.C. has a graduate fellowship from CAPES. Part of this work was done during a visit of J.W. de M.C. to the Max Planck Institut in Mu¨lheim, which was supported by the Deutsche Akademische Austauschdienst (DAAD) and CAPES. The work in Niteroi was supported by FAPERJ and CNPq. M.B. wishes to thank W. Thiel and the Max Planck Institut in Mu¨lheim for support. References and Notes (1) Somorjai, G. A. Introduction to Surface Science and Catalysis; Wiley: New York, 1994. (2) Newns, D. M.; Muscat, J. P. Prog. Surf. Sci. 1978, 9, 1. (3) Morin, C.; Simon, D.; Sautet, P. J. Phys. Chem. B. 2003, 107, 2995. (4) Morin, C.; Simon, D.; Sautet, P. J. Phys. Chem. B. 2004, 108, 5653. (5) Waddill, G. D.; Kesmodel, L. L. Phys. ReV. B. 1985, 31, 4940. (6) Netzer, F. P.; Duscheck, R.; Mittendorfer, F.; Blyth, R. I. R.; Hafner, J.; Ramsey, M. G. Chem. Phys. Lett. 2000, 318, 43. (7) Hoffmann, H.; Zaera, F.; Ormerod, R. M.; Lambert, R. M.; Wang, L. P; Tyose, W. T. Surf. Sci. 1990, 232, 259. (8) Munakata, T. Surf. Sci. 2000, 454-456,118. (9) Ohtani, H.; Van Hove, M. A.; Somorjai, G. A. J. Phys. Chem. 1988, 92, 3974. (10) Jing, Z.; Whitten, J. L. Surf. Sci. 1991, 250, 147. (11) Wander, A.; Held, G.; Hwang, R. Q.; Blackman, G. S.; Xu, M. L; de Andres, P.; Van Hove, M. A.; Somorjai, G. A. Surf. Sci. 1991, 249, 21. (12) Orita, H.; Itoh, N. Appl. Catal., A 2004, 258, 17. (13) Netzer, F. P.; Mack, J. U. J. Chem. Phys. 1983, 79, 1017. (14) Aarts, J. F. M.; Sassen, N. R. M. Surf. Sci. 1989, 214, 257. (15) Lee, A. F.; Wison, K.; Lambert, R. M.; Goldoni, A.; Baraldi, A.; Paolucci, G. J. Phys. Chem. B 2000, 104, 11729. (16) Tsai, M.-C.; Muetterties, E. L. J. Am. Chem. Soc. 1982, 104, 2534. (17) Abon, M.; Bertolini, J. C.; Billy, J.; Massardier, J.; Tardy, B. Surf. Sci. 1984, 162, 395. (18) Campbell, J. M.; Seimanides, S.; Campbell, C. T. J. Phys. Chem. 1989, 93, 815. (19) Ogletree, D. F.; Van Hove, M. A.; Somorjai, G. A. Surf. Sci. 1987, 183, 1. (20) Trimble, T. M.; Cammarata, R. C.; Sieradzki, K. Surf. Sci. 2003, 531, 8.

11076 J. Phys. Chem. C, Vol. 111, No. 29, 2007 (21) Saeys, M.; Reyniers, M.-F.; Marin, G. B.; Neurock, M. J. Phys. Chem. B 2002, 106, 7489. (22) Gland, J. L.; Somorjai, G. A. Surf. Sci. 1973, 38, 157. (23) Palazov, A.; Bonev, C.; Shopov, D.; Lietz, G.; Sarkany, A.; Volter, J. J. Catal. 1987, 103, 249. (24) Koel, B. E.; Crowell, J. E.; Mate, C. M.; Somorjai, G. A. J. Phys. Chem. 1984, 88, 1988. (25) Mittendorfer, F.; Thomazeau, C.; Raybaud, P.; Toulhout, H. J. Phys. Chem. B 2003, 107, 12287. (26) Garnett, J. L. Catal. ReV. 1972, 5, 229. (27) Bandiera, J.; Meriavdeau, P. Kinet. Catal. Lett. 1988, 37, 373. (28) Srinivas, S. T.; Rao, P. K. J. Catal. 1994, 148, 470. (29) Orozco, J. M.; Webb, G. Appl. Catal. 1983, 6, 67. (30) Tysoe, W. T.; Ormerod, R. M.; Lambert, R. M.; Zgrablich, G.; Ramirez-Cuesta, A. J. Phys. Chem. 1993, 97, 3365. (31) Morin, C.; Simon, D.; Sautet, P. J. Phys. Chem. B 2004, 108, 12084. (32) Kryachko, E. S.; Arbuznikov, A. V.; Hendrickx, M. F. A. J. Phys. Chem. B 2001, 105, 3557. (33) Futaba, D. N.; Chiang, S. Jpn. J. Appl. Phys. 1999, 38, 3809. (34) de Souza, P. R. N.; Aranda, D. A. G.; Carneiro, J. W. M.; de Oliveira, C. S. B.; Antunes, O. A. C.; Passos, F. B. Int. J. Quantum. Chem. 2003, 92, 400. (35) Carneiro, J. W. M.; de Oliveira, C. S. B.; Passos, F. B.; Aranda, D. A. G.; de Souza, P. R. N.; Antunes, O. A. C. J. Mol. Catal. A: Chem. 2005, 226, 221. (36) Carneiro, J. W. M.; de Oliveira, C. S. B.; Passos, F. B.; Aranda, D. A. G.; de Souza, P. R. N.; Antunes, O. A. C. Catal. Today 2005, 107108, 31. (37) Baiker, A. J. Mol. Catal. A: Chem 1997, 115, 473. (38) Blaser, H.-U.; Jalett, H. P.; Muller, M.; Studer, M. Catal. Today 1997, 37, 441. (39) Augustine, R. L.; Tanielyan, S. K. J. Mol. Catal. A: Chem. 1997, 118, 79. (40) Webb, G.; Wells, P. B. Catal. Today 1992, 12, 319. (41) Simons, K. E.; Meheux, P. A.; Griffiths, S. P.; Sutherland, I. M.; Johnston, P.; Wells, P. B.; Carley, A. F.; Rajumon, M. K.; Roberts, M. W.; Ibbotson, A. Recl. TraV. Chim. Pays-Bas 1994, 113, 465. (42) Bond, G.; Wells, P. B. J. Catal. 1994, 150, 329. (43) Ferri, D.; Burgi, T. J. Am. Chem. Soc. 2001, 123, 12074. (44) Bonello, J. M.; Lindsay, R.; Santra, A. K.; Lambert, R. M. J. Phys. Chem. B 2002, 106, 2672. (45) Vargas, A.; Baiker, A. Surf. Sci. 2006, 239, 220.

Cruz et al. (46) Chang, J.-R, Chang, S.-L.; Lin, T.-B. J. Catal. 1997, 169, 338. (47) Becke, A. D. J. Chem. Phys. 1992, 96, 2155. (48) Lee, C.; Yang, W.; Parr, R. G. Phys. ReV. B 1988, 37, 785. (49) Hay, P. J.; Wadt, W. R. J. Chem. Phys. 1985, 82, 270. (50) Hay, P. J.; Wadt, W. R. J. Chem. Phys. 1985, 82, 299. (51) Dunning, T. H., Jr.; Hay, P. J. Modern Theoretical Chemistry; Schaefer, H. F., III, Ed.; Plenum: New York, 1976; Vol. 3, pp 1-28. (52) Bu¨hl, M.; Kabrede, H. J. Chem. Theory Comput. 2006, 2, 1282. (53) CRC Handbook of Chemistry and Physics, 77th ed.; Lide, D. R., Ed.; CRC: Boca Raton, FL, 1996; Chapter 12, p 18. (54) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Montgomery, J. A.; Vreven, T., Jr.; 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.; 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.; and Pople, J. A. Gaussian 03, Revision B.04; Gaussian, Inc.: Pittsburgh, PA, 2003. (55) (a) de Souza, P. R. N.; Pereira, M. M.; Antunes, O. A. C.; Aranda, D. A. G.; Carneiro, J. W. de M. Braz. J. Chem. Eng. 2002, 19, 187. (b) Kaupp, M.; Kopf, T.; Murso, A.; Stalke, D.; Strohmann, C.; Hanks, J. R.; Cloke, G. N.; Hitchcock, P. B. Organometallics 2002, 21, 5021. (c) Valcarcel, A.; Clotet, A.; Ricart, J. M.; Delbecq, F.; Sautet, P. Surf. Sci. 2004, 549, 121. (56) Jasen, P. V.; Brizuela, G.; Padin, Z.; Gonzalez, E. A.; Juan, A. Appl. Surf. Sci. 2004, 236, 394. (57) Anderson, A. B.; McDevitt, M. R.; Urbach, F. L. Surf. Sci. 1984, 146, 80. (58) Majumdar, D.; Roszak, S.; Balasubramanian, K. J. Chem. Phys. 2001, 114, 10300. (59) Tian, W. Q.; Ge, M.; Sahu, B. R.; Wang, D.; Yamada, T.; Mashiko, S. J. Phys. Chem. 2004, 108, 3806.