14202
J. Phys. Chem. C 2010, 114, 14202–14207
Adsorption, Oxidation State, and Diffusion of Pt Atoms on the CeO2(111) Surface Albert Bruix,† Konstantin M. Neyman,*,†,‡ and Francesc Illas*,† Departament de Quı´mica Fı´sica and Institut de Quı´mica Teo`rica i Computacional (IQTCUB), UniVersitat de Barcelona, C/Martı´ i Franque`s 1, 08028 Barcelona, and Institucio´ Catalana de Recerca i Estudis AVanc¸ats (ICREA), 08010 Barcelona, Spain ReceiVed: May 17, 2010; ReVised Manuscript ReceiVed: July 16, 2010
The interaction of atomic Pt with the regular CeO2(111) surface has been studied by periodic density functional calculations using exchange-correlation potentials with the inclusion of the on-site Hubbard correction (LDA+U and GGA+U) or employing the hybrid HSE06 functional. Various starting geometries have been explored, and two types of electronic configurationsswith either no or two unpaired electrons in the unit cellshave been considered. All methods consistently predict that the most favorable interaction corresponds to adsorption on top of a site bridging two nearest-neighbor surface O atoms (O2 site) in a closed-shell electronic configuration with an essentially neutral or slightly positively charged Pt. However, for this site another slightly higher energy solution is found with two unpaired electrons in the unit cell. This other local minimum on the potential energy surface represents a completely different physical picture, where Pt has been oxidized to Pt+ with the concomitant reduction of a single Ce4+ to Ce3+. Calculations have also been carried out to estimate the energy barrier corresponding to diffusion of Pt atoms over the surface. The calculated values indicate that the energy cost required for this process is very small. This means that adsorbed Pt adatoms can easily nucleate into larger particles in Pt/CeO2 systems. Introduction Cerium dioxide (CeO2 or ceria) has become a material of increasing interest due to its application as a component in car exhaust three-way catalysts for the elimination of toxic gases1 and in the direct electrochemical oxidation of various hydrocarbons using a composite anode of copper and ceria2 and, more recently, due to the excellent catalytic activity of noble metals such as Au and Pt supported on ceria in the low-temperature water-gas shift reaction (WGS).3 The latter is a process widely used in industrial applications, such as H2 production for hydrogen fuel cells,4 methanol synthesis,5 or methanol steam reforming.6 It has now become clear that ceria is not a simple spectator and that it can be a chemically active component as well. Let us just mention the recent work of Park et al. evidencing that ceria plays a key role in the WGS reaction.7 Ceria can also act as an oxygen reservoir able to release it in the presence of reductive gases and to store it upon interaction with oxidizing gases.8-10 The broad use of ceria in oxidation reactions is due to the facile Ce3+ T Ce4+ redox process11 concomitant with the formation of oxygen vacancies.12,13 However, this redox process can also be triggered by the presence of adsorbates/deposits. In particular, there is compelling evidence that adsorption of Cu and Ag atoms on CeO2(111) results in a spontaneous charge transfer from the metal to the substrate in such a way that the adsorbed metal atom becomes oxidized (Cu+ and Ag+) by one of the Ce4+ cations in the underlying oxide.14 The case of Au atoms and clusters supported on CeO2(111) has received considerable attention of several experimental and theoretical groups15-20 mainlybecauseofitsoutstandingcatalyticproperties.21,22 However, in this case the oxidation state of the adsorbed Au * To whom correspondence should be addressed. E-mail: konstantin.neyman@ icrea.es (K.M.N.);
[email protected] (F.I.). † Universitat de Barcelona. ‡ ICREA.
atom is not so well established. Single-crystal experiments23 suggest that the charge of Au atoms adsorbed on perfect CeO2(111) surfaces is essentially zero and that low-coordinated Ce atoms are required to produce the redox reaction where Ce4+ is reduced to Ce3+ whereas Au0 is oxidized to Au+. This is in agreement with the theoretical study by Castellani et al.24 but contrasts with earlier theoretical studies25-27 and also with the interpretation of resonant photoemission spectroscopy experiments28 claiming that adsorption of Au (in the form of small islands) on the regular CeO2(111) surface leads to spontaneous oxidation of Au0 to Au+. The origin of the inconsistency in the theoretical studies is the energy proximity between the two electronic states corresponding to neutral and charged Au, which indeed is very sensitive to the surface model and exchangecorrelation potential (EXC[F]) used in the calculations.29 A similar scenario is found in the case of single Pd atoms adsorbed on CeO2(111). In fact, Wilson et al.30 reported that for the most stable Pd adsorption site an electron transfer process from the adatom to the surface takes place (Pd0 + Ce4+ T Pd+ + Ce3+), although these authors also find a slightly less stable (∼0.15 eV) state where Pd remains neutral. A consequence of the previous discussion is that while Cu and Ag atoms will become oxidized when deposited on CeO2(111), it is likely that the proximity between the two electronic states with adsorbed Au0 and Au+, and also between the two electronic states with adsorbed Pd0 and Pd+, will result in a statistical distribution of the two states. An immediate question concerns the oxidation state of Pt atoms adsorbed on CeO2(111), a system which is also directly relevant to catalysis.31-33 Actually, the ionization potential of Pt (9.01 eV) is quite close to that of Au (9.22 eV) and thus considerably larger than the corresponding values for Cu and Ag (7.72 and 7.57 eV, respectively), with Pd (8.33 eV) lying midway between the Pt/Au and Cu/Ag cases. A recent study using state of the art density functional theory (DFT) techniques including the
10.1021/jp104490k 2010 American Chemical Society Published on Web 07/30/2010
Adsorption of Pt Atoms on the CeO2(111) Surface on-site Hubbard correction and accounting for spin polarization effects predicts that on the regular, unreduced, CeO2(111) surface Pt adsorbs essentially as a neutral atom.34 However, no attempts have been made to locate the charge transfer state, and there is no information about the energy difference between these two possible states. Motivated by these arguments, we have undertaken a systematic study of the adsorption of Pt atoms on CeO2(111) considering the high-symmetry surface sites and different electronic states and exploring several DFT-based methods. In addition, to examine the mobility of Pt atoms important for the particle growth, we have also considered the diffusion of Pt atoms over the surface and identified the transition-state structures corresponding to the minimum-energy pathway. We present convincing evidence that, similar to the case of Au on CeO2(111), adsorbed Pt atoms can become oxidized rather easily at room temperature. We also show that diffusion of Pt atoms at this surface does not involve significant activation barriers. Computational Details We have investigated the interaction of single Pt atoms with the regular stoichiometric CeO2(111) surface by means of periodic DFT slab model calculations, as detailed in the next section, and focusing on the electronic structure and oxidation states of the most stable adsorption sites. Following previous work for Au on CeO2(111),29 we combine the local density approach (LDA) and the generalized gradient approach (GGA) for EXC[F] with the inclusion of the on-site Coulomb interaction Hubbard-like term U, leading to the so-called LDA+U and GGA+U methods.35-37 These are sometimes called DFT+U, which we feel is an incorrect term since DFT is exact while the approximation is in the form of the unknown universal EXC[F]. For strongly correlated systems, such as magnetic transitionmetal or CeOx oxides, this is a necessary requirement to correct the incomplete cancellation of the self-interaction error in LDA and GGA, which results in a qualitatively wrong description of localized states.38 In the case of CeOx, this concerns the 4f electrons of partially reduced ceria, and it has been shown that a proper description requires the use of either LDA+U and GGA+U methods39,40 or hybrid functionals.41,42 The latter are often considered to be physically more grounded because they modify the whole density, whereas the on-site Hubbard corrections apply to only a single level, here Ce 4f. The former methods suffer from the choice of the U parameter to the point that the optimum value of U may depend on the studied property.43 To circumvent this problem, Loschen et al.44 suggested using a U parameter fitted to reproduce various physical properties. Following this procedure, we use U ) 5 eV for calculations with LDA+U (hereafter referred to as L5) and U ) 3 eV for those carried out using GGA+U (or G3). In the first case we use the standard LDA VWN functional;45,46 in the second case we use the PW91 functional.47 Usually, LDA provides structures involving heavy-transition-metal atoms and their vibrational frequencies in better agreement with experiment, whereas GGA offers more accurate energies.48 This is especially the case when the bonds under scrutiny are not very weak. Hence, we use L5 for geometry optimization and G3 to obtain the energy at the L5 geometry. Following the notation introduced in our previous work, this computational scheme is referred to as G3//L5.29 To avoid any possible bias of this choice, we have also employed the more reliable, but computationally demanding, HSE06 hybrid functional,49 which properly describes reduced ceria and the particular electronic structure of the resulting 4f localized states.41,42,50,51
J. Phys. Chem. C, Vol. 114, No. 33, 2010 14203
Figure 1. Top view of the 2 × 2 supercell used in the present work. Different surface sites are labeled.
The periodic calculations with the exchange-correlation potentials mentioned above have been carried out by using the VASP code.52-54 Hence, a plane wave basis set is used to describe the valence states with a cutoff of 415 eV for the kinetic energy. The effect of the inner cores in the valence states is taken into account by the projector-augmented-wave (PAW) method of Blo¨chl55 as implemented in VASP. The valence density is defined by the Ce 5s, 5p, 6s, 5d, and 4f, the O 2s and 2p, and the Pt 5d and 6s states. Suitable Monkhorst-Pack56 grids of special k-points were used to carry out the numerical integration on the reciprocal space (see the next section). The total energy threshold defining self-consistency of the electron density was set to 10-4 eV, and geometries were optimized until forces acting on the relaxed atoms were smaller than 0.01 eV/ Å. The optimized geometries of the Pt/CeO2 systems were characterized as minimum-energy stationary points by proper analysis of the vibrational frequencies involving the Pt atom. Surface Models The interaction of atomic Pt with the CeO2(111) surface has been studied using a repeated slab model. The slab was cut from the bulk cubic (Fm3m) fluorite-type structure with a lattice parameter of 5.40 Å, the optimal value calculated at the L5 level, which nicely reproduces the experimental value of 5.41 Å.57,58 Note that using a more expanded lattice, as predicted by GGA and GGA+U, has a noticeable effect on the electronic structure of adsorbed Au.29 Our slab model has a 2 × 2 surface unit cell with a total of six atomic layers or, equivalently, two O-Ce-O trilayers with a vacuum width of ∼15 Å between the neighboring interleaved slabs. Adding three additional atomic layers does not significantly modify the studied properties. For instance, the adsorption energy varies by less than 0.1 eV. Therefore, the thickness of the slab is considered to be large enough to adequately model this surface and Pt adsorption thereon. Note that with this model the Pt coverage is 0.25 monolayer (ML) with respect to the number of surface O atoms in the supercell. The structural parameters of adsorbed Pt were obtained from geometry optimizations carried out at the L5 level starting from the different initial trial sites shown in Figure 1. These are on top of one O atom (O1), a surface site bridging two O atoms of the topmost layer (O2), a 3-fold site (O3) above an O atom of the third layer, a site above a Ce atom of the second layer (Ce1), and a site bridging two Ce atoms in the second layer (Ce2). In these calculations, the three CeO2 bottom layers were fixed at the optimized bulk positions, while the remaining atoms were allowed to fully relax. To scrutinize different possible electron configurations of the system, to the extent allowed by the Kohn-Sham formalism, geometry optimization was carried out first, forcing the system to acquire the lowest energy closedshell singlet state. This corresponds to a situation with the same number of spin-up (NR) and spin-down (Nβ) electrons and thus
14204
J. Phys. Chem. C, Vol. 114, No. 33, 2010
NR - Nβ ) 0, where adsorbed Pt remains neutral and the system has not undergone any noticeable electron transfer. A second geometry optimization calculation was carried out, forcing the system to feature two unpaired electrons or NR - Nβ ) 2. This is to acquire the lowest energy triplet-state configuration. Analysis of the Bader charges59 and magnetic moments shows that the resulting situation corresponds to an electron transfer process from a 5d orbital of the Pt atom to the 4f orbital of a single Ce4+ cation (Pt0 + Ce4+ f Pt+ + Ce3+). For each of the converged optimized geometries characterized as stationary points on the potential energy surface, a single-point calculation was carried out at the G3 level, imposing the same spin restriction of the preceding L5 calculation. For the L5 and G3 calculations, a 4 × 4 × 1 Monkhorst-Pack56 grid was used for the numerical integration of the reciprocal space. The last set of calculations was performed using the HSE06 hybrid functional. In this case, the L5 geometries were reoptimized using the HSE06 functional at the Γ point. Single-point calculations were later carried out for the converged geometries using a more suitable but less affordable 4 × 4 × 1 Monkhorst-Pack grid of k-points. For all sites, L5, G3//L5, and HSE06 adsorption energies were defined as Eads ) E(Pt/CeO2) - E(Pt) - E(CeO2), negative values corresponding to stable situations with respect to the separated systems. Bader charges and magnetic moments were evaluated from the density distribution obtained in the G3// L5 calculations. Results and Discussion Adsorption of Atomic Pt on the CeO2(111) Surface. From the different, high-symmetry, starting geometries considered as possible adsorption sites, the geometry optimization procedure converges always to a minimum-energy stationary point with respect to the distance perpendicular to the surface. However, a frequency analysis shows that several of the optimized geometries do not correspond to true minima on the potential energy surface, the convergence of the geometry optimization being biased by the underlying symmetry. We applied symmetry breaking by slight distortion of the initial starting geometry as a simple procedure to facilitate convergence to the proper local minima. Before commenting on the different possible electronic states, it is worth pointing out that the isolated Pt atom exhibits a d9s1 electronic configuration giving rise to a 3D (3D3 level if the spin-orbit interaction is included) multiplet. Upon metal interaction with an oxide surface, two scenarios are possible depending on the strength of the chemisorption bond. For weak interactions, the intraatomic exchange favors Hund’s rule as in the isolated atoms and the high-spin state is maintained, although the energy difference between the high- and low-spin states is reduced with respect to the atomic value.60 However, a stronger interaction with the underlying oxide surface results in bonding and antibonding orbitals with occupation of the lowest energy ones, concomitant electron pairing, and spin quenching.61,62 For Pt on CeO2(111) we see that high- and low-spin states are nearly degenerate and, in addition, the high-spin state involves charge transfer from the metal to the surface with subsequent reduction of one Ce4+ cation to Ce3+, a process which cannot occur in an nonreducible oxide such as MgO.63,64 In the case of the closed-shell singlet state, two proper local minima are justified by the frequency analysis. For this electronic configuration, the most stable site predicted by the L5 and HSE06 approaches corresponds to Pt above the O2 site, in agreement with the previous study of Yang et al.34 Moreover, for this site, the structural parameters obtained with the L5
Bruix et al. TABLE 1: Interatomic Distances R1-R3 (Å) Defining the Geometry of Adsorbed Atomic Pt above the O2 Site of the CeO2(111) Surface (Figure 2) for the Closed-Shell (Nr - Nβ ) 0) and Open-Shell (Nr - Nβ ) 2) Solutions as a Function of the Exchange-Correlation Potential (EXC[G]) N R - Nβ
EXC[F]
R1
R2
R3
0
L5 G3//G3 (L5) HSE06 PBE+534 L5 G3//G3 (L5) HSE06 PBE+534
2.11 2.14 2.15 2.14 1.98 2.02 1.98 -
2.56 2.60 2.56 2.60 2.70 2.75 2.70 -
2.77 2.90 2.85 2.87 2.85 2.99 2.85 -
2
potential (Table 1) are very similar to those reported by Yang et al., although they used the PBE exchange-correlation potential with U ) 5 eV (PBE+U with U ) 5 or simply PBE+5). The calculated G3//L5 adsorption energy is -2.65 eV, close to the PBE+5 value of -2.62 eV.34 This is not surprising because the PW91 and PBE functionals are both of the GGA family and lead to similar adsorption energies.65,66 Changing the type of functional does not change the prediction of the most stable site and has a modest effect on its structure with changes in the relevant distances of 2-3% (Table 1 and Figure 2). Note also that the G3 structure at the L5 lattice parameter does not significantly differ from that obtained from the L5 calculations with the same lattice parameter. This validates our choice of the G3//L5 approach to obtain the structure and energy of the Pt atom above the CeO2(111) slab surface model. However, the choice of the type of the exchange-correlation potential strongly influences the calculated adsorption energy value (Table 2). This is a common feature in the DFT description of the metal-metal oxide interface.67,68 Thus, the Eads value for the O2 site in the closed-shell electronic configuration predicted by the L5 approach is much larger in magnitude (-3.89 eV), and that predicted by the hybrid HSE06 potential is smaller (-1.96 eV). For metals on MgO it has been shown that hybrid functionals lead to Eads values which are close to those obtained by the explicitly correlated coupled cluster singles and doubles method
Figure 2. Top view of the optimized structure for the Pt atom at the most stable site (O2) of CeO2(111) as predicted from the present density functional calculations using a 2 × 2 unit cell. The most characteristic distances are defined, the values of which are reported in Table 1 for the different exchange-correlation potentials used.
TABLE 2: Adsorption Energy (eV) of the Pt Atom on Various Sites of the CeO2(111) Surface for the Closed-Shell (Nr - Nβ ) 0) and Open-Shell (Nr - Nβ ) 2) Solutions as a Function of the Exchange-Correlation Potentiala site O2 O2 Ce2 O3 a
NR - Nβ L5//L5 (L5) G3//L5 (L5) HSE06//HS06 (L5) PBE+534 0 2 0 2
-3.89 -3.83 -3.50 -3.64
-2.65 -2.34 -2.49 -1.84
-1.96 -1.90 -1.94 -1.52
-2.62 -2.47
Results are listed only for the structures characterized by vibrational analysis as local minima in the potential energy surface.
Adsorption of Pt Atoms on the CeO2(111) Surface with the perturbative correction of triple excitations (CCSD(T)).67,68 Therefore, we expect that the HSE06 predictions will be the most accurate also in the case of the ceria surface. Nevertheless, it is important to mention that the present trends and order of stability predicted by LDA+U, GGA+U, and HSE06 are in good agreement, at variance with a recent result for Au on CeO2.29 However, the latter is likely to be a special case since good agreement is also found for Cu and Ag on this ceria surface.14 For all computational methods, the Bader analysis of the charge density indicates that adsorbed Pt remains almost neutral with a charge of +0.17e only and concomitant zero values for the spin density at all sites. The stabilization comes from the response of the Pt electron density to the presence of the charges of the surface ions; the resulting chemical bond can be described as a polarized covalent bond, similar to the case of transition-metal atoms on nonreducible oxides such as MgO.61,63,64,67,68 Let us analyze the bonding interaction between Pt and the CeO2(111) surface in the electronic state with two unpaired electrons in the unit cell. First, we focus on the interaction of Pt above the O2 site, which, for the closed-shell singlet solution, corresponds to the most stable site. At the L5//L5 level, the adsorption energy is only 0.06 eV higher than that in the corresponding singlet state. The same energy difference is found for the HSE06 hybrid functional, and a slightly (0.2 eV) higher value is predicted by the G3//L5 approach. It is slightly reduced to 0.17 eV when the structure is reoptimized at the G3 level; this leads to the G3//G3 result at the L5 lattice parameter previously referred to as G3//G3 (L5).29 The energies of the two (closed- and open-shell) solutions are close enough to suggest that these solutions can coexist at room temperature. In addition, the structural difference between the two solutions is not very large (Table 1), implying that the switch from one state to the other will not be hindered by dynamical effects arising from geometry distortions. This is an important conclusion because the Bader charge analysis and spin density values clearly indicate that the open-shell solution corresponds to a completely different bonding mechanism. In fact, the Bader charge of Pt becomes 0.40e, whereas for the closed-shell solution it is only 0.17e. This is a strong indication that in this openshell low-energy excited electronic state the charge transfer between adatom and substrate is substantially larger than on the near-degenerate neutral ground state. This picture is fully confirmed by the spin density values of 0.61e on the Pt atom and of 0.91e (G3//L5 results) on one of the Ce atoms of the substrate. The conclusion from these results is rather straightforward: interaction of atomic Pt with the CeO2(111) surface can lead to spontaneous oxidation of Pt0 to Pt+ with a concomitant reduction of one Ce4+ cation to Ce3+. This charge transfer requires some additional energy, which is small enough to enable the process already at room temperature. This finding may have important implications in catalysis since the reactivity of such two adsorbed Pt species will be different. It might also have a technological application provided one is able to find the way to control the switch to this charge transfer process. To close this section, we briefly discuss the results corresponding to other adsorption sites but focusing only on those whose optimized structure is properly characterized as a local minimum by the vibrational analysis. This is because from the several starting trial geometries only those reported in Table 2 correspond to local minima on the potential energy surface. Spin polarization calculations carried out without any restriction in the number of R and β electrons in the unit cell predict the Ce2 site as the only one competitive with the most favorable O2
J. Phys. Chem. C, Vol. 114, No. 33, 2010 14205
Figure 3. Diffusion pathways A and B of adsorbed atomic Pt between equivalent O2 sites of the CeO2(111) surface.
site described at length above. At the L5//L5 and G3//L5 levels, the adsorption energy is 0.39 and 0.16 eV, respectively, smaller than that in the O2 site, whereas a difference of only 0.02 eV is predicted by the HSE06 hybrid functional. This seems to indicate that at sufficiently high coverage both O2 and Ce2 sites will be covered by Pt atoms. The Bader analysis and spin density values for the adsorption of Pt above the Ce2 site indicate that the bonding mechanism is similar to that described for the O2 site in the closed-shell electronic configuration: Pt remains neutral or slightly positively charged, but without any evidence for electron transfer to the CeO2(111) substrate. Spin-polarized calculations for Pt above the Ce2 site carried out imposing two unpaired electrons in the unit cell converged to the O2 structure. Thus, contrarily to the case of the O2 site, for the adsorption of Pt at the Ce2 site there is no possibility of near-degenerate electronic states with different bonding mechanisms. An opposite situation is found when analyzing results corresponding to the O3 site. Here, calculations imposing a closed-shell configuration also converge to the O2 structure, whereas spinpolarized calculations with two unpaired electrons in the unit cell converge to a structure very close to the starting point, although with an adsorption energy significantly smaller; therefore, it is unlikely that these sites will be occupied. Nevertheless, it is important to mention that, as in the case of the open-shell configuration for the O2 structure, the Bader and spin density analyses show a charge transfer from Pt to the surface with the appearance of a single Ce3+ cation. Diffusion of Atomic Pt on the CeO2(111) Surface. To evaluate whether Pt atoms can diffuse easily over the regular CeO2(111) surface or become essentially immobile, we have studied the diffusion of a single Pt atom from one minimumenergy adsorption site to an equivalent position by performing a pertinent transition-state (TS) search. We consider only the singlet-state electronic configuration and the O2 site. Figure 1 shows that the O2 sites have a 3-fold symmetry, meaning that there are three equivalent sites around a subsurface Ce cation in the second atomic layer or around an O atom in the third layer. Consequently, there are two equivalent diffusion pathways in direct contact with the same second-layer Ce cation allowing an adsorbed Pt atom to move from one site to another (Figure 3, pathway A) and two different equivalent pathways involving an O atom on the third layer (Figure 3, pathway B). Only pathway A has been studied since it is likely to be the one with the highest energy barrier. This is because in the alternative B pathway, the underlying O atom is in the third atomic layer and thus the Pt adatom experiences a smaller Pauli repulsion. The TS search was performed with the L5 scheme and using the quasi-Newton algorithm69 implemented in the VASP code. This is a suitable choice to locate saddle points provided the initial guess is close to the TS structure. The TS structure thus obtained was characterized by a proper vibrational frequency analysis revealing one vibrational mode with an imaginary frequency. To explore the influence of the exchange-correlation potential on the calculated diffusion energy barrier, we carried
14206
J. Phys. Chem. C, Vol. 114, No. 33, 2010
out single-point energy calculations with the G3 and HSE06 functionals at the L5 TS structure. The energy barriers thus obtained range from 0.45 eV as predicted by the L5 scheme (most probably too high to be realistic) to a significantly smaller value of 0.15 eV as predicted by the G3 scheme or an even lower value of almost zero at the HSE06 level. The results above indicate that Pt atoms on the regular CeO2(111) surface will be rather mobile. This facilitates nucleation of larger supported particles as those dealt with in experiments on Pt/CeO2 systems.70 On the other hand, the high mobility of adsorbed Pt atoms may preclude their use as a switch (as suggested above), because the energy required to change the state of Pt0 to Pt+, with a concomitant reduction of one Ce4+ to Ce3+ and the simultaneous appearance of a magnetic moment on that cation, is similar to the energy cost needed to displace the Pt atom to another site. A different situation may appear if the Pt atoms are fixed at the surface by interaction with an appropriate coadsorbed species, although this is perhaps too speculative. Conclusions The interaction of atomic Pt with the regular CeO2(111) surface has been studied by periodic density functional calculations using a variety of exchange-correlation potentials. Various starting geometries have been explored, and two types of electronic configurations, with either zero or two unpaired electrons in the unit cell, have been considered. All methods consistently predict that the most favorable interaction corresponds to adsorption on top of a surface site bridging two nearest-neighbor surface O atoms (O2 site) in a closed-shell electronic configuration and with an essentially neutral or slightly positively charged Pt, in agreement with previous findings using a similar approach.34 However, one must also point out that the adsorption energy difference between the O2 and Ce2 sites is fairly small and decreases from 0.39 eV at the L5//L5 (L5) level to 0.16 and 0.02 eV for the G3//G3 (L5) and HSE06 approaches, respectively, thus indicating that the possibility of multisite adsorption cannot be completely ruled out. Moreover, for the O2 site another rather low lying solution is found with two unpaired electrons in the unit cell, which also corresponds to an energy minimum on the potential energy surface. Bader and spin density analyses consistently show that this open-shell electronic state represents a completely different physical picture, where Pt0 has been oxidized to Pt+ with the concomitant reduction of a single Ce4+ to Ce3+. Therefore, the situation is reminiscent of that encountered for Au on the same surface where states with Au0 and Au+ appear at similar energies and different from the situation corresponding to adsorbed Cu and Ag atoms, where interaction with the CeO2(111) surface leads to their spontaneous oxidation. Note that these findings correlate nicely with the ionization potential of the atoms under study. The energy difference between the two electronic states is small enough to permit spontaneous charge transfer at room temperature or externally induced by an appropriate potential. This feature might have technological application as a switch provided the coverage and surface structure can be sufficiently controlled. However, calculations for the energy barrier corresponding to Pt diffusion over the surface indicate that the energy cost required for this process is also very small. Thus, adsorbed Pt adatoms can nucleate into larger particles in Pt/CeO2 systems. Acknowledgment. A.B. is grateful to the Spanish Ministerio de Ciencia e Innovacion (MICINN) for a predoctoral grant.
Bruix et al. Financial support has been provided by the Spanish MICINN (Grants FIS2008-02238 and ERA-CHEMISTRY/CTQ200730547-E/BQU), in part by the Generalitat de Catalunya (Grants 2009SGR1041 and XRQTC), and by COST Action D41 “Inorganic oxides: surfaces and interfaces”. Computational time has been generously granted by the Barcelona Supercomputing Center. References and Notes (1) Trovarelli, A. Catal. ReV.sSci. Eng. 1996, 38, 439. (2) Park, P.; John, M.; Gorte, J. Nature 2000, 404, 265. (3) Fu, Q.; Saltsburg, H.; Flytzani-Stephanopoulos, M. Science 2003, 301, 935. (4) Lee, S. H. D.; Applegate, D. V.; Ahmed, S.; Calderone, S. G.; Harvey, T. L. Int. J. Hydrogen Energy 2005, 30, 829. (5) Rozovskii, A. Y.; Lin, G. I. Top. Catal. 2003, 22, 127. (6) Vargas, M. A. L.; Busca, G.; Costantino, U.; Marmottini, F.; Montanari, T.; Patrono, P.; Pinzari, F.; Ramis, G. J. Mol. Catal. A 2007, 266, 188. (7) Park, J. B.; Graciani, J.; Evans, J.; Stacchiola, D.; Ma, S. G.; Liu, P.; Nambu, A.; Sanz, J. F.; Hrbek, J.; Rodriguez, J. A. Proc. Natl. Acad. Sci. U.S.A. 2009, 106, 4975. (8) Yao, H. C.; Yu Yao, Y. F. J. Catal. 1984, 86, 254. (9) Engler, B.; Koberstein, E.; Schubert, P. Appl. Catal. 1989, 48, 71. (10) Miki, T.; Ogawa, T.; Haneda, M.; Kakuta, N.; Ueno, A.; Tateishi, S.; Matsuura, S.; Sato, M. J. Phys. Chem. 1990, 94, 6464. (11) Trovarelli, A.; de Leitenburg, C.; Boaro, M.; Dolcetti, G. Catal. Today 1999, 50, 353. (12) Ganduglia-Pirovano, M. V.; Hofmann, A.; Sauer, J. Surf. Sci. Rep. 2007, 62, 219. (13) Ganduglia-Pirovano, M. V.; Da Silva, J. L. F.; Sauer, J. Phys. ReV. Lett. 2009, 102, 026101. (14) Branda, M. M.; Herna´ndez, N. C.; Sanz, J. F.; Illas, F. J. Phys. Chem. C 2010, 114, 1934. (15) Chen, M.; Goodman, D. W. Acc. Chem. Res. 2006, 39, 739. (16) Rainer, D. R.; Goodman, D. W. J. Mol. Catal. A 1998, 131, 259. (17) Henry, C. Surf. Sci. Rep. 1998, 31, 231. (18) Ba¨umer, M.; Freund, H.-J. Prog. Surf. Sci. 1999, 61, 127. (19) Shaikhutdinov, S.; Meyer, R.; Naschitzki, M.; Ba¨umer, M.; Freund, H.-J. Catal. Lett. 2003, 86, 211. (20) Starr, D. E.; Shaikhutdinov, S. K.; Freund, H.-J. Top. Catal. 2005, 36, 33. (21) Saltsburg, H.; Fu, Q.; Flytzani-Stephanopoulos, M. Science 2003, 301, 935. (22) Carrettin, S.; Concepcion, P.; Corma, A.; Nieto, J. M. L.; Puntes, V. F. Angew. Chem., Int. Ed. 2004, 43, 2538. (23) Baron, M.; Bondarchuk, O.; Stacchiola, D.; Shaikhutdinov, S.; Freund, H.-J. J. Phys. Chem. C 2009, 113, 6042. (24) Castellani, N. J.; Branda, M. M.; Neyman, K. M.; Illas, F. J. Phys. Chem. C 2009, 113, 4948. (25) Liu, Z. P.; Jenkins, S. J.; King, D. A. Phys. ReV. Lett. 2005, 94, 196102. (26) Zhang, C.; Michaelides, A.; King, D. A.; Jenkins, S. J. J. Chem. Phys. 2008, 129, 194708. (27) Cruz Herna´ndez, N.; Grau-Crespo, R.; de Leeuw, N. H.; Sanz, J. F. Phys. Chem. Chem. Phys. 2009, 11, 5246. (28) Skoda, M.; Cabala, M.; Matolinova, I.; Prince, K. C.; Skala, T.; Sutara, F.; Veltruska, K.; Matolin, V. J. Chem. Phys. 2009, 130, 034703. (29) Branda, M. M.; Castellani, N. J.; Grau-Crespo, R.; de Leeuw, N. H.; Cruz Hernandez, N.; Sanz, J. F.; Neyman, K. M.; Illas, F. J. Chem. Phys. 2009, 131, 094702. (30) Wilson, E. L.; Grau-Crespo, R.; Pang, C. L.; Cabailh, G.; Chen, Q.; Purton, J. A.; Catlow, C. R. A.; Brown, W. A.; de Leeuw, N. H.; Thornton, G. J. Phys. Chem. 2008, 112, 10918. (31) To¨rncrona, A.; Skoglundh, M.; Thorma¨hlen, P.; Fridell, E.; Jobson, E. Appl. Catal., B 1997, 14, 131. (32) Burch, R.; Ramli, A. Appl. Catal., B 1998, 15, 49. (33) Ayastuy, J. L.; Gil-Rodrı´guez, A.; Gonza´lez-Marcos, M. P.; Gutie´rrez-Ortiz, M. A. Int. J. Hydrogen Energy 2006, 31, 2231. (34) Yang, Z.; Lu, Z.; Luo, G. Phys. ReV. B 2007, 76, 075421. (35) Anisimov, V. I.; Aryasetiawan, F.; Lichtenstein, A. I. J. Phys.: Condens. Matter 1997, 9, 767. (36) Anisimov, V. I.; Solovyev, I. V.; Korotin, M. A.; Czyzyk, M. T.; Sawatzky, G. A. Phys. ReV. B 1993, 48, 16929. (37) Solovyev, I. V.; Dederichs, P. H.; Anisimov, V. I. Phys. ReV. B 1994, 50, 16861. (38) Moreira, I. de P. R.; Illas, F.; Martin, R. L. Phys. ReV. B 2002, 65, 155102. (39) Nolan, M.; Grigoleit, S.; Sayle, D. C.; Parker, S. C.; Watson, G. W. Surf. Sci. 2005, 576, 217.
Adsorption of Pt Atoms on the CeO2(111) Surface (40) Fabris, S.; de Gironcoli, S.; Baroni, S.; Vicario, G.; Balducci, G. Phys. ReV. B 2005, 71, 041102. (41) Hay, P. J.; Martin, R. L.; Uddin, J.; Scuseria, G. E. J. Chem. Phys. 2006, 125, 034712. (42) Da Silva, J. L. F.; Ganduglia-Pirovano, M. V.; Sauer, J.; Bayer, V.; Kresse, G. Phys. ReV. B 2007, 75, 045121. (43) Castleton, C. W. M.; Kullgren, J.; Hermansson, K. J. Chem. Phys. 2007, 127, 244704. (44) Loschen, C.; Carrasco, J.; Neyman, K. M.; Illas, F. Phys. ReV. B 2007, 75, 035115. (45) Vosko, S. H.; Wilk, L.; Nusair, M. Can. J. Chem. 1980, 58, 1200. (46) Ceperly, D. M.; Alder, B. J. Phys. ReV. Lett. 1980, 45, 566. (47) Perdew, J. P.; Wang, Y. Phys ReV. B. 1992, 45, 13244. (48) Go¨rling, A.; Trickey, S. B.; Gisdakis, P.; Ro¨sch, N. In Topics in Organometallic Chemistry; Brown, J., Hofmann., P., Eds.; Springer: Heidelberg, Germany, 1999; Vol. 4, p 109. (49) Heyd, J.; Scuseria, G. E.; Ernzerhof, M. J. Chem. Phys. 2006, 124, 219906. (50) Ganduglia-Pirovano, M. V.; Hofmann, A.; Sauer, J. Surf. Sci. Rep. 2007, 62, 219. (51) Ganduglia-Pirovano, M. V.; da Silva, J. L. F.; Sauer, J. Phys. ReV. Lett. 2009, 102, 026101. (52) Kresse, G.; Hafner, J. Phys. ReV. B 1993, 47, 558. (53) Kresse, G.; Furthmu¨ller, J. Phys. ReV. B 1996, 54, 1169. (54) Kresse, G.; Furthmu¨ller, J. Comput. Mater. Sci. 1996, 6, 15. (55) Blo¨chl, P. E. Phys. ReV. B 1994, 50, 17953. (56) Monkhorst, H. J.; Pack, J. D. Phys. ReV. B 1976, 13, 5188.
J. Phys. Chem. C, Vol. 114, No. 33, 2010 14207 (57) Duclos, S. J.; Vohra, Y. K.; Ruoff, A. L.; Jayaraman, A.; Espinosa, G. P. Phys. ReV. B 1998, 38, 7755. (58) Gerward, L.; Olsen, J. S. Powder Diffr. 1993, 8, 127. (59) Bader, R. Atoms in Molecules: A Quantum Theory; Oxford University Press: New York, 1994. (60) Markovits, A.; Paniagua, J. C.; Lo´pez, N.; Minot, C.; Illas, F. Phys. ReV. B 2003, 67, 115417. (61) Yudanov, I.; Pacchioni, G.; Neyman, K.; Ro¨sch, N. J. Phys. Chem. B 1997, 101, 2786. (62) Florez, E.; Mondragon, F.; Fuentealba, P.; Illas, F. Phys. ReV. B 2008, 78, 075426. (63) Neyman, K. M.; Inntam, C.; Nasluzov, V. A.; Kosarev, R.; Ro¨sch, N. Appl. Phys. A: Mater. Sci. Process. 2004, 78, 823. (64) Neyman, K. M.; Inntam, C.; Matveev, A. V.; Nasluzov, V. A.; Ro¨sch, N. J. Am. Chem. Soc. 2005, 127, 11652. (65) Rolda´n, A.; Ricart, J. M.; Illas, F. Theor. Chem. Acc. 2009, 123, 119. (66) Fajin, L.; Illas, F.; Gomes, J. R. B. J. Chem. Phys. 2009, 130, 224702. (67) Lo´pez, N.; Illas, F. J. Phys. Chem. B 1998, 102, 1430. (68) Lo´pez, N.; Illas, F.; Ro¨sch, N.; Pacchioni, G. J. Chem. Phys. 1999, 110, 4873. (69) Pulay, P. Chem. Phys. Lett. 1980, 73, 393. (70) Happel, M.; Kylhammar, L.; Carlsson, P. A.; Libuda, J.; Gro¨nbeck, H.; Skoglundh, M. Appl. Catal., B 2009, 91, 679.
JP104490K