Predicted oxidation of CO catalyzed by Au nanoclusters on a thin defect-free MgO film supported on a Mo(100) surface.

Recently there has been a surge in research pertaining to the physical and chemical properties of gold nanoclusters. Unlike supported particles of larger sizes, or extended solid surfaces1,2 small metal clusters adsorbed on support materials were found to exhibit unique properties that originate from the highly reduced dimensions of the individual metal aggregates.3-8 In particular we note here joint experimental studies and theoretical investigations5-8 using first-principles simulations on size-selected small gold clusters, Aun (20 g n g 2), adsorbed on a well characterized metal-oxide support under ultrahigh-vacuum conditions. These studies revealed that cluster morphology, dimensionality, dynamical structural fluxionality, electronic structure, and charge state, as well as the state of the metal-oxide surface (specifically, defect-rich MgO(100) surfaces containing oxygen-vacancy F centers, or defect-poor surfaces), govern the catalytic activity. Particular emphasis has been put5-8 on the low-temperature (as low as 140 K) oxidation of CO, catalyzed by Aun clusters with 7< n < 25 atoms that are adsorbed on F-center-rich magnesia surfaces. In these investigations charging of the adsorbed metal cluster through partial electron transfer from the substrate F-center defect and subsequent occupation of the antibonding 2π* orbital of O2 leading to activation of the O-O bond of the molecule adsorbed on the cluster (resulting in formation of peroxo or superoxo states), have been identified as underlying the catalytic activity. Here we show results of first-principles investigations aiming at tuning and controlling the catalytic activity of gold nanoclusters through the design of the underlying support. We show that gold clusters adsorbed on a very thin (2 layers) defect-free MgO film, which is itself supported on Mo(100),9 may serve as model catalysts for the low-barrier oxidation of CO. The origin of the emergent activity of the nanoclusters is a dimensionality crossover of the adsorbed gold clusters, from inactive three-dimensional (3D) optimal structures on thick MgO films to catalytically active 2D ones for sufficiently thin MgO films (less than 1 nm in thickness) supported on Mo(100). The increased gold wetting propensity on the MgO/Mo(100) surface originates from electrostatic interaction between the underlying metal and metal-induced excess electronic charge accumulated at the cluster interface with the metal-oxide film.10 The excess interfacial charge is predicted here to activate O2 molecules adsorbed at the interfacial periphery of the 2D gold island with the MgO/Mo(100) surface. This activation, which weakens greatly the O-O bond, lowers rather remarkably the barrier for reaction of the activated molecule with CO and the subsequent emission of CO2. Our first-principles calculations are based on a density functional theory approach11,12 with exchange and correlation energy corrections included through a generalized gradient approximation (GGA).13 Such calculations have been shown to give very accurate bond lengths (up to 1% too long), and reaction energy barriers that are accurate to within 30% (usually too low) (see page 87 of ref 8). A plane wave basis is used with a cutoff kinetic energy of 30 Ry, and ultrasoft pseudopotentials (scalar relativistic for Au)14 are employed with Γ-point sampling of the Brillouin zone. In structural relaxations corresponding to minimization of the total energy, convergence is achieved when the forces on the atoms are less than 0.001 eV/Å. In modeling the metal-supported MgO films we use a four-layer Mo(100) slab (lattice constant of 3.15 Å) of thickness 4.64 Å, which has been found to reproduce (in its middle) the bulk electronic properties of Mo.15 We adsorb the planar Au20 clusters on a MgO/ Mo(100) with 5 × 6 unit cells of the Mo(100) surface; for the 3Dtetrahedral cluster the dimensions of the MgO surface (without the metal support) are the same, and the bottom layer of the metal oxide is held fixed. In all calculations, the periodically replicated slabs are separated [in the (100) direction] by a vacuum region of 20 Å. In structural optimizations all the atoms of the adsorbed gold clusters, the MgO thin film, and the first two layers of the Mo substrate are allowed to relax. In calculations involving the tetrahedral gold structure, the MgO(100) crystalline surface was modeled by a two-layer MgO(100) slab which is sufficiently thick both to reproduce the properties of the bare MgO surface16 and to obtain converged results (with respect to the number of MgO layers) for the energetics of adsorbed Au clusters17 (for further details see ref 10). The gas-phase optimal 3D tetrahedral Au20 cluster18 maintains its structure on the MgO(100) surface, with a 1.2 eV advantage over the planar structure. However, this cluster is found to adsorb O2 only weakly (0.34 eV on the top apex atom of the pyramid with (d(O-O) ) 1.28 Å remaining close to the gas-phase value), and no binding is found at peripheral sites of the gold/MgO interface; when CO is preadsorbed to the top apex Au atom (0.7 eV) no coadsorption of O2 occurs. In light of the above inactivity of the 3D structure we focused our investigations on Au20 adsorbed on a two-layer MgO/Mo(100). Here the planar isomer is more stable than the 3D one by 3.3 eV, with the enhanced stability resulting from penetration of metal states through the thin metal oxide film and charge accumulation at the cluster/MgO interface (up to 1 e for Au20). Furthermore, in this configuration the O2 adsorbs relatively strongly (1.35 eV) at the cluster periphery (Figure 1), and the process is accompanied by transfer of electronic charge19 (about 1.3 e) into the antibonding 2π* orbital leading to activation of the O-O bond of the adsorbed molecule into a peroxo state with d(O-O) ) 1.52 Å and no spin polarization (that is a spin-singlet state). We explored two mechanism for the reaction of CO with the activated oxygen molecule. One involves a coadsorbed CO molecule (Langmuir-Hinshelwood, LH), and in the other one the reactant CO molecules approach the activated O2 directly from the gas phase (Eley-Rideal, ER). Various CO coadsorption sites were explored. First we examined adsorption to the gold cluster, yielding relaxed adsorption configurations with CO binding energies ranging from 0.8 eV (CO binding to a peripheral Au atom that is nearest-neighbor Published on Web 02/01/2007