Theoretical Insights on O2 and CO Adsorption on Neutral and

With the aim of understanding the elementary steps governing the oxidation of CO catalyzed by dispersed or supported gold nanoclusters, the adsorption...
1 downloads 0 Views 240KB Size
12240

J. Phys. Chem. B 2006, 110, 12240-12248

Theoretical Insights on O2 and CO Adsorption on Neutral and Positively Charged Gold Clusters Antonio Prestianni and Antonino Martorana Dipartimento di Chimica Inorganica e Analitica, UniVersita` di Palermo, Palermo, Italy

Fre´ de´ ric Labat, Ilaria Ciofini, and Carlo Adamo* Laboratoire d’Electrochimie et Chimie Analytique, UMR-7575, E.N.S.C.P., Paris, France ReceiVed: December 15, 2005; In Final Form: March 27, 2006

With the aim of understanding the elementary steps governing the oxidation of CO catalyzed by dispersed or supported gold nanoclusters, the adsorption of molecular species, such as O2 and CO, on model neutral and positively charged clusters (Aunm+ n ) 1, 9, and 13; m ) 0, 1, and 3) has been studied using an ab initio approach. The computed structural and thermodynamic data related to the binding process show that molecular oxygen interacts better with neutral clusters, acting as an electron acceptor, while CO more strongly binds to positively charged species, thus acting as an electron donor.

1. Introduction In recent decades, the interest toward gold-based nanostructured catalysts deposited on metal oxides has strongly increased mainly due to the wide domain of envisaged applications, ranging from environmental to industrial fields.1,2 Dispersed gold catalysts can be used to purify hydrogen obtained from industrial hydrocarbon reforming processes,3-8 since they have shown an unusually high catalytic efficiency toward the oxidation of CO (eq 1) at low temperatures (as low as 200 K).9

CO + 1/2O2 ) CO2

(1)

Nevertheless, even if several publications have been focused on this very important catalytic process, its reaction mechanism is still far from fully clarified. To this end, several experimental and theoretical studies were performed on systems in both real and ideal conditions, aimed at modeling the elementary steps of the catalytic process and, thus, providing fundamental understanding. Some experimental data support the hypothesis that a crucial role for catalytic efficiency is played by the support, in particular, by the heterojunction between the Au particles and the supporting metal oxide.10 Other experiments highlight the influence of the gold oxidation state related to a synergic effect due to the presence, on the support surface, of Au-Auδ+ couples.11 Further works have stressed the relevance of the size and shape of the clusters,12 corroborated by the fact that gold foils do not show any catalytic activity, while unsupported gold nanoparticles do.13-15 In particular, experimental studies performed on gold catalysts supported on MgO11,16 have shown that, beside the cluster dimensions, a fundamental role to increase the catalytic activity is played by cluster-support interactions, the latter being an active element in the catalytic process. Heiz and Schneider12 have demonstrated that the extremely high catalytic activity of an eight-atom gold cluster supported on MgO (100) is related to an electron transfer * Corresponding author. E-mail: [email protected].

(0.5|e-|) from the oxide surface to the metal cluster. This effect is enhanced in the presence of oxygen vacancy (F-center) defects in the MgO substrate.12,17 Recently, Heiz and co-workers18 have demonstrated through a combined experimental-theoretical study that the defects not only anchor the Au8 nanoparticles but also play the role of active sites, controlling the (negative) charged state of the gold clusters and, thus, the overall catalytic activity. On the other hand, Guzman and Gates,11 on the basis of XANES experiments performed on Au/MgO samples, have identified as active sites Au0 clusters interacting with cationic gold species at the gold-support heterojunction, thus supporting the idea of catalytically active oxidic gold. In agreement with these findings, Flytzani-Stephanopoulos and collaborators19 have shown that, in the case of ceria-supported catalysts (Au/CeO2), the active species in the water gas shift reaction (CO + H2O ) CO2 + H2) is probably a gold oxide. Finally, Wallace and Whetten20 found experimental evidence of cooperatiVe effects in the coadsorption of O2 and CO on gold anion clusters, indicating that the two molecules adsorb on different coordination sites instead of competing for the same binding site (competitive coadsorption). From the theoretical side, several ab initio calculations have been performed in order to model gold clusters of increasing size,21-27 gold surfaces and supported gold surfaces,28-30 adsorption of O2 and CO both on gold clusters17,31-38 and on surfaces,39-44 as well as their coadsorption.17,45 To this end, density functional theory (DFT)46 has mainly been applied due to its favorable cost to accuracy ratio allowing treatment of fairly large molecular aggregates, which can better simulate the real nanoparticles. Nevertheless, both post-Hartree-Fock calculations on smaller aggregates37 and studies on larger clusters, performed using semiempirical potentials or combining DFT with genetic-symbiotic optimization methods or molecular dynamics, can be found in the literature.25,26,44 Among the papers focused on the determination of the stability of increasingly large gold aggregates, several were carried out imposing the overall symmetry of the clusters,21-23

10.1021/jp0573285 CCC: $33.50 © 2006 American Chemical Society Published on Web 06/03/2006

O2 and CO Adsorption on Charged Au Clusters while a few were attempting to scan the full potential energy surface (PES) and, thus, to find the global minimum.24-26,44 As a matter of fact, although, in the case of model theoretical studies, small structured (i.e., symmetry-constrained) clusters have been used, there is experimental evidence that gold nanoparticles with diameter of 1-2 nm may present various noncrystallographic and also amorphous arrangements,47 thus suggesting the necessity of applying methods exploring the overall PES. Other authors have modeled the gold nanoparticles as surfaces and thus studied crystallographic gold surfaces.28-30 Finally, more recently, the effects of the metal oxide support and of its defects have also been taken explicitly into account studying supported gold clusters (Au/MgO) at the DFT level.18,27,44,48 Concerning molecular oxygen adsorption, the main outcome from both theoretical and experimental studies is that O2 interacts better with gold nanoparticles having an odd number of electrons35,36,38 and that, among gold clusters possessing an odd number electrons, the negatively charged ones interact more strongly than the neutral ones.33,34 In particular, the work of Ding et al.38 showed that by using hybrid functionals (i.e., B3LYP) a good agreement with experimental data for binding energies can be recovered for cationic, anionic, and neutral gold clusters of up to six gold atoms. Indeed, coadsorption of other molecular species such as OH- can activate(/deactivate) nonreactive(/reactive) bare gold clusters.49 In general, gold clusters able to donate electrons interact better with O2 and substantially activate this molecule by elongation of the O-O bond distance upon adsorption (i.e., formal reduction of molecular oxygen to superoxide, O2-). As a consequence, adsorption of molecular oxygen on small positively charged gold clusters is experimentally not observed,50,51 the related binding energy being very small (