Energetics at Metal-Support Interfaces: Effect on Water Gas Shift Intermediates

Energetic Interactions at the Metal-Support Interface: Effects on Water Gas Shift Intermediates

Metal-support interactions often have a large impact on the catalytic properties, either by direct participation of the support, indirect modification of electronic or structural properties, or through special sites at the interface. For example, the WGS activity of Au supported on TiO₂ is twenty times higher than Au supported on γ-Al₂O₃. We have found that density functional theory (DFT) based models of unsupported Pd and Pt fail to capture the experimentally observed WGS kinetics on γ-Al₂O₃, indicating that even "inert" supports play an active part in the catalysis. Identifying the different factors that control reactivity at the metal-oxide interface is thus essential for rational catalyst design and optimization. However, there is no general approach to modeling metal-oxide interfaces. Multiple interface models have been proposed in the literature, but they are mostly limited to specific cases, and very often the implications of the model choice are not discussed.

We report detailed investigations of WGS intermediates on models of supported quasi-two-dimensional transition metal nanowires. These calculations have been performed in a consistent fashion within the framework of DFT. We systematically vary the type of metal (Au, Ag, Pd, Pt), the oxide (MgO(100), α-Al₂O₃(0001)), and the number of layers in the nanowire. We have found that electronic interactions between the metal and the support are limited to near the interface (see Fig 1(a)), and metal-like behavior is recovered at the top of the nanowire with as few as 2-3 layers. One-layer surfaces show different behavior due to quantum size effects. This is also reflected in binding behavior of WGS intermediates – adsorption energies on metal sites remote from the interface are similar to those on the unsupported metal, Sites at the interface have a modest stabilizing influence on the adsorbates, shown for the case of H in Fig. 1(b-c). On the oxide, we find that binding of molecular intermediates (CO, H₂O) is unaffected by the presence of the metal. However, a dramatic metal-dependent stabilization of redox-active intermediates (H, OH and COOH) is seen, caused due to electron tunneling between the metal and the adsorbate. This is closely coupled with large structural relaxations of the oxide surface, more pronounced in the case of α-Al₂O₃(0001) than MgO(100). This behavior of odd-electron adsorbates offers opportunities for tuning catalytic properties based on the choice of oxide and the choice of metal. Isolating different energetic contributions that affect the binding of such adsorbates and their catalytic implications is the focus of our study.

In Fig. 2, we compare potential energy diagrams of H₂O dissociation near the interface of the 3-ML Pd-Al₂O₃ system to that of the isolated metal and oxide. We see that H₂O dissociation is rapid on the oxide and slow on the Pd, and only near the interface can H and OH be separated. This illustrates the "dual" nature of the interface, i.e., the combined metal oxide system can be catalytically active, even though the individual components are inactive.

Figure 1: (a) Charge density difference plot of 3-ML Au nanowire on α-Al2O3(0001) (side view) . (b) Top view of the nanowire model. Atop, bridge and hollow sites are shown as circles, rectangles and triangles respectively. (c) H binding energies referenced to ½ H2 vs. adsorption sites on Au, Ag, Pd, Pt.

Figure 2: Comparison of H2O dissociation potential energy surfaces on on alumina, Pd(111), and the Pd/alumina interface