Heterogeneous catalysis is tremendously complicated. A prototypical catalyst used in industrial reactors consists of metal nanoparticles dispersed onto a high surface area metal oxide. In this sense, the catalyst itself is not a single simple material, but a collection of many metal faces, edges, defect sites, co-adsorbed promoters, etc., all of which can contribute to the observed catalytic activity and selectivity. In contrast, well-controlled surface science measurements, as well as most computational studies, have traditionally focused on metal single crystals. The vast discrepancy between catalyst structures in these instances is known as the “materials” gap.
Instead of simply modeling catalysts with more ‘realistic’ structures to estimate reaction rates for various component mixtures, our goal is to understand the roles of the various factors (supports, promoters, etc.) in a more general, “chemical” and hopefully transferable sense. The goal is to understand the electronic structure as chemcial bonds within these solids in order to understand the chemical nature of different materials observed catalytic behavior. Thus, we are working to develop new computational analysis methodologies to address these qualitative questions, focusing on techniques that can be utilized in conjunction with plane wave density functional theory (PWDFT) calculations, which dominate computational catalysis.
Specifically, we have generalized the existing Natural Bond Orbital analysis (NBO) method to PWDFT. The NBO method yields a localized, chemical, “Lewis-like” perspective on bonding in molecules, and has been tremendously successfully in understanding the bonding and reactivity of isolated molecules. We generalized this method to work with PWDFT, addressing the need to handle both periodic boundary conditions and interface with the plane wave basis set. We hope this and other analysis methodologies will be uniquely suited to addressing these important catalysis questions.