(495d) Chemical Kinetics of "Simple" Reactions From DFT and Cluster Expansion Models

Among the many contributions of Nick Delgass to heterogeneous catalysis science and practice is strong advocacy for developing the deep connections between observed chemical kinetics, underlying mechanism, and catalyst material properties.  In principle, the knowledge of the connections between all these can be used to drive the discovery of new catalysts.  In this talk we explore these connections computationally, using as a model system catalytic oxidation on transition metal surfaces.  One of the key varialbes that is difficult to infer in a kinetic model is the sensitivity of surface reaction rates to the surrounding environment of co-adsorbates.  We first use supercell density functional theory (DFT) simulations to probe the phenomena the underlie adsorbate-adsorbate interactions.  We then show how these interactions can be captured in a cluster expansion (CE) model, taking as an example the coverage-dependent adsorption of oxygen on the late transition metal surfaces.  We show through a combination of the CE and Grand Canonical Monte Carlo how the coverage of surface adsorbates introduces a coverage-induced heterogeneity on top of any other structural heterogeneity in the material.  Using the results for coverage-dependent O adsorption combined with Brønsted-Evans-Polyani relationships between the fintal state energy and the rate of surface reactions, we develop steady-state kinetic models for both the temperature-programmed desorption of O2 and the catalytic oxidation of NO to NO2, by ensemble averaging over available surface sites as a function of reaction conditions.  We construct rate derivatives to extract apparent activation energies and rate orders from the kinetic models, and show how these observed macroscopic kinetics are related in subtle ways to the underlying state of the catalyst surface.  The results illustrate the rich kinetic behavior that can emerge even in simple, steady-state reacting systems, and how DFT-enabled modeling can help bridge the "kinetic gap" betwen experiment and reaction mechanism.