Kinetics of Nanoparticle Catalysis from First Principles

Sammanfattning: Modern society depends heavily on heterogeneous catalysis, which creates strong economical and environmental incentives to improve catalyst efficiency. Heterogeneous catalysts are often realized as metal nanoparticles (NPs) supported on oxide surfaces, and catalysts are traditionally developed by trial and error approaches. However, rational catalyst design can be enabled by understanding reactions on the atomic scale. Presently, computational power has matured sufficiently to obtain atomic scale insights into reaction kinetics, directly from first-principles kinetic simulations. This thesis develops the methodologies of first-principles kinetic simulations over NPs. Multiple factors affect modeling of reactions over NPs, such as reaction energy landscapes and entropy changes during reaction. This makes it important to investigate different methodological choices, within kinetic modeling. Herein, Complete Potential Energy Sampling (CPES) is introduced as a method to calculate adsorbate entropy. CPES directly samples the adsorbate potential energy landscape, which allows for systematic improvements over approximate models within mean-field kinetics. CPES is tested on CO-oxidation over Pt(111), where it improves agreement with experimental references. Furthermore, CPES is applied to enable accurate description of molecular entropy in zeolites. Reaction energy landscapes on NPs are challenging to calculate as NPs contain multiple different sites. Thus, NPs are commonly approximated using extended surfaces as model systems. In this thesis, the challenge of mapping out NP reaction energy landscapes is solved pragmatically using scaling relations. Kinetic Monte Carlo simulations are used to investigate the kinetics for CO-oxidation over Pt and selective acetylene hydrogenation over Pd/Cu single-atom alloys. It is found that kinetic couplings between the NP-sites govern the kinetics. The kinetic couplings influence how turnover frequency and selectivity depend on particle size, shape, and strain. Thus, the energetics of isolated sites and extended surface models are found to have limited value as descriptors for NP catalysis.