Computational Analysis and Characterization of Catalytic Reactions on Platinum NanoparticlesPublic Deposited
Computational catalysis is a challenging discipline because of the complexity of catalytic particles and the large number of reaction pathways that may occur in a given reaction. This thesis addresses these challenges with two strategies. First, an appropriate molecular model is defined that captures adequate physical and chemical details while requiring limited computational resources. Second, experimental catalytic activity and selectivity data are analyzed over general trends in computational thermodynamic parameters known as descriptors. The primary reaction of interest is the hydrogenation of acrolein on platinum nanoparticles with strontium titanate or barium titanate substrates. This reaction is computationally desirable because acrolein is a relatively small molecule, platinum nanoparticles form well-ordered crystallographic surfaces and edges, and strontium titanate or similar substrates may be strained and modified by doping. Computational data used to characterize reactions include relaxed structures, adsorption energies, and charge distributions on several reaction sites. These data are important because trends in catalytic activity and selectivity versus computationally generated thermodynamic data allow catalyst optimization for certain controllable parameters such as substrate strain. In this thesis, doped and pure candidate substrates are computationally relaxed, adsorption results of atoms and small molecules are presented, critical adsorption and catalytic reaction sites are located, and modifications of the catalyst are analyzed. It is found that selectivity of acrolein hydrogenation to allyl alcohol correlates with molecular adsorption energy of a strained platinum catalyst due to substrate A-site.