Understanding Catalysts With Density Functional Theory and Machine Learning
CitationHoyt, Robert. 2018. Understanding Catalysts With Density Functional Theory and Machine Learning. Doctoral dissertation, Harvard University, Graduate School of Arts & Sciences.
AbstractCatalysts are the cornerstone of the chemicals industry, whose products are used in nearly all human endeavors. At the core of catalysis lies the intricate relationship between their atoms and electrons, where quantum mechanics dictates interactions with reactants, products, and electromagnetic fields. This it the "electronic structure" of catalysts, and studying this structure provides deep insight into the understanding and design of novel catalytic materials. This thesis focuses on understanding a small subset of promising heterogeneous catalytic systems using density functional theory (DFT), from oxygen evolution over polyiodide-doped graphene to the nonadiabatic dissociation of hydrogen over Cu nanoclusters. Some of these studies emphasize the importance of nonadiabatic behavior, especially magnetization transitions in Cu nanoclusters upon hydrogen dissociation. Further insights into catalytic properties can be obtained by comparing DFT calculations to corresponding machine learning predictions. For example, differences between DFT and empirical data-driven kernels highlight important discontinuous quantum mechanical effects in H adsorption on dilute Ag alloys. The studies presented here are examples of how detailed electronic structure calculations can be used to develop a deeper understanding of catalysts and how they might be improved.
Citable link to this pagehttp://nrs.harvard.edu/urn-3:HUL.InstRepos:41129179
- FAS Theses and Dissertations