Exploration of current methods for modeling catalytic materials: periodic density functional theory and machine learning

Abstract

Traditional energy technologies are unsustainable in the modern world. Thus, the development of more advanced technologies is of great significance. From a chemistry perspective, an ideal area for development is within the realm of catalysis. Using catalysts leads to faster and more energy-efficient chemical reactions. Finding an optimal catalyst is of great interest in the communities of chemists and material scientists. To that end, a fundamental understanding of the properties and mechanisms of action of the current most efficient and environmentally sustainable catalysts must be obtained to enable the design of next-generation catalysts. Computational models are very efficient ways of obtaining such information and have become an invaluable component of this work. This dissertation uses computational chemistry methods, namely Density Functional Theory (DFT), to study complex catalytic systems. Data science techniques are also utilized, including machine learning (ML) methods. Many diverse systems are surveyed here, including metal–organic frameworks (MOFs) and surfaces. Section 7 focuses on the computational modeling of magnetic MOFs. Sections 8 and 9 focus on elucidating mechanisms concerning the nitrogen reduction reaction (NRR) and the methane oxidation reaction (MOR) on surfaces, respectively. Section 10 gives a brief introduction on the use of machine learning (ML) techniques in MOFs. Finally, Section 11 uses ML methods to predict adsorption energies on surfaces utilizing nominal information.