Learning Design Rules for Catalysts through Computational Chemistry and Machine Learning

Abstract

The study of transition metal catalysts with computational chemistry is essential to identify reactive intermediates and mechanisms, advancing both understanding and design. The combinatorial space arising from combinations of ligands, metals, oxidation states, and spin states mandates accelerated searches to design transition metal complexes with targeted properties. This chapter focuses on machine-learning accelerated inorganic discovery. First, we cover computational chemistry methodology and concepts that have led to more efficient traversal of transition-metal chemical space for catalysis. We demonstrate how computational catalysis coupled to machine learning makes it even faster to discover new catalysts. Next, we cover opportunities in harnessing experimental data sources and gaining insights by supplementing these data sources with computational modeling. Overall, this chapter highlights the related roles of computational catalysis, experimental data, machine learning, and optimization on improved materials design.