(477b) Finding Needles in a Haystack: Sifting through 16M Catalysts for Optimal Methane-to-Methanol Catalyst Design Under Weak Thermodynamic Scaling
AIChE Annual Meeting
2022 Annual Meeting
Catalysis and Reaction Engineering Division
Data Science & Machine Learning Approaches to Catalysis II: AI-Accelerated Modeling of Catalysts and Materials
Wednesday, November 16, 2022 - 1:06pm to 1:24pm
Due to strong oxidation and spin-state dependence on the relative energetics of reactive intermediates on the methane-to-methanol energy landscape, linear free energy relationships that are invoked during HTVS to simplify catalyst screening cannot be readily used. As an alternative approach, the absence of universal scaling relations between intermediate energetics provides an opportunity for non-linear machine learning (ML) models that can be used over a larger space of candidate materials. Rather than relying on linear relationships between quantities, ML models can be trained to directly predict catalyst reactivity on the basis of chemical composition and applied to thousands of compounds.
We sift through a large space of possible compounds to evaluate candidate catalysts. Here, we draw inspiration from experimentally synthesized macrocycles (e.g. porphyrins, cyclams, corroles, and phthalocyanines), and recombine their pieces to make new macrocycles. We use ML model driven efficient global optimization with a 2D expected improvement criterion to simultaneously optimize the thermodynamics for C-H bond activation (e.g. hydrogen atom transfer), and methanol release energetics. This creates a Pareto front of methane oxidation catalysts that demonstrate the best tradeoffs between C-H bond activation and methanol release. We use active learning to improve model performance on the Pareto front and generate novel lead candidate materials that we validate by DFT.