Design of methane oxidation catalysts using microkinetic modeling and machine learning

Abstract

As the supply of natural gas increases, the importance of technology using methane, which makes up most of the natural gas, is increasing. Methane becomes an energy source for transportation using natural gas as a fuel through combustion, and is used to generate synthesis gas through reforming and produce high value-added compounds through partial oxidation or C-C coupling reactions. However, for better utilization, new catalyst materials should be developed to solve the problems of the previous catalysts (high price, low activity and stability, etc.). In this thesis, we basically use computational chemistry to discover promising methane oxidation catalysts. Specifically, we establish a machine learning (ML) model for accurately predicting methane activation energy at a low cost. Also, the microkinetic model (MKM), which is the multiscale simulation, is constructed to confirm the activity under actual experimental conditions, and this model is reduced to the descriptor-based MKM that can predict the catalytic activity at a low computational cost. Furthermore, we efficiently perform high-throughput screening by combining the constructed MKM and ML model.