A Hybrid Modeling Framework for Catalytic Systems: Sensitivity Analysis and Estimation of Activation Energies

In recent years there has been a growing need for novel electrocatalysts to produce green ammonia via Nitrogen Reduction Reaction (NRR). Screening new catalyst materials requires a detailed understanding of the pathways underlying these reactions. To this end, coming up with a framework that allows us to easily identify feasible reaction pathways associated with the NRR reactions is of significant value. In this regard, it is important to identify and capture the underlying latent mechanisms governing these reaction pathways. To this end, hybrid models that combine machine learning techniques with first-principles models have extensively been utilized to model such complex chemical processes and capture these latent mechanisms. As of now, the application of hybrid models has stayed limited to the systems governed by simple differential equations. In this work, extending the application of hybrid models to the field of catalysis has been explored. Specifically, the ability of these models to decipher reaction pathways is further understood in great detail by comparing the same with the values from density functional theory (DFT).