Executive Summary : | Oxygen evolution reaction (OER) is a crucial aspect of energy conversion and storage technologies. Currently, costly metal-based catalysts like RuO2 and IrO2 are the best for OER. However, finding a large configuration of catalysts is a complex task. Carbon and molecular-based materials could be a potential solution for OER due to their efficiency, stability, and cost. To design the best catalyst cost-effectively, theoretical insight can help overcome the complexity in defining optimal electronic and structural descriptors. To find the best catalyst, first understand the electronic structure and correlate it with energy parameters of intermediate steps ΔGO, ΔGOH, and ΔGOOH. Then, define the best descriptors and create a predictive model equation to estimate overpotential and adsorption free energy. The third step is applying a machine learning algorithm to use the DFT estimated parameter and define a predictive model for training and testing. This approach can predict the activity of a large number of sites of various model structures of the same class. Recent work has shown that only one DFT calculation of host surface can estimate the overpotential and energy parameter of all active sites belonging to the host surface using a predictive model developed using ML technique. In this project, the data will be generated using the predictive model equation, followed by developing a general descriptor and ML-assisted predictive model for molecular catalysts towards OER. |