| Amidst an energy crisis and environmental concerns, hydrogen emerges as an efficient
energy source, especially in solid oxide electrolysis cells using hydrogen evolution reaction
(HER). This thesis proposes a novel fusion of machine learning (ML) and quantum chemistry
to develop an intelligent sampling method for high-entropy alloy (HEA) microstructures
relevant to HER catalysts. To assess the efficacy of the proposed sampling technique, our
objective is to gauge the performance of the chosen models in predicting the adsorption energy
of individual and multi-site. QUANTICS, executed sequentially, includes feature scaling,
dimensionality reduction, clustering, and representative sample extraction. The dataset
comprises 12 features with four regions in a microstructure model, each region contains three
metals. Labels are based on hydrogen adsorption energy calculated via Density Functional
Theory (DFT).
Keywords: Machine Learning, Hydrogen Evolution Reaction, Intelligent sampling |