Machine Learning-Based Optimization Method for the Oxygen Evolution and Reduction Reaction of the High-Entropy Alloy Catalysts

In recent times, high-entropy alloys (HEAs) have found application in heterogeneous catalysis, capitalizing on their vast chemical potential. Yet, this extensive chemical landscape presents significant challenges when attempting a comprehensive exploration of HEAs through traditional trial-and-error approaches. Therefore, the machine learning (ML) approach is offered to appearance into the catalytic activity (CA) of countless sensitive sites on HEA surfaces in the oxygen-lessening response (ORR) and oxygen evolution reactions (OER). In this research, a Density Functional Theory (DFT) with a supervised ML model is assembled and founded on the gradient boosting regression (GBR) algorithm that predicted the O2 adsorption energies with a high overpotential of all surface sites on the two HEAs. Initially, the HEAs Co-Fe-Ga-Ni-Zn and Al-Cu-Pd-Pt offer a framework for adjusting the composition of disordered multi-metallic alloys to regulate the activity and selectivity of the reduction of oxygen to extremely reduced compounds. This attains generalizability, high accuracy and simplicity with the proposed technique. For fine-tuning such features, HEAs provide a huge compositional space. Consequently, the research reports the custom of the Bayesian optimization model based on HEA active compositions to suppress the formation of Oxygen (O2) and with strong O2 adsorption to favour the lessening of O2. The GBR approach is applied to build a highly accurate, easily generalizable, and effective ML model. The proposed work is analysed using Python software. The findings show that the separate charities of correlated metal atoms close to the responsive site are mixed to form the adsorption energy, which is clear from a thorough analysis of the data. It is suggested that a highly effective HEA catalyst composed of Co-Fe-Ga-Ni-Zn and Al-Cu-Pd-Pt be exploited, which is an effective method for further enhancing the ORR CA of potential HEA catalysts. An instruction manual for the logical design and synthesis of HEA catalysts' nanostructures is provided by the proposed research.