Accelerated design of high-entropy alloy coatings for high corrosion resistance via machine learning

The high-entropy alloy (HEA) coating offers a promising solution by combining the superior performance characteristics of bulk HEAs with cost-effectiveness, facilitating broader application potential. Magnetron sputtering is a valuable method for producing HEA coatings, but establishing the relationship between composition, processing parameters, and performance is challenging due to the complexity of alloys with five or more principal elements. This study employed machine learning techniques to accelerate the screening and design of HEA coatings with enhanced corrosion resistance. This machine learning design framework constructed a random forest prediction model by using alloy composition ratios and key magnetron sputtering process parameters as input features, pitting potential (E_pit) and corrosion potential (E_corr) as output features, followed by multi-objective optimization via genetic algorithm. A HEA coating with excellent corrosion resistance was obtained through only four iterations and experimental verification. This approach rapidly guided the selection of components and process parameters, assisting in the development of new HEA coatings. As a result, the Ti₃₅Zr₁₄Nb₂₈Mo₇V₁₆ HEA coating was successfully prepared, demonstrating a pitting potential of 1931.1mV_SCE and a corrosion potential of 13.8 mV_SCE in 3.5 wt% NaCl solution. The passivation region (E_pit$-$E_corr, mV_SCE) was enhanced by 15 %, indicating excellent corrosion resistance. The corrosion resistance mechanism was also explained by microstructural characterization and electrochemical analysis.