Structure prediction in high-entropy alloys with machine learning

Abstract

High-entropy alloy is an alloy design concept without a principal component. This concept not only refers to the complexity of alloy compositions but also means that when the high-entropy alloy transits from a high-energy state to low-energy state, there will be more intermediate metastable states. Corresponding to different states are the changes in the degree and manner of order in the microstructure. In this study, we used machine learning to combine elemental characteristics with long-term ordering and established 87% of prediction accuracy. This data-driven method can correlate elemental characteristics and metastable states and accelerate the discovery of potential compositions.