Current status and prospects in machine learning-driven design for refractory high-entropy alloys

Abstract

Due to excellent comprehensive properties such as high strength, high hardness, and excellent high-temperature oxidation resistance, the refractory high-entropy alloys have broad application prospects and research value in the fields of aerospace and nuclear energy. However, the refractory high-entropy alloys have very complex composition features, making it difficult to perform alloy design. It seriously restricts the development of high-performance refractory high-entropy alloys. In recent years, the machine learning technique has been gradually applied to various high-performance alloys with efficient and accurate modeling and prediction capability. In this review, there was a comprehensive summary of research achievements on machine learning-driven design of refractory high-entropy alloys. A detailed review on the applications and progress of machine learning technique in different aspects was given, including alloy phase structure design, mechanical property prediction, strengthening mechanism analysis and acceleration of atomic simulations. Finally, the currently existing problems in this direction were summarized. The prospect about promoting the design of high-performance refractory high-entropy alloys was presented, including development of high-quality database for refractory high-entropy alloys, establishment of quantitative relation of "composition-process-structure-property" and achievement of multi-objective optimization of high-performance refractory high-entropy alloys.

Keywords

• refractory high-entropy alloy
• machine learning
• phase structure
• mechanical property
• strengthening mechanism
• atomistic simulation