Design of high temperature oxidation-resistant high-entropy alloys via machine learning and natural mixing process

High-entropy alloys (HEAs) have attracted significant attention for their exceptional properties, particularly their potential in high-temperature engineering applications. However, the large compositional space of HEAs presents challenges in alloy design to achieve optimal properties. In this study, we developed an integrated approach that combines machine learning (ML) and a natural mixing process to guide the design of HEAs with enhanced high-temperature stability. ML was used to assist in element selection and oxidation prediction, while the natural mixing and short-term high-temperature exposure guided the formulation of HEA compositions with superior thermal stability. Among the ML models evaluated, the Gradient Boosting Regression (GBR) model showed the highest prediction accuracy (R2= 0.94). A series of HEAs were designed using the integrated approach, and their oxidation behavior was thoroughly investigated. The designed alloy H3 (AlCrCu0.4FeNi) showed excellent oxidation resistance (kp=1.19 ×10−2 mg2/cm4·h) with a high hardness (865 HV), demonstrating its potential for high-temperature applications.