Overview: recent studies of machine learning in phase prediction of high entropy alloys

• Published in: Tungsten Vol. 5; no. 1; pp. 32 - 49
• Main Authors: Yan, Yong-Gang, Lu, Dan, Wang, Kun
• Format: Journal Article
• Language: English
• Published: Singapore Springer Nature Singapore 01.03.2023; Springer Nature B.V
• Subjects: Algorithms; Chemistry and Materials Science; Classification; Corrosion resistance; Datasets; Density functional theory; Entropy; Experimental methods; Extreme environments; Feature selection; High entropy alloys; High temperature; Intermetallic compounds; Machine learning; Materials Engineering; Materials Science; Metallic Materials; Nuclear Chemistry; Particle and Nuclear Physics; Phase diagrams; Plasma sintering; Radiation tolerance; Research methodology; Review Paper; High entropy alloys; Phase prediction; Machine learning
• ISSN: 2661-8028; 2661-8036
• DOI: 10.1007/s42864-022-00175-0

High entropy alloys (HEAs), especially refractory HEAs, have become a subject of interest in the past years due to their exceptional properties in terms of high-temperature strength, corrosion resistance, radiation tolerance, etc. under extreme environments. While the phase formation of these alloys significantly affects their properties. If the phase of HEAs can be forecasted before the experiments, the material design process can be greatly accelerated. The phase formation study of HEAs mainly relied on trial-and-error experiments and multi-scale computational simulations such as calculation of phase diagrams (CALPHAD) and density functional theory (DFT). However, those methods require massive time, man-power, and resources. As a highly efficient tool, machine learning (ML) method has been developed and applied to predict the phase formation of HEAs very recently. This review provided a comprehensive overview and analysis of the most recent research work in this area. First, we introduce ML methodologies applied in HEAs' phase prediction in terms of principles, database, algorithm, and validation. We then summarize recent applications of the ML method in the phase prediction of HEAs. In the end, we propose possible solutions to the current problems and future research pathways for various challenges in the phase prediction of HEAs using ML.