2020/11/29 by Qi, Jie, Cheung, Andrew M., Poon, S. Joseph
#Disordered Systems and Neural Networks (cond-mat.dis-nn) #FOS: Physical sciences #Materials Science (cond-mat.mtrl-sci)
paper · doi:10.48550/arxiv.2011.14403
High-entropy alloys, which exist in the high-dimensional composition space, provide enormous unique opportunities for realizing unprecedented structural and functional properties. A fundamental challenge, however, lies in how to predict the specific alloy phases and desirable properties accurately. This review article provides an overview of the data-driven methods published to date to tackle this exponentially hard problem of designing high-entropy alloys. Various utilizations of empirical parameters, first-principles and thermodynamic calculations, statistical methods, and machine learning are described. In an alternative method, the effectiveness of using phenomenological features and data-inspired adaptive features in the prediction of the high-entropy solid solution phases and intermetallic alloy composites is demonstrated. The prospect of high-entropy alloys as a new class of functional materials with improved properties is featured in light of entropic effects. The successes, challenges, and limitations of the current high-entropy alloys design are discussed, and some plausible future directions are presented.