Recent progresses on high entropy alloy development using machine learning: A review

High-entropy alloy (HEA)-based materials have attracted considerable attention owing to their exciting opportunities for fundamental research and practical applications over the past 20 years. Through alloys/materials designs, the properties of the materials can be several orders of magnitude higher than that of conventional alloys. High entropy alloys represent a special class of complex concentrated alloys with unique properties and demanding applications. This review presents a comprehensive overview of the progress in developing high-entropy materials using machine learning and examines fabrication techniques and design of HEAs based on deep learning of recent research publications. It also outlines potential for further research directions in this field and provides an understanding of the current progress of HEAs regarding the application of Machine learning (ML) unlocking the next level of alloy development. It starts with describing the general characteristics of HEAs, such as their multicomponent nature, the multiphase constitution, mechanical properties, and stability in corrosive and high-temperature environments. It has been shown how multiple data analysis methods like supervised machine learning, unsupervised clustering, deep machine learning, reinforcement machine learning, etc. can be used to predict many of the important material properties like phase stability, mechanical strength, hydrogen storage capacity, etc., as opposed to the conventional practices which are often time-consuming and less accurate based on trial-and-error approaches. Moreover, the applications of ML to computational approaches such as calculation of phase diagrams (CALPHAD), first-principle, and molecular dynamics have been addressed as a means to fortify models of HEAs. Critical challenges, like data limitations, challenges in model explainability, and computational complexity of applying machine learning to HEAs development are reviewed. This review also emphasizes the role of ML in creating a data-driven framework, optimizing HEAs properties, and moving forward with the trend of machine learning-driven materials discovery. The subsequent case studies demonstrate how, by using ML, new HEAs have been discovered, supporting the usefulness of it in materials science. A prospective outlook on future directions that will shape ML for future advances in HEAs, including data sharing, better algorithms, and stronger physics-informed constraints will be emphasized.