Autonomous materials search using machine learning and ab initio calculations for L1 ₀ -FePt-based quaternary alloys

The efficient exploration of expansive material spaces remains a significant challenge in materials science. To address this issue, autonomous material search methods that combine machine learning with ab initio calculations have emerged as a promising solution. These approaches offer a systematic and rapid means of discovering new materials, particularly when the material space is too large. This requirement is particularly important in the development of L10-structured alloys as magnetic recording media. These materials require a high magnetic moment (M) and magnetocrystalline anisotropy energy (EMCA) to satisfy the demands of next-generation data storage technologies. Although autonomous search methods have been successfully applied to various material systems, quaternary L10 alloys with optimized magnetic properties remain an open and underexplored frontier. In this study, we present a simulation-based autonomous search method aimed at identifying quaternary L10 alloys with enhanced M and EMCA values. Over a continuous 100-day search, our system suggested the FeMnPtEr alloy system as a promising candidate, exhibiting superior values for both M and EMCA. Although further experimental validation is required, this study underscores the potential of autonomous search methods to accelerate the discovery of advanced materials.