Searching for High-Performing Magnetic Materials in High-Entropy Space Using Active Learning

Advancements in magnetic recording technologies, such as Heat-assisted magnetic recording (HAMR) media, may be hidden in the configuration space of high-entropy incarnations of known high-performing magnets. The rare-earth L1₀-phase MnAl is a metastable state with experimental magnetic anisotropy energy (MAE) values ranging from 1-1.7 MJ/m³. We introduce disorder to the Mn site in pursuit of enhanced MAE, found through performing ab initio calculations within the coherent potential approximation (CPA). One Gaussian process is used to predict recipes with high MAE existing in the configurational space where (MnFeCoNi) may take occupation of the disordered site; another predicts the volumes and c:a ratios providing average relaxed lattices for these recipes. With this workflow, we can efficiently determine candidates for high performing recipes to be tested in experimental synthesis. We found that the recipe (Mn_0.92Fe_0.06Ni_0.02)Al had a calculated MAE of 3.09 MJ/m³ compared to our calculated MAE value for pure MnAl of 2.3 MJ/m³. This demonstrates the great potential to discover materials with strong magnetic anisotropy within the compositional parameter space of high entropy mixing.