Physics-informed machine learning prediction of Curie temperature of rare-earth magnetic materials

P. Singh, T. Del Rose, A. Palasyuk, Y. Mudryk
Ames National Laboratory,
United States

Keywords: Rare-earth, Machine learning, DFT, Magnetism, Intermetallics


High performance permanent magnets with high Curie temperature (TC) containing less critical materials are integral to zero-carbon energy solutions. Machine-learning (ML) model was built over an available collection of experimentally measured Curie temperatures to develop a data-driven ML approach for predicting TC of multicomponent magnetic materials containing rare earths. We chose two compositions from the pseudo-binary (Zr1-xCex)Fe2 system to experimentally validate the ability of our model to predict TC of novel compounds. We also provided a detailed discussion on correlation of Curie temperature with de-Gennes scaling factor, and its breakdown below a certain rare-earth content in intermetallic compounds. The electronic-structure calculation using density-functional theory method was performed to show that change in electronic states and electron/hole fillings at the Fermi-level directly correlates with TC. Our ML model was able to capture the interplay of key electronic-structure features correlating with trends in TC. This shows that physics-informed ML can be used for accurate design of new high-performance materials with improved properties for environmentally sustainable applications. Acknowledgements: This work was supported by the U.S. Department of Energy (DOE), Office of Science, Basic Energy Sciences, Materials Science and Engineering Division. The database used in this work was created using the support from the Laboratory Directed Research and Development Program (LDRD) of Ames National Laboratory. The research is performed at the Ames National Laboratory, which is operated for the U.S. DOE by Iowa State University under contract DE-AC02-07CH11358.