Selecting Doping Elements by Data Mining for Advanced Magnets

Compared with other material performances, magnetic properties ate more closely related to the electronic structure and sensitive to the material composition. To date, no unified approach has been reported to understand the complicated influence of composition on a material's magnetic properties. As one of the two most representative permanent magnetic systems, Sm-Co-based materials have been developed in terms of composition design mainly by empirical exploration. Taking advantage of a unique home-built Materials Genome Initiative database of Sm-Co systems, this work reports a data-driven material development that aims at high saturation magnetization, along with excellent comprehensive magnetic performance. A novel method using a unified indicator capable of precisely evaluating effects of doping elements on saturation magnetization was developed, and the priority of possible elements was predicted by extending the screening of elements to the periodic table, from which some elements that have never been tried in magnetic materials are discovered. By this approach, a top level of saturation magnetization was achieved in a series of prepared stoichiometric Sm-Co-based alloys, which also have high coercivity and high remanence in the same system. The strategy proposed here will be applicable on doping element selections for a large variety of material property modulations.