Predictions on the Phase Constitution of SmCo_7−XM_x Alloys by Data Mining

Abstract

Based on a home-built Sm-Co-based alloys database, this work proposes a support vector machine model to study the concurrent effects of element doping and microstructure scale on the phase constitution of SmCo7-based alloys. The results indicated that the doping element’s melting point and electronegativity difference with Co are the key features that affect the stability of the 1:7 H phase. High-throughput predictions on the phase constitution of SmCo7-based alloys with various characteristics were achieved. It was found that doping elements with electronegativity differences with Co that are smaller than 0.05 can significantly enhance 1:7 H phase stability in a broad range of grain sizes. When the electronegativity difference increases to 0.4, the phase stability becomes more dependent on the melting point of the doping element, the doping concentration, and the mean grain size of the alloy. The present data-driven method and the proposed rule for 1:7 H phase stabilization were confirmed by experiments. This work provides a quantitative strategy for composition design and tailoring grain size to achieve high stability of the 1:7 H phase in Sm-Co-based permanent magnets. The present method is applicable for evaluating the phase stability of a wide range of metastable alloys.

Keywords

• permanent magnets
• machine learning
• phase stability
• composition design
• grain size