ホーム > 研究活動 > 口頭発表(2016) > Computational exploration of new permanent magnet compounds

研究活動

Computational exploration of new permanent magnet compounds

日本磁気学会第40回学術講演会

2016年9月6日(火)

三宅隆

概要

I will discuss current status and challenges for permanent magnet research by information integration. Strong magnet compounds such as Nd2Fe14B, Sm2Fe17N3 and NdFe12N consist of three elements, namely rare-earth, iron and the third element. A natural question is: What is the best third element, and what about the fourth in a quaternary compound? This is an issue to be tackled by computational screening. As an example, we will present first-principles calculations of ThMn12 type iron-based compounds. However, brute-force search based on first-principles calculations is computationally demanding even if using supercomputer facilities, since the number of combinations of chemical composition increases rapidly as the number of elements in a compound is increased. Machine learning is a possible solution to improve the efficiency drastically. It is found that Gaussian process regression using 7 descriptors accurately reproduces the Curie temperatures of bimetal alloys composed of transition-metal and rare-earth elements. This technique can be utilized for virtual screening. Another issue is exploration of crystal structure. Saturation magnetization is expected to be larger as the iron content increases. Hence, the crystal structure of new iron-rich phases is of particular interest. Crystal structure prediction is a hot topic in computational materials science in the past decade, and various efficient algorithms have been developed. Recent progress and applications will be reviewed.

その他特記事項

ポスト「京」重点課題7、MI^2I、ESICMM


研究活動

文部科学省

文部科学省
元素戦略プロジェクト(活動紹介)
NIMS磁石パートナーシップ

元素戦略拠点

触媒・電池元素戦略拠点
触媒・電池元素戦略研究拠点 (京都大学)
東工大元素戦略拠点
東工大元素戦略拠点 (東京工業大学)
構造材料元素戦略研究拠点
構造材料元素戦略研究拠点 (京都大学)
高効率モーター用磁性材料技術研究組合
高効率モーター用 磁性材料技術研究組合