Searching high spin polarization ferromagnet in Heusler alloy via machine learning
Xiao Hu, Yaqiong Zhang, Shuaiyu Fan, Xin Li, Zhenjie Zhao, Chao He, Yong-Hong Zhao, Yong Liu, Wenhui Xie
Abstract
In order to search for stable ferromagnets with high spin polarization in Heusler alloys for spintronic applications, we develop an efficient machine learning workflow based on a deep neural network, whose training data were collected from the open quantum materials database and high throughput calculation by first-principle calculations. The lattice constants, formation energy and spin polarization of 10 577 candidate materials were predicted, and 192 materials with high spin polarization were selected according to a spin polarization greater than 0.87 and formation energy less than 80 meV/atom. 57 of these alloys have been reported as Half-metal (100% spin polarization) according to previous researches, and 18 have been reported as semiconductors. Especially, 6 Heusler alloys were identified as promising half-metallic ferromagnets, and some of them have high Curie temperature above room temperature. Our study suggests this approach is an efficient method for the discovery of superior spintronic materials, which should be also suitable for exploring other functional materials.