Application of first-principles determination of magnetic structures using magnetic interactions
POSTER
Abstract
For systematic prediction of magnetic structures, the ab-initio calculation based on spin-density functional theory is a powerful method. Here, we employ the formulation using the magnetic force theorem based on Wannier-orbital tight-binding model, which evaluates magnetic interactions between spin magnetic moments by assuming a magnetic structure. This formulation does not assure the consistency between the assumed magnetic structure and the resulting magnetic interactions [1]. Therefore, we can determine the ground-state magnetic structures and propagation vectors by calculating magnetic interactions with the assumption of virtual ferromagnetic structure and exploring energy minimum in classical Heisenberg model.
In this presentation, we validate this method with two results. First, we demonstrate the determination of the propagation vectors by calculating eigenvalues of magnetic interactions in reciprocal space. The obtained propagation vectors are consistent with fully ab-initio calculations or neutron diffraction experiments. Second, we show that the obtained ground-state magnetic structures reproduce experimental observations. We apply this method to 44 magnets on experimental database MAGNDATA and get accurate results for 90% of them. This precision is on par with the previous study [2], indicating high potential of this method for predicting magnetic structures.
[1] Nomoto et al., PRL 125, 117204 (2020).
[2] Huebsch et al., PRX 11, 011031 (2021).
In this presentation, we validate this method with two results. First, we demonstrate the determination of the propagation vectors by calculating eigenvalues of magnetic interactions in reciprocal space. The obtained propagation vectors are consistent with fully ab-initio calculations or neutron diffraction experiments. Second, we show that the obtained ground-state magnetic structures reproduce experimental observations. We apply this method to 44 magnets on experimental database MAGNDATA and get accurate results for 90% of them. This precision is on par with the previous study [2], indicating high potential of this method for predicting magnetic structures.
[1] Nomoto et al., PRL 125, 117204 (2020).
[2] Huebsch et al., PRX 11, 011031 (2021).
* WISE Program for AI Electronics, Tohoku University.
Presenters
-
Katsuhiro Arimoto
Tohoku University, Japan
Authors
-
Katsuhiro Arimoto
Tohoku University, Japan
-
Takashi Koretsune
Tohoku University, Department of physics, Tohoku University