Ab initio magnetic structure prediction for topological magnets

To theoretically design functional magnets having a non-trivial spin configuration, we need a reliable method for magnetic structure prediction. While the numerical cost for structure prediction is generally expensive, we recently formulated an efficient scheme based on the cluster multipole theory [1].

The approach based on cluster multipole moments was originally introduced to characterize the non-collinear magnetic structures in topological antiferromagnets such as Mn3Sn and Mn3Ge [2,3]. Although the uniform magnetization is vanishingly small in these materials, significant anomalous transverse transport phenomena such as the anomalous Hall effect [4], anomalous Nernst effect [5], and magnetooptical Kerr effect [6] have been observed. It has been shown that these phenomena can be described by octupolarization, i.e., a ferroic order of a cluster octupole moment [2,3].

Recently, we have shown that the multipole expansion of the spin configuration of magnets can be used in the ab initio magnetic structure prediction. Namely, we can systematically and efficiently make a list of initial guesses in the calculations based on spin density functional theory. In the benchmark calculation for magnets in the database (MAGNDATA), we demonstrated that the experimental magnetic structures of almost all magnets were successfully reproduced [1].

I will show that we can combine this method of magnetic structure prediction and the scheme of the high-throughput calculation with which we recently succeeded in finding ferromagnets showing a large anomalous Nernst effect [7] and make a database of anomalous transverse transport in antiferromagnets.