Linear Response Based Theories for Dzyaloshinskii-Moriya Interactions

TL;DR

This paper evaluates linear response techniques, especially the magnetic force theorem, for calculating Dzyaloshinskii-Moriya interactions from first-principles, demonstrating their effectiveness in various Cr-based materials and clarifying their relation to other models.

Contribution

It compares the magnetic force theorem with exact inverse response methods for DM interactions and introduces a downfolding procedure for effective spin interactions.

Findings

MFT is a good approximation for DM interactions when ligand SO coupling is included.

MFT and exact methods differ significantly for isotropic exchange interactions.

The spin-current model for DM interactions can be derived from MFT expressions.

Abstract

We investigate abilities of various linear response based techniques for extracting parameters of antisymmetric Dzyaloshinskii-Moriya (DM) interactions from the first-principles electronic structure calculations. For these purposes, we further elaborate the idea of Sandratskii [Phys. Rev. B 96, 024450 (2017)], which states that $z$ component of the DM vector can be computed by retaining only spin-diagonal part of the spin-orbit (SO) coupling. We start our analysis with the magnetic force theorem (MFT), which relies on additional approximations resulting in the linear dependence of the exchange interactions on the response tensor, and compare it with the exact approach formulated in terms of the inverse response. We propose the downfolding procedure transferring the effect of these ligand spins into parameters of effective interactions between the localized spins. These techniques are applied for the series of CrCl$_3$ and CrI$_3$ based materials, including bulk, monolayer, bilayer, and three-layer systems. Particularly, we discuss how the DM interactions are induced by the inversion symmetry breaking at the surface or by the electric field. As long as the SO interaction of the heavy ligand atoms is taken into account in the calculations of the response tensor between the Cr $3d$ states, the MFT appears to be a good approximation for the DM interactions, being in contrast with the isotropic exchange, for which MFT and the exact method provide quite a different description. Finally, we discuss the relevance of our approach to other techniques ever proposed for calculations of the DM interactions. Particularly, we argue that the spin-current model for the DM interactions can be derived from the MFT based expression and is the relativistic counterpart of the double exchange, occurring in metallic systems in the limit of infinite exchange splitting.