Benchmark for Ab Initio Prediction of Magnetic Structures Based on Cluster-Multipole Theory

M.-T Huebsch,R Arita,T Nomoto,M.-T Suzuki

doi:10.1103/physrevx.11.011031

M.-T Huebsch, R Arita + Show 2 more

Open Access

https://doi.org/10.1103/physrevx.11.011031

Copy DOI

Abstract

The cluster multipole (CMP) expansion for magnetic structures provides a scheme to systematically generate candidate magnetic structures specifically including noncollinear magnetic configurations adapted to the crystal symmetry of a given material. A comparison with the experimental data collected on MAGNDATA shows that the most stable magnetic configurations in nature are linear combinations of only few CMPs. Furthermore, a high-throughput calculation for all candidate magnetic structures is performed in the framework of spin-density functional theory (SDFT). We benchmark the predictive power of CMP+SDFT with $2935$ calculations, which show that (i) the CMP expansion administers an exhaustive list of candidate magnetic structures, (ii) CMP+SDFT can narrow down the possible magnetic configurations to a handful of computed configurations, and (iii) SDFT reproduces the experimental magnetic configurations with an accuracy of $\pm0.5\,\mu_\text{B}$. For a subset the impact of on-site Coulomb repulsion $U$ is investigated by means of $1545$ CMP+SDFT+U calculations revealing no further improvement on the predictive power.

Highlights

The grand challenge in first-principles calculation for magnetic materials is whether we can predict the experimental magnetic structure for a given material
We benchmark the predictive power of CMP þ spin-density functional theory (SDFT) with 2935 calculations, which show that (i) the CMP expansion administers an exhaustive list of candidate magnetic structures, (ii) CMP þ SDFT can narrow down the possible magnetic configurations to a handful of computed configurations, and (iii) SDFT reproduces the experimental magnetic configurations with an accuracy of Æ0.5μB
(b) Can the experimentally determined magnetic configuration be found among all SDFT results? Note that the similarity between two magnetic configurations is expressed by the overlap defined in Eq (12)

Summary

Introduction

The grand challenge in first-principles calculation for magnetic materials is whether we can predict the experimental magnetic structure for a given material. First-principles calculations with the generalized gradient approximation (GGA) in the framework of spin-density functional theory (SDFT) for magnetic materials have a problem: It is still an open question how accurately SDFT GGA can reproduce the experimental magnetic ground state. While SDFT has been widely used in studies on various magnets [8], there has been no systematic benchmark calculation for noncollinear AFM materials. In regard to noncollinear AFM materials, highthroughput calculations have been limited to setting the experimentally determined magnetic configuration as an initial guess [13]. To search for all the (meta)stable states, we need an exhaustive list of physically reasonable magnetic configurations for which first-principles calculations can be performed

Methods

Results

Conclusion