Systematic Magnetic Structure Generation Based on Oriented Spin Space Groups: Formulation, Applications, and High-Throughput First-Principles Calculations

Abstract

We propose a framework for generating magnetic structures, inspired by the concept of oriented spin space groups (SSGs): magnetic structures are first generated as totally symmetric representations of an SSG and are then rotated such that they belong to the maximal magnetic space group of the SSG, which we term spin-symmetry-adapted (SSA) structures and oriented SSA structures, respectively. This is a natural framework to enforce fixed magnetic moment magnitudes on the symmetry-equivalent sites as well as to exploit the spin-orbit coupling (SOC)-induced hierarchy of energy scales. To examine the present scheme, we analyze the MAGNDATA database and find that 77% of the reported structures are reproducible at the SSG level, among which 82% are fully reproduced within the oriented SSG scheme, regardless of their spin-only group types or propagation vectors. To quantitatively assess computational and predictive performance, we perform spin density functional theory calculations for 283 materials, first carrying out self-consistent calculations for SSA structures without SOC, followed by fixed-charge calculations including SOC for the descendant oriented SSA structures. The experimental magnetic structures are reproduced as energetically most stable in 82% of cases at the SSG level without SOC and in 76% of cases at the oriented SSG level with SOC, showing that the fixed-charge scheme enables accurate evaluation of SOC-induced energy differences at low computational cost. The characteristic energy scale among oriented SSA structures is only $\sim$0.29 meV per magnetic atom, about 300 times smaller than that of distinct SSA structures. These results demonstrate that oriented SSG-based enumeration, combined with the two-step calculations for SSA and oriented SSA structures, provides an efficient and robust route for large-scale magnetic-structure prediction.