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 with 2935 calculations, which show that (i) the CMP expansion administers an exhaustive list of candidate magnetic structures, (ii) 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 . For a subset the impact of on-site Coulomb repulsion is investigated by means of 1545 calculations revealing no further improvement on the predictive power.
- Received 1 September 2020
- Revised 17 November 2020
- Accepted 11 January 2021
DOI:https://doi.org/10.1103/PhysRevX.11.011031
Published by the American Physical Society under the terms of the Creative Commons Attribution 4.0 International license. Further distribution of this work must maintain attribution to the author(s) and the published article’s title, journal citation, and DOI.
Published by the American Physical Society
Physics Subject Headings (PhySH)
Popular Summary
In a magnetic material, some of the atoms in the crystal carry magnetic moments, which align themselves with their neighbors in very specific formations. The smallest building block of a magnetic material—the magnetic structure—fundamentally determines the properties of a magnet. Yet predicting the magnetic structure is highly nontrivial because of the many variables in the system. Here, we show that state-of-the-art simulation of many interacting particles can predict the magnetic ground state of a material from first principles if we provide a sophisticated list of possible magnetic structures.
The list of magnetic structures is found via combinations of special functions defined on a sphere—analogous to how atomic orbitals are constructed—but highly adapted to the crystal symmetry. We then perform a high-throughput calculation with close to 3000 possible magnetic structures for 131 materials and find many stable and metastable magnetic structures. We compare our prediction to the gold standard: experiments, which we aim to reproduce. This kind of high-throughput benchmark calculation shows that our approach can significantly narrow down the possible properties of a material.
With this approach, the exploration of magnetic materials is pushed to enter a new paradigm of material design that is well matched to the era of big data.