# Crystal field coefficients for yttrium analogues of rare-earth/transition-metal magnets using density-functional theory in the projector-augmented wave formalism

## TL;DR

This paper introduces a DFT-based method to calculate crystal field coefficients for RE-TM magnets by using yttrium analogues, avoiding complexities of 4f electrons, and successfully predicts magnetic anisotropy consistent with experiments.

## Contribution

The paper presents a novel DFT approach utilizing Y-analogues and projector-augmented wave formalism to accurately compute crystal field effects in RE-TM magnets.

## Key findings

- Accurately predicts easy magnetization directions in tested materials.
- Shows good agreement with previous open-core approximation calculations.
- Potential for large-scale screening of magnetic materials.

## Abstract

We present a method of calculating crystal field coefficients of rare-earth/transition-metal (RE-TM) magnets within density-functional theory (DFT). The principal idea of the method is to calculate the crystal field potential of the yttrium analogue ("Y-analogue") of the RE-TM magnet, i.e. the material where the lanthanide elements have been substituted with yttrium. The advantage of dealing with Y-analogues is that the methodological and conceptual difficulties associated with treating the highly-localized 4f electrons in DFT are avoided, whilst the nominal valence electronic structure principally responsible for the crystal field is preserved. In order to correctly describe the crystal field potential in the core region of the atoms we use the projector-augmented wave formalism of DFT, which allows the reconstruction of the full charge density and electrostatic potential. The Y-analogue crystal field potentials are combined with radial 4f charge densities obtained in self-interaction-corrected calculations on the lanthanides to obtain crystal field coefficients. We demonstrate our method on a test set of 10 materials comprising 9 RE-TM magnets and elemental Tb. We show that the calculated easy directions of magnetization agree with experimental observations, including a correct description of the anisotropy within the basal plane of Tb and NdCo$_5$. We further show that the Y-analogue calculations generally agree quantitatively with previous calculations using the open-core approximation to treat the 4f electrons, and argue that our simple approach may be useful for large-scale computational screening of new magnetic materials.