Improving the prediction of magnetic interactions from DFT methods: Non-collinear magnetization and self-interaction*

Electronic structure methods are routinely used to predict magnetic properties of diverse materials, including molecular nanomagnets, clusters, and crystals. Density functional theory (DFT) has been one of the workhorses of modern electronic structure methods, and as such, it has been used extensively for the prediction of magnetic properties. In this talk I will summarize our efforts to improve the prediction of magnetic exchange couplings J from DFT methods in two fronts. First, I will show a proposed method to extract J based on differential rotations of the local atomic magnetization. This method avoids the explicit evaluation of energy differences, which can become impractical for large complexes and clusters. Our approach is based on the evaluation of the transversal magnetic torque between magnetic centers using constraints of the local magnetization direction via Lagrange multipliers and involves non-collinear spin DFT.[1] This, combined with a local partitioning of 〈S²〉, [2] makes possible the evaluation of J couplings entirely from first principles without ad-hoc nominal spin values. Second, I will show how self-interaction error (SIE) impacts the prediction of J couplings using an efficient implementation for SIE removal based on Fermi-Löwdin orbitals (FLOSIC).[3] Using this method, removing SIE from the simple local spin density approximation improves calculated J couplings,[4] in line with previous observations for small model systems.

[1] J. J. Phillips and J. E. Peralta, J. Chem. Phys. 138, 174115 (2013).
[2] B. Abate, R. p. Joshi, and J. E. Peralta, J. Chem. Theory Comput. 13, 6101 (2017).
[3] M. R. Pederson, A. Ruzsinszky, and J. P. Perdew, J. Chem. Phys. 140, 121103 (2014).
[4] R. P. Joshi, K. Trepte, K. P. K. Withanage, K. Sharkas, Y. Yamamoto, L. Basurto, R. R. Zope, T. Baruah, K. A. Jackson, and J. E. Peralta, J. Chem. Phys. 149, 164101 (2018).