Nontrivial Topology and Anomalous Hall Effect in Half-Metallic Double Perovskites: A Combined Machine Learning and Ab Initio Study

With a goal to propose candidate metallic oxide materials hosting nontrivial topology and intrinsic Anomalous Hall effect, we explore double perovskite compounds of the general formula A2BB′O6. Based on an exhaustive data set of 468 double perovskites, several potential compounds are identified from a set of known double perovskites, both characterized as well as yet-to-be-characterized, and a computationally predicted set of unknown, yet-to-be-synthesized compounds. A two-step machine learning is applied to screen magnetic double perovskite compounds having a finite uncompensated moment. Accurate first-principles and model Hamiltonian approaches, elucidating the electronic and magnetic properties of the screened compounds, unravel half-metallic ferromagnetism of these compounds. The magnetism is governed by a two-sublattice double-exchange mechanism, with core spin of the B site inducing negative spin polarization at the B′ site determined by energy level positioning of B, B′ states, and B–B′ hybridization. The presence of strong spin–orbit-coupled 4d or 5d transition-metal states at the Fermi energy, on the other hand, gives rise to nontrivial topology, which, together with time-reversal-breaking from a finite magnetic moment, results in a finite intrinsic Anomalous Hall conductivity. The strongly spin-polarized nature of the compounds, combined with the reasonably high magnetic transition temperature and high anomalous Hall conductivity, provides a technologically attractive solution to topological spintronics.