Machine-learning Assisted Prediction of Magnetic Double Perovskites

In the present study, we use a combination of computational tools; machine learning technique for screening of stable candidates, evolutionary algorithm for crystal structure determination, and first-principles calculations for characterization of
electronic and magnetic states, to make predictions on a set of new magnetic double-perovskites of composition
A₂BB'O₆. Out of 412 scanned candidates composition with 3d and 4d or 5d transition metals at B and B' sites, 33 compounds are found to form in stable double-perovskite structure, 25 of which are further considered for characterization of their crystal structure and properties. This exercise provides 21 new double-perovskites of varying magnetic and electronic properties, which range from ferromagnetic half-metals to ferri- and antiferro-magnetic insulators to ferromagnetic metals and rare example of antiferromagnetic metals. Our computational study is expected to help in discovering new magnetic double perovskites.