Using Quasiparticle Self-Consistent GW and Density Functional Theory with Machine-Learned Hubbard U Corrections in Search of Half-Metallic Heuslers

Half metallic Heusler compounds are of interest for spintronics. For device fabrication, compounds that can be grown on top of III-V semiconductors are particularly attractive. We conduct a first-principles investigation of seven Ni-based, Co-based, and Ru-based Heusler compounds that are lattice matched to InAs and GaSb. The results of density functional theory (DFT) using semi-local and hybrid functionals are compared to quasiparticle self-consistent GW (QPGW). We find that the degree of spin polarization at the Fermi level, and in some cases even the dominant spin channel (majority or minority), depend strongly on the choice of method. For the purpose of computational efficiency, we examine whether DFT with machine learned Hubbard U corrections [npj Computational Materials 6, 180 (2020)] can reproduce the QPGW results. We formulate a Bayesian optimization (BO) objective function to find the U values that reproduce as closely as possible the QPGW band structure and magnetic moments. We find that DFT+U(BO) can adequately reproduce QPGW for all but one of the compounds studied here. The Ru-based compounds are found to have a very high spin polarization, as well as a high density of states around the Fermi level. Hence, they would be intriguing candidates for experimental investigation.