Oxygen magnetic polarization, nodes in spin density, and zigzag spin order in oxides

Abstract

Recent density functional theory (DFT) calculations for Ba2CoO4 (BCO) and neutron scattering experiments for SrRuO3 (SRO) have shown that oxygen develops a magnetic polarization. Moreover, DFT calculations for these compounds also unveiled unexpected nodes in the spin density, both along Co-O and Ru-O. For BCO, the overall antiferromagnetic state in its triangular lattice contains unusual zigzag spin patterns. Here, using simple model calculations supplemented by DFT we explain and extend these results. We predict that ligands that in principle should be spinless, such as O 2− , will develop a net polarization when they act as electronic bridges between transition metal (TM) spins ferromagnetically ordered, regardless of the number of intermediate ligand atoms. The reason is the hybridization between atoms and mobility of the electrons with spins opposite to those of the closest TM atoms. Moreover, for bonds with TMs antiferromagnetically ordered, counterintuitively our calculations show that oxygens should also have a net magnetization for the super-super-exchange cases TM-O-O-TM while for only one oxygen, as in Cu-O-Cu, the Opolarization should cancel. Our results for the TM antiferromagnetic case are more general: for an even number of ligands between antiparallel TM spins, all ligands will develop nonzero polarization albeit of opposite signs, while for an odd number of ligands only the central one will have zero net polarization while the rest will have a nonzero polarization. Our simple model also allows us to explain the presence of nodes based on the antibonding character of the dominant singly occupied molecular orbitals along the TM-O bonds. Finally, the zigzag pattern order becomes the ground state mainly due to the influence of the Hubbard U , that creates the moments, in combination with a robust easy-axis anisotropy that suppresses the competing 120 • degree antiferromagnetic order of a triangular lattice. Our predictions are generic and should be applicable to any other compound with characteristics similar to those of BCO and SRO.