Structural and Correlation Effects in the Itinerant Insulating Antiferromagnetic Perovskite NaOsO3

TL;DR

This study investigates the metal-insulator transition and magnetic ordering in NaOsO3, revealing a Slater-type transition driven by structural effects and Coulomb interactions, with spin-orbit coupling playing a secondary role.

Contribution

It provides a detailed ab initio analysis showing that NaOsO3's MIT is a Slater transition influenced by structural distortions and Coulomb repulsion, clarifying the roles of SOC and hybridization.

Findings

MIT occurs at U_c ~2 eV in AFM state

SOC doubles the critical U_c needed for gap opening

Calculated magnetic moment ~1 μ_B consistent with experiments

Abstract

The orthorhombic perovskite NaOsO3 undergoes a continuous metal-insulator transition (MIT), accompanied by antiferromagnetic (AFM) order at T_N=410 K, suggested to be an example of the rare Slater (itinerant) MIT. We study this system using ab initio and related methods, focusing on the origin and nature of magnetic ordering and the MIT. The rotation and tilting of OsO6 octahedra in the GdFeO3 structure result in moderate narrowing the band width of the t_{2g} manifold, but sufficient to induce flattening of bands and AFM order within the local spin density approximation (LSDA), where it remains metallic but with a deep pseudogap. Including on-site Coulomb repulsion U, at U_c ~2 eV a MIT occurs only in the AFM state. Effects of spin-orbit coupling (SOC) on the band structure seem minor as expected for a half-filled $t_{2g}^{3}$ shell, but SOC doubles the critical value U_c necessary to open a gap and also leads to large magnetocrystalline energy differences in spite of normal orbital moments no greater than 0.1$\mu_B$. Our results are consistent with a Slater MIT driven by magnetic order, induced by a combination of structurally-induced band narrowing and moderate Coulomb repulsion, with SOC necessary for a full picture. Strong p-d hybridization reduces the moment, and when bootstrapped by the reduced Hund's rule coupling (proportional to the moment) gives a calculated moment of ~1 $\mu_B$, consistent with the observed moment and only a third of the formal $d^3$ value. We raise and discuss one important question: since this AFM ordering is at q=0 (in the 20 atom cell) where nesting is a moot issue, what is the microscopic driving force for ordering and the accompanying MIT?