Metallic Oxides and the Overlooked Role of Bandwidth

TL;DR

This work asks when oxides become metallic by focusing on the bandwidth $W$ of conduction states near the Fermi level. Using first-principles DFT calculations, it shows that a dispersive conduction band with $W$ typically around $2\,\text{eV}$ is common in metallic oxides, and that $W<1\,\text{eV}$ almost precludes metallicity. The analysis, illustrated across diverse oxide families (from ReO$_3$ to copper-oxide superconductors), attributes metallicity to extended covalency and multidimensional connectivity that yield wide bands, while highlighting that band gaps alone do not determine metallic behavior. The findings provide a transferable framework for screening materials and caution against over-interpreting narrow flat bands as signatures of emergent superconductivity, with implications across oxides and related frameworks.

Abstract

Oxides exhibiting metallic conduction are crucial for various applications, including fuel cells, battery electrodes, resistive and magnetoresistive materials, electrocatalysts, transparent conductors, and high-temperature superconductors. Oxides that approach metallicity also play significant roles in switching applications, where the metal-insulator transition phenomenon is utilized across a range of technologies. This perspective, motivated by the question of when oxides are metallic, employs electronic structure calculations on metallic oxides to identify the typical feature in the electronic structures that promote metallic behavior. The critical factor of the bandwidth of the electronic energy bands near the Fermi energy is emphasized since it has been somewhat overlooked in the literature. For example, bandwidth considerations would suggest that the recently proposed phosphate "LK-99" would never be a suitable target for superconductivity. From the analysis performed here, we learn that if the width of the conduction band as obtained from density functional theory-based electronic structure calculations is less than 1 eV, then the likelihood of obtaining a metallic compound is vanishingly small. This survey of representative oxide metals highlights the essential elements of extended covalency that lead to wide bands. A key takeaway is that oxyanion compounds such as borates, carbonates, silicates, sulfates, nitrates, and phosphates are unlikely to exhibit metallic conduction at ambient pressure. While the focus here is on oxides, the general findings should apply across various material families, extending to organic crystals, polymers, and framework materials.

Figures (11)

• Figure 1: Crystal structures of the different compounds described in this perspective. (a) Cubic ReO3, (b) the defect rock-salt structure of TiO, (c) the layered delafossite $3R$--PdCoO2, (d) the rutile structure of CrO2, (e) low-- and high--$T$ variants of (part) of the structure of rutile VO2 displaying metal--metal bonding in the low--$T$ state, (f) the layer-ordered rock-salt structure of $3R$--LiCoO2, (g) the structure of the olivine phosphate LiFePO4, (h) the tetragonal ground state structure of perovskite SrTiO3, (i) the rhombohedral structure of perovskite LaNiO3, and (j) the triple-CuO2-layered, tetragonal structure of HgBa2Cu2Cu3O8 superconductor parent.
• Figure 2: Local bonding environments within $M$O$_6$ octahedra. (a) Partial molecular orbital diagram highlighting antibonding frontier orbitals given occupation of $M$ d orbitals (adapted from Housecroft2012). (b) Orbital configurations for $t_{2g}$ and $e_g$ filling based on corner-sharing or edge-sharing $M$O$_6$ octahedra.
• Figure 3: Electronic structure of ReO3 displaying Fermi-level states with bandwidths $W$ = 5 eV. The DOS and $-$COHP indicate that these states are covalent Re 5d and O 2p states with antibonding character as expected for materials with $M$O6 octahedra and partial filling of the d orbitals.
• Figure 4: Electronic structures of (a) TiO and (b) PdCoO2 showing disperse bands and states at the Fermi energy associated with direct metal--metal interactions.
• Figure 5: Electronic structures of selected rutile materials highlighting a variety of conduction behavior. (a) TiO2 is a band insulator, (b) CrO2 is a ferromagnetic half-metal, (c) RuO2 is a metal, and (d) SnO2 is a semiconductor. With a completely empty or full d shell, TiO2 and SnO2 show O 2p valence states and no $M$--O interactions at the Fermi level. CrO2 and RuO2 possess partially filled d-orbitals and, therefore, have mixed $M$ d and O p states at the Fermi level with antibonding interactions indicated in the $-$COHP.