Metal-insulator transitions in GdTiO3/SrTiO3 superlattices

TL;DR

This study uses density functional theory with U correction to explore the electronic phases of GdTiO3/SrTiO3 superlattices, revealing how layer thickness and interactions induce metal-insulator transitions with distinct charge and orbital orderings.

Contribution

It provides a detailed phase diagram and identifies two different insulating states driven by layer thickness and orbital ordering, highlighting the role of many-body effects and potential ferroelectricity.

Findings

Metallic phases occur with large SrTiO3 layers or small U.

Two insulating states are identified with different charge and orbital orderings.

Critical U for single-layer insulator is about 2.5 eV.

Abstract

The density functional plus U method is used to obtain the electronic structure, lattice relaxation and metal-insulator phase diagram of superlattices consisting of $m$ layers of Gadolinium Titanate (GdTiO$_{3}$) alternating with $n$ layers of Strontium Titanate (SrTiO$_{3}$). Metallic phases occur when the number of SrTiO$_3$ layers is large or the interaction $U$ is small. In metallic phases, the mobile electrons are found in the SrTiO$_3$ layers, with near-interface electrons occupying $xy$-derived bands, while away from the interface the majority of electrons reside in $xz/yz$ bands. As the thickness of the SrTiO$_3$ layers decreases or the on-site interaction U increases a metal-insulator transition occurs. Two different insulating states are found. When the number of SrTiO$_{3}$ layers is larger than one, we find an insulating state with two sublattice charge and orbital ordering and associated Ti-O bond length disproportionations. When the number of SrTiO$_{3}$ units per layer is one, a different insulating phase occurs driven by orbital ordering within the quasi one-dimensional $xz/yz$ bonding bands connecting Ti atoms across the SrO layer. In this phase there is no sublattice charge ordering or bond disproportionation. The critical U for the single-layer insulator is $\sim$ 2.5 eV, much less than critical U $\sim$ 3.5 eV required to drive the metal-insulator transition when the number of SrTiO$_3$ is larger than one. Inconsistencies between the calculation and the experiment suggest that many-body correlations may be important. A local inversion symmetry breaking around Ti atoms suggests the possibility of in-plane ferroelectric polarization in the insulating phase.