Enhanced orbital anisotropy through the proximity to a $\mathrm{SrTi}{\mathrm{O}}_3$ layer in the perovskite iridate superlattices

We have used angle-dependent soft x-ray absorption spectroscopy (XAS) at the O $K$ edge and first-principles calculations to investigate the electronic structures of iridate-based superlattices ${\left(\mathrm{SrIr}{\mathrm{O}}_3\right)}_m/\left(\mathrm{SrTi}{\mathrm{O}}_3\right)$ ($m=1$, 2, 3, and ∞). We focus on the pre-edge Ir $5d t_{2g}–\mathrm{O} 2p$ orbital hybridization feature in the XAS spectra. By varying the measurement geometry relative to the incident photon polarization, we are able to extract the dichroic contrast and observe the systematic increase in the anisotropy of Ir $5d$ orbitals as $m$ decreases. First-principles calculations elucidate the orbital anisotropy coming mainly from the enhanced out-of-plane compression of $\mathrm{Ir}{\mathrm{O}}_6$ octahedra in the $\mathrm{SrIr}{\mathrm{O}}_3$ layers that are adjacent to the inserted $\mathrm{SrTi}{\mathrm{O}}_3$ layers. As $m$ decreases, the increased volume fraction of these interfacial $\mathrm{SrIr}{\mathrm{O}}_3$ layers and their contact with the $\mathrm{SrTi}{\mathrm{O}}_3$ layers within the ${\left(\mathrm{SrIr}{\mathrm{O}}_3\right)}_m/\left(\mathrm{SrTi}{\mathrm{O}}_3\right)$ supercell lead to enhanced orbital anisotropy. Furthermore, the tilt and rotation of $\mathrm{Ir}{\mathrm{O}}_6$ octahedra are shown to be essential to understand the subtle orbital anisotropy in these superlattices, and constraining these degrees of freedom will give an incorrect trend. Our results demonstrate that the structural constraint from the inserted $\mathrm{SrTi}{\mathrm{O}}_3$ layers, in addition to other electronic means such as polar interface and charge transfer, can serve as a knob to control the orbital degree of freedom in these iridate-based superlattices.

Figure 1

(a) Schematic plot showing the stacking of $\mathrm{SrIr}{\mathrm{O}}_3$ (yellow) and $\mathrm{SrTi}{\mathrm{O}}_3$ (blue) layers to form the ${\left(\mathrm{SrIr}{\mathrm{O}}_3\right)}_m/\left(\mathrm{SrTi}{\mathrm{O}}_3\right)$ superlattice. (b) The RHEED patterns of (001)-oriented STO substrate (top) and the $m=3$ SL (bottom). (c) Time dependence of the RHEED intensity curve of [01] diffraction spots [orange box in panel (b)] of the $m=3$ SL. (d) The XRD θ-2θ scans of SLs with $m=1$, 2, 3, and ∞. Arrows indicate the superlattice peaks. (e) Schematic plot showing the experimental geometry and sample angles (θ and χ) relative to the incident x-ray beam. The red arrow denotes the linear photon polarization. (f) TEY mode of O $K$-edge XAS spectra from the $m=\infty$ SL (yellow curve) and $\mathrm{SrTi}{\mathrm{O}}_3$ (blue curve) substrate. The inverted triangle denotes the Ir $t_{2g}\text{−}\mathrm{O} p$ hybridization feature.

Figure 2

O $K$-edge XAS spectra recorded in (a) TEY and (b) TFY modes. The inverted triangles mark the Ir-O hybridization feature. The red (30°) and black curves (90°) are from samples with photon polarizations projected largely out-of-plane and completely in-plane, respectively. (c) Orbital anisotropy calculated from the XAS recorded in the TEY (blue circles) and TFY (yellow squares) modes.

Figure 3

Calculated density of states (DOS) for SLs with $m=\infty$ [(a), (e), (i), (m)], 3 [(b), (f), (j), (n)], 2 [(c), (g), (k), (o)], and 1 [(d), (h), (l), (p)]). (a)–(d) The total DOS (gray line) and PDOS on the Ir (blue line) and O (red line) elements. (e)–(p) Projected Ir DOS onto its five $5d$ orbitals [$xy$ (red solid line), $yz$ (gray solid line), $xz$ (green solid line), $x^2-y^2$ (black dashed line), and $3z^2-r^2$ (orange dashed line)] for three cases: (e)–(h) for SLs close to the surface region where no constraint is applied to either the in-plane rotation/tilt degrees of freedom or lattice constants; (i)–(l) for SLs close to the STO substrate where the in-plane lattice constant of SLs is set to that of the pristine $\mathrm{SrTi}{\mathrm{O}}_3$, but the in-plane rotation/tilt degrees of freedom are permitted during the structural relaxation; (m)–(p) with constraint only on the in-plane rotation and tilt degrees of freedom of $\mathrm{Ir}{\mathrm{O}}_6$ and $\mathrm{Ti}{\mathrm{O}}_6$ octahedra. The Fermi energy is set to zero (black dashed line). The total DOS and PDOS are normalized by the number of oxygens in the supercell.

Figure 4

(a) Side view and top view of $m=1$ (left) and 3 (right) SLs with fully relaxed structure without strain. The shaded blue, orange, and yellow octahedra denote the $\mathrm{Ti}{\mathrm{O}}_6, \mathrm{Ir}{\mathrm{O}}_6$ (interfacial), and $\mathrm{Ir}{\mathrm{O}}_6$ (sandwiched), respectively. The in-plane rotation angles are listed in the figure. (b) Calculated total energies (red triangles and black squares for results with and without $\mathrm{Ir}{\mathrm{O}}_6$ octahedra tilt and in-plane rotation, respectively, left axis) and their difference (green open circles, right axis) as a function of $m$ after the structural relaxation. The total energy is normalized by the number of oxygens in the supercell. (c) Calculated orbital anisotropy as a function of $m$ for the following cases: with tilt and rotation, no strain (red triangles); without tilt and rotation, no strain (black squares); with tilt, rotation, and strain (blue diamonds). (d) Ir $5d$ PDOS for the interfacial (top) and sandwiched (bottom) SIO layer in $m=3$ SL shown in panel (a). The Fermi energy is set to zero (black dashed line). The PDOS is averaged over all Ir sites lied in the same layer in the supercell.