On the electronic ground state of two non-magnetic pentavalent honeycomb iridates

TL;DR

This work probes the electronic ground states of two Ir$^{5+}$ honeycomb iridates, Sr$_3$CaIr$_2$O$_9$ and NaIrO$_3$, using XAS and high-resolution RIXS to determine whether they realize a $J_{\rm eff}=0$ Mott insulator or a band-insulating state. Sr$_3$CaIr$_2$O$_9$ is found to be well described by a nearly ideal $J_{\rm eff}=0$ state despite strong ligand distortions, with disorder broadening the RIXS features and indicating a large gap. In contrast, NaIrO$_3$ exhibits a broad, featureless RIXS response inconsistent with a conventional $J_{\rm eff}=0$ excitation scheme and aligns with a flat-band, narrow-gap $S=0$ band insulator as supported by DFT+SOI calculations. Together, the results highlight the crucial role of ligand environment and local structure in shaping the Ir 5d electronic landscape on the honeycomb lattice and demonstrate RIXS as a powerful diagnostic to distinguish Mott vs band insulating ground states in non-magnetic insulators.

Abstract

We investigate the electronic structure of two Ir$^{5+}$ honeycomb iridates, Sr$_3$CaIr$_2$O$_9$ and NaIrO$_3$, by means of resonant x-ray techniques. We confirm that Sr$_3$CaIr$_2$O$_9$ realizes a large spin-orbit driven non-magnetic $J = 0$ singlet ground state despite sizable tetragonal distortions of Ir coordinating octahedra. On the other hand, the resonant inelastic x-ray spectra of NaIrO$_3$ are drastically different from expectations for a Mott insulator with octahedrally coordinated Ir$^{5+}$. We find that the data for NaIrO$_3$ can be best interpreted as originating from a narrow gap non-magnetic $S = 0$ band insulating ground state. Our results highlight the complex role of the ligand environment in the electronic structure of honeycomb iridates and the essential role of x-ray spectroscopy to characterize electronic ground states of insulating materials.

Figures (9)

• Figure 1: Schematic of two limiting cases for an octahedrally coordinated Ir$^{5+}$ ion: (a) $J = 0$ state and (b) $S = 0$, with $\lambda >> \Delta$ and $\Delta \approx \lambda \approx J_H$, respectively.
• Figure 2: (a) Ir L$_3$ edge XAS intensity in Sr$_3$CaIr$_2$O$_9$ and ,(b), in NaIrO$_3$. (c) Ir L$_2$ edge XAS in Sr$_3$CaIr$_2$O$_9$ and, (d), in NaIrO$_3$. All data was taken at $T \!=\! 300$ K. Red line is a fit to a Lorentzian peak and an arctangent step (dotted black line), as explained in the main text. Direct comparison of the (e) $L_3$ white line and (f) $L_3 + L_2$ integrated intensity of Sr$_3$CaIr$_2$O$_9$ (SrCa) and NaIrO$_3$ (Na) to that of IrO, $\alpha$-Li$_2$IrO$_3$ ($\alpha$) and Ag$_3$LiIr$_2$O$_6$ (Ag). Dotted blue line is a linear fit.
• Figure 3: (a) Ir L$_3$ RIXS intensity at $E_i \! = \! 11.217$ keV, in Sr$_3$CaIr$_2$O$_9$ at $T \!= \!20$ K. Vertical dotted line indicates $E_{\mathrm{loss}}$$=\! 0$. Red line is a fit to a Voigt elastic line (dotted line) and four Gaussians peaks (grey shaded). (b) Calculated powder average RIXS intensity for two Ir$^{5+}$ sites with nonequivalent tetragonally distorted octahedral environments as described in the main text (black dashed line). The blue and red dotted lines show the corresponding RIXS intensities when including random tetragonal fields to account for static disorder. The total spectrum is shown in orange.
• Figure 4: Ir L$_3$ RIXS intensity, $E_i \! = \! 11.217$ keV , in NaIrO$_3$ at $T \!= \!20$ K. Vertical dotted line indicates $E_{\mathrm{loss}}$$\!=\! 0$. Red line is a fit to a Voigt elastic line (dotted line) and a set of Gaussian peaks (grey shaded).
• Figure 5: Excitation spectrum for a single Ir$^{5+}$ ions with $J_H \! =\! 0.285$ eV as a function of (a) atomic spin orbit coupling $\lambda$, (b) trigonal field, $\Delta$, (c), tetragonal field ,($\delta_1$, $\delta_2$), and (d) non-cubic fields ($\Delta_{NCF}$). We also show the excitation spectrum as a function of exchange field for fixed values of (e) trigonal, (f) tetragonal, and (g) non-cubic crystal fields.