Abstract
We studied $Sr_2IrO_4$ and $Sr_3Ir_2O_7$ using angle-resolved photoemission spectroscopy (ARPES), making direct experimental determinations of intra- and intercell coupling parameters as well as Mott correlations and gap sizes. The results are generally consistent with LDA+$U$+spin-orbit coupling calculations, though the calculations missed the momentum positions of the dominant electronic states and neglected the importance of intercell coupling on the size of the Mott gap. The calculations also ignore the correlation-induced spectral peak widths, which are critical for making a connection to activation energies determined from transport experiments. The data indicate a dimensionality-controlled Mott transition in these 5$d$ transition-metal oxides.
Introduction
In the study of condensed matter physics, the "Mott transition" remains a central puzzle: why do certain materials, which should be metallic based on their electron count, behave as insulators? This phenomenon is typically driven by strong electron-electron repulsion. While extensively documented in 3$d$ transition-metal oxides like copper-based superconductors, the 5$d$ perovskite iridates offer a different regime. In these materials, the electrons are more spatially extended, reducing direct repulsion, but this is offset by massive Spin-Orbit Coupling (SOC) due to the high atomic number of iridium.
This research explores how increasing the dimensionality of these systems—moving from the single-layer $Sr_2IrO_4$ to the bi-layer $Sr_3Ir_2O_7$—serves as a "dial" to control the Mott transition. By increasing the coupling between layers, the material is pushed from a robust insulating state toward a metallic one.
Experimental Methodology
The researchers employed high-resolution Angle-Resolved Photoemission Spectroscopy (ARPES) to map the electronic band structures of high-quality single crystals. These experiments were conducted at the Synchrotron Radiation Center and the Advanced Light Source.
A critical component of the methodology was the tuning of incident photon energy (ranging from 80 eV to 140 eV). This allowed the team to probe the $k_z$ dispersion, effectively measuring how electrons move between the crystal layers rather than just within a single 2D plane. These experimental "maps" were then compared against standard theoretical models, specifically Local Density Approximation (LDA) plus Hubbard $U$ and SOC calculations.
Key Results and Insights
The Impact of Dimensionality
The transition from a single-layer to a bi-layer structure introduces significant inter-cell coupling, measured at approximately 100 meV in $Sr_3Ir_2O_7$. This coupling acts as an internal pressure that narrows the Mott gap. The experimental data revealed a dramatic reduction in the "pole energy" (the peak position of the electronic state):
- Single-layer ($Sr_2IrO_4$): ~282 meV
- Bi-layer ($Sr_3Ir_2O_7$): ~97 meV
Furthermore, the bi-layer material exhibited a "splitting" of electronic bands of about 0.25 eV, a direct consequence of the interaction between adjacent iridium-oxide layers.
Theoretical Discrepancies: "X" Marks the Spot
While the overall results support the $J_{eff}=1/2$ Mott insulator model, the study identified a significant breakdown in current theoretical predictions. Standard LDA+U+SOC models predicted that the lowest energy states would reside at the center of the Brillouin zone (the $\Gamma$ point). However, ARPES data clearly located these states at the corner (the X point). This suggests that the actual Spin-Orbit Coupling in these iridates is stronger than theorists previously assumed, and that current models fail to capture the momentum-dependent self-energy of the electrons.
The Transport Paradox and Spectral Widths
One of the paper's most significant contributions is addressing the discrepancy between spectroscopic "gaps" and transport "activation energies." Historically, the energy required for a material to conduct electricity (activation energy) has been much lower than the gap measured by spectroscopy.
The authors propose that electrical conductivity is not determined by the center of the spectral peak, but rather by the "onset" or leading edge of the electronic spectral weight. The width of these peaks is caused by complex electron correlations, such as polaronic effects, which are ignored in standard calculations. By accounting for these spectral widths, the researchers provide a more accurate bridge between microscopic spectroscopy and macroscopic electrical behavior.
Conclusion and Future Outlook
The study demonstrates that the Mott transition in perovskite iridates is dimensionality-controlled. The addition of a second layer provides sufficient inter-cell coupling to significantly reduce the Mott gap, placing the bi-layer system on the precipice of a metal-insulator transition.
Future work is required to resolve the "periodicity mystery" observed in $Sr_3Ir_2O_7$, where the experimental $k_z$ periodicity was twice the expected value. Additionally, the researchers suggest that moving to tri-layer or full 3D structures would likely cause the spectral peaks to overlap with the Fermi level, potentially creating a unique metallic state similar to the "pseudogap" observed in high-temperature superconductors. This pathway offers a promising route for engineering new electronic phases in 5$d$ transition-metal oxides.
Summary updated