An optical feature known as an excitonic resonance shifted and its fitted binding energy rose before the reported metallic transition in palladium-substituted thin films, according to a manuscript. Across x = 0 to 0.4 in PdxCu1-xCrO2, the E1 resonance moved from 3.4 eV to approximately 3.2 eV while the fitted optical gap stayed nearly unchanged. Over the same range, the fitted binding energy rose from 489 meV to about 730 meV, a change of about 241 meV. The authors leave metallic proximity as one possibility, rather than a settled interpretation of the pattern.
The optical peak sharpened
The optical feature also changed in width and intensity. With increasing palladium concentration, the fitted full width at half maximum (FWHM), a measure of peak width, decreased while volume-normalized intensity (VNI) increased. In the reported analysis, the excitonic feature was therefore narrower and more intense.
The metallic-network transition was placed near x ≈ 0.5. At that point, the real part of the dielectric function became negative at low frequencies and a Drude response developed in ε₂. Conductivity was observed for x ≥ 0.5, and the onset coincided with the theoretical site-percolation threshold for a two-dimensional triangular lattice.
How the result was measured
The work used epitaxial PdxCu1-xCrO2 thin films synthesized on (0001)-oriented sapphire substrates by pulsed laser deposition. The dielectric response was fitted with a Tanguy model for E1 and Lorentz oscillators for E2, higher-energy features and Urbach tails.
For the CuCrO2 parent material, the study used first-principles density-functional theory (DFT) and many-body GW–BSE calculations. The GW–BSE result reproduced the excitonic peak with a redshift of about 500 meV relative to the non-interacting spectrum. The parent E1 resonance had a Tanguy-derived fitted binding energy of approximately 489 meV.
A possible mechanism, with limits
The manuscript also compared the pre-percolation result with a variational electrostatic image-charge model. In that calculation, conventional screening was allowed when a metallic region was much larger than the exciton. When the two sizes were comparable, the calculated binding energy could increase by up to 7%.
An effective-mass and dielectric estimate put the exciton size at about 2 Å and suggested possible sensitivity to local electrostatic interactions from nanoscale metallic regions.
The model's enhancement was smaller than the measured change of about 241 meV. The authors say the calculation establishes only a physical possibility and does not establish microscopic coupling between the metallic and excitonic states. They also leave open the possibility that palladium-derived metallic states and copper–chromium excitonic states could evolve largely independently.
After the network formed
For x > 0.5, isolated metallic regions were described as merging into a continuous network while the exciton resonance was quenched. The authors regarded this pattern as consistent with conventional screening, but said the observation did not single out screening as the unique interpretation.
Taken together, the reported sequence is a change in the fitted optical response before the network threshold, followed by quenching after the metallic regions were described as continuous. The analysis does not identify a microscopic link between the metallic and excitonic states and leaves open the possibility that the two sets of states evolve largely independently.
Paper data and sources
Original title: Anomalous stabilization of excitons by metallic proximity
Authors: Jeongkeun Song, Uksam Choi, Shan Lin et al.
Journal/Repository: arXiv
Status: Preprint, not yet peer-reviewed
First online: 2026-08-24
DOI: Not available
Original paper · Full text