Can a deep dz2 band hybridize with Fermi-level bands to enhance iron-based superconductivity?

A band hundreds of meV below the Fermi level

In iron-based superconductors, the bands usually credited with electron pairing are the dxz/dyz and dxy orbitals that sit around the Fermi level, EF. The dz2 orbital band sits several hundred meV below EF and has generally been treated as irrelevant to superconductivity. This paper asks whether that assumption is too narrow.

The authors work on monolayer FeSe grown on SrTiO3. In this system the dz2 band is unusually sensitive to the in-plane lattice constant, which gives the experiment a handle: if the lattice can be stretched, the band can be moved. The paper's claim is that moving it far enough turns a nominally inert deep band into a participant in pairing.

Using tip force as a lattice strain knob

The experiment uses a scanning tunneling microscope (STM) at 0.4 K. Rather than only measuring, the tip acts as a mechanical tool: increasing the tunneling current at fixed bias reduces the tip-sample spacing, applying a repulsive force that expands the FeSe lattice. The dz2 band, being strain-sensitive, shifts upward toward EF while the dxz/dyz and dxy bands shift slightly away from it.

Tracking wide-bias dI/dV spectra across currents from 30 pA to 9 nA, the authors see the dz2 feature move rapidly toward EF and merge first with the dxz/dyz band at current I3 (2 nA) and then with the dxy band at I4 (4 nA). The merging is marked by increased bending of the dxz/dyz and dxy bands away from EF, which the authors read as hybridization. They note that the schematic band positions are inferred, because local spectroscopy has limited momentum-space resolution.

Two stages of gap growth, with kinks at each crossing

The superconducting gap itself grows in two stages. Before the dz2 and dxy bands cross, the gap rises from 17.8 meV to 23.6 meV, which the authors attribute to enhanced electronic correlations under tensile strain and to renormalization of the pairing-active d bands. Once the bands intersect, a stronger enhancement sets in.

The spectroscopic signature is a pair of new kinks on the coherence peaks, appearing precisely at I3 and I4. The authors interpret these kinks as the response of Bogoliubov quasiparticles near EF to hybridization with bands lying around -90 meV and -77 meV. They state that the origin of the kinks needs further study, but treat the change in gap-growth slope at both currents as direct evidence that band reconstruction drives the amplification.

Momentum-space confirmation from quasiparticle interference

Local spectroscopy alone can only show correlation, so the authors add quasiparticle interference (QPI) imaging. Fe vacancies in the film act as scattering sources; real-space dI/dV maps and their Fourier transforms reveal ring-like scattering wave vectors from intra-pocket scattering within the electron-like pocket at the M point.

Fitting the parabolic dispersion of these wave vectors across 11 currents yields the band bottom and effective mass. Both change sharply near I4: the band bottom shifts downward and the effective mass decreases. The authors stress that this mass reduction is a local effect, a pulling down of the dxy band near EF through hybridization with the upshifted dz2 band; whether the bandwidth changes across the whole Brillouin zone is not resolved by this QPI analysis. A nearly identical normal-state Fermi wave vector across all currents indicates doping is unchanged, supporting band reconstruction rather than carrier change as the cause of the gap enhancement.

A second QPI signature appears at I3: a hole-like dxy band visible near -70 meV at lower currents disappears once the dz2 band merges with the dxz/dyz band top, consistent with intensified hybridization.

Bogoliubov dispersion ties the gap to superconductivity

To show the enlarged gap is genuinely superconducting, the authors map Bogoliubov QPI, the interference pattern of superconducting quasiparticles. These dispersions are suppressed by temperature and magnetic field, the behavior expected of a superconducting state rather than a trivial spectral feature.

The quantitative gap function extracted from these collective excitations matches local STS and prior ARPES data in both amplitude and anisotropy. That agreement across a local probe and a momentum-resolved probe is the paper's central argument that tip-enhanced spectral weight reflects a real superconducting gap, not an artifact of the tip.

Where the causal claim stops

The evidence is strong but bounded. The hybridization picture is supported by QPI-derived band parameters and by strain-dependent density functional calculations, which the authors describe as qualitative only, since a full theoretical description of monolayer FeSe/STO remains difficult. The schematic band positions in the figures are not exact, given limited momentum resolution.

Spatially resolved data add a caution: at a fixed current, the absolute dz2 band energy does not correlate directly with local gap size, and the gap still varies by about 3 meV at high current, so other factors such as microscopic inhomogeneity also matter. What remains systematic is the joint evolution with current, reproduced at multiple locations.

Deep-band engineering as the design priority

The paper's practical message is a priority shift rather than a new material recipe. In this system, the decisive variable is not only which orbitals sit at EF but whether a deep band can be brought into contact with them. The threshold is the band crossing: gap growth accelerates once dz2 meets dxy at I4, and the authors identify that dz2-dxy channel as the pivotal one, since the I3 crossing with dxz/dyz produces a kink in local STS but no abrupt change in the BQPI gap trend.

For readers weighing this against conventional orbital-selective pictures, the relevant comparison is between the two crossings: the dz2-dxy hybridization is where both local and momentum-resolved probes agree on a stronger enhancement. The open question the authors leave is how far this deep-band strategy transfers beyond monolayer FeSe/STO to other multiorbital correlated materials.

Sources used in this answer