Flat-band states—electronic features tracked at particular energies—showed a spatial energy pattern tied to the material’s moiré superlattice, but the pattern appeared only once the moiré period rose above roughly 10 nanometres in the tested samples. At smaller periods, the partially filled flat-band feature had a nearly uniform splitting of about 15 meV; at larger periods, that splitting oscillated with the moiré pattern.
The underlying question was what role the moiré potential—the local electronic landscape associated with the repeating superlattice—plays in forming the fractional quantum anomalous Hall effect, or FQAHE. To investigate it, the researchers used gate-tunable scanning tunnelling microscopy and spectroscopy, or STM/STS, to image spatial flat-band modulation on the moiré-distant side, the surface away from the aligned interface, of four- and five-layer rhombohedral graphene/hBN devices. The devices were fabricated on SiO2/Si with a silicon back gate and measured at about 5 K.
The signal followed the stacking sites
At zero displacement field, the spectra contained two flat-band peaks separated by roughly 18 meV and carrying comparable spectral weight. When the flat band was fully filled in a sample with a twist angle of 0.23 degrees, its position shifted downward at the CBN and CN stacking sites and oscillated with the moiré period.
Across the shown samples, the largest CBN-to-CB flat-band shift was about 15 meV. Remote bands—features outside the flat-band peaks—shifted in the same direction by a similar amount, which the authors interpret as local chemical-potential variation.
The pattern changed with period
Researchers compared a series of samples with moiré periods from 7.7 to 14.4 nm. The flat-band offset emerged only when the period exceeded about 10 nm, corresponding to twist angles—the slight rotations between layers—of roughly 1 degree or less. That makes the threshold a boundary observed in this tested series, not a universal cutoff established for graphene devices.
That size dependence also appeared in correlated states. With partial filling, the sample below 10 nm showed a nearly uniform splitting of about 15 meV, while the sample above 10 nm showed a splitting that rose and fell periodically across the moiré lattice.
Energy-resolved maps supplied another view. On the lower-energy side of the flat-band peak, states were localized at CBN and CN; on the higher-energy side, their localization shifted toward CB. The maps show that changing the energy moved the dominant signal between stacking sites rather than leaving it fixed at one place.
What the model suggests
To explore what might produce this hierarchy, the study used twist-angle-dependent calculations. The model associated atomic corrugation—a structural variation across the stack—with site-dependent on-site energy, meaning the energy assigned to a location in the model. It estimated a variation of several meV below 1 degree and found no comparable effect above 1 degree.
In the same small-angle calculation, the model produced an isolated Chern band with |C| = 1, while the band was absent in the large-angle case. A Chern band is a calculated band carrying a topological index; here, that result is a theoretical prediction rather than a direct measurement of topology.
The topological link remains untested
The authors connect the spectroscopic pattern to the fractional quantum anomalous Hall effect, or FQAHE, because the approximately 10 nm threshold for moiré modulation matches the threshold associated with reported FQAHE transport. They present the shared threshold as a possible microscopic link to topological order.
That connection was not tested directly in this work. The study did not measure FQAHE transport, so the STM/STS observations cannot show that a period near 10 nm is necessary or sufficient for the effect, or that the measured shifts alone generate topological order. The proposed mechanism also depends partly on modeling, including the corrugation calculation and the predicted Chern band.
The direct result is narrower but concrete: in the studied R4G/R5G graphene/hBN systems, the surface spectra changed across stacking sites, and the change depended on moiré period and energy. Whether that real-space reshaping tracks a transport phase boundary remains an open question that would require measuring both in the same device.
Paper data and sources
Original title: Visualizing flat-band spatial renormalization in rhombohedral graphene superlattices
Authors: Peng-Cheng Pan, Shihao Zhang, Yang Zhang et al.
Journal/Repository: arXiv
Status: Preprint, not yet peer-reviewed
First online: 2026-08-25
DOI: Not available
Original paper · Full text