A state that depends on the route
An anomalous resistance peak that appears at zero in-plane magnetic field changed with the route used to prepare a twisted double bilayer graphene device, according to a preprint. The pattern points, in the authors' interpretation, to metastable magnetic domains, or magnetic configurations that can persist after preparation.
The work examined one micron-scale Hall-bar device made from two Bernal-stacked bilayers twisted by about 1.34 degrees. Researchers tracked longitudinal resistance while taking controlled gate-voltage routes into a quarter-metal state, described in the study as having full spin and valley polarization, and while applying magnetic fields in and out of the graphene plane. The typical alternating-current bias was 1.4 nanoamperes. Most data were collected at 100 millikelvin, while longitudinal-resistance maps were taken at 25 millikelvin and Hall data at 300 millikelvin.
Different paths, different outcomes
Preparation history made a sharp difference. Path alpha produced no zero-in-plane-field resistance peak over 20 repetitions, whereas path beta frequently produced one, although the exact frequency was not reported. The researchers averaged the resistance for 10 seconds after arrival via path beta and compared it with a peak-height measure: the resistance at zero in-plane field minus the resistance at 10 millitesla. They presented the comparison as a correlation, so it shows an association between the prepared state and the later peak rather than establishing a cause.
Field conditions during the gate sweep were also associated with different outcomes. After the measured filling value passed 3.4, traces split into high- and low-resistance trajectories. With the x-directed in-plane field at zero, the high-resistance outcome occurred about half the time. At 0.6 millitesla, the sample consistently entered the low-resistance state. The study reports this branching descriptively, without a formal probability model.
The approach-field response was nonmonotonic, meaning it did not move in one direction as the field grew. The large-peak state disappeared for in-plane approach fields of roughly 0.4 millitesla or more, then reappeared when the in-plane field exceeded about 175 millitesla. It persisted to out-of-plane fields six times larger than the smaller in-plane threshold. The unusual return is central to the authors' interpretation of a field-sensitive, history-dependent state.
What the pattern may mean
Once measured, the peak showed a different kind of selectivity. It was isotropic within the graphene plane to experimental resolution, while the millitesla-wide feature was unaffected by an out-of-plane field up to 10 millitesla. In almost all cases, it appeared alongside the quarter-metal state's full spin and valley polarization. These are associations, not a direct measurement of a domain pattern.
The authors interpret the history dependence as evidence for metastable magnetic domains, specifically domains of valley polarization in which spin is linked to valley polarization through spin-orbit coupling. They propose that existing domain walls, the boundaries between such regions, could affect resistance and that current-driven domain-wall motion may contribute to the peak. They also state that the microscopic origin has not been identified unambiguously.
Supplemental measurements supplied supporting clues. A sharp differential-resistance peak appeared at about 1 nanoampere of DC bias near zero in-plane field and reached about 1.5 nanoamperes at plus or minus 0.6 millitesla, with the threshold shifting higher as the in-plane field magnitude increased. In a temperature scan at zero in-plane field, a peak near 0.7 nanoampere was sharpest and tallest at 20 millikelvin and became very weak by 200 millikelvin. The domain state was unstable during that supplemental cooldown, so the current dependence could not be studied systematically.
A narrow result with open questions
The evidence is narrow. The completed history-dependent measurements came from one primary device, and the domain picture was inferred from transport history and correlations rather than directly imaged. Some measurements also involved residual or uncalibrated field offsets: the residual in-plane approach field was estimated at 0.45 plus or minus 0.05 millitesla, while the magnet setting corresponding to true zero out-of-plane field was minus 1.8 millitesla. No inferential statistical test was reported, and the field thresholds were approximate.
For now, the preprint constrains possible explanations for the zero-in-plane-field peak in this device, but it does not establish that magnetic domains cause the peak, that current-driven domain-wall motion is the mechanism, or that the finding extends beyond the device and tested protocols. The open questions are how domain walls generate the resistance peak, how spin-orbit coupling enters the transport, and whether the proposed current dependence can be tested in a more stable sample. The manuscript is arXiv version 2, dated 2 September 2026.
Paper data and sources
Original title: Metastable magnetic domains and the anomalous $B_\parallel=0$ resistance peak in twisted double bilayer graphene
Authors: Zhenxiang Gao, Christopher Coleman, Silvia Folk et al.
Journal/Repository: arXiv
Status: Preprint, not yet peer-reviewed
First online: 2026-08-26
DOI: Not available
Original paper · Full text