A signal that appears only in a narrow state
An arXiv version-1 preprint dated 26 August 2026 reports electrical patterns in twisted graphene that the authors interpret as signatures consistent with current-driven spin-domain-wall dynamics. The study examines an association between current-driven spin-transfer torque, domain-wall precession and a ferro-Josephson-like transport response. The practical question is whether motion of a magnetic boundary could leave a distinctive electrical signal.
The main experiment used a twisted monolayer-trilayer graphene stack, called t1+3, with a twist angle of 1.29 degrees. The measurements centered on a narrow valley-polarized region around the state labeled ν=3, with the response compared as current, field direction and preparation history changed.
Researchers used four-terminal lock-in transport. Standard longitudinal and Hall resistance measurements used a 1.4-nA AC current at frequencies below 42 hertz, while differential-resistance traces, written dV/dI, added a 10-to-80-pA AC modulation to a DC bias. This allowed the team to map both the usual resistance and the finer field-dependent features.
The peak was tied to bias
At zero in-plane field, the longitudinal resistance Rxx formed a narrow peak, but only in the valley-polarized sliver of the t1+3 state. It dropped sharply once the field rose above a fraction of a millitesla. The reported result does not specify one numerical value for the peak's width or for the field at which the drop occurs, so the sharpness is qualitative rather than a precisely quantified threshold.
The peak also depended strongly on the AC bias. In the reported series, its width and shape changed with current, and the feature disappeared for biases below 0.4 nA. That dependence is why the authors discuss a current-driven mechanism, although the transport result is an association rather than a direct measurement of the proposed magnetic motion.
Resonances shifted as the field changed
Differential resistance made the pattern more detailed. At zero in-plane field, several dV/dI peaks approached zero bias. Increasing the in-plane field shifted the first peak to higher bias, and the peak vanished above a field that the paper describes without a numerical value. The authors compare this combination of multiple peaks and a field-dependent threshold with a ferro-Josephson-like response.
The direction of the field mattered as well. The spectra were approximately, though not perfectly, isotropic when the field stayed in the plane, while out-of-plane fields that were orders of magnitude larger had little effect. The contrast narrows the possible magnetic-field response, but it does not by itself identify the underlying domain structure.
Where the resonance dispersed cleanly, its threshold shifted by a few nanoamps per millitesla. A phenomenological estimate in the paper translated that slope into a wall containing about 200 precessing moments. That figure is model-based, and the slope varied between data sets and field directions, so it should not be read as a direct count of spins in the device.
The device remembered how it was prepared
The signal also changed with the device's preparation history. Repeated gate-history measurements in the valley-polarized range from 2.8 to 3.2 were followed by a discrete domain flip and large magnetoresistance on both sides of ν=3. The observations associate the electrical response with a changing domain configuration, but the exact structure of those domains and the mechanism of the flip were not directly measured.
Out-of-plane-field training produced another history effect. After B⊥ was ramped past approximately -130 millitesla and then brought back toward zero, magnetoresistance was absent. It reappeared when the field crossed zero. This behavior is consistent with a field-trained domain configuration, although that configuration remains inferred rather than directly imaged.
An energy scale that complicates a simple picture
The reported magnetoresistance was robust at 700 millikelvin, even though the thermal energy kB T exceeded the tens-of-neV Zeeman scale by more than three orders of magnitude. The authors use this comparison as an energy-scale clue, not as a direct measurement of an activation barrier.
What the result does and does not say
The authors read the current dependence, moving resonances, field-direction response and history effects as consistent with their proposed ferro-Josephson picture. In that picture, spin-transfer torque would reorient moments in a domain wall and, above a field-dependent current threshold, drive them into precession. The resulting collective motion is proposed to produce a Josephson-like voltage response. The study tests an indirect transport signature of this process, not the precession itself.
But the result has a clear boundary. The evidence concerns transport signatures in the primary t1+3 device, while the wall size and precessing-moment count come from a phenomenological estimate. The mechanism therefore remains a proposed explanation for the observed association, rather than an established account of the material's behavior.
Still a preprint
The document is an arXiv version-1 preprint dated 26 August 2026. Source data are available, and other supporting data can be obtained from the corresponding author on request. The authors declare no competing interests.
The UBC experiments received support from Canadian agencies and programs, the Max Planck-UBC-UTokyo Centre, the Canada First Research Excellence Fund and the European Research Council under EU Horizon 2020. U.S. NSF and Department of Energy programs supported device preparation and twisted-graphene sample development.
Paper data and sources
Original title: Signatures of a ferro-Josephson effect in twisted graphene
Authors: Ruiheng Su, Zhenxiang Gao, Christopher Coleman et al.
Journal/Repository: arXiv
Status: Preprint, not yet peer-reviewed
First online: 2026-08-26
DOI: Not available
Original paper · Full text