A near-monochrome collage of seven laboratory photographs cut into diagonal panels: a cryostat and vacuum chamber trailing cabling and steel flanges; an optical table with posts, a glowing lamp and a circuit board; a bright ring of light seen through a dark chamber viewport; a chamber interior with a finned copper disc, translation stages and stepper motors; a rack of long cryostat probe inserts with stacked baffles; a vacuum chamber wrapped in aluminium foil with pneumatic valves; and a chamber with a brilliant glowing plume beneath a circular electrode.

Bias-tunable magnetoresistance in a CrSBr junction, and the mechanism behind it

The short version

Devices that read out magnetic state electrically normally need two magnetic electrodes with an insulator between them. This one has no magnetic electrodes at all: the magnetism sits in the insulating layer itself, and the resistance still changes by a factor of four and a half depending on how that layer's internal magnetic order is arranged. The size of the effect is set by the operating voltage, and we identify what causes it.

Status: manuscript prepared for submission to Nano Letters. Not yet peer-reviewed and not yet published, so there is no DOI.

A magnetic tunnel junction normally reads magnetic state through two ferromagnetic electrodes. This device inverts that arrangement: both electrodes are non-magnetic few-layer graphene, and the magnetism sits in the 12 nm CrSBr barrier between them, whose own antiferromagnetic order sets the barrier height.

The junction reaches a magnetoresistance of 350% at 20 K, with a distinctive M-shaped dependence on bias voltage peaking near ±0.5 V. That shape has been reported across the 2D magnetic junction literature and explained in several incompatible ways. By rotating the layer magnetisations continuously with a hard-axis field rather than only switching between the two collinear states, we find the band-edge offset moves linearly with cos(θ/2), the first-order signature of spin-dependent interlayer hybridisation. That linearity excludes both the Jullière model and a spin-filter projection, which predict a quadratic dependence.

The practical consequence: bias voltage, not electrode spin polarisation, is the design knob, and the resistance change can be programmed by where the operating point sits relative to the two band edges. The absolute energy scale rests on an empirical bias-to-energy calibration, so we compare it on scale rather than treating it as exact.

Two panels. Left: an optical micrograph of the device with the h-BN, top and bottom few-layer graphene and CrSBr flakes outlined in yellow, black and blue with a 15 micrometre scale bar, above a three-dimensional cutaway schematic of the same stack showing h-BN capping over the graphene electrodes, the CrSBr layer and gold contacts on a substrate. Right: magnetoresistance ratio in percent against bias voltage from minus 1 to plus 1 volt, plotted for six temperatures from 20 to 70 K. Every curve is M-shaped with a deep minimum at zero bias and two maxima near plus and minus 0.5 volts, the 20 K curve peaking near 350 percent and each successive temperature lying lower.

The graphene/CrSBr/graphene junction and its M-shaped bias dependence. The two maxima near ±0.5 V are the field-emission onsets of the two magnetic configurations; they merge and vanish above a crossover near 90 K.

Shared first authorship with Sadeed Hameed. I built the high-impedance cryogenic measurement and transport-spectroscopy pipeline this study runs on, described under CryoSoft, and the band-structure interpretation was developed with a first-principles collaboration.