Researchers at the National University of Singapore (NUS), led by Assistant Professor Ahmet Avsar of the university's Centre for Advanced 2D Materials, have combined a record-fidelity graphene spin-transport platform with magnetic-proximity band engineering to move graphene closer to practical spin-logic and spin-memory devices, across two complementary studies. The work targets one of graphene spintronics' core limitations: interfacial disorder at the electrical contacts that inject and detect spin, which has historically scrambled spin information before it can be read out electrically.
The first study rebuilt the graphene spin-device fabrication process around an inert-glovebox van der Waals assembly, laminating and cleaning the stack to produce atomically flat hexagonal boron nitride (h-BN) tunnel barriers rather than the oxide barriers more commonly used in graphene spin valves. That interface quality translated directly into device performance: nonlocal spin signals reached up to 1.6 kΩ at 2.5 K, spin polarization approached 90% (89% in the lead device), spin lifetime measured about 2.04 nanoseconds with a spin diffusion length of about 4.74 μm, and gate-tunable magnetoresistance exceeded 80%. Critically for eventual device use, the effect persisted at room temperature, where the same device retained a nonlocal spin resistance of about 160 Ω and roughly 42% spin polarization.
The second study used that clean-interface platform to probe how a nearby magnetic material reshapes graphene's spin-dependent electronic structure. Placing cobalt contacts near twisted bilayer graphene moiré devices (graphene/h-BN twist angles of about 0.6° and 0.8°, yielding moiré wavelengths of 13.9 nm and 10.7 nm) induced a magnetic proximity effect predicted to reach exchange splitting of nearly 10 meV even through several h-BN spacer layers. The resulting spin-resolved band structure let the team extract injector spin polarizations of about 47% and 25% at different charge-neutrality points, and, notably, drive a complete reversal of the spin signal, from maximum positive to maximum negative, using a gate-voltage swing of only about 300 mV.
That narrow switching window is the result with the clearest device implications: it demonstrates spin-dependent bands in graphene that can be flipped with a small electrostatic gate rather than an external magnetic field, the operating principle a gate-controlled spin transistor would need. Paired with the first study's improvements in spin-injection and spin-detection fidelity, the two results address both halves of the graphene spintronics problem, transporting spin information with less loss and controlling its polarization electrically, within the same materials platform.
The NUS team positions the combined results as a step toward graphene-based spin-logic and spin-memory architectures for ultra-low-power electronics, an application spintronics has targeted for graphene given the material's long spin lifetimes and weak intrinsic spin-orbit coupling.
