image: (Left) Schematic of an orbital torque-driven magnetic tunnel junction. Orbital angular momentum is generated and subsequently converted into spin angular momentum to switch the magnetic memory. (Right) The device is confirmed to be switchable by 28.7‑ps electrical pulses generated from the home‑built picosecond pulser.
Credit: ©Science Bulletin
While spin-orbit torque magnetic random-access memory (SOT-MRAM) is rapidly approaching commercialization, the technology has hit a severe industrial roadblock. The industry standard relies on β-phase tungsten (β-W) write electrodes, which suffer from high electrical resistance and strict thickness limitations. For chip manufacturers, these physical constraints translate directly to poor device consistency, high power consumption, and a hard limit on how small the memory cells can be scaled.
Now, a research team led by Professor Weisheng Zhao from Beihang University, and collaborators from Truth Memory Corporation, Aix-Marseille University and Université Paris-Saclay, has demonstrated a breakthrough that bypasses these limitations by harnessing the hidden degree of freedom, the orbital angular momentum. Published in the journal Science Bulletin, their work details a highly efficient orbital torque-driven magnetic tunnel junction (MTJ) that pushes orbitronics out of theoretical physics and into industrial application.
The researchers found that a precisely tuned ruthenium-tungsten (Ru/W) bilayer write electrode could serve as a highly efficient charge-to-orbital-to-spin converter. Next, they demonstrated that this novel write electrode interacted seamlessly with core magnetic layers in MTJs, maintaining strong perpendicular magnetic anisotropy and tunneling magnetoresistance (TMR) even after high-temperature manufacturing processes. After fabricating these devices on an industrial 8-inch MRAM pilot line, they verified that the new architecture solved critical yield bottlenecks, demonstrating highly uniform resistance across the wafer. Finally, performance tests confirmed this design as a highly viable commercial candidate because it dramatically lowered writing voltage and enabled ultra-fast 28.7-picosecond switching, while unlocking a shared-electrode structure that can shrink device area by an impressive 45% for massive future power and density improvements.
The team says further studies are needed, but this report represents the first step toward moving orbitronics from theoretical physics into real-world chips. “In light of these results, it will be necessary to further investigate the integration of these devices into large-scale memory arrays, moving this technology toward practical applications and providing a brand-new technical route for next-generation universal magnetic memory,” says Zhao
Journal
Science Bulletin