Spintronics seeks to replace conventional electronics, where growing computational demands hit the limits of heat dissipation and von Neumann architecture. Spin-current logic promises in-memory computing, but so far it has relied on metallic magnets such as cobalt-iron-boron, which are hard to scale. Van der Waals magnets like Fe3GaTe2 are attractive for their low magnetization, flexible layer stacking, and clean interfaces without dangling bonds. However, no device based on them had operated at room temperature without an external magnetic field or had measured thermal stability.
A Fe3GaTe2 crystal was grown by the self-flux method, then a 37-nanometer-thick layer was exfoliated and transferred onto an oxidized silicon substrate in an argon glovebox. On top, 4 nanometers of platinum were sputtered as a spin-current source. A double Hall cross with electrodes was etched from the bilayer. The asymmetric cross geometry breaks the transverse inversion symmetry of the electric field, generating perpendicular spins and an effective magnetic field strong enough to switch the magnetization without an external field. Switching was tested with current pulses of varying duration, and the anomalous Hall resistance was measured.
The device shows stable '0' and '1' states at room temperature with a coercivity of 20 mT. Without an external field, switching occurs only in the asymmetric crosses, with the left and right crosses switching in opposite directions. Measurements of the switching current versus pulse duration gave an intrinsic critical current density of 1.8×10⁷ A/cm² and a thermal stability factor of 50. This is the first report of such parameters for a homogeneous van der Waals magnet. For industrial channel dimensions of 100×50 nm², the estimated write energy is 1.5 fJ per bit for a 1 ns pulse, which is 40 times lower than for optimized tungsten/cobalt-iron-boron devices.
This work shows for the first time that a homogeneous van der Waals magnet with a platinum channel can serve as the basis for scalable spin logic at room temperature. Unlike neighboring studies that used composition gradients, antiferromagnetic layers, or low-symmetry crystals, here field-free switching is achieved through electrical asymmetry in highly conductive platinum. This opens a path to dense, energy-efficient computing chips with in-memory computation.
Further reduction of write energy is possible by thinning Fe3GaTe2 to a few layers and using platinum alloys with higher spin-current efficiency. Adding tunnel magnetoresistance for highly sensitive readout is also promising. Developing such devices could lead to full spin logic circuits integrated with CMOS technology.
The results will influence the development of energy-efficient chips for artificial intelligence and the Internet of Things, where density and heat dissipation are critical.
Optimize the magnetic layer thickness and the composition of the spin-Hall metal, and test the device on arrays of nanoscale bits.
For spintronics, the simultaneous achievement of high thermal stability, low switching current, and scalability at room temperature remains an unsolved problem. This work shows that van der Waals magnets can solve this trilemma, but verification on nanoscale elements and with sub-nanosecond pulses is required.
This section is addressed to the author of the paper and their colleagues. Neighbouring works were found by meaning-based search across our archive, not by word overlap. We do not review: we show the field around the work.
What the work does
This work demonstrates a spin logic device based on the layered magnet Fe3GaTe2 with a platinum channel. The device switches with current pulses at room temperature and without an external magnetic field, showing thermal stability with a factor of 50 and a write energy of 1.5 fJ per bit.
Where its strength is
The strength of this work lies in achieving field-free switching in a uniform van der Waals magnet, thanks to an asymmetric geometry and the high conductivity of platinum, which opens a path to scalable spin logic with in-memory computing.
Where it could go next
Worth trying
Why this may matter
If the approach can be transferred to other materials and scaled up, it could lead to energy-efficient chips for in-memory computing that reduce heat dissipation and bypass the limitations of the von Neumann architecture.
Nearby works in our archive
The archive is sparse around this work: the closest review is only 0.53, against a usual 0.64. It looks like few have drilled nearby.
An energy of 1.5 fJ is comparable to the energy of a single visible-light photon, yet it is enough to switch a magnetic bit.