One problem hindering the progress of spintronics is that it is hard to transfer currents of spin-polarized electrons over distances greater than a micrometre in conductors like copper. The advantage of spin waves – collective oscillations of stationary spins in a magnetic insulator – is that they can travel millimetres or even centimetres in some materials with very little loss.

Spin-Hall effect

The technique was developed by Eiji Saitoh and colleagues at Tohoku University, Keio University and the FDK Corporation. They built their device from a 1.3 µm thick rectangular strip of the magnetic insulator Y3Fe5O12. Platinum electrodes just 15 nm thick were deposited at either end of the strip, with 1 mm of bare Y3Fe5O12 between the two electrodes.

The team sent an electrical current through one of the electrodes causing "spin-up" electrons to collect at the interface between the platinum and Y3Fe5O12, while spin-down electrons collected at the opposite surface of the platinum. This behaviour is well known to physicists and is called the spin-Hall effect.

Although electrons in the platinum cannot flow into the insulator, they can exert a torque on spins in the Y3Fe5O12 that are close to the interface. These stationary spins then exert torques on their neighbours, which do the same to their neighbours and the excess spin ripples through the insulator in the form of a spin-wave.

When this wave reaches the other platinum electrode, the reverse process occurs – the spin-wave transfers spin across the interface, creating a surfeit of spin-up electrons. This creates a spin-Hall voltage that causes electrons to flow in the second electrode.

Towards 'spin wiring'

Although the existing device does not input a spin current from an external source – nor does it output a spin current – Saitoh told nanotechweb.org's sister website, physicsworld.com that the principle could be used to create devices that do both. This could be exploited to create sources of spin current, or to transfer spin current over large distances.

Saitoh also believes that the device could be used to create "spin wiring" that could someday replace conventional wires in integrated circuits. And because the spin-waves oscillate at gigahertz frequencies, the device could be adapted as a source of microwaves.

The team is now trying to optimize their design by trying different combinations of materials.

The work is reported in Nature 464 262.