In their work, the researchers, led by Caroline Ross and Geoffrey Beach, switched the magnetization in an 8 nm thick thulium iron garnet film by passing an electric current through a layer of a platinum metal adjacent to it. The technique produces a spin current thanks to the spin Hall effect that is perpendicular to the charge current.

The spin Hall effect is a spin transport phenomenon mediated by strong spin-orbit coupling, in which opposite spins are deviated in opposite directions while propagating inside a conductive material. In heavy metals, this effect can be used to convert charge current into pure spin current, which can then be injected into an adjacent ferromagnet to exert a torque (known as the spin-orbit torque, or SOT). The SOT has been used to manipulate the magnetization in metallic ferromagnets, but it had not been successfully used to do the same thing in magnetic insulators - until now. However, this is possible because although charge current cannot flow in these materials, spin current can.

Useful for new types of memory or logic devices

“The spin current interacts with the magnetic moment of the garnet, exerting a spin torque on it, and this torque (the SOT) is strong enough to switch the garnet’s magnetization,” explains Ross. “And although we need to apply an in-plane magnetic field to the material, this can be small and constant and so is easily applied.”

The technique could be useful in new types of memory or logic devices made from magnetic insulators, she tells “It is good to use magnetic materials because they ‘remember’ their (non-volatile) state, but switching them with a magnetic field is inconvenient. Switching with an electric field is much easier and gives the best combination of performance, and in principle the energy required to do so can be very small.

“We can also use electric effects to read back the state of the magnetic material, which allows us to make an all-electrical magnetic device,” she adds.

The researchers, reporting their work in Nature Materials doi:10.1038/nmat4812, say that they now need to better understand how their technique works – for example, how does the SOT act on the domain walls in the garnet? And what happens when the garnet is patterned into small structures rather than as a continuous film, as in this work?