Researchers in the US have found yet another use for the "wonder material" graphene. Instead of exploiting the material's exceptional ability as an electrical conductor, the team has found a way to use graphene as an extremely thin "tunnel barrier" to conduction. The team says that this new application is particularly suited to developing spintronics – a relatively new technology that exploits the spin of an electron as well as its charge.

Graphene is a sheet of carbon just one atom thick and ever since the material was first isolated in 2004, researchers have been trying to create electronics devices that make use of its unique properties. Most of this effort has focused on how electrons flow in the plane of the sheet – which can behave both as a conductor and semiconductor. But now Berry Jonker and colleagues at the US Naval Research Laboratory (NRL) have shown that graphene can serve as an excellent tunnel barrier when current is directed perpendicular to the plane of carbon atoms. The spin polarization of the current is also preserved by the tunnel barrier, a finding that could have important implications for spintronics.

Low-energy switching

The spin of an electron can point in an "up" or "down" direction and this property could be used to store and process information in spintronics devices. Circuits that employ a spin current – electrons with opposite spins moving in opposite directions – could, in principle, be smaller and more efficient than conventional electronic circuits that rely on switching charge alone. This is because switching spins from up to down can be done using very little energy.

Spintronics devices are typically made from ferromagnetic materials and semiconductors. Ferromagnetic metals, such as iron or permalloy, have intrinsically spin-polarized electron populations – that is, different numbers of up-spin and down-spin electrons – and thus make ideal contacts for injecting spins into a semiconductor. However, ferromagnets and semiconductors have a large conductivity mismatch, so spin is injected via a tunnel barrier – an electrically insulating barrier through which electrons tunnel quantum mechanically. The problem is that the oxide barriers normally employed as tunnel barriers introduce defects into the system and have resistances that are too high – factors that adversely affect device performance.

Enter the graphene tunnel barrier

To overcome this problem, Jonker and colleagues decided to employ single-layer graphene as the tunnel barrier, because the material is defect resistant, chemically inert and stable. These properties can be exploited to make low-resistance graphene spin contacts that are compatible with both the ferromagnetic metal and semiconductor.

The researchers began by mechanically transferring graphene grown by chemical vapour deposition onto hydrogen-passivated silicon surfaces. They achieved this by floating the graphene on the surface of water and bringing the silicon substrate up from below. This common technique ensures that there is no oxide layer between the silicon surface and the graphene. The team then injected electron spins from a ferromagnetic nickel–iron alloy into the silicon via the graphene tunnel barrier. The voltage arising from the resulting spin polarization in the silicon was then measured using the Hanle effect, a method that is routinely employed by spintronics scientists.

Beyond Moore's law

"Our discovery clears an important hurdle to the development of future semiconductor spintronics devices – that is, devices that rely on manipulating the electron's spin rather than just its charge for low-power, high-speed information processing beyond the traditional size scaling of Moore's law," Jonker says. "These results identify a new route to making low-resistance-area spin-polarized contacts, which are key for semiconductor spintronics devices that rely on two-terminal magnetoresistance, including spin-based transistors, logic and memory."

Using graphene in spintronics structures may provide much higher values of the tunnel spin polarization thanks to so-called spin-filtering effects that have been predicted for selected ferromagnetic metal/graphene structures, Jonker adds. "Such an increase would improve the performance of semiconductor spintronics devices by providing higher signal-to-noise ratios and corresponding operating speeds, so advancing the technological applications of silicon spintronics," he says.

The work, which was supported by programs at the NRL and the US Office of Naval Research, is reported in Nature Nanotechnology.