Atomically thin layers bring spintronics closer to applications

'Spin' is a magnetic property of electrons, which can take the values 'up' or 'down'. It could be used to store, transport and manipulate information, but is difficult to handle. For example, it loses direction over time, and thus far no one has managed to create more than a few percent of spin polarization -- in other words, the difference between the number of 'up' versus 'down' spins is small.

Spin injection

Much of the research in Van Wees's lab is directed towards gaining a better understanding of spin behaviour in different materials. His lab has already managed to transport spin signals over record distances at room temperature. The latest experiments focused on spin injection and detection. Injection means getting electrons with polarized spins into a device. In a normal electron current, the number of up and down spins are the same. 'Spin polarization is achieved by sending the electrons through a ferromagnetic material', Van Wees explains. This creates an excess of one type of spin.

The device used in the latest experiments was a sandwich of different materials. At the core was a layer of graphene, just one atom thick. 'Graphene is a very good material for spin transport, but it doesn't allow you to manipulate the spins', says Van Wees. The graphene rests on an insulator layer of boron nitride, which rests on a silicon semiconductor. On top of the graphene is a very thin layer, just a few atoms thick, of boron nitride, which protects the electrons in the graphene from outside influences.

Unexpected

'To inject spins into the graphene, you have to make them pass through the upper layer of the boron nitride insulator. This can be achieved with quantum tunnelling', says Van Wees. The initial construction was one atom thick, but it proved too thin and failed to shield the electrons in the graphene from outside influences. A three-atom layer provided enough protection and allowed normal spin injection. But a two-atom layer caused something totally unexpected to happen. 'We observed a very strong spin polarization of up to 70 percent, ten times what we usually get.'

It was always assumed that polarization was the result of the electrons' passage through a ferromagnet. But in that case, the polarization should have had a fixed value. In Van Wees's devices, the polarization increased with the voltage. 'We have no idea why this happens', says Van Wees. He also found a similar tenfold increase in spin detection in the same device. 'So overall, the signal increased by a factor of 100.'

Flagship Project

This creates many possibilities. 'We can now inject a spin into the graphene and measure it easily after it has travelled some distance. One application would be as a detector for magnetic fields, which will affect the spin signal.' Another possibility would be to build a spin logic gate or a spin transistor. As the experiments with the new device were conducted at room temperature, such applications are quite close. 'However', Van Wees warns, 'we use graphene which we obtained by exfoliation, using Scotch tape to peel monolayers off a piece of graphite. This is not suitable for large scale production.' Techniques to make the right quality of graphene on an industrial scale are still under development.

The work on spin in graphene is part of the European Union's ten-year Flagship Project that started in 2013 with a budget of EUR 1 billion. Van Wees is leader of the Spintronics Work Package, which has met all its goals so far and is set to continue for another two years at least. Working with industrial partners to translate the lab results into applications is an important aim at this stage.

Van Wees has already managed to increase spin transport in graphene and manipulate the direction of transport. Now he has dramatically increased the signal. 'We now have to work on both understanding the physics and developing the technology to integrate these devices into bigger systems. But we also have to think about entirely new applications which may become possible.'