Spintronics makes computers more efficient

Electron spin-states can make computing less energy hungry...
02 April 2019

Interview with 

Ben McAllister, Naked Scientists


this is a picture of some computer chips


Scientists at the University of Cambridge believe they’ve come a step closer to revolutionising how computers work, and making them much more powerful and energy efficient. At the moment, chips work by pushing electrical charges through semiconductor materials like silicon to do calculations. But, as we shrink the sizes of the circuits to pack more onto a chip to increase computing power, it becomes harder to push electricity through them, and more energy is wasted. An alternative is to transfer information by using a property called the “spin-state” of electrons. But the challenge has been to design a material that can allow this information to be transmitted over a sufficient distance to be useful, but without messing up the spin information, and this is what the Cambridge team think they’ve done using materials called organic semiconductors. Here to explain how this works is Naked Scientist Ben McAllister...

Ben - It's a great question because it's something that is sort of difficult to conceptualise is this quantum mechanical property, the sort of bizarre thing subatomic particles have like electrons, for example, which is what we’re talking about here. They have this property known as "spin", sort of like intrinsic momentum that they have and it can take one to states: it has to either either up or down. And the fact that it has to be up or down means that you might start thinking about ways to use it in computing because anyone who has heard of binary before, which is the language that computer's speak and the language that they use to do their operations relies on a series of ones and zeros. You essentially need to states to do any kind of computing based on binary and when you have electrons that have this property that can either be up or down, you can make one of those one and one of them zero and then do computing that way.

Chris - How do you register the spin on the electron, whether it's an up or down spin?

Ben - So it kind of depends on the context that you're working in. In this case they're using something called the "inverse spin hall effect", another quantum mechanical process which basically means the spin creates a small voltage across some sensor that they can read out.

Chris - And the problem has been that you can impart this spin to an electron but you can't send it anywhere, over any meaningful distance in order to convey a message, and the Cambridge team are saying that's what they think they've surmounted?

Ben - Yeah. So specifically they’re talking about this class of materials called "organic semiconductors", and semiconductors are really important materials in computing, they're what we typically use to do computing. Typically we use what are called "inorganic semiconductors" so things like the silicon which is a chemical you find in sand and other things like that. If we want to use organic materials instead to make semiconductors, they're much cheaper and easier to produce, so there's been a lot of interest in doing that in recent years but yes, as you say, in these organic semiconductors that this team is working with, they've found that transporting the spins to use for computing is really difficult. They basically don't travel far enough and they don't stick around for long enough - they kind of defuse they call it.

Chris - And what's their solution?

Ben - So what they found, if you pump these organic semiconductors, if you put a bunch of additional spins in there it enters the strange regime where the spins start basically travelling a lot easier within the organic semiconductors. So they don't defuse as quickly, they can travel longer distances and they stick around longer inside the material as long as you provide this artificially increased spin density.

Chris - Is that a bit like if I was at a loud party and I just turned the music up that I did want to listen to a bit louder, the people in the room playing something different I drown them out. Is that it or is there something else to it?

Ben - I think it might be kind of similar in the sense that you're providing more of the things that you do want into the material, but it's really something quite strange and interesting going on in the material where if you provided enough spins seems that the way that they move around in the material actually sort of changes; it's like a mechanism for it. So it would be like if you turned up your music and when it reached a certain level all of a sudden you also got some headphones and then you could hear it a lot better.

Chris - Now when we design computer chips at the moment, one of the constraints is that we have shrunk the components, the transistors which are what are actually doing the logic, the noughts and ones of binary, to such an extent now that they are not much bigger than just a few atoms across and that means that obviously the energy that it takes to push electricity through there is very high. Also they're closely packed now that they begin to interfere with each other if we go much smaller so we've hit this silicon wall where we can't shrink them any further. What does this mean if we can pull off these new forms of semiconductors - these organic semiconductors - where does this take our computing?

Ben - It really is a completely new regime because we sort of would not necessarily need to think about using traditional semiconductor-based transistors and making them smaller and smaller and smaller because you wouldn't be using those transistors to encode your information and that's what's going on at the moment. Transistors that are either on or off, so that's how you read your ones and zeros and so you can only have as many ones and zeros as you can have these little transistors which are a few atoms across. If you were to use spins in these organic semiconductors instead they can be much, much, much smaller so we could basically keep scaling up the number of ones and zeros and bits you can feed into your chip. So yeah, we really could get around this problem that were having with miniaturisation.

Chris - What would the chip of tomorrow, using this technology, actually look like? I'm pretty comfortable with how a microchip at the moment works. It's got a leg or a pin and I can send some electricity in their and it finds way through all these various circuits inside the chip, so are these new organic semiconductors computer chips going to be completely different architecture?

Ben - Well the interesting is that those traditional semiconductors that you're talking about, the transistors that we use to do computing today, they're relying on transport of charge, so basically moving the electrons around inside them and then based on how much charge is moving around, how many electrons there are calling that one or a zero. This is actually reading something entirely different, which is the electron's quantum mechanical spin. So really the architecture, yeah is going to be quite different. It's quite a bit of a change of paradigms and I believe there is still quite a lot of work that needs to be done. It's really like getting these fundamental technologies sorted out, like chips that we can actually transport the spins around in, that's going to allows to figure out what the future looks like.

Chris  - And when people talk about quantum computing, is this it question?

Ben - So this is a good question. It's computing that relies on quantum mechanics. To understand how it works. We're talking about quantum mechanical properties like the spin of electrons. When people are typically talking about quantum computing, what they're really talking about using a different thing which is called a "qubyte" or a "quantum byte" which is a byte that exists in like a super position of being up and down at the same time. Then you can do all kinds of other interesting operations there that work a lot faster. In this context they're not talking about using them for quantum computing even though they are quantum properties that are being read out. So no is the real answer but it is kind of on the blurry line there.


Add a comment