X
Innovation

Spintronic doctors get silicon injection

More news on spintronics, my current favourite bet for what Moore's Law does next. Researchers at the US Naval Research Laboratory have published a paper saying that they've demonstrated one of the more important inventions needed to make the stuff work commercially - the creation of spin currents in a silicon device.
Written by Rupert Goodwins, Contributor

More news on spintronics, my current favourite bet for what Moore's Law does next. Researchers at the US Naval Research Laboratory have published a paper saying that they've demonstrated one of the more important inventions needed to make the stuff work commercially - the creation of spin currents in a silicon device.

A spin current is like an ordinary electrical current in that it's a signal which is carried around the place by electrons. It's unlike an ordinary current in that the electrons don't actually have to move.

Imagine you've got an old bagatelle game, one in which there's a sloping wooden board with lots of nails sticking up. You put a ball bearing in at the top and it careers down, zig-zagging across the board as it bounces off the nails - with luck, ending up in one of the holes at the bottom which carry a score. It's a noisy business and the ball takes a lot longer to reach its destination than if there were no nails.

You can think of an electric current through a material to be like those balls cascading through the forest of nails. It's a crude analogy, but it has some merits - if you increase the number of balls you get more through, like increasing the current, and if you tip the board more the balls move faster, like increasing the voltage. In either case, the process gets noisier and more energy gets lost, like a conductor heating up, and you get more balls going where you don't want them to go.

Now, imagine that the board is perfectly horizontal, and that the little steel balls are magnetised. The balls would just sit there - no noise, to be sure, but no useful work gets done. Instead, they'll settle into a state where the various north and south poles attract each other and the balls are in alignment. Flip one ball, however, and the others will settle into a new alignment, the change rippling through the board.

If your intention is to get a physical flow of balls, then this is pretty useless - you won't get a high score - which means, electrically, you can't light a bulb or turn a motor. But if you're just trying to represent and manipulate information, it works perfectly well - and, in fact, this is what happens when you record data onto a magnetic medium.

That's the idea behind spintronics. Electrons have spin like tiny rotating balls (they're not balls and they don't actually spin, but that's the closest the human brain can get to imagining what's actually happening. Unless you're a quantum physicist, which I'm not, and if you are, I apologise). That spin IS the root cause of magnetism, though, and the way electrons couple spin to each other does lie at the heart of why some materials become magnets while others do not.

And because the electrons don't actually move, you don't get the electrical losses which are the biggest single bugbear for solid-state engineers trying to make electronics do more, faster, in a smaller space with less power. Suddenly, all the rules change - it really is as big an advance as when we discovered the transistor and could move on from thermionic valves.

The problem is, it's very difficult to manipulate spin the same way we do electron movement. We can easily do it on the large scale - bung a magnetic field at the electrons, and they'll obediently flip their spins - but as generating and detecting controllable magnetic fields generally means moving electrons through a conductor, you don't get the big win (works great on hard disks, though).

So what the Naval Research Lab team has done in demonstrating how to controllably set uniform spin states in something akin to an ordinary transistor is absolutely key to unlocking the next twenty years. Moreover, and just as importantly, they've done it using simple structures and very mundane materials such as iron, aluminium oxide and silicon, entirely congruent with the sort of techniques the industry is already using.

I wish I could tell you more, but the paper is (as usual) only available to subscribers to the right journals and I can't get down to the British Library today. Doubtless I'll get a copy sooner or later, and then I'll be able to see details of what sort of temperature the device operates at, the potential and achievable efficiencies, and all the little aspects that make it possible to have a stab at what happens next and when.

And yes, that sound you hear is a journalist coupling his own spin moment to the page.

Editorial standards