Making brainier computers
Professor Stuart Parkin, inventor of the magnetic disk drive, was recently awarded the Millennium Prize for his work in spintronics. The IBM Fellow and Max Planck Director spoke to James Perkins about his research.
What have you been working on lately?
I have been working on what I call cognitive materials, which essentially means materials whose properties evolve as you use them. Can we imagine building devices with architectures that could enable computing, which mimics in some sense how we ourselves think, and do this in an energy efficient manner? Our brain, for one given operation, uses one million times less energy than we do today in silicon computer chips.
What materials are you using to achieve this?
It is a certain class of oxides called correlated electron oxides. I am using vanadium dioxide, but have seen the same behaviour in other oxide systems. These are materials that are initially insulating, and by applying large electric fields you can make them conducting and metalise them. When you apply large enough electric fields to the surface of a thin film of one of these correlated insulating oxide systems, you remove tiny amounts of oxygen from the films, which metalises them. I argue that this is a very exciting new means of creating low-energy devices by the manipulation of a small number of ions.
You have been developing Racetrack Memory for a decade, what is it?
Racetrack Memory has the potential to replace magnetic disk drives and many forms of today’s solid-state memories in computers. It is an innately 3D structure where information is stored in the form of tiny magnetic regions in tall nanowires. It can potentially store 100 times as much data as any conventional or proposed solid-state memory at roughly the same cost, but using less energy.
How does it work?
The underlying concept is the possibility to move the information up and down these nanowires, not by moving atoms, but rather by rotating the magnetisation of atoms within the nanowire by passing current along it. The current is innately spin-polarised, and the flow of the electrons, whose spins are oriented along the local direction of magnetisation of the nanowire, causes the magnetic boundaries – the domain walls – to move along the racetrack in concert as the magnetic orientation of the atoms near the domain walls is reversed. When those electrons cross the boundaries between two adjacent magnetic regions with different magnetisation directions (e.g. up and down), the magnetisation in one region changes direction and the boundary moves. The magnetic information moves together just like soldiers in an army. This is what I proposed 10 years ago as a concept and recently we have shown that we can move these bits of information at up to 1,000 metres per second.
What other projects are in the pipeline?
I plan to continue work on Racetrack Memory, but also other aspects of spintronics. One project is the conversion of charge current to spin current, through something called the spin hall effect. There are also magnetic structures called skyrmions – objects within the magnetic layer whose magnetisation is reversed in a special way. One of my other main interests will further exploration of cognitive materials. I will also concentrate on novel superconducting materials, in particular looking to find materials that could superconduct at room temperature, which would make them of great practical use. All these topics are great, grand challenges that if we are successful in meeting could change the world.