Published online 12 January 2006 | Nature | doi:10.1038/news060109-12

News

Magnetic thinking

Credit-card strips could get a mind of their own.

A cross of magnetic particles forms a gate, while three magnets outside tell the gate what to do.A cross of magnetic particles forms a gate, while three magnets outside tell the gate what to do.© Alexandra Imre/Notre Dame.

Computers already use magnets to store their memories. Now scientists are trying to make their whole 'brain' go magnetic.

Conventional computers do their 'thinking' by shuttling electrons through arrangements of transistors called logic gates. But in order for those thoughts to be stored as computer memories, the electrical signals have to be translated by bulky components into magnetic fields on the metallic grains that cover your hard drive. This additional step takes up extra room in a computer.

What's more, transistors get so hot that it is becoming increasingly difficult to pack more of them on to silicon chips without melting something important.

The solution? Space-saving, cool-headed, magnetic logic gates.

Alexandra Imre and her colleagues from the University of Notre Dame, Indiana, have created such a logic gate from a pattern of tiny nickel-iron magnets, each about 100 nanometres long. Their magnetic fields can point in one of two directions, which represent the computer bits '1' or '0'. Because each nanomagnet influences the state of its neighbour, a range of input signals at one side of the pattern can trigger a predictable outcome at the other.

At the moment, the team have only made isolated logic gates as a proof of principle. But linking them together could build much more complex computing devices that rely on a whole cascade of flipping fields. They unveil the system in this week's Science1.

If successful, such computers could have the added bonus of boosting security on things such as magnetic strips in credit cards. If these strips are capable of performing calculations rather than simply carrying a password, says Mark Welland, a nanotechnologist at the University of Cambridge, UK, this would make the card much harder to copy.

Neither nor

Although Imre's invention is not the first magnetic logic gate2, it is certainly the most versatile, because it carries a switch that can completely change the gate's function. This could allow very different computations to be run on the same chip.

Known as a MAJORITY gate, its electronic equivalent is already widely used. Such gates can behave as either NAND or NOR gate (the inverse of AND and OR functions) depending on how they are set. Any other logic system can be built from a handful of these gates.

The scientists calculate that their gates would use about 100 times less power than those built from transistors, thus reducing troublesome heating in the circuit. Unlike conventional electronics, each gate does not need to be powered separately.

They also suggest that their logic gate could potentially use magnetic memory bits directly as an input. At the moment, the input has to be laboriously constructed, by painstakingly arranging other magnetic particles with their north or south poles facing the gate. But other teams are currently tackling this hurdle, says Dan Allwood, a materials scientist at the University of Sheffield, UK.

The researchers at the University of Notre Dame are now trying to join up their gates, and find a way to interface them with a conventional electronic device.

Stiff competition

ADVERTISEMENT

So will your computer be going magnetic anytime soon? "Whether or not this will compete with conventional microelectronics is hard to say," says Wolfgang Porod, part of the Notre Dame team.

Manufacturing of silicon chips has become so streamlined that any new advance not based on the same processing methods must offer serious advantages to make them cost effective.

But magnetic chips are making some ground already. Intel, the world's largest computer chip manufacturer, is looking at integrating some of the different approaches to magnetic computing in to its next generation of chips, says Allwood. "Each [approach] has an application where it can beat electronics in some way," he says. 

  • References

    1. Imre A., et al. Science, 311. 205 - 208 (2006).
    2. Allwood D. A., et al. Science, 309. 1688 - 1692 (2005). | Article | PubMed | ISI | ChemPort |