Attractive proposition

An international project led by Bath University could make computers up to 500 times faster by dispensing with the need for wires to carry electric currents in silicon chips.

A three-year international project led by BathUniversity could, it is claimed, make computers up to 500 times faster by dispensing with the need for wires to carry electric currents in silicon chips. Research will instead focus on using nanotechnology to create magnetic fields to produce electric signals.

Computer power doubles on average every 18 months — a process known as Moore’s Law — as technology allows developers to reduce the size of integrated circuits. However, over the next few years it is believed researchers will hit a limit imposed by the need to use electrical wiring, which weakens signals sent between computer components at high speed.

Work at Bath aims to establish if there is a way of carrying electric signals without wiring. Wi-fi internet systems and mobile phones use wireless technology now, but the electronics that create and use the signals are too large to be used within individual microchips successfully.

Central to the project will be discovering a way to produce microwave energy on a small scale by firing electrons into magnetic fields produced in semiconductors that are only a few atoms wide and layered with magnets.

The process, called inverse electron spin resonance, uses the magnetic field to deflect electrons and modify their magnetic direction. This creates oscillations of the electrons that make them produce microwave energy, which can then be used to broadcast electric signals in free space without the weakening caused by wires.

Dr Alain Nogaret at Bath’s physics department pioneered this project, which begins in October and will be the first time the theory is put into practice. ‘This is advantageous for computer technology,’ he said, ‘because we want to replace physical wires as currently the physical, copper wires lose energy.’

He claimed that getting more power from silicon chips by shrinking their components is a limited approach, with conventional technology a finite resource.

The technology will also allow the development of integrated circuits that will continue to work even if some of its connections fail. The system can be programmed to re-route itself so it can continue working. Currently, should wiring fail at any point, an integrated circuit will fail, meaning the new approach offers significant cost savings.

‘One other area for the technology is that if we have a wireless connection rather than a fixed one, we could build flexible architecture
that could be routed around defects on the chip. At the moment if there was one transistor defective out of one million the chip has to be thrown away,’ said Nogaret.

Such flexibility would be welcomed by an industry in which manufacturers of integrated circuits have no room for error.
The advantage of the new, flexible system is that only 95 per cent of the electric components would need to operate for the chip to function, so it would be significantly cheaper to produce.

Over the life of the project Nogaret will work with colleagues to assemble and fabricate the device at Bath, before testing the electrical and microwave emissions and properties.

Not only could this technology improve the speed of computers by 500 times, Nogaret believes it may assist in medical diagnostics, by raising the speed and accuracy with which data can be gathered from health monitoring sensors.

In this instance, the microwave emitters are small enough to be integrated on portable biological sensors. Nogaret admitted,
however, that this is very much a long-term aim.

The university will receive £463,000 for the project. NottinghamUniversity will receive £65,000, and LeedsUniversity £27,000. St AndrewsUniversity in Scotland, and Antwerp University, Belgium, will also take part, as will the Centre National de la Récherche Scientifique in Grenoble, France.