Innovation: Computers see the light: A British discovery could lay foundation for a new generation of processors

A BRITISH discovery is laying the foundations for the next generation of computers, which run on light rather than on electricity.

Dr Leigh Canham at the Defence Research Agency in Malvern, Worcestershire, has developed a technique for making silicon emit light - a breakthrough that raises the possibility of building on existing micro-electronics technology to create optical computers.

The optical computer - a thinking machine that runs on light - is the dream of the information industry. There are several potential advantages from using pulses of light rather than electricity to transmit information. In principle, light- based switches will be extremely fast. Light beams can also carry more information than electrical pulses and do not heat up the medium through which they travel in the same way that electricity heats up wires.

The transition to light is already well underway in the communications industry, with optical fibre cables carrying telephone signals, television pictures and computer data under land and sea.

Integrating light-based circuitry into computers is another matter, however. Everything must be made at a size too small for the eye to see - the light source, waveguides for transmitting the light, and detectors for reading the signals at the other end.

And, crucially, it must be possible to incorporate these components into silicon chips. It is possible to fabricate detectors and waveguides from silicon, but a suitable light source has proved elusive.

In the past decade or so, researchers have focused on two composites of semiconducting materials as the basis for chip- based light sources: the alloys of indium and phosphorous, and of gallium and arsenic. But incorporating these into silicon technology is difficult.

The process developed by Dr Canham - which he has spent the past four years refining - negates the need for this kind of hybrid, because silicon itself provides the light source. Silicon is processed so that it glows with visible light when a voltage is applied to it, a property known as electroluminescence.

The trick is to make silicon highly porous, by etching into it a complex network of channels. This is done by immersing it in a bath of hydrofluoric acid. The acid eats away at the silicon, opening up channels until the material consists mostly of empty space - a kind of silicon sponge.

The material that is left forms an interconnected web of fine silicon wires, as little as a few millionths of a millimetre thick: about 70,000 times thinner than a human hair. This form of silicon is electroluminescent. The attraction of the method, says Dr Canham, is that 'you can do it in a bucket'.

The material was first discovered in 1956 but its discoverers, who were more interested in making smooth, polished silicon surfaces, regarded the porous material simply as a nuisance. The reason why porous silicon emits visible light while bulk silicon does not is down to quantum mechanics - the theory of very small systems.

Electrons in thin silicon wires behave very differently from those in normal silicon, in that they can lose energy by emitting light at visible frequencies. The colour of the light emitted depends on the size of the region within which they are confined - a so-called quantum-confinement effect. Because of this, the colour of the light emission from porous silicon can be tuned by altering the thickness of the wires, which in turn depends on the extent to which the material is etched.

So far, emission of red, yellow and green light has been achieved. In principle, blue emission should be possible if the wires can be made thin enough, but the silicon sponge starts to become very frail at this stage. This fragility is manifested in particular during the drying process after etching - as the hydrofluoric acid evaporates, the network crumbles.

Last year, Dr Canham and colleagues at the DRA (an offshoot of the Ministry of Defence) showed that a technique called supercritical drying, which is similar to the process used to remove caffeine from coffee, can be applied to make ultra-thin wires.

This kind of innovation has kept the DRA team ahead of the field so far, but the acid test will come with the transition from bench-top to market place. Dr Canham says that industrial companies are watching and waiting at this stage. But there is, needless to say, competition from large micro- electronics firms such as AT&T's Bell Laboratories in New Jersey, in the US, and the Japanese telecommunications company NTT in Tokyo. However, IBM's financial difficulties have put paid to its research efforts at the company's Yorktown Heights research centre in New York.

Dr Canham says the DRA's strong patents on the method for making the silicon wires and on the supercritical drying process will ensure it can capitalise on future developments.

(Photograph omitted)