Holey Silicon Lights Way for New Computers

Thin and iridescent as soap bubbles, porous silicon wafers hardly seem like the stuff of a future generation of computers. But they do one thing that run-of-the-mill silicon chips can't: They glow. A new process for fabricating porous silicon chips, described in the current issue of Physical Review Letters, could someday make it possible to mass produce computer circuits that use light, rather than electricity, to communicate.

When porous silicon was discovered in 1954, "it was just a curiosity," says David Lockwood, an optical spectroscopist at the National Research Council (NRC) in Ottawa, Canada. Then in 1990, British physicist Leigh Canham discovered that, because of the unusual quantum effects that result when electrons are confined to small spaces in a hole-ridden wafer, electric current makes porous silicon glow bright red. This finding prompted researchers to try making light-emitting diodes (LEDs) from silicon instead of rarer and more expensive materials like gallium arsenide, a quest that finally succeeded last year.

But before silicon LEDs could hope to compete with conventional LEDs, which have been used in everything from sneakers to traffic lights, engineers had to figure out how to create precise porous silicon features on a silicon wafer. The etching techniques that create the pores in silicon don't leave crisp edges, and the resulting fuzziness would wreak havoc in a light-based computer: It would be like trying to draw a pixel with a can of spray paint.

Lockwood, electrochemist Patrick Schmucki of the Swiss Federal Institute of Technology, and ion implantation specialist Lynden Erickson of the NRC found a way to replace the paint can with a precision brush. First they injected extra silicon ions into precise locations in a silicon wafer, disrupting its crystal lattice. Next, they put the wafer in hydrofluoric acid and ran a current through it. Only the disrupted portions of the silicon dissolved, leaving behind a wafer with the desired pattern of porous silicon.

"This is a very nice, important result," says Philippe Fauchet of the University of Rochester, who produced the first porous silicon LED last year. But he says that practical problems of porous silicon--it's 10 times less efficient than gallium arsenide at converting electrons to photons, and it burns out more quickly--remain unsolved. For these reasons, "People in industry don't believe that porous silicon will be useful," Lockwood says. But he notes that further advances could easily change their minds.