A dream just came true for the chip industry. For decades, engineers have
tried to coax ordinary silicon to emit light. That way, silicon could be used not only for computer chips but also for optical devices such as
microlasers and light-emitting diodes (LEDs), which now are built with
pricier materials like gallium arsenide. Being able to combine electronic
and optoelectronic functions on the same chip would also simplify design and cut equipment costs.
In telecommunications, for example, it would end the need for both silicon and gallium chips in the switch boxes on fiber-optic networks: gallium to send and receive the pulses of light that carry data, and silicon to direct the data streams to their proper destinations.
IBM has now found this long-sought universal material. But it isn't
silicon. As company researchers report in the May 2 issue of Science, it's the carbon nanotube -- a wee whisker of carbon almost 100,000 times thinner than a human hair.
A CLOSED CIRCLE. Nanotubes have long been viewed as silicon's likeliest successor by around 2015, when it's expected to run out of steam. The new materials' big advantage is that they can be engineered to function either as a conducting wire (replacing the copper or aluminum circuit lines on today's chips) or as a semiconductor (for the on/off switches called transistors). Adding photonic capabilities "closes the circle," says Phaedon Avouris, IBM's manager of nanoscale research. "Now, everything can be done with nanotubes."
What Big Blue unveiled is a red nanosize LED. The researchers fire electrons into one end of a 1.4-nanomter-wide tube, and positively charged electrons, called holes, into the other end. When the electrons and the holes meet, they annihilate each other in bursts of infrared light. This light happens to be the same wavelength as the pulses now zipping through fiber-optic systems, but other colors could be generated with bigger or smaller nanotubes, says Avouris.
Creating light through the collision of oppositely charged electrons and holes is nothing new, notes Avouris. "The problem has always been that the light yield is very limited" -- too dim and intermittent to be useful, he says.
CAN'T MISS. To generate a steady glow, thousands of particles must smash continuously into each other. Aiming a tiny electron directly at an equally tiny hole is really tough. But inside a nanotube, says Avouris, "there's no way they can miss each other."
Another key to this breakthrough is the novel ambidextrous transistor that IBM researchers developed in 2001. Conventional silicon transistors can handle only holes or only electrons, not both. But IBM's new silicon transistor can. With a nanotube inserted smack in the middle of this so-called ambipolar transistor, electrons and holes stream into opposite ends when the juice is turned on.
That's important because it shows that nanotube LEDs could be combined with today's silicon circuitry. And that may push nanotubes onto commercial chips even before 2015, when the semiconductor industry expects to bump into physical limits that will end its amazing record of doubling the power of chips every 18 months or so. By Otis Port in New York