http://cn.newmaker.com
9/20/2006 10:07:00 AM
佳工机电网
SAN FRANCISCO, Sept. 17 — Researchers plan to announce on Monday that they have created a silicon-based chip that can produce laser beams. The advance will make it possible to use laser light rather than wires to send data between chips, removing the most significant bottleneck in computer design.
As a result, chip makers may be able to put the high-speed data communications industry on the same curve of increased processing speed and diminishing costs — the phenomenon known as Moore’s law — that has driven the computer industry for the last four decades.
The development is a result of research at Intel, the world’s largest chip maker, and the University of California, Santa Barbara. Commercializing the new technology may not happen before the end of the decade, but the prospect of being able to place hundreds or thousands of data-carrying light beams on standard industry chips is certain to shake up both the communications and computer industries.
Lasers are already used to transmit high volumes of computer data over longer distances — for example, between offices, cities and across oceans — using fiber optic cables. But in computer chips, data moves at great speed over the wires inside, then slows to a snail’s pace when it is sent chip-to-chip inside a computer.
With the barrier removed, computer designers will be able to rethink computers, packing chips more densely both in home systems and in giant data centers. Moreover, the laser-silicon chips — composed of a spider’s web of laser light in addition to metal wires — portend a vastly more powerful and less expensive national computing infrastructure. For a few dollars apiece, such chips could transmit data at 100 times the speed of laser-based communications equipment, called optical transceivers, that typically cost several thousand dollars.
Currently fiber optic networks are used to transmit data to individual neighborhoods in cities where the data is then distributed by slower conventional wire-based communications gear. The laser chips will make it possible to send avalanches of data to and from individual homes at far less cost.
They could also give rise to a new class of supercomputers that could share data internally at speeds not possible today.
The breakthrough was achieved by bonding a layer of light-emitting indium phosphide onto the surface of a standard silicon chip etched with special channels that act as light-wave guides. The resulting sandwich has the potential to create on a computer chip hundreds and possibly thousands of tiny, bright lasers that can be switched on and off billions of times a second.
“This is a field that has just begun exploding in the past 18 months,” said Eli Yablonovitch, a physicist at the University of California, Los Angeles, a leading researcher in the field. “There is going to be a lot more optical communications in computing than people have thought.”
Indeed, the results of the development work, which will be reported in a coming issue of Optics Express, an international journal, indicate that a high-stakes race is under way worldwide. While the researchers at Intel and Santa Barbara are betting on indium phosphide, Japanese scientists in a related effort are pursuing a different material, the chemical element erbium.
Although commercial chips with built-in lasers are years away, Luxtera, a company in Carlsbad, Calif., is already selling test chips that incorporate most optical components directly into silicon and then inject laser light from a separate source.
The Intel-Santa Barbara work proves that it is possible to make complete photonic devices using standard chip-making machinery, although not entirely out of silicon. “There has always been this final hurdle,” said Mario Paniccia, director of the Photonics Technology Lab at Intel. “We have now come up with a solution that optimizes both sides.”
In the past it has proved impossible to couple standard silicon with the exotic materials that emit light when electrically charged. But the university team supplied a low-temperature bonding technique that does not melt the silicon circuitry. The approach uses an electrically charged oxygen gas to create a layer of oxide just 25 atoms thick on each material. When heated and pressed together, the oxide layer fuses the two materials into a single chip that conducts information both through wires and on beams of reflected light.
“Photonics has been a low-volume cottage industry,” said John E. Bowers, director of the Multidisciplinary Optical Switching Technology Center at the University of California, Santa Barbara. “Everything will change and laser communications will be everywhere, including fiber to the home.”
Photonics industry experts briefed on the technique said that it would almost certainly pave the way for commercialization of the long-sought convergence of silicon chips and optical lasers. “Before, there was more hype than substance,” said Alan Huang, a former Bell Laboratories researcher who is a pioneer in the field and is now chief technology officer of the Terabit Corporation, a photonics start-up company in Menlo Park, Calif. “Now I believe this will lead to future applications in optoelectronics.”