From JPost, 18/9/06, by Avi Krawitz [full article available to subscribers]....
Intel's Jerusalem research and development center and the chip maker's Santa Clara headquarters has developed a new concept to answer a foreseen bottleneck in data communication transfer, which will increase the speed of future generation computers a thousandfold. "This technology removes one of the last major barriers to producing low-cost, highbandwidth silicon-based optical devices for use inside and around future computer and data centers," Intel said ahead of the release of the new technology on Monday. The company carried out the research in partnership with the University of California, Santa Barbara, to develop a "proof of concept" for an electrically powered hybrid silicon laser using silicon manufacturing processors [sic - should be "processes" - SL] - a world first, Intel said.
In layman's terms, Nahum Izhaky, photonics technology development manager at Intel Jerusalem and head of the Israeli team involved in the project, explained that the technology allows data to be transferred from chip to chip via light, or laser electrons, rather than copper wiring, thus overcoming major limitations expected to come to the fore from the current system. "The question will soon arise on how much information you can send through copper lines and at what speed," Izhaky told The Jerusalem Post. "While the foreseeable problems can be overcome with the current technology this would be very complicated, inefficient and prohibitively expensive." Intel's breakthrough, he explained, will make that process cheaper and several thousand times faster.
The researchers developed a method to combine the light-emitting properties of Indium Phosphide with the light-routing capabilities of silicon into a single chip. "When voltage is applied to the chip, light generated by the indium phosphide-based layer enters the silicon waveguide where it is contained and controlled, creating a hybrid silicon laser," the company said. While the previous use of indium phosphide-based lasers has been very expensive, the new method's use of silicon drastically reduces that cost. "This marks the beginning of highly integrated silicon photonic chips that can be massproduced at low cost," said John Bowers, a professor at UCSB. One of the great advantages of this, Jerusalem's Izhaky said, is that it will provide the ability to produce hundreds of lasers in the same process.
As the demands of data transfer increase, the limitations of copper wiring on transistors becomes ever more pressing. "We don't see limitations on the transistors themselves and it will take about 20 years to get there," Izhaky said. "The complications from using copper wiring in the interconnections between chips within the transistor however will limit our capabilities in a few years' time, probably around 2012." And with the proof of concept in place, the Jerusalem team can continue its work to bring it to the next stage of development in creating a suitable architecture thus allowing for the development of commercial product by then. "While still far from becoming a commercial product, we believe dozens, maybe even hundreds of hybrid silicon lasers could be integrated with other silicon photonic components onto a single silicon chip," said Mario Paniccia, director of Intel's photonics technology lab in Santa Clara. "This could enable low-cost, terabit optical 'data pipes' in future computers, making possible a flood of new applications for high-speed computing."