Scientists have long sought to build lasers from silicon. Such an advance would enable engineers to incorporate both electronic and optical devices onto cheap silicon chips rather than being compelled to employ costly-to-make lasers based on "exotic" semiconductor materials such as gallium arsenide or indium phosphide. Silicon lasers could lead to affordable light-based systems that harness photons instead of electrons to shuttle huge amounts of data swiftly--at multigigabit-per-second rates. Two research groups, one at the University of California at Los Angeles and the other at Intel Corporation, have recently reported success in making silicon emit continuous laser light.

This much anticipated feat came despite silicon's dogged resistance to serving as a lasing medium. In a good lasing material, electrons that are pumped up with energy release that energy in the form of coherent photons of light. In silicon, however, excited electrons are more likely to vibrate, thus generating heat instead. "There have been many attempts, but no one had been able to get silicon to lase before now," notes Bahram Jalali, the physicist who led the U.C.L.A. team.

Jalali and his group solved the problem last fall by making clever use of some of the very vibrations that undermined silicon's suitability for lasers in the first place. In particular, they focused on the Raman effect, a process in which the wavelength of light lengthens after it scatters off atomic vibrations. The U.C.L.A. researchers matched the scattered light with the pump energy from another laser in a way that created constructive feedback, resulting in a net amplification of light.

Intel reported its own success in creating a silicon Raman laser several months afterward. The chipmaker's scientists fed light from a separate laser into a waveguide (or light pipe)--basically an S-shaped ridge the engineers sculpted onto a 15-millimeter-square silicon chip--and Raman laser light emerged. Naturally, the task was not that easy. The power of a silicon Raman laser typically hits a limit as photons sporadically collide with silicon atoms and release free electrons. "Unfortunately, the free electron cloud absorbs and scatters light, so you get diminishing returns as you pump the device harder," explains Mario Paniccia, director of Intel's photonics technology laboratory. The team therefore positioned two electrodes on either side of the waveguide, forming a kind of diode. "Placing a voltage across the diode sucks the free electrons away like a vacuum cleaner," he says, and thus keeps the light flowing through the chip.

"This and related research should lead to many useful applications," says Philippe M. Fauchet, an electrical and computer engineer at the University of Rochester. A laser beam generated continuously through silicon could overcome cost and size limitations in lasers that could be used in surgical procedures, for example. The technology could also detect tiny amounts of chemicals in the environment, jam the sensors of heat-seeking missiles or enable high-bandwidth (high-capacity) optical communications.

Looking a bit farther afield, Paniccia believes that the new laser technology could serve as a building block for high-bandwidth photonic devices constructed almost entirely of inexpensive silicon in existing semiconductor foundry and micromachining facilities. "We've already developed the other necessary components of such a system," including fast modulators (optical encoders), light guides and photodetectors, he notes.

Of course, many in the industry hope that this technology will eventually lead to fully optical computers--superspeedy digital systems in which photons rather than electrons serve as 0s and 1s. Paniccia is certainly optimistic about the recent progress: "This work constitutes not only a scientific breakthrough but also a psychological one, because nobody thought it could be done."

Comments