In the next-generation silicon Raman laser Jalali is developing, pump light travels along an optical waveguide to a micro-disk resonator, circulates around the disk interacting with the atomic vibrations of the silicon, and continues along the waveguide at a new wavelength.
Silicon lasers working in the mid-IR could have new applications in sensing and military defense, and open the door to optical computing. The speed of computer chips is limited by how much data they can push through copper interconnects, and using photons instead of electrons would overcome that limit. The difficulty is that the materials and processes for making semiconductor lasers are incompatible with the process for making silicon chips. Researchers at the University of California, Los Angeles, (UCLA) recently have found a way to generate light using common silicon structures.
Bahram Jalali, a professor of electrical engineering, and Ozdal Boyraz, a postdoctoral student at UCLA persuaded silicon to emit pulses of photons at 1675 nm, using stimulated Raman emission."This is something people have been trying to do since the first semiconductor laser was demonstrated in the early '70s," says Jalali, whose work was funded by the Defense Advanced Projects Research Agency.
Unlike the III-V compounds, such as gallium arsenide, generally used to make semiconductor lasers, silicon has an indirect bandgap. That means the momentum of the charge carriersnegative electrons and positive holesdo not match, and when they combine they are more likely to produce a vibration than a photon. Jalali and Boyraz found they could work with the atomic vibrations of the silicon to produce new wavelengths of light.
They pumped the material with a mode-locked fiber laser operating around 1540 nm with a 25-MHz repetition rate. The vibrations in the atoms scatter the light at another frequency, the Stokes frequency. The Stokes wave interacts with the pump wave to create a new frequency that drives the atomic vibrations. "Atomic vibrations scatter more of the pump light into the new wavelength," Jalali says. "It's essentially a feedback process." The researchers found a sudden increase in emission of 1675-nm photons when the pump power reached 9 W, indicating that lasing was occurring. Deep Into the Infrared
While there is no easy way to make an electrically pumped silicon Raman laser, the device can still prove useful, Jalali says. The next step will be to build the device with a cascading laser cavity in which the feedback process produces a series of increasing Stokes frequencies, causing the emission to hop to longer and longer wavelengths. That will bring the laser emissions into the mid-IR, between 2 and 10 µm, where ordinary semiconductor lasers cannot reach.
Existing fiber Raman lasers also provide emission at essentially any wavelength. But glass fiber is an insulator without the semiconductor properties that make electronic switching on a chip possible. Silicon Raman lasers provide the kind of control semiconductor lasers have, but at wavelengths beyond the near-IR, Jalali says.
"It's a fundamentally new device," he says. "This is a range where there are a lot of critical applications, but there are no practical lasers that operate in the mid-infrared." Such devices could improve detectors for biological molecules, almost all of which have fundamental vibrational frequencies in the mid-IR. Current detection schemes rely on harmonics of those frequencies in the near-IR. The military could use the lasers to blind the detectors of heat-seeking missiles. Jalali is considering forming a company to commercialize the device.
The research is "a significant step forward," says Lorenzo Pavesi, a physicist who has long worked on silicon lasers at Università di Trento (Povo, Italy). "This work is a significant achievement, as it demonstrates for the first time that lasing is possible in silicon." Pavesi does wonder about practical applications because of the optical pumping of the laser, and he worries that continuous wave emission might prove difficult. Jalali, however, believes continuous wave emission will be achieved "fairly soon."