PORTLAND, Ore. -- Intel and academic electrical engineers (EEs) recently demonstrated the world's first mode-locked silicon evanescent laser, a device capable of performing optical functions on CMOS chips. This is instead of translating from optical-to-electrical then back from electrical-to-optical, as is standard procedure for telecommunications applications of lasers today. The silicon laser emitted 40 billion pulses of light per second (40-Gbit/sec), and was built on the hybrid silicon/indium phosphide platform developed last year.
The joint-development effort between Intel Corp. (Jerusalem) and the University of California (UCSB; Santa Barbara) produced a new silicon laser that delivered highly stable ultra-short pulses of laser light, which can be used in a variety of ways: for high-speed data transmissions at multiple wavelengths, for remote sensing using Lidar (Light Detection and Ranging), for processor-to-processor optical communications on multi-core chips, and for highly accurate optical clocks.
"We have improved upon Intel's original design to enable it to emit extremely short, four pico-second pulses with very low jitter and extinction ratios above 18 dB," said electrical engineer Brian Koch. "Going along with this is its capability to produce a wide spectrum of wavelengths that could potentially be separated to provide multi-channel communications capabilities from a single laser, instead of having to use an array of lasers, as is the case today." Koch, a doctoral candidate working under EE professor John Bowers at UCSB, performed the work in cooperation with Intel researcher Oded Cohen.
According to Koch, Intel is already producing experimental optical modulators, but lacked a stable silicon pulsed laser as a multi-spectrum laser light source. Now the two should be able to be used together to create all-silicon optical communications chips with multiple channels on the same chip.
Intel's silicon laser architecture, which was previously reported to be capable of a continuous wave, was modified by the current research group by adding elements to the laser cavity that permitted it to emit extremely short pulses at intervals that were both even and ultra precise. By pulsing the laser at regular intervals, the researchers were able to pack more power into each one for easy handling of optical time-division multiplexing (OTDM) and wavelength-division multiplexing (WDM).
"Because our silicon laser is so powerful and its pulses so well timed, we could combine, say, four 40-Gbit-per-second lasers into a single 160-Gbit-per-second signal," said Koch. "One other thing we hope to do is separate all the wavelengths coming out of each pulse, to create an array of wavelengths that could be separately modulated then recombined for multi-channel communications over a single fiber."
For the future, Intel and UCSB plan to cooperate on designing chips that implement some of the possible applications for a silicon laser, with Intel supplying the necessary on-chip silicon waveguides and UCSB implementing the architecture to create the rest of a telecommunications-application chip.
Beyond telecommunications, Koch predicts that silicon lasers could also be used on multi-core processors for internal core-to-core communications at speeds impossible for electrically switched signals.