These are heady times for silicon lasers and their developers. Shortly after a team at the University of California, Los Angeles, reported the first silicon laser last fall, a group from Intel Corp. in Santa Clara, Calif., and Jerusalem reported a silicon laser that avoided several of the drawbacks inherent in the device. Now the university researchers have reported modulating their laser electronically, and the Intel group has reported the first CW silicon laser.Silicon is the material of electronics, and any integration of electronics and photonics would be hugely enhanced by the existence of a silicon laser. Unfortunately, as a semiconductor, silicon has an indirect bandgap, which means that phonons, rather than photons, are created when electrons and holes combine.For years, researchers around the world have investigated sophisticated means to create a silicon-based laser. But the breakthroughs in recent months have all involved silicon Raman lasers, whose physics is fundamentally different from that of conventional semiconductor lasers. In a Raman laser, the laser photons are not created when electrons and holes combine, but when other laser photons lose part of their energy to a lattice vibration. Thus, a Raman laser must be optically pumped by another laser.At Intel, the first CW silicon laser resulted from refinements to the resonator of the previously reported pulsed laser. The earlier version of that resonator, fabricated as a rib waveguide on an undoped 2 × 16-mm silicon-on-insulator substrate, had a dielectric mirror on the rear facet but depended on the Fresnel reflection from the front facet as the output coupler. To achieve CW lasing, the researchers applied a dielectric coating to the front facet as well. This not only brought the output coupling closer to its optimum value, but also provided resonant feedback of the 1550-nm pump power. By tuning the pump to a longitudinal mode of the rib-waveguide resonator, they saw an enhancement of the intracavity pump power by more than a factor of two under some circumstances.The scientists achieved significantly greater average power from their CW laser than from the earlier pulsed version. With optical pumping of 600 mW (180 mW in the waveguide), the CW output was ~9 mW (Figure 1). Lower average pump powers -- less than a milliwatt -- had produced only about 40 µW from the pulsed laser.Figure 1. With a 25-V bias on the PIN diode, Intel scientists observed 9 mW from their CW silicon Raman laser. The roll-off at higher power results from the generation of free carriers at a rate faster than they can be swept out by the bias on the PIN diode. The slope efficiency for the 25-V bias (prior to the roll-off) is 4.3 percent. ©Nature Publishing Group.A key to the success of the earlier laser was an integrated PIN diode that swept photon-absorbing free carriers out of the laser resonator. The bias on this diode played a crucial role in laser operation. In refining the design for CW operation, the scientists also made improvements to the diode. They note that, in the future, the bias on the diode could be modulated to directly modulate the output power of the laser.That is precisely what the University of California researchers did to modulate their laser, except they applied a forward bias to a diode integrated into the laser gain medium to generate free carriers, which increased the intracavity loss and extinguished the laser. Their previous laser normally avoided free carriers by operating in the very short pulse regime so that the Raman pulse was generated and exited the laser before there was a significant buildup of free carriers.By turning the current through the diode on and off with a 200-ps rise/fall time, the researchers observed a rise time of about 1 µs and a fall time of about 500 ns in the laser's output (Figure 2). These switching speeds are too slow for many telecom and information-processing applications, and they result primarily from the long buildup and drain-out time for photons in the physically long (~15-m optical length) fiber optic resonator. The scientists foresee replacing the fiber resonator with a microring resonator. Lead researcher Bahram Jalali expects that data rates of 10 Gb/s eventually can be obtained.Figure 2. By flooding the intracavity optical path with photon-absorbing free carriers generated by a forward-biased diode, researchers at the University of California, Los Angeles, modulated their silicon laser with rise and fall times on the order of 1 µs and a modulation depth of 30 dB. The individual spikes correspond to the pump pulses from the mode-locked laser. ©Optical Society of America.After the dust has settled, an overarching question will remain about these Raman lasers: Because they are optically pumped, do they really provide a potential solution to the photonics/electronics integration problem? One still needs a semiconductor laser, fabricated from exotic materials, to get things started."[These lasers] will not replace conventional diode lasers used in DVD/CD players or telecom equipment," Jalali acknowledged. Instead, he envisions the silicon Raman laser becoming a powerful and rugged source of mid-infrared radiation, for which applications abound -- from molecular spectroscopy to optical communications to military countermeasures.Intel is in none of these markets, and it sees the silicon Raman laser as enabling the continuation of Moore's law into the next generation of computers. "Think of [the] power supply that exists in every PC," said Mario Paniccia, director of the company's Photonics Technology Lab. A single optical power supply similarly could power dozens of silicon Raman lasers in a computer.In the long term, Paniccia sees the silicon Raman laser enabling telecom applications such as routing, switching, amplification and wavelength conversion, as well as many other applications in computing, medicine and spectroscopy."CW operation opens entirely new opportunities for a silicon laser that previously did not exist," he stated.