03 Apr 2007
Hybrid AlGaInAs-silicon laser specialist Alexander Fang talks with optics.org about how developers are addressing manufacturing issues.
Intel and UCSB have announced major improvements to their hybrid AlGaInAs-silicon laser platform that was unveiled to the world last year. The latest design features a "racetrack" shaped laser cavity that is entirely defined by silicon processing and now also includes on chip photodetectors. (Optics Express 15 2315)
The dream is to come up with an affordable toolkit that gives chip makers the basic building blocks for creating high-speed photonic integrated circuits. By taking a monolithic approach, the group hopes to move away from a dependence on wafer dicing and facet polishing that limited the appeal of its original hybrid laser to manufacturers.
"We wanted to show that our hybrid silicon waveguide can be used as an integration platform," Alexander Fang of UCSB's department of electrical and computer engineering told optics.org. "By placing photodetectors on the output of the laser we can test the chip without the cost of additional fabrication steps."
As well as tackling manufacturing issues, the team has been busy improving device performance. "Last year, the laser had a maximum operating temperature of 40 degC and now it works at up to 65 degC," revealed Fang. "The hope is to have silicon lasers that operate at 85 degC and beyond to reduce the amount of power that is required for thermoelectric cooling."
To make the device, the researchers bonded an AlGaInAs quantum well structure to a low-loss silicon rib waveguide. The electrically-pumped evanescent laser emits up to 29 mW of optical power at a wavelength of 1590 nm.
Fang and his colleagues have also demonstrated a discrete amplifier based on the same waveguide structure, which represents another step towards building practical transmitters and receivers on silicon. (IEEE Photonics Technology Letters 19 230)
Critics are concerned about the scalability of the group's wafer bonding technique, but Fang remains upbeat. "We're in the process of studying the effects of moving to larger bonding areas on bonding quality and device performance," he commented. "From a device perspective, it could be feasible to see our technology in applications like board-to-board interconnects in five years or so."