John Bowers, a professor of electrical and materials at UC Santa Barbara, pioneered a way to integrate a laser onto a silicon wafer fifteen years ago. This technology is now widely used with other silicon photonics devices to replace copper-wire interconnects that once linked servers at data centers. It dramatically increases energy efficiency, which is essential when data traffic grows by approximately 25% annually.
The Bowers group has worked with Tobias J. Kippenberg, at the Swiss Federal Institute of Technology (EPFL), for several years. This collaboration is part of the Defense Advanced Research Projects Agency’s (DARPA) Direct On-Chip Digital Optical Synthesizers (DODOS). The Kippenberg group discovered “Microcombs.” These are a series of low-noise and highly stable laser lines. Each line of the laser comb can contain information, increasing the data that can quickly be sent using a single laser.
Recent demonstrations showed that several teams could create compact combs by placing a silicon nitride ring resonator and semiconductor laser chips close together. The laser and resonator were devices made separately and placed close to each other. This time-consuming and costly process needed to be more easily scalable.
The Bowers laboratory has collaborated with the Kippenberg laboratory to create an integrated on-chip semiconductor resonator and laser to produce a microlaser comb. The labs’ achievement in becoming the first to achieve this goal is described in a paper entitled “Laser Soliton Microcombs Heterogeneously Integrated on Silicon,” published in the latest issue of the journal Science.
Soliton micro combs, which emit optical frequency lines in mutually coherent phases, are optical frequency combs with the ability to produce laser lines that are constant and unchanging relative to one another. This technology can be used in visual timing, metrology, and sensing. Recent field demonstrations include multi-terabit-per-second optical communications, ultrafast light detection and ranging (LiDAR), neuromorphic computing, and astrophysical spectrometer calibration for planet searching, to name several. This powerful tool requires high-power lasers, expensive optical coupling, and sophisticated optics.
Chao Xiang (postdoctoral researcher) explained that a distributed feedback laser produces only one laser line. The line passes through an optical phase control and enters the microring resonator. This causes the power intensity of the light to increase as it travels around the ring. Non-linear visual effects can occur when the intensity exceeds a certain threshold. This causes the laser line to produce two identical lines on each side. Each of these “sidelines” creates another, resulting in a cascade generation of laser-line generators. “You end up having a series of mutually coherent frequency combs,” Xiang stated — and a greatly expanded capability to transmit data.
This research allows semiconductor lasers to integrate with low-loss optical micro-resonators seamlessly. “Low-loss” is because light can travel through the waveguide without losing intensity. Electricity controls the device entirely, and no optical coupling is necessary. The new technology can be scaled commercially because it can make thousands of devices from a single wafer using industry-standard complementary metal oxide semiconductors (CMOS-compatible) techniques. Researchers stated that their approach “paves the way to large-volume, low-cost manufacturing of chip-based frequency combiners for next-generation high capacity transceivers and datacenters, as well as mobile platforms.”
The main challenge when making the device was that the semiconductor laser (which generates the comb) and the resonator (which creates it) had to be constructed on different materials platforms. Lasers cannot be made with materials other than those listed in the Periodic Table’s III and V groups. The best combs are made from silicon nitride. “So, we had the challenge of putting them all together on one wafer,” Xiang said.
The researchers used UCSB’s heterogeneous process to make high-performance lasers on a silicon substrate and their EPFL collaborators’ ability to create record-setting high-Q silicon-nitride microresonators using the “photonic Damascene process.” This wafer-scale process, different from making individual devices and combing them one by one, allows thousands of devices to come out of a single wafer measuring 100 mm in diameter. It also allows scale-up production levels beyond the 200-mm or 300-mm industry standards.
The device must function properly if the laser, resonator, and optical phase are controlled to create a coupled system based on “self-injection locking.” Xiang explained how the micro-resonator partially back-reflects the laser’s output. The laser is locked to the micro-resonator when a particular phase is reached between the laser’s light and the back-reflected light of the resonator.
Back-reflected light is unsuitable for laser performance, but it is essential to generate the micro comb. The laser light lock triggers soliton formation within the resonator. It also reduces frequency instability or laser light noise. This transforms something terrible into something positive. The team created the first integrated laser soliton micro comb on a single chip and the first narrow linewidth laser sources with multiple channels on one chip.
“The field is increasing and is exciting. It has applications in optical clocks, high-capacity optical networks, and many spectroscopic uses,” Bowers, Fred Kavli Chair for Nanotechnology and director of the College of Engineering’s Institute for Energy Efficiency, said. “The missing element was a self-contained chip, including the optical resonator and the pump laser. This key element was demonstrated and should allow for rapid adoption.
Xiang said, “I believe this work will become extensive.” He said that the potential of this technology reminds him of how putting lasers onto silicon 15 years ago helped both research and commercialization of silicon photonics. He noted that Intel had shipped millions of transceiver units annually because this transformative technology was commercialized. Future silicon photonics that uses co-packaged optics will likely be a powerful driver for transceivers with higher capacities and a wide range of optical channels.
Xiang stated that the current comb produces approximately twenty to thirty usable comb lines and aims to increase that number. “Hopefully, one hundred combined lines will be possible from each laser-resonator with low power consumption.”
“Based on the soliton’s low energy consumption and ability to provide a large amount of high-purity optical comb line lines for data communications,” said Xiang. “We believe our achievement could be the backbone to applying optical frequency comb technology in many areas, including efforts in keeping up with fast-growing data traffic and hopefully slowing down the growth in energy consumption in mega-scaled datacenters.”
