Optical Frequency Combs on Wafers

Artist’s idea illustration of electrically managed optical frequency combs at wafer scale. Credit score: Brian Lengthy

Fifteen years in the past, UC Santa Barbara electrical and supplies professor John Bowers pioneered a way for integrating a laser onto a silicon wafer. The know-how has since been broadly deployed together with different silicon photonics gadgets to interchange the copper-wire interconnects that previously linked servers at knowledge facilities, dramatically rising power effectivity — an necessary endeavor at a time when knowledge site visitors is rising by roughly 25% per yr.

For a number of years, the Bowers group has collaborated with the group of Tobias J. Kippenberg on the Swiss Federal Institute of Expertise (EPFL), inside the Protection Superior Analysis Initiatives Company (DARPA) Direct On-Chip Digital Optical Synthesizer (DODOS) program. The Kippenberg group found “microcombs,” a sequence of parallel, low-noise, extremely secure laser traces. Every of the numerous traces of the laser comb can carry info, extensively multiplying the quantity of information that may be despatched by a single laser.

Just lately, a number of groups demonstrated very compact combs by putting a semiconductor laser chip and a separate silicon nitride ring-resonator chip very shut collectively. Nevertheless, the laser and the resonator have been nonetheless separate gadgets, made independently after which positioned in shut proximity to one another completely aligned, a expensive and time-consuming course of that isn’t scalable.

The Bowers lab has labored with the Kippenberg lab to develop an built-in on-chip semiconductor laser and resonator able to producing a laser microcomb. A paper titled “Laser soliton microcombs heterogeneously built-in on silicon,” printed within the new challenge of the journal Science, describes the labs’ success in turning into the primary to attain that aim.

Soliton microcombs are optical frequency combs that emit mutually coherent laser traces — that’s, traces which can be in fixed, unchanging part relative to one another. The know-how is utilized within the areas of optical timing, metrology and sensing. Latest subject demonstrations embrace multi-terabit-per-second optical communications, ultrafast gentle detection and ranging (LiDAR), neuromorphic computing, and astrophysical spectrometer calibration for planet looking out, to call a number of. It’s a highly effective device that usually requires exceptionally excessive energy and costly lasers and complex optical coupling to perform.

The working precept of a laser microcomb, defined lead creator Chao Xiang, a postdoctoral researcher and newly minted Ph.D. in Bowers’s lab, is {that a} distributed suggestions (DFB) laser produces one laser line. That line then passes by way of an optical part controller and enters the micro-ring resonator, inflicting the facility depth to extend as the sunshine travels across the ring. If the depth reaches a sure threshold, non-linear optical results happen, inflicting the one laser line to create two further, equivalent traces on both aspect. Every of these two “aspect traces” creates others, resulting in a cascade of laser-line technology. “You find yourself with a sequence of mutually coherent frequency combs,” Xiang mentioned — and a vastly expanded capacity to transmit knowledge.

This analysis allows semiconductor lasers to be seamlessly built-in with low-loss nonlinear optical micro-resonators — “low-loss” as a result of the sunshine can journey within the waveguide with out shedding a major quantity of its depth over distance. No optical coupling is required, and the machine is fully electrically managed. Importantly, the brand new know-how lends itself to commercial-scale manufacturing, as a result of 1000’s of gadgets could be produced from a single wafer utilizing business customary complementary metallic oxide semiconductor (CMOS)-compatible methods. “Our method paves the best way for large-volume, low-cost manufacturing of chip-based frequency combs for next-generation high-capacity transceivers, datacenters, house, and cell platforms,” the researchers said.

The important thing problem in making the machine was that the semiconductor laser and the resonator, which generates the comb, needed to be constructed on totally different materials platforms. The lasers could be made solely with supplies from the III and V teams on the Periodic Desk, comparable to indium phosphide, and the most effective combs could be made solely from silicon nitride. “So, we needed to discover a technique to put them collectively on a single wafer,” Xiang defined.

Working sequentially on the identical wafer, the researchers leveraged UCSB’s heterogeneous integration course of for making high-performance lasers on silicon substrate and the power of their EPFL collaborators to make report ultra-low-loss high-Q silicon nitride micro-resonators utilizing the “photonic damascene course of” they developed. The wafer-scale course of — in distinction to creating particular person gadgets after which combining them one after the other — allows 1000’s of gadgets to be produced from a single 100-mm-diameter wafer, a manufacturing degree that may be scaled up farther from the business customary 200-mm- or 300-mm-diameter substrate.

For the machine to perform correctly, the laser, the resonator and the optical part between them have to be managed to create a coupled system based mostly on the “self-injection locking” phenomenon. Xiang defined that the laser output is partially back-reflected by the micro-resonator. When a sure part situation is achieved between the sunshine from the laser and the back-reflected gentle from the resonator, the laser is alleged to be locked to the resonator.

Usually, back-reflected gentle harms laser efficiency, however right here it’s essential for producing the microcomb. The locked laser gentle triggers soliton formation within the resonator and reduces the laser gentle noise, or frequency instability, on the identical time. Thus, one thing dangerous is reworked right into a profit. Because of this, the workforce was capable of create not solely the primary laser soliton microcomb built-in on a single chip, but in addition the primary narrow-linewidth laser sources with a number of obtainable channels on one chip.

“The sphere of optical comb technology may be very thrilling and shifting very quick. It’s discovering purposes in optical clocks, high-capacity optical networks and lots of spectroscopic purposes,” mentioned Bowers, the Fred Kavli Chair in Nanotechnology and the director of the Faculty of Engineering’s Institute for Vitality Effectivity. “The lacking ingredient has been a self-contained chip that features each the pump laser and the optical resonator. We demonstrated that key ingredient, which ought to open up speedy adoption of this know-how.”

“I believe this work goes to turn out to be very huge,” mentioned Xiang. The potential of this new know-how, he added, reminds him of the best way placing lasers on silicon 15 years in the past superior each analysis and industrial commercialization of silicon photonics. “That transformative know-how has been commercialized, and Intel ships hundreds of thousands of transceiver merchandise per yr,” he mentioned. “Future silicon photonics utilizing co-packaged optics will seemingly be a robust driver for higher-capacity transceivers utilizing a lot of optical channels.”

Xiang defined that the present comb produces about twenty to thirty usable comb traces and that the aim going ahead can be to extend that quantity, “hopefully to get 100 mixed traces from every laser-resonator, with low energy consumption.”

Based mostly on the soliton microcombs’ low power use and their capacity to supply a lot of high-purity optical comb traces for knowledge communications, mentioned Xiang, “We consider that our achievement may turn out to be the spine of efforts to use optical frequency comb applied sciences in lots of areas, together with efforts to maintain up with fast-growing knowledge site visitors and, hopefully, gradual the expansion of power consumption in mega-scale datacenters.”

Reference: 1 July 2021, Science.

By Rana

Leave a Reply

Your email address will not be published. Required fields are marked *