Lytek Takes On Laser Challenge
Lytek Corp., a small startup from Phoenix, believes it could leapfrog the competition and be first to market with Vertical Cavity Surface Emitting Lasers (VCSELs) operating at 1310 nanometer wavelengths -- a type of laser that should prove to be easier to manufacture and therefore cheaper than standard edge-emitting lasers (see Laser Blazers).
Lytek has recently had a breakthrough in its labs, says founder and CEO Yong-Hang Zhang. The company has produced single-mode devices with output powers of 0.25 milliWatts and a threshold current of 0.5 milliAmps. Preliminary reliability tests show no degradation in performance at 70 degrees Celsius. Lytek will present a post-deadline paper detailing these results at the Photonics West conference in January.
These results are sufficient for the startup to plan sampling of devices early in 2003, with the aim of moving to production in the second quarter of the year, Zhang says.
If Lytek could keep up with this schedule, there's a pretty good chance it would be first to market with 1310nm VCSELs. Despite plenty of promises, wannabe VCSEL vendors have all missed their targets for delivery. In fact, industry pundits tell Light Reading that it's not possible to buy a 1310nm VCSEL today.
As those early hopefuls have already discovered, making the leap from R&D to manufacturing is never as simple as it sounds. So what makes Lytek think it will succeed where others have failed?
Two things have hampered the commercialization of VCSELs. One of them, of course, is the industry downturn, which has forced companies to can projects that won't bring in immediate revenues and also pushed startups into bankruptcy before they've completed product development.
Some of the VCSEL vendors that have gone out of business include: Gore Photonics Inc., which has closed down and is in the process of auctioning off its assets (see Gore to Sell Fiber Optic Biz); and Cielo Communications, which sold its assets to Optical Communication Products Inc. (OCPI) (Nasdaq: OCPI) (see OCPI Acquires Cielo). OCPI plans to continue developing VCSEL technology, but industry insiders question its ability to do that without the expertise of key engineers involved in materials growth, who have reportedly been hired by competitor Picolight Inc..
Industry sources also say that Agilent Technologies Inc. (NYSE: A) had started the process of transferring its 1310nm VCSEL technology from its R&D lab to production, but the company recently backtracked, laying off the six or so people involved in production. The decision reportedly came after Agilent decided to cut all programs that weren't making a profit in an attempt to boost its quarterly earnings results. Agilent Labs spokesperson Cynthia Smith wouldn't confirm this directly, but did say: "The 1310nm VCSELs project remains -- as it has always been -- a long-term research project" (emphasis added).
Rumor also has it that Infineon Technologies AG (NYSE/Frankfurt: IFX), one of the leading manufacturers of short-wavelength VCSELs, cancelled development of 1310nm VCSELs this summer, after having announced a "breakthrough" in 2001 (see Infineon Unveils VCSEL). The research scientists in charge of the project have left the company, according to sources.
Infineon, however, disputes this. "Infineon's work on 1310 nm VCSEL technology continues with an unchanged high priority and is on schedule to be used in a broad range of products that will be introduced later this year," says spokesperson Remee Vargas. "The development team based in Munich is fully intact."
Lytek has benefited from this shakeout, selectively hiring key personnel away from former competitors. It recently filled the position of senior product development manager by hiring Dr. Stewart Field, a former Cielo employee. It has also hired Dr. Bing Liang, former president of AXT Inc.'s VCSEL and optoelectronics division, as its new VP of engineering.
In Zhang's view, the second reason for the delay in bringing 1310nm VCSELs to market is the material. Most vendors in the field -- including Agilent, Cielo, and Infineon -- are developing devices based on a material called "gallium arsenide nitride" (GaAsN). "There is a material lifetime problem [with nitrides]," Zhang contends.
Plenty of vendors are betting that these problems can be solved and continue to work on nitride-based devices. These vendors include Emcore Corp. (Nasdaq: EMKR), E2O Communications Inc., Honeywell International Inc. (NYSE: HON), and Picolight. At Photonics West, a number of other research papers will cite progress in this regard. The bottom line, however, is that there's a lot more work to be done than was anticipated.
Emcore and Picolight did not return calls requesting information on this issue. Picolight staffers were probably too busy restoring its Website (see Picolight Site Hacked).
One of the reasons for Lytek's confidence is because its technology is not based on GaAsN, says Zhang. Instead, Lytek uses antimonide-based material -- and it's probably the only vendor with this approach. "The original perception was that antimonides have an even lower gain than nitrides," he says. "But our theoretical studies show that's not true. It can have very high gain." In other words, it should be possible to produce high-power lasers with this technology.
The manufacturing process is virtually identical to that for 850nm VCSELs -- devices currently in production -- making it very low-cost, he adds. Lytek is claiming virtually 100% yield on a wafer, with power fluctuation from device to device on the same wafer of 20%. "I'm told this is close to [the yield of] 850nm VCSELs a couple of years ago," says Zhang.
Equally importantly, Lytek claims that its devices have shown no signs of failure in testing so far -- proving that they won't suffer from the same problems as nitrides. "Based on our preliminary reliability tests, we have not seen any degradation in output power," Zhang claims.
He does acknowledge, however, that reliability testing hasn't been going on for long enough to be conclusive. So far, Lytek has tested its devices for hundreds of hours at room temperature, and for "a shorter time" at 70 degrees C. Thousands of hours of testing on different batches of wafers will be required before customers have confidence to deploy a device in the field -- and success, at this point, is far from a foregone conclusion.
— Pauline Rigby, Senior Editor, Light Reading