 |
|
||
|
||||||||||||||||
Toyoda Gosei and TridonicAtco to Establish Joint Venture for Development and Production of White LEDs |
||||||||||||||||
|
||||||||||||||||
Commentary & Perspective...
The Long Awaited DARPA GaN AwardsMarch 14, 2005...In 2003, then again in 2004, rumors were that DARPA was going to have a GaN program that rivaled the old $5xx (?) million "MIMIC Program" that put GaAs monolithic microwave IC (MMIC) technology on the USA's military map. GaAs epi-based HEMTs and HBTs began their rise to what is now an incredible installed base in both military and commercial systems throughout the world. The MIMIC program, which was run by DARPA back when "dual use" was a DoD mandate in the USA, meaning all target applications had to be equally relevant to civilian and military use, resulted in fueling a then-fledgling compound semi industry. And DARPA, in the name of the American taxpayer, earned the well deserved credit for an admirable "mission accomplished." About the only folly of that program was its name. "MIMIC" was the acronym for the program. "MMIC" and pronounced exactly the same as the established acronym for the technology, "MMIC". They both meant basically the same thing. When industry insiders heard rumbles that a similar initiative for GaN for electronic applications was on DARPA's radar screen, everyone became extremely excited. While GaN was doing great in the LED sector and brightening the entire blue spectrum, electronic applications for this particular wide bandgap (WBG) semiconductor family was sorely in need of an R&D boost. DARPA had invested heavily in SiC through the years, and I have a hunch that was at the heart of the delay. Couple that with a severe depression in the communications industry and the fact that it's difficult for even the military branches to find budget monies while the US Congress knew good and well that the American deficit spending was reaching an all time high. Behind the scenes, GaN was gaining on SiC in the labs, with companies like Nitronex showing impressive reliability data on their GaN on Silicon runs. Nitronex co-founder Kevin Linthicum reported eye-opening X-Band MMIC results at Outlook in December. Nitronex 100-mm runs of X-Band GaN on silicon at TriQuint's two stage X-band pHEMT foundry were extremely impressive! And that work was done with no backing by DARPA whatsoever. Nitronex proved that they could run 451 MMICs per wafer and they produced metrics of: Vd: 30V, Pout: 25.7W (2.3 W/mm) PAE: 20.7% and Gain: 14.8dB. While it wasn't likely that GaN on silicon would get a crack at the taxpayer pie, GaN on SiC performance by epiwafer fab pros at companies like Emcore and Cree was enough to make prime contractors like Northrop Grumman and Raytheon want to get in on the act. A major WBG initiative was in the wings once the giant defense contractors got in on the act. Raytheon was one of the GaAs MIMIC primes, co-oping with Texas Instruments. Here's where the history gets interesting for those of you not on the scene at the time. TI dropped their compound semi work in favor of DSPs and TriQuint picked up the group. Thus TriQuint Texas was born out of MIMIC. Northrop Grumman was the first company the DoD funded to make SiC substrates, eventually selling off their SiC group. The other MIMIC program prime was TRW, which eventually became part of Northrop Grumman. So here we have the best of the old guard.... three primes who have a history of spending DARPA's funds fairly efficiently: Raytheon, Northrop Grumman, and TriQuint. Of the three, only TriQuint is a merchant foundry. That means that whatever they produce will be openly used by their foundry customers. NG and Raytheon will undoubtedly again stay "closed" and stick to DoD only business. Two very familiar names are key in these contacts, Tom Cordner, VP and GM of of TriQuint Texas and Dwight Streit of NG. Tom was at TI and my prime contact during the MIMIC program and Dwight was my prime contact at TRW during the MIMIC program. If Jim DiLorenzo was still at Raytheon, it'd be like old home week. I was the only technology reporter to track the MIMIC Program, I've been a major cheerleader for the WBGs, so I guess I better gear up to track this follow-on DARPA GaN on SiC Wide Bandgap Semiconductor Initiative, which is its official name and how it should be spelled. That underscored blue link will take you to DARPA's official postings on the program. The first to announce their award was Northrop Grumman on March 9th (ref: our headline news coverage). Due to the brevity of the NG news release, at first I thought this was simply another token award for GaN by DARPA. Then TriQuint's release came over the wire and I called my good friend Tom Cordner for details and confirmation that this may really be the start of something big... the something we've all been waiting impatiently for. Here was relatively little TriQuint playing toe-to-toe with two huge defense contractors. Impressive. According to Tom, the key reason TriQuint went for this contract, that actually began in February with TriQuint footing the costs directly until the DARPA monies kicked in, is because of their open foundry offering policy. Whereas the other DARPA GaN on SiC contracts now kicking in are led by closed manufacturers, Raytheon and Northrop Grumman, TriQuint's foundry is truly open to everyone, which appeals to two of TriQuint's partners in this extraordinary and exciting endeavor, Lockheed Martin and BAE Systems, two very large USA defense contractors, who normally compete with one another (thus proving the strength of the open foundry concept). TriQuint is already BAE's commercial foundry of choice and they've worked closely with Lockheed in the past as well. Both bring especially interesting and potentially lucrative systems applications to the GaN on SiC HEMT program. The supply side of the contract team includes II-VI Inc., which will provide the SiC substrates. II-VI's Wide Bandgap Materials (WBG) Group has two SiC wafer production facilities, one in Pine Brook, New Jersey and the other in Saxonburg, Pennsylvania. The SiC substrates will then move on to Emcore Corporation's epiwafer foundry in Somerset, New Jersey where some of the best GaNzilla's still reside. Then Nitronex will contribute their reliability knowledge. For the GaN device modeling they will employ none other than the great Michael Shur (who co-founded Sensor Electronic Technologies Inc.(SETI) with Remis Gaska...) SETI is indisputably ahead of any other company at this stage in their GaN device work). "Why II-VI instead of Cree," I asked Tom? Because Cree is on Raytheon's team. We haven't yet received the details from Raytheon, but that contract should be in the same basic ballgame. And we have a call in to Dwight Streit to get the details of who's playing on his team, and will report more when we get it. They each seem to be getting about $16-17 million for the first phase with potentially lucrative follow-on options should they produce as expected. So that would mean about $45 million plus upfront, with double or triple that in the longrun. Coupled with all the millions that have already been doled out to a variety of USA universities, startups and established defense contarctors, in my book, that puts the WBG Initiative in the ballgame with the old MIMIC Program. DARPA and the supporting DoD services and agencies with whom they work are largely run by seasoned professionals who know how to spend the taxpayer's monies wisely, so this should be just the boost GaN electronics needs right now to get it off the ground and deployed in a variety of useful systems. Now... if we can just get the checks to clear the bank. But hey, it's the US Treasury. What's a few more millions in deficit spending? If you have news or
views to share about the compound semiconductor, LED or solid
state lighting industries |
||||||||||||||||
