元器件 > 高级半导体博客



The Death of GaAs (?)...IMS2013 Panel

by Eric Higham | 6月 24, 2013

I hope that you were all able to attend the IMS2013 in Seattle and everyone got a chance to sit in on the Death of GaAs (?) panel session. I had the distinct pleasure of organizing and chairing a great group of panelists that included my colleague Chris Taylor, Dan Green from DARPA, Rodd Novak of Peregrine Semiconductor, Yongxi Qian with RFaxis, Doug Teeter of RFMD, Dean White of TriQuint Semiconductor, Paul Hurwitz of TowerJazz Semiconductor and David Danzilio of Win Semiconductors. Not surprisingly and perhaps like any group relationship, the participants hit some bumps along the way, but made some good points and in the end everyone was happy!

The main premise behind the panel session is that the cellular segment of the GaAs device market accounts for more than $3 billion dollars and nearly 60% of revenue. This segment used to contain switches and amplifiers (and even converters if you go back far enough). Over the last couple of years, companies like Peregrine Semi have been very successful at changing the switch technology to silicon and in addition to numerous start-up companies, Qualcomm recently threw their sizable hat into the CMOS PA ring with a multi-mode, multi-band offering targeting LTE applications. The question, now is how the GaAs community responds to the silicon challenge with $3 billion of cellular revenue at stake. In the larger sense, the balance of the GaAs device revenue is also seeing competition as technologies like GaN and SiGe evolve and capture market share.

The silicon companies on the panel all showed some very impressive results. Peregrine Semi has been in the forefront of process and performance improvements with their UltaCMOS process that have dramatically reduced die size, while improving all the performance metrics relative to GaAs switches. They have efforts underway to improve frequency range and power handling performance and they have recently demonstrated a CMOS PA prototype aimed at LTE applications. TowerJazz detailed their SoI efforts and the applications for SiGe in amplifiers and high-speed circuitry. RFaxis showed several of their product successes using CMOS for high-volume Wi-Fi applications.

Representing the GaAs side, we had industry giants RFMD, TriQuint and Win Semiconductors. These companies are more than just “GaAs companies” as they have processes and products that use other technologies, in addition to GaAs. I specifically asked TriQuint to focus on GaN, because of their long heritage of success in that area. GaN has appreciable performance advantages and is well entrenched in military applications, but this technology is just now starting to see widespread commercial adoption and developers of this technology are looking to expand the market applications. While GaAs is in the sights of this technology, the main target for GaN is likely to be LDMOS. RFMD has placed bets on most every technology, so they are able to pick the best technology for the function. Both companies stated that there are some applications and circumstances that favor GaAs and some will favor silicon-based technologies. GaAs may do better with less mature, moving target requirements, while silicon may be better with applications having more mature performance requirements that are able to tolerate longer manufacturing cycle times. GaAs foundry leader WIN Semiconductors closed the presentation portion of the panel by eloquently reminding us that every purchasing decision is based on a number of factors and GaAs is still the choice in a significant portion of these decisions. WIN’s view is the fast evolution of the handset and other electronic devices bodes well for GaAs devices and the death of GaAs is premature!

 So, what was the verdict? There was no clear winner. The performance of silicon-based and GaN-based devices is impressive, but GaAs has a very strong foothold in the market and the manufacturers won’t give this up without a fight! The stakes are high, as evidenced by $5.3 billion in GaAs device revenue in 2012 and it is not surprising that this is attracting the attention of other device technologies. I don’t think the final chapter of this story is anywhere close to being written. One of the results of this competition will be to spur innovation across all the technologies and consumers will be the ultimate winners.

So, thanks, again to all the panelists for their time and efforts and stay tuned as this important question continues to attract close attention from everyone!

-Eric

Previous Post: It’s spring and GaN is in the air! | Next Post: Semi-Insulating GaAs Bulk Substrate Market Hits a Rough Patch
Leave a comment