GaAs Foundry's Pure-Play to Win
Yu-Chi Wang received his BS degree in physics from the National Central University, Chung-Li, Taiwan, in 1989, and his PhD degree in materials science and engineering from Rutgers University, New Brunswick, NJ. He joined Bell Laboratories, Lucent Technologies, Murray Hill, NJ, where he was a Member of the Technical Staff, involved in the compound semiconductor device design and process development for wireless and fiber-optic ICs. He later joined WIN Semiconductors Corp., Taoyuan, Taiwan. He is currently the General Manager of the MMIC Business Unit at WIN Semiconductors.
MWJ: Back in 2007, there was a shake-out in the pure-play GaAs Foundry with the departure of Suntek, GCTC and Knowledge*On. How is the 2010 pure-play GaAs foundry business today compared to the conditions three years ago?
YCW: WIN is continuing to grow very strongly. The market itself is growing, but we are also increasing our market share. The last few years have shown that the pure-play foundry model does work for GaAs and we showed excellent growth in 2007, 2008 and even 2009. We feel we are well placed to show continued double digit growth for the next three years. There is space in the market for a few suppliers and we feel that WIN has attained the scale necessary to maintain our growth rate and continue to gain market share over the next few years.
MWJ: The position of a foundry is often measured by its capacity. How does WIN’s capacity and technology measure up against other GaAs foundries?
YCW: WIN has been the largest pure play foundry for several years now, both in terms of wafer production and revenue. Our second factory, Fab B, is now undergoing an expansion and this added capacity will make us one of the leading GaAs manufacturers in terms of wafer shipments, This level of capacity has advantages for our customers as economies of scale are driving down costs, and the ability to produce wafers in either of our factories provides increased flexibility and security of supply. WIN’s substantial capital investment also allows us to develop and offer the industry a suite of MMIC technologies to meet a variety of markets across a very broad range of customer applications from 1 to 100 GHz. For example, our portfolio includes a high voltage, linear HFET process optimized for CATV markets, a range of 2um HBT for PA MMICs for GSM, CDMA and WLAN applications, 0.5um D-mode process for high throw handset switch applications, and a range of pHEMT technologies for EW and infrastructure, from 0.5um down to 0.15um. We released our BiFET technology H2W (ED pHEMT plus HBT) that is gaining traction for integrated front end modules. Our future portfolio includes a 1um HBT process, with an integrated varactor, optimized for VCOs and a new 0.1um pHEMT. Not many foundries can match our range of technology, all on 150mm wafers.
MWJ: Do you know what the current size and expected grow rate of the global GaAs foundry market over the next four to five years?
YCW: The market data WIN has shows the GaAs foundry business at approximately $320 M in 2008 – roughly 30% year-on-year growth. 2009 began as a down year, and for most IDM companies ended down relative to 2008. Despite this, the GaAs foundry market grew to about $375M in 2009. The projections are for CAGR of 9% over the next 4 years, and we expect to grow faster than the market.
MWJ: What applications are consuming the most GaAs products?
YCW:The main driver for GaAs remains mobile communication devices. There is an ongoing increase both in unit sales and GaAs content per unit, and this is reflected in the continued growth of WIN’s business.
MWJ: What are the major benefits of a chip market based on a fabless strategy?
YCW: Our customer’s benefit from the broad range of WIN technologies as well as the competitive pricing we can offer given our manufacturing scale. We now have a variety of optimized HBT, BiFET, pHEMT and HFET technologies that few other fabs can support or offer the market. Additionally, WIN remains a pure play foundry and this business model eliminates any concerns regarding competing with our customers. Of course, the benefit of not having to cover the fixed costs of an internal fab also helps drive profitability for our customers.
MWJ: How does the Silicon vs. GaAs battle play out over the next few years? Where do you see losing or gaining some ground?
YCW: Silicon continues to advance the state of the art and we are seeing low-cost, lower performance amplifiers and switch products entering the market. Competition for slots is intense and WIN continues to innovate and offer advancements in device performance and manufacturing efficiency, which translates to more competitive products for our customers. Ultimately the market will chose the best value for a particular application and WIN is focused on enabling our customers to win those slots with GaAs products.
MWJ: Your process roadmap calls for introducing a new WINHBT4 High efficiency/linearity process. What kind of performance gains will foundry customers see from this new technology?
YCW: This is an advanced technology for handset and WiFi applications, targeting 3.5G and 4G markets. The process will be a new offering for WIN from the ground up, including improved epitaxial designs for best in class linearity and ruggedness, compact unit cell design and high density capacitors for chip size reduction, thick plated metal interconnect for low loss, and Cu pillar technology for flip chip assembly.
MWJ: How about for your 0.1 um e-beam gate pHEMT?
YCW: We are making fast progress with this technology and have demonstrated market leading performance. Or technologists are continuing to refine and prepare the process for full manufacturing release late in 2010. We are presently engaged with several strategic customers for device evaluation and early prototype development. WIN’s advanced 0.1um pHEMT technology will enable our customers to develop products aimed at very performance, high frequency applications using a cost effective 150mm wafer process platform.
MWJ: Have you had success in optical lithography processes for .25 micron and below?
YCW: WIN has released two stepper based 0.25 μm process – PP25-00 and PP25-10 for Ku- and Ka-band applications. The process flow for both is identical, with just a difference in the epi design to provide higher performance for the Ka-band version. We have a lot of interest in these technologies as a stepper based 150mm foundry process offers a low cost, high volume approach for our customer’s products. There is also global interest in these platforms as they are free of ITAR restrictions, which limits the use of some of our competitor’s technologies. We have more than ten customers now engaged with us at various stages of product development. We predict a substantial level of production on these technologies in 2011 on beyond.
For below 0.25um, we have added a second high speed e-beam direct writer and have adequate capacity to serve the entire industry for the foreseeable future. Electron beam lithography is a low risk, mature technology, providing excellent yield and uniformity at a competitive cost compared to the complex higher risk optical approaches. We have delivered thousands of 0.15um wafers fabricated using e-beam lithography and see it as the technology of choice for high performance and high reliability applications.
MWJ: What markets (applications) are currently experiencing the highest growth?
YCW: We are seeing excellent demand in the high volume markets for handset and WLAN, where we manufacture pHEMT switch products, HBT power amplifiers and products built in our BiFET technology. WIN has positioned our customers to succeed in their markets and the advantage we provide has enabled many of our customers to increase their penetration. As evidence of their success, a number of our customers are increasing their production volumes as their business continues to grow faster than the market. There is also good growth in infrastructure in response to the shift to data rich applications on mobile devices.
MWJ: Do you have design engineers that work directly with your customers, providing design advice, validating active and passive models, etc. and how do you decide which vendors to work with?
YCW: WIN is a pure play foundry and will remain that way – we don’t want to compete with our customers at the product level. As an added service we assist our customers where we can on design issues and product optimization, and provide Process Design Kits in both Agilent’s ADS and AWR’s Microwave Office. We have expanded our team for Design Kit development but also collaborate extremely well with both AWR and Agilent to enhance the functionality of the kits to help our mutual customers succeed in their markets.