Competing fab strategies rely on advanced automation
08/01/2005
No corner of the semiconductor industry is expanding manufacturing capacity faster than Asia, where rising competition between established players such as Taiwan and the emerging Chinese semiconductor industry are helping to drive growth. The former is rapidly raising state-of-the-art 300mm fabs to maintain its regional leadership in capacity growth, while the latter is refining a comparatively trailing-edge manufacturing infrastructure to tap its own potentially explosive market for both semiconductor devices and certain classes of fabrication equipment.
By some counts, China’s domestic demand for semiconductor chips may reach $50 billion by 2008. If these estimates prove accurate, the Asian mainland could become the largest chip market in the world, surpassing the US and Japan, and justifying the growing interest of investors around the globe. It is this market - rather than China’s manufacturing capabilities - that could make the country a significant player in the industry.
Politically, economically, and technologically, China and Taiwan are worlds apart, but long-range trends indicate they are on a converging, if not competitive, course. Taiwan has a clear technological lead, but China’s consumer market signifies a compelling engine for increased manufacturing capacity. Wherever the chips fall in the coming years, these two countries spell opportunity for equipment suppliers around the world.
China rising
Among China’s biggest backers are the Chinese themselves. Three domestically run companies have either started construction or announced plans to launch five new 300mm plants in the near future. Two of these projects, however, appear to be indefinitely stalled, and a third probably won’t begin production for another 18 months. Of the remaining two fabs, one is currently operating and the other is expected to open its doors by the end of the year.
If China’s 300mm build-out seems restrained, however, consider the rate at which its smaller fabs have sprung up over the past five years: From 2000 to the present, the country’s manufacturing infrastructure went from a single 200mm fab and two 150mm plants to ten 200mm fabs and a half-dozen 150mm plants. Figure 1 illustrates the breakdown of China’s foundry capacity by wafer size and technology.
Many of these 200mm facilities are operating with used or refurbished equipment - a strategy that doesn’t appear to favor technological supremacy. But China’s focus on trailing-edge technology is not entirely by choice. The international community has been extremely cautious about exporting strategic technologies, engineering talent, or cutting-edge equipment to Chinese manufacturers.
When viewed from a different perspective, however, China’s relatively low-end entrance strategy appears more plausible. First of all, it holds the keys to its own singularly expansive domestic market. Second, its late entry into the market provided it with the opportunity to leverage the experience of Taiwan and other industry veterans around the Pacific Rim. Third, an abundant used-equipment market has given China the opportunity to quickly ramp new 200mm fabs with lower equipment costs and speed their return on investment (ROI). Last, advanced isolation technologies and software have significantly extended the processing performance and cost-effectiveness of 200mm manufacturing.
Put another way, the biggest difference between the Chinese business model and the semiconductor industry’s traditional approach is that cost-efficient manufacturing takes precedence over implementing the latest, greatest manufacturing technology. This departure from tradition is minor, even typical for a new player trying to break into a fiercely competitive high-tech market.
Lessons from Taiwan
When Taiwan joined the industry in the early 1980s, it sought to speed ROI by integrating isolation technology with place-and-go automation into its fabs. (Place-and-go automation is the automatic reading of the lot ID, enabling recipe downloads and loading of the SMIF-Pod onto the tool.) This approach led to the early adoption of standard mechanical interface (SMIF) technology, the benchmark for its foundries since the days of 150mm wafer manufacturing. By enclosing wafers inside portable ultraclean environments, Taiwan’s early fabs achieved equivalent Class 1 cleanroom environments for their wafer processes, while maintaining overall levels of fab cleanliness at ISO Class 5 (Fig. 2).
This technique significantly reduced up-front and operating expenses, and enabled Taiwanese fabs to accommodate new generations of design-rule shrinks without expensive facilities changes. Isolation technology led to better yields and enabled fabs to efficiently run multiple lots and recipes. These capabilities formed the bedrock of Taiwan’s 200mm foundry business, and they have evolved to help improve profit margins for its 300mm processes. Today, the early implementation of SMIF technology is one of the most significant advantages that Taiwan fabs have over their counterparts in Asia and other regions.
Taiwan’s problem is that the massive consumer market on the mainland is difficult to resist, much less ignore - particularly as other global players jockey for position there. So the unwritten strategy among Taiwanese players is to migrate their 200mm equipment into China to establish a presence there, and to capitalize on lower production costs. Eventually, these facilities may be upgraded to 300mm capability. But for now, Taiwan’s 65nm processes - possible only with 300mm equipment - are based entirely on its own shores. China, for its part, is happy to implement 200mm technology, thanks in part to a flood of used and refurbished manufacturing equipment on the market. As fabs elsewhere upgraded to 300mm operations or closed unprofitable operations, some 200mm equipment ended up on the selling block.
Estimates for building a new 300mm fab today run around $3 billion, vs. the $1 billion required to raise a 200mm fab with all new equipment. By buying used and refurbished equipment that runs approximately one-quarter to one-third of the cost for new tools, China continues to save capital for potential investment in leading-edge technology. Despite a gradual rise in used equipment prices to compensate for the increased demand, the used equipment market remains competitive, thanks to a wealth of equipment manufacturers and independent companies that offer refurbishing services and used equipment with performance guarantees.
China has also paid heed to lessons Taiwanese fabs learned two decades ago about the value of isolation technology. Although Chinese fabs saved capital by buying used equipment, they recognized the island reaped solid financial returns by investing in SMIF technology. Today, SMIF-based isolation methods are the de facto standard for nearly all of China’s 200mm plants.
Implementing used equipment on this scale is not a trivial task, given the disparate origins and generations of used processing tools. Although data collected from these tools often share the same exchange protocol, they embody different and often proprietary formats. This presents a greater challenge for 200mm fabs than it does for Taiwan’s 300mm processes, which are compatible with equipment information bridge (EIB) technology based on Semi’s new Interface A standard. This standard allows tool modeling to use data from different hardware platforms and software languages from an entire processing line. In addition to optimized tools, it enables processes such as e-diagnostics, “on-the fly” tuning, and real-time statistical process control.
Taiwan’s 300mm fabs have leveraged Interface A technology and advanced material-handling systems to improve overall production efficiencies and increase profit margins. The benefits, indeed the necessity, of such automation technologies increase as capacity nears 100% because automation’s underlying function is to ensure that a foundry can consistently meet or even exceed its projected cycle times.
Leading the pack
Comparing the semiconductor industries in Taiwan and China is like comparing apples and oranges. Although both are based on the foundry model, China is currently geared toward its comparatively low-end, high-volume domestic markets, whereas Taiwan’s chip technology is aimed toward delivering high-end chips to global markets, including the mainland. If China’s foundries hope to attract global market share from Taiwan, they will need to provide competitive technology at competitive market value pricing. Today, chips produced in Chinese foundries do not exact the country’s 17% tariffs, giving domestically produced chips a tremendous cost advantage to mainland consumers, and providing Chinese fabs an edge on the massive market there.
These cost advantages, however, mean less to customers off of the mainland’s shores, and China’s 200mm technology is not likely to attract outside IC designers who wish to transition their chips to smaller geometries. That business will likely stay with Taiwan’s foundries, which all have strong backgrounds in fabless production, allowing them to quickly ramp designs and turn out advanced chip technologies. So it appears likely that for now, Taiwan’s fabs will maintain their lead over China and other regional competitors. Cutting-edge automation methods will continue to help as the semiconductor industry there moves to 300mm.
China’s entrance strategy, however, appears to have put it on the road to success - at least when measured by optimized ROI. Its overall infrastructure continues to receive substantial government backing. In some respects, China’s 200mm fabs are not far behind Taiwan’s state-of-the-art facilities. Their widespread implementation of SMIF technology has improved manufacturing capabilities and cost performance. Plus, the support of process development by equipment vendors has helped China come on-line quickly and leapfrog technology nodes. The country has effectively covered ground in five years that took Taiwan’s leading fabs a decade. Remaining hurdles are the political climate and the lack of engineers who are capable of running 300mm fabs.
For now, 200mm manufacturing may well continue to be a great niche market for China’s chipmakers. SMIF technology is more than capable of satisfying the current internal needs of the country’s rapidly growing IC market. Also, the domestic demand for consumer electronics emphasizes devices with smaller areas and higher linewidths, which may not benefit from 300mm wafer sizes. Figure 3 shows the division of China’s semiconductor market by device from 2004 and to 2008.
 Figure 3. China’s semiconductor market by device: 2004 vs. 2008. (Source: Gartner Dataquest, “China Offers Opportunity for Its Foundry Industry,” K.-Y. Tan, Feb. 2, 2005) |
It’s probable China’s end markets will evolve rapidly, with manufacturing capabilities not far behind. It won’t be hard to find players willing to invest $3 billion to build more 300mm fabs. If, and when, that happens, the next Asian market leader will be the one who best leverages automation technologies to optimize capacity and ROI.
Acknowledgments
Asyst is a registered trademark of Asyst Technologies Inc. Asyst SMIF-Pod is a trademark of Asyst Technologies Inc. All rights reserved.
Contact Richard Dow at Asyst Technologies Inc., 48761 Kato Rd., Fremont, CA 94538; ph 510/661-5000, e-mail [email protected].
Norma Riley and Kenneth Lo, Asyst Technologies Inc., Fremont, California