Time for enlightened self-interest?
10/01/2003
Chipmaking has been able to track Moore's "Law" since the 1960s because of the coordinated efforts of every part of the infrastructure, from wafermaking to patterning and lithography, and on to final test. Chris Mack of KLA-Tencor has pointed out that the industry has been following the trajectory of a classic learning curve (see SST, July 2003, p. 51), lowering costs steadily as the industry gains more knowledge.
It hasn't always been an easy task to keep all parts of the infrastructure on target, which is the reason that the industry has adopted a roadmap projection mechanism that is now worldwide. One goal of the roadmap is to keep all players informed of their particular targets in terms of specifications and timetables. It has proven to be a crude tool, often missing on the technologies that will be needed to meet goals, usually by underestimating the ability of the industry to find clever ways to push existing technology beyond what anyone thought possible.
To keep every part of the infrastructure in tune is an economic problem as well as a technological one, however. This becomes ever more so as the industry pushes toward the nanotechnology range. In the past, some top companies in the semiconductor community looked at the problems ahead, and decided that it would be in their enlightened self-interest to make investments with the companies developing portions of chipmaking technology — the tools, resists, maskmaking, materials, or other key elements.
Research funds from sources such as DARPA in the US went toward future technologies, such as next generation lithography, but the chipmakers knew that critical parts of the needed infrastructure would come from a few companies working to push the state-of-the-art, and that was where they made investments. IBM, which had the biggest cash horde at the time, helped fund the switch to 6-in. wafers, and Intel did the same for the transition to 8-in. wafers. In tough times, IBM even helped Intel, which made the x86 microprocessors at the heart of IBM PCs, by buying 20% of the company (which it later sold off). These targeted investments helped the whole industry to progress, while also serving the interests of the investing company. While IBM and Intel benefited from lower manufacturing costs, they also helped widen the markets for semiconductor-based products. This expanded the market for everyone.
This same approach was not taken for the transition to 300mm wafers. Wafermakers almost went bust, and maskmakers and resist materials suppliers were not far behind. Toolmakers were forced to spend multimillions for one false move to 300mm, and then had to retool all over again at great expense when the shift really came. In more recent times, early investments in organizations like Sematech in the US and Selete in Japan were aimed more at protecting national interests than in pushing global chipmaking capabilities. Because lithography is a pivotal technology, much investment and emphasis has gone into very costly research on next-generation exposure tools, with roadmappers continually projecting the demise of optical lithography. But overemphasizing the tools has provided an expensive lesson to the industry. Optical tools for 193nm were developed well before ancillary technologies, such as resists, were ready, so that costly resolution-enhancement mask technology had to be employed with 248nm tools to meet Moore's Law targets.
Is there a shift from enlightened self-interest to simply self-interest by the companies with the resources that could push development in the weaker parts of the supply chain? Wafermakers skirted bankruptcy to design and build equipment for 300mm wafers. Now if there is a major upturn in electronics markets (it could happen!), chipmakers may face wafer shortages if 300mm capacity falls short. As the industry pushes toward immersion lithography and deals with problems such as data overload in future heavily RET-based maskmaking, shouldn't the chipmakers that will benefit most be helping to ensure that the solutions are there when needed? Intel, for one, is sitting on over $10 billion in cash, but recently shifted to in-house maskmaking, seeing this as a competitive advantage for its own manufacturing. In Japan, a group of top chipmakers are jointly investing in a foundry operation and suggesting that the new proprietary technology they develop for future chips will be offered for sale to others.
The national consortia have moved toward global cooperation in areas such as 300mm standardization because they recognize that the incredible development costs for future chipmaking technology is too huge for even one country or region. Some major companies seem to be looking inward to their own needs rather than ensuring that they will have a strong cadre of future suppliers capable of meeting every facet of their manufacturing needs. A new round of enlightened self-interest might prove to be a more beneficial strategy, for the whole industry as well as their own fortunes.
 |
Robert Haavind
Editor in Chief