Issue



Resistance at 450mm


04/01/2006







According to the 2005 International Technology Roadmap for Semiconductors (ITRS), the industry should plan on shifting from 300mm to 450mm wafer production by the year 2012. But adopting the change by this date appears to be overly optimistic, at least according to presenters addressing the subject at the recent Industry Strategy Symposium (ISS) at Half Moon Bay, CA.

Although the transition to 450mm wafers will not take place for many years, deciding exactly when and how to implement the change was among the most pressing and contentious issues raised at the conference. Indeed, while major chipmakers lobby for larger wafers as a means to improve their competitive advantage, the rest of the industry, particularly equipment suppliers, are likely to support the change later rather than sooner because of the associated development costs.

In terms of equipment development, the total amount spent doubled during the transitions from both 125mm to 150mm and from 150mm to 200mm, according to VLSI Research. But during the most recent transition from 200mm to 300mm wafers, tool costs jumped nearly tenfold. If this level of spending is extrapolated linearly, the cost of moving to 450mm would be about $100 billion. Moreover, if spending were to rise at the same rate of change, the figure would approach $1 trillion, obviously far exceeding the industry’s resources.

Another way to examine the trend is to compare the relative areal cost of capacity for the last three wafer sizes. The analysis shows that while equipment development costs rose gradually from $15/sq cm for 150mm wafers to $17/sq cm for 200mm, they more than doubled to nearly $35/sq cm for 300mm wafers, according to VLSI Research. Yet, during the same period, fab costs also rose from about $7/sq cm for 150mm to $12/sq cm for 200mm, but suddenly dropped to < $10/sq cm for 300mm wafers.

This comparison indicates that chipmakers suddenly began building fewer fabs and spending less money per square inch of silicon, explained ISS featured speaker Dan Hutcheson, VLSI Research analyst. “That’s why equipment industry revenues have been flat, and why the chipmakers think that 300mm is such a great deal.”

Sharing that view from the equipment-supplier vantage point, ISS keynote speaker Mike Splinter, CEO of Applied Materials, analyzed options for how the industry could spend its development dollars and determined that the benefits of further wafer size increases are shrinking. “We should focus on things other than wafer size conversion and dismiss a shift to 450mm wafers in the current time frame,” he proclaimed. In fact, comparing the costs and benefits of investing R&D dollars in future technology versus the next substrate size, Splinter showed that while both approaches would significantly increase the number of devices per wafer, focusing on a substrate size change would entail significantly higher capital and variable costs as well as higher implementation risks.

So when will the shift to 450mm wafers occur? It’s unlikely that it will take place according to the ITRS timetable. In fact, looking back at the shift to 300mm wafers and extrapolating the development times, it’s clear that it will be virtually impossible to pull it together in that time frame, said Hutcheson. Moreover, to build a new lithography platform, alone, typically takes up to 10 years, so there’s no way the work could be completed by 2012.

A more reasonable scenario is that the transition will occur much later, perhaps sometime between 2020 and 2025, according to VLSI’s analysis. That should allow enough time for companies to get funding, develop standards, and the like.

Even so, we as an industry will need to avoid the mistakes committed during the transition to 300mm wafer production. Success will depend on a shared vision and commitment among chipmakers and tool suppliers. “We wasted way too much time fighting each other and focusing on what the cost was going to be,” Hutcheson concluded. Instead, it will boil down to planning, execution, and how well we can work together, he said. “The equipment guys are not going to pay for it this time.”

Click here to enlarge image

Phil LoPiccolo
Editor-in-Chief