Technologists Debate 450MM Wafer Transition at SEMICON Taiwan

Bookmark and Share

Technologists Debate 450MM Wafer Transition at SEMICON Taiwan

Cost High, Benefits in Doubt

TAIPEI, Taiwan – September 12, 2006 – Economic and cost barriers, rather than technical challenges, are more likely to slow down technology advancement in future, according to speakers at the SEMICON Taiwan 2006 CTO Forum.

By 2012, the high cost of keeping up with Moore’s Law could create an R&D gap of $20 billion in the semiconductor equipment and materials industry. A move to 450mm wafers would exacerbate that situation, some speakers said.

“The industry is already tight on resources and will get tighter even without 450mm,” said Iddo Hadar, chief technology office and chief marketing officer for the Foundation Engineering Group at Applied Materials. “We need to be more frugal in how we allocate them.”

Jack Sun, vice president of R&D for TSMC, singled out mask costs as an area that has gotten out of control, with a 90-nanometer mask set costing $1 million. He added that mask inspection tool costs were “astronomical”, in the tens of millions of dollars.

In the past, physics was the limit to technology advancement in the industry. “If we are not careful then cost and economics will limit us first,” said Sun.

The rising complexity and cost of lithography may be arrested with the introduction of extreme ultra-violet (EUV) technology, according to Martin van den Brink, executive vice president, technology and marketing, ASML. “Successful EUV seems to be the lowest cost promise,” he said. “The problem with EUV is that to make it work is quite a challenge. It is not a done deal yet.”

During the CTO forum much of the debate over the high cost of R&D centered on whether or not the industry could afford a transition from 300mm to 450mm wafers.

“A diameter transition at this time would be detrimental to the industry,” said Shakar Sadasivam, senior vice president, R&D, for MEMC Electronic Materials. Instead of moving to larger diameter wafers, the industry should “use physics to maximize efficiency and lifetime of existing platforms,” he said.

Sadasivam said 200mm wafers were still the “workhorse” of the industry, accounting for 52 percent of all wafers in 2005. By 2008, 200mm will still account for 43 percent of the total wafer market.

The shift to a consumer driven electronics market has resulted in a higher product mix for fabs and shorter life cycles. These requirements are better served by more productive use of existing 300mm fabs, according to Hadar. This concept, dubbed “300mm Prime”, is suited to small lot manufacturing, tight equipment characterization, intelligent system control and universal single wafer processing. Hadar pointed out that 450mm wafers provide no overall benefit in performance and would not necessarily be a cost driver. “450mm benefits are limited to microprocessors,” he said.

A slightly different perspective on the 450mm debate was given by the TSMC CTO. “I would love to see a new wafer size if it can be done cost effectively,” said Sun. “The single product, high volume guys will benefit first, and then it will trickle down to the others,” he said.

Charles Kau, president of DRAM maker Inotera and moderator of the CTO forum, pointed out that the optimum output for a 300mm fab was 80K wafers/month. For a 450mm fab, the optimum could be in the range of 120K to 150K wafers per month, resulting in an investment of $12-$15 billion per fab. Very few single companies could afford that type of investment, resulting in more collaborations among device makers in the 450mm era.

In any case, a move to 450mm wafers is unlikely to happen any time soon, according to MEMC’s Sadasivam. Based on the timetables of past wafer diameter transitions, 450mm won’t happen until 2020, he noted.

SEMICON Taiwan 2006 runs September 11-13 at the Taipei World Trade Center, Taipei, Taiwan.

SEMI is a global industry association serving companies that provide equipment, materials and services used to manufacture semiconductors, displays, nano-scaled structures, micro-electromechanical systems (MEMS) and related technologies. SEMI maintains offices in Austin, Beijing, Brussels, Hsinchu, Moscow, San Jose (Calif.), Seoul, Shanghai, Singapore, Tokyo and Washington, D.C. For more information, visit


Scott Smith/SEMI