By J. Robert Lineback, Senior Technical Editor
Ready or not, the semiconductor industry is attempting to work up the gumption and early development strategy for the next big wafer challenge: 450mm diameter silicon substrates. Nearly all experts agree that the industry is dragging its collective heels in organizing an effort to define and create R&D programs needed to put 450mm wafers into production by 2011-2012, which is the official target of the current International Technology Roadmap for Semiconductors (ITRS).
But even the ITRS itself isn’t completely clear about the timing of putting 450mm wafers into production. Various chapters in the ITRS show 450mm wafers being introduced by 2010, 2011, 2012, and 2015. The industry roadmap concludes that if 450mm wafers were to be launched by 2011, wafer-manufacturing development should have begun in 2003. That really didn’t happen, but efforts are now underway to start up some early activities in 450mm planning.
Chip consortium Sematech Inc., which oversees annual updates to the ITRS, has begun the groundwork to identify issues for manufacturing on 450mm substrates, said Scott Kramer, director of the International Sematech Manufacturing Initiative (ISMI) subsidiary. While speaking at the first ISMI Symposium on Manufacturing Effectiveness in Austin, TX, in late October, Kramer said early planning efforts have begun to look at funding and business models for the next wafer size as well as the viability of a Sematech-led international program to help launch 450mm wafers.
Specifically, ISMI is exploring the possibility of initiating a program similar to Sematech’s i300i effort, which cooperated with Japan’s Selete consortium to set standards and benchmark 300mm equipment during the late 1990s. While i300i has been widely credited with helping to establish international cooperation in 300mm wafers, automation, and tool sets, the program also became entangled in a major dispute between chipmakers and unhappy fab equipment suppliers that had invested more than $4 billion in 300mm technology but found few takers in the late 1990s. Demand for 300mm technology was delayed in the late 1990s by a combination of uncertain market conditions and the ability to accelerate device shrinks on 200mm wafers. It wasn’t until the 2002-2003 recovery that 300mm demand began to build strong momentum.
Many industry analysts now believe the next wafer size generation has less than a 50% chance of getting off the ground. “The big issue with 450mm is: Who will pay for equipment and process development?” said Bob Johnson, principal analyst at Gartner Inc.’s Dataquest unit in San Jose, CA. “With the industry consolidating and the huge amount of money it will take to develop tools at a new wafer size, there are only going to be a few companies with the resources to do it. These companies are going to be asking [chipmakers] to make the effort worthwhile.”
In organizing industry cooperation for 300mm programs, IC manufacturers “tried to push more development down to the equipment companies,” Johnson added. “Many suppliers are not willing to admit it, but they simply are not interested in 450mm.”
The lack of collaboration between chipmakers and equipment suppliers in setting targets for the i300i program was a problem that probably won’t be repeated, said analyst G. Dan Hutcheson, president and CEO of VLSI Research Inc., Santa Clara, CA. In particular, he said, i300i did not back “bridge-tool” platforms which could be used on 200mm and 300mm wafers. Resistance to bridge tools drove up R&D cost because tool suppliers had to work on separate 200mm and 300mm platforms for each process node, Hutcheson noted.
“If they do something like ‘i450i’ and it’s a true partnership, 450mm could be enormously successful,” said Hutcheson. “Done right, the cost of 450mm development could total $10 billion. If it’s done wrong, I don’t think it will happen.
“The biggest issue first is whether the chip companies can generate a consensus [about 450mm]. Right now, there are a lot of skeptics in the chip industry,” he added.
During the Austin symposium, ISMI’s Kramer was candid about the doubts surrounding 450mm wafers. In fact, he was somewhat hesitant to call the next generation wafer size “450mm” and instead simply called it “the next wafer size.” Kramer outlined a number of market and industry factors that could drive the timing of larger wafers sooner or later. He said the acceleration or deceleration of process nodes and device shrinks could move out or pull in the need for larger wafers.
But assuming historical growth rates for IC units and the lack of disruptive technologies, Kramer cited two scenarios that show a need for hundreds of 300mm fabs by the second half of the next decade. One projection showed 400-500 300mm fabs would be needed worldwide by 2018, if 450mm wafers failed to take off. Even if 450mm wafers were to ramp into production in 10 years, “several hundred 300mm fabs will still be needed,” Kramer said. Many industry analysts question these two scenarios, suggesting that the number of 300mm fabs in 10-15 years could be half the projected amounts, but no solid estimated are available.
“Assuming that the next wafer size goes into full-scale production in 2012 and assuming that the industry uses a [cooperative] model similar to the one experienced at 300mm, a program like i300i within ISMI might take place about five years before that — which would be 2007,” Kramer told the symposium audience. That’s three years from now, but he emphasized the need to begin the planning today, in 2004. “Some of this work has begun within ISMI now — modeling work and work in conjunction with the ITRS effort to look at high-level alternatives and to identify the issues needed to put the wafer size in production.” — J. Robert Lineback