A stop along the way to 450 mm silicon


By Hank Hogan

In late May, semiconductor manufacturers confirmed that the jump from 300- to 450-millimeter wafers was too broad to be covered in one step. So, they committed to developing something in between, 300 mm prime (300P), to ease the transition. When complete, the 300P guidelines will impact the design and operation of new cleanrooms. However, there are some questions about whether or not it’s possible to pull it off.

Joe Draina is associate director of International SEMATECH Manufacturing Initiative, ISMI (Austin, Texas), the manufacturing consortium that’s spearheading 300P development. He notes that one of the differences between 300 mm wafer fab designs and those of 300P lies in their relative suitability for present and future wafer sizes. “We want optimum,” elaborates Draina. “It needs to be scalable to 450, applicable and the right thing for 450 in a 300 mm fashion.”

Scott Kramer is the director of International SEMATECH Manufacturing Initiative, ISMI (Austin, Texas).
Click here to enlarge image

Standing in the way of this, however, are a host of complications, which have to be dealt with now. For example, the transition from 300 to 450 mm may take place as early as 2012 or as late as 2020 or beyond. Given the time it takes to design and build a fab, as well as its operating lifetime, new cleanrooms must soon account for that switch.

The good news is problems of past transitions aren’t likely to reappear. Because 300 mm fabs use automated material handling systems (AMHS) to move wafers around, the required overhead space and necessary ceiling load bearing for 450 mm AMHS have largely already been accounted for. Since wafers are now sealed in pods, the overall contamination level of the cleanrooms also shouldn’t change with the wafer size switch.

However, the need to maximize efficiency and flexibility means that 300P will require wafers to run at high volume in a small-lot, high-product-mix environment. This implies faster transport mechanisms, possibly different stocking scenarios, and more sophisticated scheduling. Achieving high volume and mix won’t be easy, says Scott Kramer, director of ISMI, but notes the problem is being addressed. “We have fab simulations that are an integral part of our 450 millimeter program to test out the various scenarios and combinations,” he says.

To pull off this balancing act, one possibility is a move from batch to near single-wafer processing. This will in turn lead to more locally directed wafer movement using operator-controlled robots, with the attendant impact on cleanroom design and contamination control.

Such details may be left up to companies that implement 300P’s fab vision and should be clarified over the course of the next year or so while 300P is hammered out between semiconductor manufacturers and tool makers. The first public unveiling of 300P guidelines is expected to take place around October. Initial plans may be somewhat vague but will grow more specific in 2007.

Bob Johnson, research vice president with technology market analysts Gartner Inc., notes, “With tools sitting idle 30 to 40 percent of the time, there’s room to boost efficiency through better scheduling. And with the long return on investment tool makers continue to suffer with 300 mm, there’s a need to get some return on 450 mm investment as soon as possible. Nevertheless, it’s not completely guaranteed that the effort will succeed at all. Says Johnson, “The project may be a good and even necessary step but that doesn’t make it easy given the technical and other challenges. It’s ambitious.”