by Debra Vogler, Senior Technical Editor, Solid State Technology
Speaking at last week’s ISSM, Lam Research’s Nick Bright asked a point-blank question: why should equipment suppliers invest in 450mm when there are so many opportunities elsewhere (e.g. solar, FPDs, LEDs, and MEMS)? Perhaps the answer lies in whether or not enough equipment companies are able to find answers to IC manufacturers’ calls to improve fab agility at a reasonable rate of return on investment.
For AMD’s Les Marshall, senior program manager, manufacturing technology development, getting wafers through the fab faster is a necessity, otherwise “all we’ll be doing is carrying the waste with us when we go to 450mm,” he said during his presentation. Without a fundamental change in the way the industry manufactures, he believes that the lack of business agility that IC manufacturers experience now at 300mm will just follow them to 450mm.
Marshall is optimistic about the future of 450mm, noting that “at AMD, we like to say that we have a 20/20 vision” — meaning that 2020 “is probably about the right time” for the next wafer-size transition. Today, though, “productivity gains from a 450mm transition have been pitched to the industry, but little work has been done to validate the base assumptions,” he said, “and the lack of due diligence has been questioned by the equipment supplier community.”
Marshall described the company’s efforts to work with ISMI to identify and address the productivity factors that are dragging down the industry and need to be addressed before 450mm manufacturing can begin. One of the issues that stood out was traditional 25-wafers/lot manufacturing, which “constrains our agility,” he said. With vacuum systems, material movement, and fab layouts in general revolving around the 25-wafer lot, a paradigm change is needed, Marshall said. If lot size could be made more agile, e.g. down to single-wafer processing throughout, he believes that the long cycle times, high inventories, and poor customer responsiveness that now dog the industry could be addressed. To that end, AMD has been studying other industries, especially the automotive industry, for ideas. (See related story: “SEMICON WEST REPORT: Do lean principles apply to semi manufacturing?“)
By reducing the number of wafers in a FOUP, cycle times will be shortened, and replacing batch tools with single wafer process tools will further reduce cycle time. Because the cascading behavior of tools today is very much dependent on 25 wafer lots, AMD is looking for innovative changes to reduce dependence on cascading. Pointing out that some sequential tools are not optimally matched — citing tracks and scanners as one example — Marshall observed that, “Maybe we’ve gone too far