|
|||
|
|
|
The chip quandary Twice the throughput at lower cost eyed by semi makers By Craig Matsumoto, EE Times San Francisco -- Equipment suppliers last week gnawed at the riddle posed by 21st-century semiconductor manufacturing: how to squeeze twice the throughput out of a wafer fab while cutting manufacturing costs to the bone. To be sure, semiconductor makers will need plenty of new equipment to handle 300-mm wafers. But beyond that, said exhibitors at the Semicon/West trade show here, next century's factories will need unprecedented levels of automation to satisfy chip makers' and foundries' craving for added productivity and reliability. The industry timetable calls for pilot 300-mm factories to be online by the end of 1998 and first full-production fabs to be up and running by 2000. Last week at Semicon two consortia offered a common document--"Global Joint Guidance for 300-mm Semiconductor Factories"--toward meeting that goal. Its authors were the International 300 Initiative, which includes 13 semiconductor houses in Europe, Korea, Taiwan and the United States, and the 10 Japanese semiconductor companies in J300. Their agreement to work toward common standards means that in theory equipment will be interoperable, thereby allowing better computer-controlled management of the entire suite of processing tools needed to make next-generation chips. The transition to 300-mm wafers is not trivial and is estimated to be a $14 billion endeavor for equipment and materials companies. The idea is to cut device-manufacturing costs by as much as 20 percent to 30 percent and produce up to 2.5 times as many chips as with 200-mm wafers. "Automation is a must in 300-mm semiconductor factories," said George Lee, director of the 300-mm Initiative for the Semiconductor Equipment and Materials International trade group. "We're reaching certain limitations from a capital-productivity point of view," said Ferdinand Seemann, vice president of sales and marketing for Mattson Technology Inc., a Fremont, Calif.-based supplier of etch and thermal-processing tools. "It seems that we're about to reach a plateau in productivity." Productivity, in turn, is a way to squeeze extra revenues out of a fab costing $1 billion to $2 billion, a particularly sensitive issue given the speed of devaluation. "You write off a fab in three years--that's the norm in the United States. That's depreciation at $1,600 a minute," said Thomas Comstock, vice president of marketing for software developer Promis Systems Corp. (Nashua, N.H.). Chip makers are reaching their limits in eking out better yields, Seemann said. At the same time, wafer throughput is expected to continue eroding and equipment costs--some tied to the equipment and materials companies' $14 billion bill for the 300-mm transition--may triple. Chip makers and equipment vendors alike believe automation is the only way to cut device-manufacturing costs the requisite 20 percent to 30 percent while doubling chip production. The ultimate goal is a "lights-out" fab--one that runs without human intervention. Not surprising, automation companies were unusually popular at Semicon. "For 300 mm, it's a massive integration problem," said Robert Postle, vice president of marketing and sales for PRI Automation Inc. (Billerica, Mass.). "We're trying to double the scope of a fab. What's surprised us is the speed with which automation has taken off." Automation content in an average fab is $15 million to $20 million but will soon rise to around $40 million, he said. "You look at the automation guys, their booths are more crowded than Applied [Materials Inc.'s]," said Brett Hodess, analyst with investment bank Montgomery Securities. The equipment industry, however, is a hodgepodge of vendors, and a fab often uses the wares of several dozen companies. Despite the good intentions being tossed around Semicon, it remains unclear who will knit these tools together in an automated fab, or how. Legend says it's been tried before--by Bechtel Corp. and "the Japanese"--without much success. For starters, individual tools will have to become smarter by using feedback loops for real-time monitoring of processes. For example, Lam Research Corp. says it needs increased monitoring and control for its etch processes. Currently, plasma is ignited in the chamber to create a firestorm that stabilizes after a few seconds and hopefully completes the etch. With 300 mm, the industry wants a more controlled process. That means adding feedback loops to monitor the pressure in the chamber and provide the igniting spark at a precise moment. Lam's answer was to add a second PowerPC processor to handle the feedback loop. The added expense is made up for in greater efficiency. Likewise, Novellus Systems Inc.'s future deposition tools need multiple processors and fast lines of communication to keep up with the precise physics they're expected to execute. (Next article.)
[This article comes from EE Times in a joint cooperative effort with the Motley Fool. For more articles like it, please look at Fool's Gold every weekend or simply go to the Fool's Gold Mine and page through our back issues, which all have clever and cool EE Times articles in them.]
|
|
|
© Copyright 1995-2000, The Motley Fool. All rights reserved. This material is for personal use only. Republication and redissemination, including posting to news groups, is expressly prohibited without the prior written consent of The Motley Fool. The Motley Fool is a registered trademark and the "Fool" logo is a trademark of The Motley Fool, Inc. Contact Us
..
..
..
..
..
..
..
..