300mm Semiconductor Wafers and the revolution that almost wasn't

Summary : All technology revolutions are not so easy. other than in 20:20 hindsight. The industry's transition to 300mm Semiconductor Wafers almost stalled, when the Asian Financial Crisis hit. Chip companies wanted it but were unwilling to pay for it. Semiconductor Equipment companies wanted to comply only if one would buy the tools.This is the story of how it all came about in 1998.

June 30, 1998: From the front lines . . .

I300I is ground zero for a revolution in 300mm and it will be ugly.  It won’t be the tech kind of revolution you read about in candy coated press releases.  No, it will be the French kind, where blood flows freely from the guillotine.  Applied Materials, who is about to cut out of I300I, will lead the revolt and take most of the equipment makers with them.  Why?  Because in these days of excess capacity, 300mm has become irrelevant.  No one is buying.  At least, not at significant levels. Applied and TEL have a combined total $1.2B in, what is now, worthless 300mm equipment inventories and they’re not happy about it.  R&D costs for 300mm are overwhelming the equipment industry as revenues, prices, and profit margins fall.  Equipment companies are spending between 5 and 10% of revenues for programs that won’t see a return until 2005 at the earliest and maybe as late as 2010.  This was OK when revenues were rising and profits were good.  But now the picture has changed and I300I is not changing with it.

I300I and several of its members say the problem is that the equipment has failed the tests.  Tests they themselves set.  They want more tests, set to another unrealistic schedule, which will cost $1-2M per tool.  One huge issue is that they want the equipment makers to supply the wafers.  They keep throwing down the gauntlet by saying, “when we’re ready, we will go with the vendors who are ready, if you’re not, well then that’s just tough.” 

But every equipment maker knows this, they are big boys and accountable for their own decisions. The whole purpose of I300I was to provide planning so the industry could get there faster and cheaper.  So, if I300I’s view is “let the market decide,” who needs them?

The equipment makers are blaming I300I and its ‘let them eat Gantt charts’ attitude.  They claim that I300I wants tests it knows will fail because the automation standards have not been met.  One the major problems that has yet to be solved is that the FOUP system is too slow for systems with throughputs greater than 100 WPH.  Equipment makers want the tests postponed until the known bugs have been worked out.  They would also like I300I to pay for the test wafer costs.  They equipment industry simply can’t afford to waste money. 

Chip makers point to the inclusion of three Applied Materials executive’s into the San Jose Mercury News’s top ten highest paid executives in the valley and scoff at the idea that the equipment industry might be hurting.  Chipmakers think the equipment makers are just posturing like they have been doing since the early nineties.

They’re wrong.  Equipment makers are serious.  Many have already put the guillotine to 300mm R&D -- and more cuts are following.  They see little to be gained in 300mm development.  The major buys won’t occur until 2002 at the earliest.  So they won’t see significant revenue stream until 2003.  Moreover, some seriously question if 300mm makes sense for any but the largest of manufacturers, given the trend to small lot sizes; which is the result of semiconductor company moves to differentiate their products.  At the same time, it really is not in the semiconductor industry’s interest to waste equipment development dollars.  They will pay for it one way or the other. 

The real question here is, do I300I and its members want to be involved in deciding when 300mm will available?  Selete is going along swimmingly.  While I300I is being caught between rapidly changing market decisions and its archaic decision making apparatus.  Those poor guys are on the front line and they have to wait for member meetings to occur before they can publicly take any positions.  At the same time, meetings go on between members and vendors, leaving I300I in a duplicitous position.  I don’t know what they paying the people at I300I, but it is not enough.  In the current market situation, by the time a new policy is approved, market conditions have changed and I300I becomes the whipping boy again.  This can be easily fixed

But first, a high level meeting between the top executives of the chip and equipment industries is needed to start the peace process.  Realistic goals for I300I need to be set.  Otherwise, 300mm will just take longer and cost more to implement.  To be honest, I have begun to question if 300mm will ever make it.  In the future, 300mm may only be used as a textbook case of how technology goes too far.  We may find it in the chapter about supersonic transport.

 

July 9, 1998: On the Guillotine . . .

 

Reader Replies regarding the June 30 article on the revolution in 300mm . . .

 

>Dan, How goes the battle?  Bastille Day approacheth.  <

— I feel like the engineer who on the guillotine, pointed out how to fix it.  This has turned out to be one of the most controversial things I have written.  Instead of repeating all the replies in a linear fashion, the replies have been broken up into areas by topic.  I have also tried to paraphrase the many conversations I have had on this topic. — Dan

 

>Many thanks for the insight-filled 300 mm story.   I think you are right.   I am only ashamed that in the euphoria of 1996-early '97 I did not have the guts to ask the 300 mm cheerleaders more forcefully if they had done a pro forma P & L, cash flow and ROCE for a hypothesized "typical" 300 mm line and compared it with a 200 mm line using realistic investment and variable cost assumptions for wafers.   Your observation about technology for its own sake -- i.e., moving way ahead of market and financial considerations -- is right on. <

 

>Dan, Thanks for stirring up the pot.   Here are a few additional comments.  In the days before SEMATECH there was a fellow who thought that 5 inch wafers were as big as would ever be cost effective.  Conventional wisdom is that he was in error.  I think he was only wrong about the timing.  Later there was a professor who was convinced that at some point the cost of getting to lithographically smaller dimensions would reach some unaffordable limit.  We can no longer deny these limits even if we are not sure what they are.  They are somewhat clearer in economic terms than in technology terms. 

Still later was my hypothesis that eventually the burden of development would have to be shared by the suppliers and users of equipment and materials, i.e. that the hump in the cost times risk curve as we go from research through development to production will have to flatten. Others

have pointed out that when costs exceed what is affordable by the industry, users, plus suppliers, of a single country is the time to seek another itinerary on the roadmap.  Three hundred millimeter wafers exceeds this limit.  We are currently attempting to determine if this is the limit or if we can go to planetary affordability.  While working on this level of cooperation is healthy, we should also explore other itineraries to the same goal on the roadmap.<

— Wow, my tunnel view was only that the industry was at a key branch point, in which change had to come or we would never see 300mm come to pass.  These people are addressing a much broader question: are the tensions surrounding cost a sign that we have reached some fundamental economic limit? — Dan

 

>Dan, Things are not “going swimmingly at Selete,” this is an equipment industry bias.  Their demo’s are slipping schedules and they have spent a lot more per tool (>$10M) than I300I).  The main difference is that the equipment industry is not expected to shoulder so much of the cost.<

 

>We are having similar problems with Selete.  They keep pushing 300mm ahead and yet there is no way we, or any other Japanese manufacturer, will build a fab on their time line.  There is no money to do it.  They are not changing their plan to industry needs, just like I300I.<

 

>As an equipment supplier, we find Selete about equivalent to I300I.  They are having similar schedule problems and it costs a lot more to support them.  The big plus is that both are setting common standards.  The minus is that the standards keep changing.  SEMI standards are a big issue here and it’s way too slow.  The problem we have with I300I is that they tend to be one-sided, favoring only their members and not listening to the equipment industry.  The people who participate don’t see the overall business picture, only the technical issues.< 

 

>Dan, you were close on your comments about automation standards, but not on target.  Some of the FOUPs don’t cut it, but others do.  Check with Frank Robertson, he has the data.  However, the real problem is that automation and other standards keep changing.  They have to go through this long round-robin decision-making process that includes I300I, SEMI standards, SEMATECH and Selete.  So, we are pushed to complete meaningless (but costly) reliability tests on tool designs that cannot be finalized. This is a waste because every chip maker will still do their own eval’s.<

— I did check with Frank and he stated that there were FOUPs that could run in excess of 100 WPH.  Implanters are the only ones that are hard to keep up with now and this is being worked on.  He did state that tests were needed to demonstrate the reliability of the FOUPs and to work out any bugs. — Dan

 

>How could you call the I300I decision-making process archaic? First, you were not in on the committee meetings and second, it is a process that has served SEMATECH for years.<

— I plead guilty to first argument, I have not attended any I300I meetings, and you are correct that it is a process that has served SEMATECH well for years.  I have participated in other SEMATECH consensus-building meetings where a similar process is used and they have worked when the world moved at a slower pace.  In this case, however, I still believe that things have moved too fast this for I300I to respond effectively.  Otherwise, they would not have to respond to criticisms with curt statements like “our member companies decided that in 1994 and we will have to meet to change our directives.”  Webster’s Encyclopedic Unabridged Dictionary of the English Language defines archaic as, “marked by the characteristics of an earlier period; antiquated.”  Certainly, consensus decision making has worked in the past, but it often fails in times of rapid change.  In today’s internet-time driven world, a decision-making process that gives senior executives little authority, but lots of responsibility is archaic. — Dan

 

>Dan, I300I is holding a meeting next week to look at these issues.  However, the problem is that there needs to be some high level decisions and commitments made by the semiconductor companies, rather than positioning by low-level people.  Before your chip insider, the level attending this meeting was too low.  Since then, something happened, because some high level management guys are coming now.  < 

Great, I had hoped this would happen. — Dan

 

>Dan, I take exception to your inflammatory remarks on June 30th.  It was like saying fire in a crowded theater.  You don’t know how many people read this.  It gets forwarded everywhere and some take it like words from god.  I got young engineers working on 300mm who are just about ready to slit their wrists and upper management who wants to cut the funding after reading your missives.  Your forecasts can be self-fulfilling prophesies.  Also, we are a foreign firm and many don’t understand the subtle nuances of your English.<

I knew I was yelling fire, but the smoke coming from burning equipment makers did not allow me to see that the theater was crowded.  All I could see was the chip people standing by the fire hose, not hearing the shouting, as they were enraptured by the marvelous sounds of the 300mm band playing the theme song to 2001: A Space Odyssey.  It was just before the scene were the equipment executive says, “Hal, open the pod bay door.”  The consortia, er computer answers,  “I’m sorry Dave, I can’t do that.”  I was just trying to get the industry’s attention before 300mm winds up in a movie titled “Lost in Space.”  

Regarding the English usage, I know the usage is complex, the vocabulary wide ranging, and the explanations brief, all of which makes it difficult for foreign readers.  My objective has been to keep it short, informative, not dry, but fun to read.  This is what all of the primary readers I have talked to enjoy most about The Chip Insider.  One of the readers from Japan told me, “I really like your Chip Insider.  But is very difficult to read.  I now have a dictionary next to my computer. . . No don’t change it.  I don’t want a boring market research report.  I like the language.  It is fun to read and I am learning so much about English.”

 

Nevertheless, if you have questions, please ask.  It helps me to know what works and what does not.  Also, please be careful about who you forward it to.  If you forward it to someone else in your company who is not a regular reader, or might misunderstand, try to give them a warning.  Something along the lines of “R-rated: Intended for mature audiences only, contains explicit language” or “If you have a heart condition do not take this ride” would work.

 

Regarding forecasts as self-fulfilling prophecies, it is often the nature of forecasts to do this or to serve as a warning.  My purpose was to make a loud and clear warning that if something was not done about 300mm, its very future would be in jeopardy.  Clearly it got people riled up, so my methods may have worked.  Don’t let them cut their wrists now.  The main reason this has become so emotional is that people (myself included) really care about making 300mm successful.   

 

You may like this also:

Access to and use of this Website is subject to VLSI's Terms of Use (including Copyright Policy & Claims) and Privacy Policy. By accessing or using this Website you agree to VLSI's Terms of Use (including Copyright Policy & Claims) and Privacy Policy.

Copyright © 2021 VLSI Research Inc. All rights reserved.