There are numerous precedents of how trailing-edge manufacturing has received a cost benefit from moving up to a large wafer size but usually this is done after equipment has been developed to service the leading-edge digital chip manufacture.
Even while it was pulling back from developing leading-edge digital processes in 3Q09 Texas Instruments announced it was opening RFAB, the industry's first 300-mm wafer fab for analog ICs in Richardson, Texas. By November 2010 TI was ramping up production at RFAB and rival Maxim Integrated Products Inc. has recently qualified and shipped production analog product built on 300-mm wafers. Maxim and TI separately claimed that 300-mm manufacturing would give them a competitive advantage over their rivals.
Infineon Technologies AG is soon to begin volume production of power devices on thinned 300-mm wafers in Dresden, which could provide a similar competitive advantage.
STMicroelectronics, which closed a number of 6-inch wafer fabs during the last decade, has focused a lot of analog manufacturing on 200-mm wafer fabs in Singapore. While those moves may have helped with cost structures in the past ST may now be at a disadvantage and need to move manufacturing on to 300-mm wafers at least.
"Could you use 450-mm to replace older wafer fabs and get significant cost reductions?" asked Penn. Some people argue that because many operations in chip manufacturing are step-and-repeat operations, the area benefits are not so large. And it requires highly engineered wafer tables that can accelerate and decelerate; all things that would push up the cost of equipment and which can usually only be amortized with a move to higher-value chip production.
Penn responded: "450-mm is going to happen. It will split the industry between the haves and the have nots. So the question is what do you do about it? If you are fabless, or fablite-going-fabless, you don't care," said Penn. But it is a key enabling technology, it does matter."
Penn agreed that if a number of older wafer fabs were replaced with a single 450-mm More-than-Moore wafer fab it would be likely that the fab would have to cope with a diverse mix of processes and products. "I don't see it as a problem. A variety of processes can be run. It depends how you set it up."
Agreed. It seems ludicrous to think that you could build a multi-billion dollar to make analog parts, and compete with depreciated and well characterized 200mm fabs. TI's RFAB was built almost entirely with 2nd hand equipment, so it not a good bit of evidence for 450mm more than Moore. Maxim is not running their own 300mm fab, either - so they're agnostic at best.
Would those who make just low margin products like analog or RF ID chips be able to afford a new Fab for 450 mm ?
Even if they had the resources would it not be a better use of their money to upgrade Fabs for high margin products like processors ?
The "analyst" quoted here needs to make a reality check.
This seems to be a case of where the "research" fits the hypothesis given to it by it's sponsors. I don't see the financial merit in the concept and at best an attempt at hand waving to get the conclusion to remotely fit. I went thru this type of thinking at a once proud and formidable semiconductor manufacturer and it ended in their ruin. There are three strategic ingredients to a wafer size change:
1) Capital leverage
2) Manufacturing scaling leverage
3) Performance leverage
It's utter lunacy to say any of those three are invalid and a "new" economic model has dawned.
@KB3001: I agree with your arument on More-than-Moore as the lead to 450mm. The industry is still grappling with handling of ultra-thin wafers in the assembly flow of W-to-W stacking. This problem gets exponentially amplified with 450mm wafers.
What I would like to see is a study by the proponents of 450mm on whether or not the 'have's can survive the transition to 450mm when many 'have-not's choose to stay with 300mm for multiple reasons, economics being the major one!
Agreed. I don't see the cost justification at anything other than the leading nodes. You can't get the leverage you need. I wonder if instead of investing in larger wafers if more fabs will instead start investing in 3D-IC technology to get density scaling using the 3rd dimension. It would probably be lower cost.
Join our online Radio Show on Friday 11th July starting at 2:00pm Eastern, when EETimes editor of all things fun and interesting, Max Maxfield, and embedded systems expert, Jack Ganssle, will debate as to just what is, and is not, and embedded system.