Electronics Weekly Electronics Design & Components Tech News
What can you get for the cost of an Olympic Games? Answer: A transition to 450mm wafers. Both cost between $25 billion and $30 billion.
But while governments fight each other to fund the Olympic Games, the only three people who want 450mm wafer processing – Intel, Samsung and TSMC – don’t seem prepared to put their hands in their pockets for it.
“A true collaboration from industry, where key stakeholders have contributed substantial funding toward supplier R&D efforts, would be a key sign that the proponents of 450 mm are very serious about moving it forward with some ‘skin in the game’ we have not seen this in any significant way,” says John Ellis of the semiconductor equipment manufacturing trade body SEMI.
Back in July 2008, when Intel, Samsung and TSMC proposed an industry effort to develop 450mm wafers, Dr Mark Pinto, CTO of Applied, told me: “If we’re going to go to 32nm and 22nm and beyond, and do 450mm as well, the costs are just huge.”
Asked why Intel, Samsung and TSMC were pushing for it. Pinto replied: “Because no one else in the industry wants 450mm wafers.” Asked what might justify 450mm in the future, Pinto replied: “If scaling ends”.
Well, scaling isn’t ending. Or not soon. The route to 11nm seems reasonably well understood.
Industry analysts, and SEMI, estimate the R&D investment for developing 450mm wafer processing equipment at $25 billion to $30 billion.
But the equipment makers have an R&D budget of about $3 billion a year, and that needs to be spent on shrinking geometries as well as increasing wafer sizes.
Moreover there are doubts about the benefits of a move to 450mm. SEMI reckons the increase in throughput could be only 1.24x, and the decrease in cost might be only 8%.
So is 450mm a dead duck?
I think you’re spot on so far as the embedded market is concerned, dbs, hence Intel’s move on it.
I guess it all comes down to the volume of wafers predicted to be processed over the lifetime of the fab. The ever-growing PC market justified such expansion over the last few decades, but that market is now somewhat saturated.
The embedded chip market still has a long way to go — beyond cars and appliances, and into the total integration of the household for example. Also RFID, and you can probably name others. But I feel the demand curve for chips is flat at this point because we haven’t yet engineered just how to use such devices in a way that would move the demand curve upward.
We seem to be waiting for the “next big thing”, but we’re just not there yet.
In respect of the cynicism, you’re probably absolutely right, dbs, but what happens if shrinking becomes impossible or too expensive? Since the primary reason for shrinking is to reduce cost, if shrinking stops then would not moving to a larger wafer size keep microelectronics on a downward cost curve?
If 450mm is not yet a dead duck, it probably should be!
I’m neither a hardware engineer nor an accountant, but I don’t see the point in all this upward progression in wafer size. It’s frightfully expensive, so only the big players can even entertain the idea. But the economic situation has diminished, and will probably coast along at a slower rate for a number of years. Even the big boys should re-examine the basic premise of ever-increasing wafer size while struggling with the ever-difficult problems of ever-smaller feature sizes. Maybe the 300mm or even 200mm size makes more economic sense!
Of course the issue of Corporatism may be at play here — where the point is to make conversion mandatory but so expensive that Global Foundries and other players cannot stay in the game. But maybe that’s just the cynic in me. And another thing: Why wasn’t Applied Materials invited to the meeting?