Semi, others, I´m trying to get an idea what it means to convert fab capacity from one node to the next. The case I´m having in mind here is AMD. Simplified, they are currently at about 50% of wafer starts at 90nm. In 5 months, they want to be at about 90% wafer starts at 90nm. What does that mean for capacity in the meantime? When would they take e.g. 130nm line "A" offline for conversion so that it can start 90nm production in 5 months at 90nm? TIA for any input you can give.
Chipguy had a good explanation. Usually, from one process generation to the next, you can reuse about 70% of the tools. The hardest part is switching over the Litho Tools. Most likely the critical layer steppers, which Chipguy refers to as "A" Steppers, will need to be taken off line, and either upgraded, or more likely, replaced. Most of the time, they're replaced, and an "A" Stepper then becomes a "B" stepper in the new process, and so on.
The biggest issue, is not in the reconfigured "B" and "C" class steppers, because they can most likely handle the non-critical Layers in the shrunken process, with adjustments to the Focus and Exposure, and without any major upgrades to the Lenses, but in any case, some sort of requalification of the "B" and "C" steppers will need to be done. If you get lucky, it might only require running a ton of split up lots, to compare previous technology results, to shrink results.
The Critical Layer "A" type Steppers are the pain. From what I've seen, from installation, to qual, to pilot runs, takes in excess of 6 months "Off Line", for each stepper, if everything goes right. I've never seen it done in less time. If however, you have the resources and space to do the stepper installations/qualifications in Parallel, so much the better. In any case, New Litho Stepper Masks will most likely be needed all across the board, for all layers.
I think the second most difficult might be thin films, since material thicknesses usually decrease with new processes, and the thinner the materials, the harder it is to maintain thickness and uniformity. There might be a need for a few deposition tools, maybe not. Etch will hopefully just need to adjust the chemical flows, and recipe times. Unless you have old etchers, which can't etch the thinner lines and spaces.
If AMD wants to go from 50% to 90% in the next 5 Months, those Critical Steppers need to be close to, if not already qualified, since Poly is relatively early in the Process Flow. You could probably be able to get away with the Metal1 Steppers for a few weeks more.
If the question being asked is do I think that any Fab can go from switching over from a 50/50 split on an old/new process, to switching over to 90% production on the new process "in the same fab" in 5 months..... maybe, if everything goes right (cough). It would be something to see, since I've never seen that aggressive a switchover ramp schedule in a single fab, in that timeframe, come to pass.
AI
