>>I-line sales will be greatly reduced...>>
I do not want to dispute ASMLF 60% DUV number (3:2 DUV / i-line ratio), but it seems pretty high. The question is what does this tell us about the coming year?
Some time ago James Word had expressed an opinion that right now an economically designed DRAM production line should have i-line/DUV ratio around 7:1, and more complex IC line should have i-line/DUV ratio around 4:1, because DUV consumables are expensive.
This suggests that with i-line/DUV ratio of 2:3 we are clearly talking about a year in which most of DUV tools will go to upgrade existing lines, where previously installed i-line equipment base could be reused. Gives one an idea that 300mm fab pushouts are for real and that 1998 outlook is subdued.
Even for existing fabs upgrades, i-line/DUV ratio of 2:3 is kind of low. New DUV tools have higher productivity, the question therefore arises how will the new line be engineered.
One approach is to use slower old i-line tools in parallel, so that their combined throughput matches the DUV stepper throughput. In this case, additional i-line tool purchases will be needed - unless .25 technology allows to process more layers in one exposure . I recall looking at one example, where it turned out that without reduction in a number of steps, ratio for new orders will be around 2 i-line for 1 DUV.
Another approach resulting in lower throughput and lower costs
is to use the DUV stepper for several operations, so that production line throughput is determined by slower old i-line steppers. For example, if old i-line equipment has throughput of 40wph, and new DUV stepper can do 80wph, the latter could be used for two steps without becoming the bottleneck. In this case, manufacturer sacrifices the throughput, but can get into .25 production with minimal cash outlay. May look good for cash-strapped DRAM manufacturers that want to get into 64Mb, but do not expect fast ramp-up. Maybe that's the route being pursued initially?
I guess we'll have to ask James, whether .25 technology allows to reduce the number of production steps. In this case, I-line/DUV ratio of 2:3 is reasonable.
Also, would be interesting to check how does 3:2 DUV/i-line number for new equipment fits the VLSI forecasts. I recall that they predicted 40% of DUV tools for year 2000, not sure though was this 40% of tools produced in Y2000, or 40% of DUV tools in the installed base.
just a few thoughts, comments are welcome
Y.
|
