Silicon Investor (SI) -- The First Internet Community

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Politics : Formerly About Applied Materials -- Ignore unavailable to you. Want to Upgrade?

To: Katherine Derbyshire who wrote (58439)	1/5/2002 3:57:23 PM
From: Kirk ©	Read Replies (1) \| Respond to of 70976

Not really, at least not in this case. The design rule, or "smallest feature size" is roughly equal to the size of the key "components," the transistors. Pretty much all of the larger circuit features are just wiring, not active components. I disagree. .13um is the channel length, correct? You need source and drain contacts which are probably much larger. Even if you assume they are equal to the gate, then the active area, the area under the gate, is only 33% of the total chip area. You also need a bit of room for isolation so perhaps 25% would be my guess at a very best case efficiency. I would guess that the average size MOS transistor uses far more area than 0.13um x 0.13um... Do you have the die size for a state of the art microprocessor? Divide the die size by the number of transistors and you get a rough approximation... but I believe 2/3rds of the chip might actually be for interconnect, sense amps and large driver devices to move electrons around the chip. The actual "active area" is probably still a TINY percentage of the total chip area. Back in the dark ages on 1979, I built a working 1mm x 1mm MOS transistor in the lab at Berkeley. Even then, bond pads and interconnect consumed more of the die than actual active area. at best, I'd say the minimum transistor is 0.13 x 0.39 assuming you can stack a gate, source and drain as close as possible and all at minimum feature size. I'm not sure if people use (W/L)=1 even for memory devices.. when I was designing analog chips, I'd usually use much wider devices as they really didn't consume much extra chip area... and you got much performance for a bit of W (keeping Lmin=0.13um). For the memory guys, perhaps they go with W/L=1 for a minimum size device for most of their chip....but again, that still needs a good deal of room for the source and drains and those used to use more area than gates back when I used to work with them. Since they are just scaling, I doubt much has changed in that regard. Kirk (I'm starting to ramble) out