Did you read the papers on gate oxides presented at IEDM?
Bob,
I was at IEDM and attended the session where IBM made their statement (my Ph.D. thesis included research on oxide reliability, so this is right up my alley). Oxide reliability is a big deal, and it has always been a big deal. However, along with my friends from a variety of companies, we found IBM's results to be Chicken Littlesque. On the other hand, IBM certainly knows how to get good free press!
How thick would the gate oxide be on a 0.07 process?
Well, you really can't go below a mono-layer, so I'd have to say thicker than 5Å. Then again, maybe you use Si3N4 or an even higher-epsilon insulator which lets you go thicker for the same 'bang', or perhaps you find a way to operate below 1V with low-threshold transistors (with some way to reduce the subthreshold/standby current).
On another related note, people often forget that just because you can resolve down to 0.08 microns (or smaller), it doesn't necessarily mean you can make a decent transistor. There is an art to process engineering--it's not just a shrink of the last design.
Steve |
