To: Merlin who wrote (910 ) 12/15/1999 9:11:00 AM From: Zeev Hed Respond to
Merlin, the whole area of low dielectric for ULSI is addressed with non SiO2 based materials (TRKN has a new low dielectric, I believe it is an organic fluoride) they should work over different substrates. I agree with you that GaAs based system for most logic chips is not cost effective, one of the main reasons, as I mentioned is the factor of 20 in raw material costs (another important factor is that you still cannot get GaAs substrates with the low defect density, essentially zero, you get in Si). GaAs was is and probably will still be "the material of the future" (G), isotopically pure Si will have a similar if not worse fate, IMHO. Last year when I addressed this issue here, I tried to get an idea of the cost premium of isotopically pure Si, the E-P transaction gives us such number, and in my opinion, the premium (a factor of about 1000) is not going to fly, when all you get is a 50% improvement in thermal conductivity. The truth is that you should get improvement in mobility as well (but not that much) due to lower phonon electron scattering. In any event, you can get the same improvement in thermal conductivity, or at least close to it, by simply keeping the chip at a lower temperature. Since very high density circuits will need to address dynamic cooling anyhow, they might as well keep the ambient circuit temperature 50 C lower and get a similar increase in thermal conductivity.
