To: Bill Jackson who wrote (1003 ) 12/7/2000 9:15:09 PM From: Zeev Hed Read Replies (1) | Respond to of 1099 Bill, the heat eventually has to go to the heat sink. If you have "thermal resistors", the thermal gradients are equivalent to voltages" in an electrical circuit small resistors (high thermal conductivity) will have small gradients and large resistors (low thermal conductivity) will have large gradients. Using Si28 for epi will help with spreading heat laterally between devices on a chip (but there is very little real estate between devices on a chip), but will not help in driving the heat generated to the heat sink which typically is on the bottom side of the substrate (the one without devices, or the support structure). Of course, since the heat is generated in the junctions (that is where most of the switching voltage "fall"), that is the material with the lowest thermal conductivity. This, even if the junction was originally made with Si28, since doping broaden the phonon spectrum, that "breadth" is what causes normal Si to have a lower thermal conductivity than isotopically pure Si. IMHO, there is a greater advantage to be found in the fact that charge carriers mobility in isotopically pure Si is greater than in normal Si. Yet, even if Si28 is used only for EPI, the factor of 1000 in cost of materials will make it very difficult to justify for main stream chips. Just look at GaAs, it is used only if Si simply cannot do the job (very high frequencies and optoelectronics), and is losing market share back to Si as the Si technology (smaller features and copper lines) improves and extend its frequency domain. Why, because GaAs is simply intrinsically more expensive, by a factor of at least 20 or so. Zeev