To: John Cuthbertson who wrote (3132 ) 3/15/1998 8:51:00 AM From: Zeev Hed Respond to
John, both diamond and diamond like carbon have excellent thermal conductivity (the former much more that the latter), but their use is not for the device itself but for the heat spreading substrate. It is contemplated for some military applications, and in some display schemes as the electron emitting medium itself. For the semiconducting devices themselves, I think we are still a long way, mostly because diamond thin film, to the best of my knowledge cannot be formed as single crystals. As for diamond like carbon, it is much cheaper to make, but, allas, the properties are not very consistent, and many of these are not even crystalline but amorphous. There are some steps (like the oxide layer in current devices) that could eventually be replaced with diamond like carbon (but not with Diamond film, since the temperature of nucleation and growth of diamond films is a little excessive). In any event, the normal silicon wafer on which all the various layers from epi on are deposited will probably reign for quite some time. The critical performance parameter of a semiconductor is the "switching time" multiplied by the "energy dissipation per switching cycle". This parameter is improving in silicon based device everytime the design rules are being reduced. The problem, is of course that the improvement in not sufficient to compensate for the higher device density resulting from the reduction in design rules. Thus the overall energy dissipation per square inch of active devices is increasing. I have been told some time ago that Intel was starting to look at method to handle heat withdrawal rates in the range of 70 Watt per chip (for future applications).
