From: Robert J. Bradbury (bradbury@aeiveos.com)
Date: Mon Oct 08 2001 - 08:48:28 MDT
Anders asked:
> How would this work?
in response to my comment that we might be able
to get a factor of 10^9 using ballistic heat removal.
As you are probably aware, I'm basing this on comments made
by Michael Franks from his paper that was from a Foresight
Conference a few years back (cited in my primary MBrain paper).
It looks like nanotubes will do ballistic transport, e.g.
Quantum electronics: Nanotubes go ballistic
CARTER T. WHITE AND TCHAVDAR N. TODOROV
Nature 411:649-651 (2001)
http://www.nature.com/cgi-taf/DynaPage.taf?file=/nature/journal/v411/n6838/full/411649a0_fs.html
The distances are still short 10^2-10^3 nm, but it provides hope
for improvement.
There is also this:
http://www.foresight.org/Conferences/MNT9/Abstracts/Leyse/index.html
(which to be honest, I'm somewhat skeptical about...)
If a way to turn heat into electron velocity can be worked out
(a not-so-small detail I'd guess), then one can replace the
phase-change fluid coolant (patented by Henson & Drexler) with
buckytube heat conducting tubes. I suspect (though calculations
would need to be done to be sure) that this allows the 1 cm^3
nanocomputer to shrink quite a bit giving you a fairly nice
bump in throughput.
I asked Michael Franks once about how to magically convert
the heat into ballistic electron (or proton I suppose)
velocity but he never provided an answer that I was
satisfied with. I suppose its going to be a very interesting
question to resolve. If heat is simply the vibration of atoms
and you have a buckytubes encased in a diamondoid matrix and the
buckytubes are connected to solid H2 heat sinks what the
heat removal capacity will be. Also of interest would
be whether flowing liquid He through the interior of the
buckytubes would do any better. I think the problem hangs
on the resistance to heat transfer at the interface surfaces,
e.g. diamondoid-to-buckytube or buckytube to lHe.
One thing that does occur to me is that there seems to be
a lot of work going on using carbon nanotubes as electron
emitters, e.g.:
http://www.foresight.org/Conferences/MNT9/Abstracts/Brenner/index.html
You could envision a design of an emitter tube inside of
a larger evacuated nanotube. If you can setup magnetic
fields such that the electrons reliably fly down the evacuated
tube (kind of like a linear accelerator), then I could see
the transfer of heat to the emitter and thence to the
electrons as a ballistic heat removal device. The
questions then become how do you channel the heat within
the nanocomputer *to* the emitters; *what* would the
temperature of the emitters; and what the velocity of
the electrons be? [This is a problem for Spike I think.]
Robert
This archive was generated by hypermail 2.1.5 : Sat Nov 02 2002 - 08:11:16 MST