Re: Hot computers [was Re: NEWS: From Genes to Difference Engines]

From: Robert Bradbury (
Date: Sun Apr 16 2000 - 01:40:27 MDT

On Sat, 15 Apr 2000, Spike Jones wrote:

> Robert Bradbury wrote:
> > liquid beryllium if you wanted the lowest mass (lower pumping costs)
> > or liquid tungsten if you wanted the highest operating temperature.
> Liquid tungsten would be *too* hot. Likely it would cause the
> diamondoid break down. And it would melt anything it touched
> that wasnt diamondoid. spike

That was the point Spike. There isn't any "diamondoid" in a rod-logic
computer built out of hafnium carbide. The M.P. of W is 3683 K, while
HfC is solid up to 4163 K. Now ~480K is a pretty wide margin if HfC
doesn't lose too much rigidity as it approaches its M.P.

I've got the data for corundum (AL2O3) [1] and it has a linear decline in
bulk modulus from ~260 GPa to 225 GPa from 400 to 1800 K (M.P. is 2345K).
If HfC follows the same pattern, then it sould have a fair amount of
strength left at 3700K. If it turns out that HfC gets too soft, Be has
a very wide liquidity range from 1551 to 3243 K, so it should do quite
nicely as a general coolant. You could get a really nice phase change
coolant using liquid Be with Mo "ice" particles in it at around 2883K.

The rarity of Hf or Ta for HfC or TaC computers means that it is more
likely that you would use TiC (3413K M.P.) or TiN (3203K M.P.) in which
case beryllium might be in short supply, so switching to liquid Al, Si
or Fe for a coolant migth be better. Fe "ice" in liquid Al or Si is
probably your most abundant coolant, but it only runs at 1808K. :-(

Something I don't know (if you want to make yourself usefull... :-;)
is what pressures would be required for a liquid carbon coolant.


(1) Anderson, O.L. & Takayasu, G., "Measurement of elastic constants of
    mantle-related minerals at temperatures up to 1800 K.", Physics of
    the Earth and Planetary Interiors 55:241-253 (1989).

This archive was generated by hypermail 2b29 : Thu Jul 27 2000 - 14:09:21 MDT