[Arowana]

Just an idea... Diamond is the most heat trasfering material in the world (or so i read), would i be possible to make a thermal paste out off diamond dust?

The dust aint to expecive, after all its a by product...

Diamond (6-50w/cm-k, dependent on purity)
Silver (4.173w/cm-k )

Sources:
ai.mit.edu
pyrographite.com
http://www.overclockers.com/articles1088/index03.asp

[Keiron]

Depends on how well crushed it was, and what you mixed it with, I think.

[CandymanCan]

Diamond doesnt melt like silver does so it wouldnt really get into the micofibers of the cracks like silver does.

[Arowana]

Originally Posted by Keiron

Depends on how well crushed it was, and what you mixed it with, I think.

I think that diamond dust is very well crushed because diamonds are so hard to grind, therfore the dust will be very fine

Its very easy to test: White goo vs White goo mixed with diamond dust

[Keiron]

I would gues AS3 mixed with the Diamond would be a better choice to test with, since it's mostly silver.

[qualhiveldorf]

Originally Posted by CandymanCan

Diamond doesnt melt like silver does so it wouldnt really get into the micofibers of the cracks like silver does.

Silver's melting point is ~961*C. The silver in silver pastes don't melt it only depends upon particle size and the silver particles are extremely small so they act as a liquid of sorts and diamonds can do the same.

[Dukemurmur]

yes in theory they should do a nice job...but there must be a reason why it isn't getting used...also some types of Graphite have very good thermal properties.....but there must be reasons why CO like AS and such havn't used them....

[assemblage]

Originally Posted by Dukemurmur

yes in theory they should do a nice job...but there must be a reason why it isn't getting used...also some types of Graphite have very good thermal properties.....but there must be reasons why CO like AS and such havn't used them....

It's because graphite is black and black thermal paste would be messy.

[skou]

Both diamonds and graphite depend on the way the atoms are aligned, in the crystal structure, for good heat transfer. It would be IMPOSSIBLE to get the crystals of either, in a paste, to line up.

Both of them only have the great heat transfer, in only 1 direction, (well 2 directions) in the crystal structure. They both have a plate structure, and the best transfer occurs when the heat is going perpendicular to those plates.

Silver, on the other hand, is a relatively soft metal, and will mash a slight amount, in a TIM installation. That will fill the voids with silver, and insure some decent thermal transfer.

steve

[bobad]

Originally Posted by Arowana

Just an idea... Diamond is the most heat trasfering material in the world (or so i read), would i be possible to make a thermal paste out off diamond dust?

The dust aint to expecive, after all its a by product...

Diamond (6-50w/cm-k, dependent on purity)
Silver (4.173w/cm-k )

Sources:
ai.mit.edu
pyrographite.com
http://www.overclockers.com/articles1088/index03.asp

Arowana, never mind heat transfer. Where can I get cheap diamond lapping paste? It's the mother of all lapping compounds.

[SpaceRangerJoe]

I believe people are already conducting test with diamonds as a thermal transfer material. I think there was something on the front page about it a while back. The prototype TIM's had a higher thermal conductivity than the metals they were connecting I believe. Also, some sort of synthetic diamond compound would probably be a little more reasonable than crushing lots of diamonds, which is also being done.

[rbwilton]

You can get 50,000 or 100,000 grit diamond dust from a lapidary supply house for a reasonable price. I have some 50,000, but I'm not sure if I want to experiment making thermal paste out of it.

[mjw21a]

I believe that since diamonds are so abrasive, using it as a thermal compound would be a bad idea. Slight vibrations from the heat sink fan would cause the diamond particles to move, causing a sanding effect on the processor core.

You'd soon end up with a dead Athlon XP processor, however CPU's with heat spreaders should be ok......

[Bensa]

A thermal paste is only as good as its worst component. You would be much better off doing extreme precission lapping and having nothing in between. Maybe a micro layer of diamond dust could be used to fill in the nanometer sized crevaces.

Also, a diamond TIM won't help much with a copper heatsink, since the copper will be the bottleneck.

The best cooling is direct die, no material hindering transfer of heat.

[subtotal]

what like CPUs with built in heatsinks?

[sandman001]

Originally Posted by Bensa

The best cooling is direct die, no material hindering transfer of heat.

Not really, direct-die watercooling does't give that great of results.

[man_utd]

You guys ever heard of carbon black? It was basically carbon nanotubes (a tad stronger than diamonds actually), and performed a tad worse than arctic silver, with some improvement if you mixed the two. However, it never really took off.

[mjw21a]

Yeah, I read up on it once. Then I never heard anything about it ever again....

[rbwilton]

Originally Posted by bobad

Arowana, never mind heat transfer. Where can I get cheap diamond lapping paste? It's the mother of all lapping compounds.

I'm not sure if I recall correctly but from some lapidary work I used to do, I think that the diamond particles will actually embed themselves in the copper heatsink. If you start with course grit diamond dust you will never be able to get a good finish under that assumption.

[Bensa]

Originally Posted by sandman001

Not really, direct-die watercooling does't give that great of results.

If the die was as large as a WB block. In theory its the best, in practice there is something that hinders the performance (not as much as a TIM though).