Diamond thermal paste?

Well, Every once in a while, somebody posts something like, "man, how great would it be if we had diamond heatsinks because the heat transfer properties of diamonds are so high!" Obvoisly until the artificial manufacture of diamonds become much cheaper, this is unfeasible. But, since the best thermal pastes use Silver particles because the heat transfer is better than copper (a.k.a. Arctic Silver), why aren't there thermal pastes being made using diamond particles?
I was searching through www.harborfreight.com the other night and found this interesting product.
http://www.harborfreight.com/cpi/ctaf/Displayitem.taf?itemnumber=45773
It's a super fine grit polishing material that comes in a syringe like thermal paste and is oil soluble. So, couldn't you make thermal paste of this and get excellent results?
In a few weeks I'll have a testing rig consisting of an old AMD T-Bird 850, and other crappy stuff, so maybe I'll try it out and see what happens. Anyone else have any ideas on this?

good deal too....less than arctic silver...let us know how it works!

I remember reading that the folks at Arctic Silver had indeed tested various combinations of diamond powder mixed with other "stuff." Their results were not that good at all. They theorized the problem was due to diamond not being able to deform like silver particles - thus preventing the surface-to-surface contact that is needed for conduction.

A few weeks back, the front page reviewed an research article that used carbon black as the main ingredient in TIM. The performance was somewhat linked to its ability to flow and deform under pressure.

Interesting, very interesting.
Well found!

Deformation of the diamond should not be much of a problem. Under adequate pressure the diamond would in fact imbed into the heatsink creating a self leveling surface.

The diamond powder is available in many mesh sizes (mesh is another sizing system that is finer than grit counts). I've got 50,000 grit diamond that I use in lapidary work.

In theory you could make your oil base, add the desired grit size of diamond powder and all the diamond particles would touch both the heat sink and the cpu.

Now the question is, what grit or mesh size is the most appropriate?

prankstar008 said:

good deal too....less than arctic silver...let us know how it works!

Click to expand...

It wont work well, and would be a detrimental mistake to even attempt to use this product, its a cleaner not a thermal material

eaglescouter said:

The diamond powder is available in many mesh sizes (mesh is another sizing system that is finer than grit counts). I've got 50,000 grit diamond that I use in lapidary work.

Click to expand...

Even at that particle size it's not suitable for a thermal filler. Not many people have the facilities to pressurize the product, now smelting the diamond powder is something that has me curious, but then again it requires a bit of heat and pressure to melt diamond dust. But even then the problems are smaller pits and valleys wiould still be an issue

Silversinksam said:

Even at that particle size it's not suitable for a thermal filler. Not many people have the facilities to pressurize the product, now smelting the diamond paste is something that has me curious but then again it requires a bit of heat to melt diamonds.

Click to expand...

What size is appropriate? I was thinking that larger might be better so that the same diamond particle contacts both the sink and the cpu with no oil barrier.

eaglescouter said:

What size is appropriate? I was thinking that larger might be better so that the same diamond particle contacts both the sink and the cpu with no oil barrier.

Click to expand...

Its immaterial as the sized no matter how minute will result in pits and valleys that make the material problematic, sure diamonds have the best W/cm-K, but using them so far has not been realistically accomplished in a cost effective manor


Thermal Properties of Materials
Thermal Conductivity, W/cm-K
Metals
Aluminum 2.165
Beryllium 1.772
Beryllium-copper 1.063
Brass 70% copper, 30% zinc 1.220
Copper 3.937
Gold 2.913
Iron .669
Lead .343
Magnesium 1.575
Molybdenum 1.299
Monel .197
Nickel .906
Platinum .734
Silver 4.173
Stainless Steel-321 .146
Stainless Steel-410 .240
Steel, low carbon .669
Boron Nitride (hot pressed) .394
Diamond (room temperature) 6.299



Now we saw that a few people had mediocre results with those 1" or so diamond 'pads', but the cost was very high.

Wouldnt diamond, due to its hardness, not only embed itself into the heatsink but also into the core itself? I would think that would not be a good thing.

Diamond abrasive will easily press into copper. If graded to 600 grit (or even 80 grit) and pressed into copper using a lapped steel plate, then the diamond should have a resulting profile that matches the shape of the press plate.

This copper heat sink would be extremely abrasive as the face is diamond impregnated. Gaps will exist between the diamonds, these gaps could be filled with thermal paste of your choice.

This heat sink should make positive contact with the P4 heat spreader and should in fact dig into that heat spreader.

While an abrasive heat sink might pose a problem for an AMD with it's exposed core, the P4's should have no problem.

What am I missing?

Side note: Esentially I am describing the process used to make Diamond Laps for faceting gemstones. The historic method was to press the diamond abrasive into the copper face, in modern times the diamond abrasive is nickle bonded to the face, this stops the diamond from moving deeper into the lap when it is in use. Normal abrasive range is 60 grit to 12,000 grit, finer laps are made by a different process (50,000 to 100,000 mesh).

here's some interesting reading http://www.intel.com/technology/itj/q32000/articles/art_4.htm

And even better: http://www.toshiba-tmat.co.jp/tmat/eng/material/e_cera02.htm where they have a W/m-K of 200! (and yes I realize that this is a different calculation, so how do we convert it?)

And here is a larger list of thermal properties, and it lists thermal grease as ....(I'll let you look it up and be surprised!) http://www.ai.mit.edu/people/tk/tks/tcon.html

No, peeps, the ticket is to convince chip manufacturers to use monocrystalline diamond heat spreaders on the CPUs.

And then, of course, we should all complain about the new CPU price.