Heat transfer capabilities?

[DaWiper]

Do anyone know what the heat transfer capabilities of micro-silica is? I'm wondering cuz the particles of micro-silica is extremly small. 1/50 to 1/100 the size of sement grains(web info). If the thermal capabilities are good than MAYBE this could be used for homemade thermal paste???
Anyone?

 

[L337 M33P]

Silica? Isn't that the stuff they make blast furnace thermal protection bricks out of?

 

[DaWiper]

micro silica is a dust extracted from the smoke when making/melting ferro-silisium. Currently used to make concrete stronger and waterproof.

 

[Skulemate]

... Currently used to make concrete stronger and waterproof.

Umm, sort of. However, that's beside the point, eh?

I just recently attended a talk given by a leading researcher in the field of materials science, and she noted that when dealing with thermal interface materials (TIMs) the conformity of the paste is often of much greater interest than the paste's thermal conductivity. In fact, she found that a simple carbon black paste out performed many of the pastes offered up as a premium TIM by industry. In some cases the available pastes were so bad that it was better to go without than to use the pastes, though I doubt that this is the case regarding the TIMs that are common to our community.

[Edit] Oh, should you happen to get a hold of some silica fume to make your homemade paste, be sure that you wear some sort of protection when dealing with it. Due to its very high fineness, it can be harmful if inhaled (the fume is finer than cigarette smoke). Though it is not as well known, silicosis is still a very serious respiratory illness (similar to asbestoses). [/Edit]

 

[DaWiper]

Well my point is that because of the small particles it will penetrate deep into the micro cracks of the metal making the contact between the cpu and the hs closer to 100%... Anyway, this is just a theory of mine. If the heat transfer capabilities of MicroSi. is crap then it can't be used. But if it is good...
The reason I'm asking is that I can get hold of microsilica... I'm currently packing the stuff (as a job) in 1200kg bags.

 

[Skulemate]

... I'm currently packing the stuff (as a job) in 1200kg bags.

Wow, you must be Superman to be able to handle those bags.

What I meant with my last post is that you may be on to something. Due to the very fine particle size the silicon will have a good conformity, which may allow it to perform adequately even if its thermal resistance is not the greatest.

 

[DaWiper]

Pretty close to superman...the truck atleast...

What do you think will be the best thing to mix it with? Thinking of using ordinary white silicone paste. Have some stuff called Silicone heat AK-100 from Plowstar(?)
<edit>:The stuff I can get is not the stuff thats sold. The density is much finer:
-ordinary densified is 550-650 g/L
-ordinary undensified is 350-450 g/L
-real undensified is about 100-150 g/L. This is the stuff that the other stuff is made of... (lots of stuff,stuff,stuff,lol)

 

[Skulemate]

That's worth a shot, though you'll want to try to get as much of the silica fume into the paste as you can without it becoming overly gritty. Perhaps you could try to find out what the base is for a paste like, say, Arctic Alumina.

 

[DaWiper]

You mean that it shouldn't be too thick... Have thought about it a lot. What I need is something that doesn't dry with thermal capabilities...

 

[Skulemate]

That's correct. Make it too thick and you will lose some of the conformity that is giving you the advantage of the silica in the first place.

 

[DaWiper]

Tomorrow I will try to make a mixture with my current TIM to see if the stuff gets too thick or not. Any ideas on how to test it. I don't feel like putting it on my main rig before I know if it works. Thinking of putting two hs's together with a soldering iron on one and a temperature sensor on the other to see the differences.

 

[Skulemate]

You don't have an old machine to play around with, eh?

 

[DaWiper]

Have some old machines, but they are too old. No sensors... I could attach a sensor... What do you think would give the most "real" result?