Carbon Black Experiment?

[RoadWarrior]

Okay, few observations about fillers in a heatsink compound, say if the best metallic conductor there when rendered into particle form made pyramid shaped particles, at one side of the interface or other there would basically only be point contact. Therefore a substance that is less conductive but has a more convenient particle shape to offer more surface area of each particle to the mating surfaces and/or a smaller particle size and thinner layering is going to do better.

Now, copper as you know is a good conductor, but say we made it into copper wool and stuffed it round something hot, it would probably then be a fair enough insulator. It's not the material it's the air it traps.

Saying you can't make an insulator out of a fairly good conductor, or a thermal interface material out of a fairly poor conductor is on the same level as saying you can't make a boat out of steel because a block of steel sinks.

Zinc oxide paste which is still a production TIM, and is still just a small percentage away performance wise from exotic current compounds uses a very poor conductor, Zinc Oxide. Zinc oxide by itself is really truly a bad conductor of heat, it's probably worse than the average microwave safe coffee mug. So if it's performing within a few percent of silver compounds, there's obviously something not quite as straightforward about TIM design as you would like to beleive.


Anyway, on to patenting. Well a quick search on google revealed evidence of prior art using carbon black as a thermal filler, so probably only the specific formulation can be patented, I don't see much hope of there being a very broad patent issued, which leaves us guys considerable leeway.


Interesting observations about evaporation, that hadn't occured to me. I was presuming that the application was just as good when dried out as long as the joint was not disturbed.

regards,

Road Warrior

 

[MILLTHERM]

Zinc oxide paste which is still a production TIM, and is still just a small percentage away performance wise from exotic current compounds uses a very poor conductor, Zinc Oxide. Zinc oxide by itself is really truly a bad conductor of heat, it's probably worse than the average microwave safe coffee mug. So if it's performing within a few percent of silver compounds, there's obviously something not quite as straightforward about TIM design as you would like to beleive.

Excellent point and I think the papers authors point was one of contact.

Steel wool between two metal screens 1/10 of an inch thick is used on some jet engine exhausts 800 F on one side the other you can rest your hand on it.

Steel wool mixed with epoxy or fiberglass also gives you a really tuff composite.

 

[Freeloader]

By Owenator:
I'll try to explain this phonomena. I don't think evaporation is excatly the correct description. A better word would be boiling. At the temperatures a die reaches the water will be boiling. The effects of boiling on cooling are actually very good because water changing to a gas (what boiling is) is a great way to remove heat. I work at a power plant and we use this idea daily. We lwt the water just start to boil on the surface of our fuel this removes orders of magnitude more heat that simple conduction. In effect you have a convective cooling as well as the energy absorbed by the phase change from liquid to a gas. It is similar to the phonomena in a refrigerant system where the working fluid boils to absorb heat in the evaporator.

Are You saying that water boils at 40C (water temp is 20C for the waterblock). If the die reached the temps at which water boils, it would already be fried (100C)

 

[Owenator]

You are right water does boil at 100C. I was thinking of the temperatures that the paper used (150C) which made me think of the boiling phonemena. My mistake.

But, localized temperatures on the die surface can be higher than 40C. As I understand it there are local "hot spots" on CPU's that may well be at higher temperatures. I could be wrong. Your test may not have made the water boil but could have made the water convect, or move because of the temperature differential. I am just guessing this. I do also think that you could still get some evaporation based cooling because the water would be a little warmer than the surrounding air gap. Unless the air was already full of water (100% humidity) you will always get some evaporation. That's why things dry in air, like laundry a table top etc. Unless you test these liquids in a vacuum there will be evaporation. Higher temperatures add to this effect.

I guess my point is that liquids are much less viscous than grease so they don't behave the same. They move about on a molecular level much more and this can lead the TIM actually providing some cooling effect other than purely filling in the gap for heat conduction.

Still I must say that testing with water as a TIM was a bold and commendable test. I don't think I would be as brave.

O

 

[Nanotherm]

Water works as a TIM due to its high Specific Heat (Specific Heat = 1, everthing else is 0.XX). Water has a tremendous ability to be able to store energy and then release energy. Consequently, water is one of Nature's best "Phase Change Materials." Not only does it "Phase Change" from a solid to a liquid, but it also changes phase from a liquid to a gas. Great stuff - you just can't live without it.

The problem is - as already noted - that water evaporates at any temperature above the freezing point. Water evens "sublimes" at temps below freezing. The bottom line is that water goes away no matter what. Therefore, it is not a great choice for a long term TIM solution, in and of itself. If you add some things to it, though, then it becomes a whole different story...

 

[Lord_MiL]

I have been reading into this foil by RTN and I'm a little confused. I understand how it would be beneficial to have a "local heat source" as they put it that would allow solder to melt and join the chip and heatsink without damaging the chip, but they also state that this would prevent the "thermal mismatch" which occurs when two bonded materials expand and contract different amounts when heating/cooling. They don't ever say how this is accomplished. As I mentioned before, I've looked into Indium solder which is supposed to solve this exact problem. Indium allows thermally mismatched materials to be soldered with ease.

Anyway, reading their webpage has again encouraged me to attempt the ultimate TIM..........soldering a heatsink to a chip. Now, once I get $40 I'm gonna buy a 1700 and see what happens!

 

[Freeloader]

By Nanotherm
The bottom line is that water goes away no matter what.

Water only evaporates untill the surrounding air reaches 100% humidity. In a extremly small sealed area, that should occur with very little loss to the total volume of water used.

I disagree that the benefit I noticed using water had much to do with it's specific heat. As I understand it, specific heat indicates how many degrees a substance of a certain volume will rise when it absorbs a certain amount of heat. The lower the specific heat rating of a liquid, the more degrees it will rise. This should have very little effect on a T.I.M. joint

 

[Nanotherm]

Specific heat or specific heat capacity is only part of the equation, as is thermal conductivity. Both are minor parts of the equation, mind you, when compared to the ability and efficiency of a material in "wetting-out" the surfaces, filling in the macro / microscopic imperfections on the contacting surfaces and gapping Air out of the thermal junction. These factors probably account for about 50 - 60% (or more) of the efficiency and effectiveness of any thermal compound, irregardless of the material's thermal conductivity, specific heat or what have you. Hence, the fact that water works quite well as a TIM - at least for the period of time that it perform the functions above - and before it evaporates as it eventually will (since the inside of your system will never run at 100% humidity, 100% of the time - and you don't want it to). If you would like me to elaborate on this point, I would be more than happy to do so.

 

[Freeloader]

I think you failed to realize that in my test, I sealed the processor to the bottom of the waterblock by using a bead of thick grease about 1 1\2 inches in diameter encircling the die. The humidity inside the case is not important. I was only refering to the humidity of the very tiny space within the seal.

As far as water having a great ability to fill in the micro crevices, I've being saying that from the start. There still needs to be better tests to prove or disprove that this is enough to overcome water's lower thermal conductivity though. Then we need to find a suitable replacement as I'm sure most people aren't too keen on shorting out their motherboards.

 

[Nanotherm]

Got you - understand. Sealed-off - OK. There are, in fact, some other alternative TIMs out there that might fit the bill along the lines that you are investigating.

 

[Owenator]

I think you failed to realize that in my test, I sealed the processor to the bottom of the waterblock by using a bead of thick grease about 1 1\2 inches in diameter encircling the die. The humidity inside the case is not important. I was only refering to the humidity of the very tiny space within the seal.

Ah, I see what you were doing now. That is quite a clever idea! If I follow you correctly you were essentially trying to have a little water "pocket" or "gap" that would be the heat transfer interface between teh cpu and heatsink, ingenious! I have been thinking of trying to make a heat pipe but they still use a sealed container that still needs thermal compound to contact the cpu. I think what you describe is sort of a fluid filled interface between the cpu and heatsink. This could have some interesting possibilities. The water has direct contact with the cpu and can transfer the heat by conduction but will probably also move around to transfer the heat by convection. If you could keep the water "sealed in" so that none escapes it could work for longterm cooling. Kind of an active TIM that in itself helps transfer heat. How long of term tests have you made?

O

PS You should patent this idea if it hasn't been already!

 

[Nanotherm]

It's a great thought - as long as you can keep it sealed - and perhaps you include some thermally conductive particles in there to promote heat transfer via direct thermal conductivity between the surfaces. Of course, there is already a liquid TIM out there there does this (and more) - and you don't have to seal the "edges" to prevent evaporation and/or oxidation of the TIM.

 

[xgman]

I think in the meantime, we can take the lessons of the research which came to the pretty obvious conclusion that "the performance of TIM's is mainly governed by the conformability and speradability and conformabitlity of the material rather than the thermal concuctivity."

This would lead to the conclusion that TIM's like Shin Etsu would have higher performance if they were simply easier to spread.

 

[Nanotherm]

That is an appropriate and "spot on" conclusion, as our friends across The Pond would say.