Specific Heat Capacities in relation to heatsinks....

[gone_fishin]

Here's a link on a study of microjet impingement for IC's. Note the high velocity/low volumetric flowrate.
Click here

There are many avenues being researched for quite some time, some involve integration into the substrate itself.
Click here

 

[seamadan000]

i heard some mentions of bucky balls, and I think that carbon compounds have a bunch of potential, especially concidering the high thermal capacity of most and the fact that there are more naturally and unaturally occuring variations on the arrangment of carbon atoms than there are known chemical compounds. A pattern of these variations could lead to some interesting solutions, as carbon makes it much easier to channel heat with the many variations on atomic structure, like you guys said, graphite only works in certian dimentions.
________
Hurt From Avandia

 

[ookabooka]

I had a thought. Since diamonds of a large size are way too expensive. What about diamond dust perhaps, it is realatively inexpensive. Then you could use another compound like glue, also with a high thermal capacity, say silver. I am not too sure about the physics of it. It seems that it would atleast make a marginal difference.

edit: anothing promising idea, Manufacture of diamonds has also become a great possibility. They are made by subjecting graphite to high temperatures and pressures, and a bit of aluminum (I think) to absorb excess nitrogen (again. . . i think) that may discolor it during production. Not only are they much cheaper, but also more "perfect". They actually gain the ability to luminesce (sp?) under UV light. They also glow after the light source has been turned off, but I cannot recall what the technical name for it is. . .

 

[JoeC]

Here's an interesting article to ponder:

Electroosmotic Microchannel Cooling System for Microprocessors

 

[RedDeathDrinker]

Now that's the sort of technology we're looking for. If a laptop system can shift 130W/hour, then imagine what a mid-tower one could do........

 

[Malpine Walis]

Nevin (the guy who makes AS3) actually tested diamond dust as a transfer material. His opinion after mixing several batches is Here. Bottom line is it will not work.

As far as manufactured diamonds go, the aluminium binds the nitrogen atoms together so that they do not discolor the finished product.

Color not being a major issue, we might be able to get by with the yellow synthetic product. However, due to the hardness, a single piece diamond heatsink may be a long time coming. If there is a source for blocks of the stuff, it might be interesting to see if it could be used for peltier cold plates.

 

[Loud]

Y'all are still stuck in the box. How about immersing the heat source (CPU) directly in a fluid to disperse / remove the heat?

Large mainframes have been running chilled water directly over hot boards for years. Granted, it makes things interesting - everything has to be sealed to keep the water and electricity apart.

Of course, there's also the famous Cray II, which was immersed in a cryo bath.

I saw something recently about using a mineral oil bath to cool a mobo. It actually looked promising - the temps were good, and the oil was so inert that it ran fine. He even took the mobo out of the bath, made a mod (swapped out the video card, I think), and put it back without incident.

I'll look for it so I can post a link.

EDIT: Okay ... I've read up on this. There seem to be a LOT of posts on mineral oil. I guess it's not the best cooling solution (pun intended ). But I submit this theory is still a good one. We seem to need a better fluid, and / or a way to immerse only the CPU (not the entire mobo).

 

[Malpine Walis]

I am not sure what you consider "outside the box" thinking (or even why that belongs here). However I did some goggle searches to see if superconductivity might have cooling applications.

I found that thermal superconductivity has just recently been discovered. At the moment it is still limited to the very low temperature type 1 superconductors. However, with the advances that have come about in the past five years it is certainly concieveable that the situation might change.

Also, several labs are working on superconducting microprocessors. With the current record for type 2 superconductors being -138c, such a chip will need to be immersed in liquid nitrogen to make it work.

Finally, I found a company in California that sells synthetic diamond heat spreaders. The website does not have prices, which leads me to suspect that they are probably priced to be less than an optimal engineering solution. I will make a point to call them this week just to find out what the buggers cost.

 

[JigPu]

On the issue with buckyballs, there may be another promising type of substance known as buckytubes. I don't know the specifics of conductivity or specific heat, but I would bet that they would be better than buckyballs for heat transfer because of their tubular stucture. Instead of the heat traveling through a "pile" of buckyballs, it could move vertically along many individual buckytubes kind of like a heatpipe.

JigPu

 

[Caffinehog]

Heatpipes are great at moving lots of heat, but they need high differences of temperature before they are particularly effective.
Copper moves heat in three dimensions. HOPG, a type of graphite moves heat well in two dimensions, but poorly in the third.
Buckytubes would be great at moving heat in one dimension, but poorly in the other two.

This means that the heatspreading ability is related to the 3rd power of the size of the metal, the 2nd power of the size of the HOPG, and the first power of the size of the buckytubes.
So, to get the same effect as twice the copper, you'd need about 2.8 times the graphite, or 8 times as much buckytubes.
So the optimal shape of a metal heatspreader is a disc, the optimal shape of a HOPG heatspreader is a highly elongated oval, and the optimal shape for a buckytube heatspreader is a really long, narrow cone, with the thinnest part on the cpu.
Taking into account the extra volume requirements, you'll see that metal is the easiest material to fit. And some heatsinks are currently approaching humongous size.

 

[Malpine Walis]

Based on a bunch of searches on google and alltheweb, I have found that doped fullerenes have been found which are (type 2, high temperature) electrical superconductors. Whether they are thermal superconductors as well remains to be seen.

In any case, fullerenes are at the cutting edge of science. Practical devices are a long way off.

Fullerenes fall into two categories. Balls and tubes but considering that they are discreet objects on the condensed matter level; they challenge our concept of solids and liquids (if there are any condensed matter physicists around here fell free to correct me). A sample of fullerenes large enough to be relevant to what we are doing could behave more like a liquid or more like a solid depending on how they are prepared.

Absent some decent condensed matter software, I am not going to hazard a guess as to how they relate to cooling but from what I have seen, it seems as if a sample of pure bucky-balls would probably assume a liquid form at the macroscopic level and would have a thermal resistance lower than pure diamond.

 

[Cathar]

I want to see some evidence first that we're even close to doing the best that we can do with conventional cooling methods. It seems to me that we're starting to get close with air cooling with some very optimised designs coming onto the market, but for water cooling we're still quite a ways off from a practically achievable "best" point.

I'm all for advanced materials for use in moving heat, but I'm also still of the belief that many of these efforts are attempting to solve problems that are more related to the inappopriate use of the more conventional materials that are in use today, rather than actually hitting a real limit as a result of the use of these conventional materials.

Of course I look forward to the day where I am proved totally wrong on this, but until then...