Has anyone ever heard of or tried a Carbon Fiber or Carbon\Carbon cpu waterblock

[rocketcity]

This is my first post, so "Hello to All!"

Carbon Fiber & to a greater extent Carbon\Carbon are used in many applications where it is necessary to dissipate huge amounts of heat. I used to build & fly high power rockets & noticed that while heating one side of a Carbon Fiber fin with a heat gun, the other side of the fin cooled to almost room temps as soon as the heat gun was removed. Formula 1 cars use it for their brake rotors & pads, as well as the now defunct Space Shuttle used it on the leading edges of the wings, etc. It is not the cheapest material on the planet for sure, but we are talking about a water block, not a house, so the price of it shouldn't be completely over the top. I have looked around & have not seen anything about them, but just because no one has tried something should not be a reason to never try it. Unfortunately I developed a severe allergy to the epoxy that one would use to bind it & can not be around the stuff (uncured epoxy) any more, unless I am ready to have my last BSOD. Thanks in advance for your opinions/ideas.

 

[wagex]

good idea but its less that 10% as conductive as aluminum which aluminum is less conductive than copper what most blocks are made of :\
http://www.christinedemerchant.com/carboncharacteristics.html

Air .024
Aluminium 250
Concrete .4 - .7
Carbon Steel 54
Mineral Wool insulation .04
Plywood .13
Quartz 3
Pyrex Glass 1
Pine .12
Carbon Fiber Reinforced Epoxy 24

 

[Conumdrum]

The parts on the shuttle, the heat tiles are a great insulator. Brakes are carbon because they absord and dissipate the heat, they don't transfer it well.

We want massive heat trasnfer from the CPU to the copper block, the better we can get.

So it's backwards from what we need/use.

 

[Bobnova]

It's an interesting concept, nobody has tried it that I know of. Would it work? I don't think it'd work all that well.

On the brakes it's mostly that they can survive insanely high temperatures, glowing orange to yellowish. Anything else at that temperature is turning gummy, which isn't ideal for brakes

Now a diamond CPU block would transfer heat like nobodies business, but it'd be a bit expensive. The extra connections between carbon atoms in diamond changes how it conducts heat significantly.

 

[rocketcity]

Thanks for all of your replies. Most of the heat tiles on the Shuttle are insulators, but they are more of a ceramic. In over 50 years of hands on building racing engines, I have not felt any material that transfers heat like carbon fiber. Brakes turn velocity into heat, then they must be able to dissipate that heat or turn into a hot jelly. I scratch built many rockets out of carbon fiber, & used kevlar to reinforce different places. Carbon Fiber is not very good at absorbing extreme shock, but I don't think a water block would be subject to much shock. I wish that I wasn't so allergic to the really strong epoxies, then I could just build one up & see how it works. The type & pattern of the weave as well as the type of epoxy used all make a big difference on the properties of Carbon Fiber, I find it difficult to put a number on "Carbon Fiber" as one category, much different than a solid metal. A Diamond is basically a piece of carbon.

Have a fantastic day!

 

[Ivy]

Highest possibility might be carbon tubes filled with silver while the surface is made with diamond dust meltet into a polished silver compound. But who can afford it?

Thermal conductivity:
Copper: ~400
Silver: ~420
Diamond: ~2500
Carbon Nanotubes: ~6000

Your idea is not entirely wrong, there is in fact certain carbon tubes who are able to transfer heat 2.5 times better than a diamond is capable of. But the problem is how to attach them to the surface and then how to take them over to the sink, thats pretty advanced stuff and silver is surely very useful to make it work. Carbon is the most complicated material on earth and the one with highest potency, even for cooling, but its the hardest stuff to implement, probably still to advanced for todays technology.

But thats just theory because even if it does exist, all we currently got are prototypes, its just to advanced. Its made with nanoparticles, a very challenging technology.

Measurements show that a SWNT (single waved nanotubes) has a room-temperature thermal conductivity along its axis of about 3500 W·m−1·K−1; compare this to copper, a metal well known for its good thermal conductivity, which transmits 385 W·m−1·K−1

Besides, on water cooling. Water itself is not a good heat conductor, but its able to dissipate huge amount of heat once it arrived (so its simply a buffer). Still, most advanced cooling is probably to transfer the heat to a extremely huge sink, over nano tubes. And then at the sink a fluid such as water (or LN2 for benchers) can be used to be way more effective than air. Im not even sure if its more effective to bring a die or IHS directly into contact with LN2 because the surface is simply to small. It could be even more effective to transfer the heat to a very huge block over carbon tubes and then cool down the entire sink using LN2 in a large area. When the heat dissipation is large enough, it can go down to extemely deep temps. Although the weak spot is how to attach the IHS or DIE to the nano tubes and the entire construction itself.

Raw LN2 is a bad conductor but its a very agressive dissipater, way more agressive than water because of its reaction which is known as vaporizing and creating a huge cooling effect. But it could be ineffective on dies who are continuously getting smaller and smaller. Someday we will need a better solution than a raw vaporizing effect and the question on how to transfer heat will even hit the extreme benchers i guess.

 

[Conumdrum]

I think in these exteme ideas all of us, if not done before need to understand the properites of Unobtainium. It's a big Wiki material issue, Wiki it!

Make it so #1!

 

[nightelph]

Now a diamond CPU block would transfer heat like nobodies business, but it'd be a bit expensive. The extra connections between carbon atoms in diamond changes how it conducts heat significantly.

You can buy a cubic centimeter of synthetic lab diamond for around $10. So a 10x10x1cm block would be a $1000. I'm sure it'll get cheaper. Imagine that for the base of a ln2 pot?

 

[Bobnova]

Whoa, that's cheap. Finally! It's not like the things are all that rare.
Think 10x10x5 would be cheaper?

 

[TsunamiJuan]

I would think that the level or required precision needed for it to designed for the high carbon/carbonfibre composites would probably outweigh its availability. At least at this point in time. Not to mention the cost and availability. Would have to look into this more it sounds promising. I would also think that one would need to put some decent time into studying the epoxy since that will effect the finished product quite a bit.

Though I wonder how well it would handle when you took it cold. Bad enough when your memory pops out of the sockets when you freeze your ram, would suck to crack the base of your pot due to an impurity or manufacturing mistake.

 

[nightelph]

Whoa, that's cheap. Finally! It's not like the things are all that rare.
Think 10x10x5 would be cheaper?

I'm sure it'd be cheaper. And I'm willing to bet its the future of cooling. 5-10 years?