Carbon or Copper??

[coolmiester]

Just wondering if anyone had any experience of using carbon as a direct replacement over copper for the base plate on a waterblock.

I know its a heavily debated subject but if anyone has first hand feedback i'd be interested to know either good or bad.

[Tony]

Been there done that, copper is better

[majestik]

Been there done that, copper is better
/thread
hehe

[Tony]

I had a base made for the new OCZ block, tested it at length....copper killed it hands down. Carbon does not give up its heat to water well, although the carbon moved heat from one end to the other much faster.

20C difference, was the outcome, totally wierd, we came to the conclusion the water did not interact with the carbon very well.

[Zaskar]

Copper is better then normal carbon unless you can get your hands on real carbon nanotube plates. And at that cost you could get silver cheaper (though carbon nanotubes I think even beats silver)

[Tony]

Ohh I was using good stuff ;) trust me; its not the thermal conductivity you need to question, its the ability to shed that heat to water

[coolmiester]

Ohh I was using good stuff ;) trust me; its not the thermal conductivity you need to question, its the ability to shed that heat to water

Sorry, i should of made it more clear..............i'm after calibrated results.

Can you point me to something with more substance - maybe you logged your findings??

[twwen2]

Ohh I was using good stuff ;) trust me; its not the thermal conductivity you need to question, its the ability to shed that heat to water

I thought that carbon nanotubes were going to be the next best thing in waterblocks, but not anymore! :p:

Perhaps you could use a copper plate on the top of the carbon, so the water only interacts with the copper but you still get the superior conductivity of the carbon?

(However i'm sure you've thought of all this already;))

[Webby]

How thick would this copper plate be? It would need to be really thin or else it will become a bottleneck (it may anyway) in moving the heat and the gains from the carbon will be negated.

[twwen2]

I imagine it would definately be a bottleneck. I'm not sure if thinner is necessarily better, i think there's a balance. I would think on the thinner side though...

[Webby]

Well you have three thermal resistances in play the resistance from carbon to copper then the resistance of the copper and finely the resistance from the copper to the water the thinner the copper the lower its thermal resistance and in this case the better however, if the three resistances add up to more than the resistance of moving the heat directly from carbon to water you will only damage your performance.

[Sadasius]

Perhaps you could use a copper plate on the top of the carbon, so the water only interacts with the copper but you still get the superior conductivity of the carbon?



That just may be a brilliant idea...Instead of a copper layer how about a carbon/copper mix material block with a thin copper layer on top. This way the copper in the mix can channel the heat to the copper layer to be removed by the water.

I think I found a patent on the mixing of copper and carbon along with it's process.

http://www.freepatentsonline.com/6238454.html

[NaeKuh]

I had a base made for the new OCZ block, tested it at length....copper killed it hands down. Carbon does not give up its heat to water well, although the carbon moved heat from one end to the other much faster.

20C difference, was the outcome, totally wierd, we came to the conclusion the water did not interact with the carbon very well.

WOW!

Tony thank you for solving this riddle.

I was also curious on how it would perform.


Well you have three thermal resistances in play the resistance from carbon to copper then the resistance of the copper and finely the resistance from the copper to the water the thinner the copper the lower its thermal resistance and in this case the better however, if the three resistances add up to more than the resistance of moving the heat directly from carbon to water you will only damage your performance.

Tony said its there. The only problem is getting the heat out of the tubes and into the water. I am too riddled by his failed experiment. I really thought it would work.


That just may be a brilliant idea...Instead of a copper layer how about a carbon/copper mix material block with a thin copper layer on top. This way the copper in the mix can channel the heat to the copper layer to be removed by the water.

I think I found a patent on the mixing of copper and carbon along with it's process.

http://www.freepatentsonline.com/6238454.html
A hybrid? oh man, thats too expensive for our poor vendors.

[coolmiester]

I kinnda knew the original question would bring all sorts of weird and wonderful answers but sorry to be fickle and just to re-iterate the original question....

Just wondering if anyone had any experience of using carbon as a direct replacement over copper for the base plate on a waterblock.

I'm not really talking of a hybrid at this point but more of a direct replacement.

[Tony]

Sorry, i should of made it more clear..............i'm after calibrated results.

Can you point me to something with more substance - maybe you logged your findings??

I told you all you need to know, 20C difference, same load , same rad, same pump, same block, 1 base Vs the other, same design on both...1 copper 1 carbon (ultra conductive)

I would show you the base but I destroyed it trying Twwen2's idea pretty much, theory being it may give up its heat to copper better than water.

Edit: found some pics, I can't show you the top of the base as we will release soon, but you can see the underside is exactly the same.

[coolmiester]

I had a base made for the new OCZ block, tested it at length....copper killed it hands down. Carbon does not give up its heat to water well, although the carbon moved heat from one end to the other much faster.

20C difference, was the outcome, totally wierd, we came to the conclusion the water did not interact with the carbon very well.


I told you all you need to know, 20C difference, same load , same rad, same pump, same block, 1 base Vs the other, same design on both...1 copper 1 carbon (ultra conductive)

I would show you the base but I destroyed it trying Twwen2's idea pretty much, theory being it may give up its heat to copper better than water.

Edit: found some pics, I can't show you the top of the base as we will release soon, but you can see the underside is exactly the same.

OK Tony, i appreciate your time :up:

So just to reiterate, your saying you got a 20C drop from Carbon base to Copper base in a like for like test??

........and even though there is a 20C difference, your still going ahead and releasing a Carbon base or a new Copper base??

Thanks

[Zaskar]

Tony that may be a high quality piece conductive carbon, but im almost positive you diddnt use a carbon nanotube plate.

When I said expensive were talkin like $80-$120 a gram. The molecules of the carbon are engineered to create complex micro tubes that carry heat or electricity incredibly well.

Quick pic depicting the molecular arrangement of a carbon nanotube:

http://upload.wikimedia.org/wikipedia/commons/7/76/Kohlenstoffnanoroehre_Animation.gif

With carbon nanotubes there is a specific thickness that is optimal due to the length of the nanotubes, the more separate non continuous strands between to 2 surfaces you have, the lower the efficiency of the heat transfer, much more so then a metal like coper or silver.

[Sadasius]

Funny I was just reading about carbon nanotubes sheets being pressed with copper for better thermal and electric conductivity for laser diodes.

[Navanod]

Diamond base plate FTW~~!!!! LOL

But on a serious note, it may be that the surface charges on the carbon repels water molecules due to hydrophobic interactions? Adjustments to the pH of the water may improve things but that means only specific coolants can be used

[NaeKuh]

OK Tony, i appreciate your time :up:

So just to reiterate, your saying you got a 20C drop from Carbon base to Copper base in a like for like test??

........and even though there is a 20C difference, your still going ahead and releasing a Carbon base or a new Copper base??

Thanks

CM he said the carbon sucked hardcore. He saw a 20C disadvantage over the copper.

[Zaskar]

Diamond base plate FTW~~!!!! LOL

But on a serious note, it may be that the surface charges on the carbon repels water molecules due to hydrophobic interactions? Adjustments to the pH of the water may improve things but that means only specific coolants can be used

Nice :D

Also the carbon nanotube plates dont move heat in the way metals do, they transfer it along the grain of the tube, so its prettymuch directly from 1 side to the other without spreading outward much, metal draws the heat away in all directions.

[NaeKuh]

Tony that may be a high quality piece conductive carbon, but im almost positive you diddnt use a carbon nanotube plate.

When I said expensive were talkin like $80-$120 a gram. The molecules of the carbon are engineered to create complex micro tubes that carry heat or electricity incredibly well.


http://www.hardwaresecrets.com/article/437
The base is made of carbon nano-tubes with a built-in pump and a regular fan on its other side – i.e. inside HydroJet.

So yes im pretty sure he used something along that grade.

Dont underestimate Tony. :rofl:


Now i think we know why it never came out.

[twwen2]

Yeah that OCZ thing never showed up. I was wondering how it would perform.
:shrug:

[Zaskar]

Ehh first time I read that I figured they were BSing on the carbon nanotube thing and were just using some carbon plate :P
That or they used some kind of nanotube piece thats as thin as tinfoil and glued it to a copper base, just so they could say it had a carbon nanotube base.

That stuff is just too expensive to be includded in anything right now id think.

Just throw that item up there with their phase change unit, OCZ sucks :P

[Kayin]

My CF/chopped fiber packed resin plates were utter :banana::banana::banana::banana:e. Amazing thermal transfer, well above copper, but as was postulated they are hydrophobic, and my experimentation with coolants was limited at the time. Stuff was an ever loving :banana::banana::banana::banana::banana: to make too. Never logged it, nearly went to the hospital cause I didn't observe ventilation protocols.

I have two projects in various states of work, and a third that I have the materials for that I'm not afraid to talk about openly. The third is working with CuSil, if I can get a torch that will make it. propane didn't cut it, and I think that MAPP and a better crucible will cut down on the oxidization I was seeing. Now all I need is a vacuum pump to combat oxygen trapped int he alloy and we should be good.

Both of the other ones are cold-cast, non-electrically conductive but may not perform as well as traditional blocks. The benefits would be extremely rapid prototyping, workability increases and utterly insane casting detail with off-the-shelf stuff as well as ultimate repeatability with almost zero lead time.

I've dropped enough hints that someone in the know would be able to figure it out. If you do, you get a cookie. I really think I have little to fear from anyone here, but I still am reluctant to just blurt it all out cause you never know who's not registered and still reading...

[Fairydust]

Maybe a thin layer open cell copper foam is worth a look, although I guess that it will work far better as a material for radiators on the airflow side.

[Tony]

Tony that may be a high quality piece conductive carbon, but im almost positive you diddnt use a carbon nanotube plate.

When I said expensive were talkin like $80-$120 a gram. The molecules of the carbon are engineered to create complex micro tubes that carry heat or electricity incredibly well.

Quick pic depicting the molecular arrangement of a carbon nanotube:

http://upload.wikimedia.org/wikipedia/commons/7/76/Kohlenstoffnanoroehre_Animation.gif

With carbon nanotubes there is a specific thickness that is optimal due to the length of the nanotubes, the more separate non continuous strands between to 2 surfaces you have, the lower the efficiency of the heat transfer, much more so then a metal like coper or silver.

You guys keep missing my point here, the carbon is awesome at moving heat, the issue is the molecules of water don't remove the heat from the carbon very well...the two just do not work well together. Hydrojet may have suffered from this also...being honest i was not on that project but I do know weight of the unit was one of the reasons we dropped it, it was just tooooo heavy ;)

So you could use nonotubes,the overlying issue though is while they wick heat from the cpu really well dumping that heat to water within the small footprint of the block base is hard. If you had a foot long heatsink with 6 inches of submerged fins in a cooling medium it may work, the carbon transfers the heat along its length really well and with you having so much of it the heat may dump to what ever cooling you have better than it did under my tests.

I just wanted to show you guys we do try all sorts of weird a wacky experiments at OCZ, trying to push the envelope is what we are about, often though stuff like this does not work.

[Tony]

CM he said the carbon sucked hardcore. He saw a 20C disadvantage over the copper.
thanks man, for a min i thought i had said it wrong, you are correct

[Webby]

During my PhD I was working with porous adsorbents many of which were hydrophobic (activated carbons and polystyrenes) however we wanted to use them in water to do this we would first wet the material in Acetone and then slowly replace the acetone over time with acetone water mixtures of increasing water content finely you would just have water and the surface is perfectly wetted :) This assuming you didn't do it may have been the problem with the carbon water transfer a microscopic layer of air between the carbon and water would make the heat transfer a lot worse.

[karbonkid]

During my PhD I was working with porous adsorbents many of which were hydrophobic (activated carbons and polystyrenes) however we wanted to use them in water to do this we would first wet the material in Acetone and then slowly replace the acetone over time with acetone water mixtures of increasing water content finely you would just have water and the surface is perfectly wetted :) This assuming you didn't do it may have been the problem with the carbon water transfer a microscopic layer of air between the carbon and water would make the heat transfer a lot worse.

Interesting - so you're saying that the carbon naturally forms a layer of air above its surface? The treatment you talk about - would it last long enough for a consumer application, or would the air packets reappear as soon as the block dried out again?

Also, when we say carbon, what exactly are we talking about? Graphite? Certainly not carbon-fibre or carbon nanotubes, right?

@Tony - Thanks for a little insight into the OCZ way of working. I simply did not believe that such a large company could be focussed on innovation and not just profits. Seems like it would be a great place to work (You aren't in need of any architects, are you? :rolleyes:).

[Webby]

The carbons I was using are activated carbons which were made in house. Basically if you have ever broken open a water filter the filtering material inside is likely to be activated carbon (they ship them pre-wetted) these have tiny pores in the material which can allow 1 gram of the material to have upwards of 1000m2 of surface area. The wetting problem comes from the fact that the water would much rather be next to other water than the carbon so small holes/groves on the surface (when I say small I'm talking nanometre sized) do not get wetted and trap a small pocket of air. The acetone treatment would only be effective if the carbon was kept wetted if it was dried then the process would have to be repeated.

[eXa]

But if the only problem is getting the water to really get contact with the carbon, couldnt u coat it in copper or silver or something. Kinda like nickle plating. But maybe that wouldnt get into the nano sized holes\groves either? I would believe that if that can be done the heat transfer of the carbon would outperfomr the loss made by going from copper->carbon->copper in the block.

[Webby]

The most efficient method would be to chemically modify the surface of the carbon to incorporate hydrophilic (water loving) groups this would eliminate the need for any additional layers of material and should make the wetting process a non-issue. Not a chemist (a chemical engineer :)) but i'm 99% sure it can be done, the issue would be if the modification would effect the transfer of heat from the carbon to the water my initial thought would be not but I wouldn't be able to say for sure.

[Hijack]

The problem with carbon, AFAIK, is has nothing to do with surface area, its the fact that its thermal conductivity is orientation based, as it is with most crystaline based materials (minerals). It's conductivity is great when running in parrallel with it's crystaline structure, but crappy when it is perpendicular. That's why nanotubes are great structures, as they can be oriented in specific directions and offer the perfect structure.

The whole trick is not only getting a material that is thermally conductive, but also has good emissitivity (releases its absorbed thermal energy). The two things (again, AFAIK) are usually operating in opposition. Those that are thermally conductive, don't wanna give up their heat once absorbed.

I'm only learning about this stuff now from a practical application point, so please don't crucify me for my remedial knowledge on the subject. At any rate, I work for an organization where I have access to the highest grades of high emissive materials available. Even before seeing this thread, I was starting to think about testing the application of these materials to highly conductive materials for thermal transfer. Basically, coating the cold side of waterblocks and radiators. The downside is, this crap is massively expensive ($2000 US per quart). I'll be sure to post here any and all findings.

[Mankz_91]

Would the carbon base work better with a different coolant, like mineral oil or Glycol or something?

[HESmelaugh]

Would the carbon base work better with a different coolant, like mineral oil or Glycol or something?

How about air? Did I miss something or has there been no mention on how carbon would do on an air-cooler?

[mike8913]

what about sandwiching carbn between two thin strips of copper. You can utilize the conductivity of carbon t transfer the heat more quickly, but the copper is sort of like the on and off ramp to get on the thruway (carbon)

[Sadasius]

what about sandwiching carbn between two thin strips of copper. You can utilize the conductivity of carbon t transfer the heat more quickly, but the copper is sort of like the on and off ramp to get on the thruway (carbon)

That is exactly what I was thinking. This would create a super thermal pump. The carbon gets it started quickly and transfers it to the copper then to the water. So basically you start a sandwich with the thickest carbon layer on the bottom and then a copper one then another carbon and then another copper. The carbon sheets are so thin they are practically invisible so it will require many layers. Now to make the block work correctly it requires a press that can put 60000lbs for 4 hours minimum. After that you CNC the channels in the copper top and voila you have a super thermal conducting block that will have cost you about $2000. Give or take a few bucks and tax and shipping....:rofl: Oh and this has been tested and there is a patent. The US Department of Defense put a lot of money into this research and supposedly these heatsinks work better then diamond. Food for thought!

http://www.freepatentsonline.com/y2007/0050870.html

[NaeKuh]

guys then your transfer is limited to that sp of copper.

your not improving anything.

your going back to the same debate in trying to prove alu/copper hyrbid sinks are better then pure copper sinks.

:rofl:

[mike8913]

no, carbon, unlike aluminum, is good at accelerating energy, not expending it. If you can match the speed of Carbon heat transfer with the expenditure potential of carbon you could have something really efficient on your hands.

But like Sadasius pointed out...it doesn't seem to be economically feasible.

[Sadasius]

That's is the biggest problem right there. In the patent it is proven to have a better thermal conductivity than copper alone. So it will work. Just the price of materials is ridiculous and then having to manufacture it afterward would require much more xtreme members then here to buy them that have very thick wallets just to gain what....maybe another 5 degrees of OC potential headroom? Here is the price for 100 grams of Single Walled Carbon Nanotubes....$8000.00. Seriously! Here is the linky with the price list of the materials. http://www.cheaptubesinc.com/cntpricelist.htm. Luckily a waterblock only needs 25% mixture of this stuff to copper being the rest.

[NaeKuh]

Bah..

im sad... i really wanted this to work.

Seems like water is truely at a dead end. :\

[Zaskar]

Hey its new tech, give it a few years. Right now most that stuff is probably made in a lab specifically for your order :P

Everything gets massed produced given enough time.

[STEvil]

What about mixing carbon and copper dust together then compression forming it?

[Hijack]

You can't form both material together into a usable block by pressure alone (at least nothing short of geologic), especially together. It would be like wet sand, even with a binder.

[leuler]

The problem with carbon, AFAIK, is has nothing to do with surface area, its the fact that its thermal conductivity is orientation based, as it is with most crystaline based materials (minerals). It's conductivity is great when running in parrallel with it's crystaline structure, but crappy when it is perpendicular. That's why nanotubes are great structures, as they can be oriented in specific directions and offer the perfect structure.

The whole trick is not only getting a material that is thermally conductive, but also has good emissitivity (releases its absorbed thermal energy). The two things (again, AFAIK) are usually operating in opposition. Those that are thermally conductive, don't wanna give up their heat once absorbed.

I'm only learning about this stuff now from a practical application point, so please don't crucify me for my remedial knowledge on the subject. At any rate, I work for an organization where I have access to the highest grades of high emissive materials available. Even before seeing this thread, I was starting to think about testing the application of these materials to highly conductive materials for thermal transfer. Basically, coating the cold side of waterblocks and radiators. The downside is, this crap is massively expensive ($2000 US per quart). I'll be sure to post here any and all findings.

Emissivity is a measure of how well a material absorbs and emits radiant
thermal energy.Not much use for a waterblock.

Effusivity ( or thermal inertia ) is a measure of how well a material absorbs
and emits thermal energy via conduction. Effusivity determines how easily heat
can be exchanged at the surface of a material. Diamond has the highest
effusivity, followed by copper, then silver.

[leuler]

That's is the biggest problem right there. In the patent it is proven to have a better thermal conductivity than copper alone. So it will work. Just the price of materials is ridiculous and then having to manufacture it afterward would require much more xtreme members then here to buy them that have very thick wallets just to gain what....maybe another 5 degrees of OC potential headroom? Here is the price for 100 grams of Single Walled Carbon Nanotubes....$8000.00. Seriously! Here is the linky with the price list of the materials. http://www.cheaptubesinc.com/cntpricelist.htm. Luckily a waterblock only needs 25% mixture of this stuff to copper being the rest.

In the patent, it seems to say that the nanotubes become oriented so that
the longitudinal axes are parallel to the surface. This would improve the
heat spreading in the heatsink, but it would reduce the ability for heat to
leave the surface of the heatsink. So more surface area would be needed.
The most important feature of combining copper and carbon nanotubes
seems to be the ability to match the thermal expansion of the heatsink
with the device it is cooling without degrading the conductivity too much.

[leuler]

Bah..

im sad... i really wanted this to work.

Seems like water is truely at a dead end. :\

There is alot of research being done investigating nanofluids, especially
their thermal properties.

[Nater]

When I said expensive were talkin like $80-$120 a gram. The molecules of the carbon are engineered to create complex micro tubes that carry heat or electricity incredibly well.

You could make a base plate out of cocaine for that kind of money! But seriously, I have a friend who messes around with carbon for heat dissipation applications (non computer). His opinion pretty much backs up what everyone has said here, at least when dealing with water.

I don't know about nanotubes, but I'm not sure that the price would really be worth the benefit. I don't think that water cooling is at a dead end per say. Except for perhaps some big gains in the early days from better designs, water cooling has always advanced through small improvements over time. Most technologies aren't compatible with Moore's 'Law'.