Pyrolytic Graphite Cooling Solution

[lucas122478]

It's expensive, but has anyone tried real world applications with heat dissipation with this method? How would you go about making your own heat sink? I've found a website where you can buy pyrolytic graphite adhesive as well as pyrolytic graphite plates.

Its much more light weight than aluminum and copper, its like 6 or 8x better - its hardness is like that of diamond. So this will suck heat out of the CPU much quicker than copper or aluminum So designs on how to utilize this? these blocks at the site are 1/4" thick and vary in Width. Would a fan mounted on top be good enough to provide better results at cooling than the typical copper or alum setups? Placing copper or aluminum fins bonded with the graphite adhesive to the graphite block be worthwhile or pointless?

Let's say buying the graphite adhesive in replacement instead of artic silver, and buying the 1/4"x2x2 graphite block and mounting it on top of the CPU, with the adhesive inbetween it and the CPU, then placing another spread of the adhesive on top of the graphic block, and mounting a heatsink on top of that, would we see a massive improvement in heat transfer from the CPU? Any geniuses out there that can give some insight on this?

 

[DerwenArtos12]

sorry but, it's AB thermal conductivity is almost identical to copper and lower than most heatpipes.

 

[wwswimming]

i've seen this used for digital modules that were conduction cooled
& part of a larger radio. very expensive.

 

[aigomorla]

this this thing sumular to carbon nanotubes?

If it is, thanks for the link, i sent it to a few custom builders to see what they think of it.

If its good, you can expect me getting some custom blocks milled out of it within a month to try it out on water.

 

[lucas122478]

Keep me updated. Everything I've read points towards this stuff being better than copper. And I also came across articles about research groups trying to find a cost effective way to create lab diamonds on a mass productive and cheaper scale to start using artificial diamonds as a cooling alternative.

At scifitoys or scitoys forgot the website, they had a kit you could buy for this stuff, to experiment with. And in the description it said to press firmly against a cube of ice and within seconds you would feel the ice on this stuff. it weighs 25% that of copper. Looking at $170 for every 1/4x2x2 - So maybe if could get a custom slab of it at 2x2x2 that would be like $800-1000 maybe.. lol. If you got the money to burn why not (if it really works)

Here's another link for you guys to check out and get more information on this.
http://www.minteq.com/fileadmi...F/M-PY-640-AT-PB-7.pdf

Tailored thermal conduction from 450 -2000 W/m°K for Pyrolytic Graphic
Thermal conduction for Copper 401 W·m-1·K-1

Minteq claims their PYROID HT Pyrolytic Graphite typically has a 1200-1400 W/mk matching that of CVD Diamond, and maximum of 2000 W/mK - So if this stuff averages 3x yield better than copper and up to 5x yield better than copper at 1/4th the weight of copper. That will be the ultimate Heat Sink, unless you get a 50 carat Blue diamond tailored into a heat sink, lol.

Also wanted to point out to Derwen, think you were looking at Natural Graphite its plane thermal conduction is 440-500W/mk which is still a little better than copper. But Pyrolytic Graphite is different. Different Graphites vary in thermal conduction.

 

[DerwenArtos12]

I was actually refering to their spec sheet here

If they have differing informatin elsewhere I just must not have seen it. If you've got the money to burn(aigo) I don't see why not, seems to me that it would work exceptionally as a directional thermal transfer material but, you'd more than likely need to use another material for the fins were it an air cooling solution. Also, the standard channel or pin grids wouldn't function the same in a waterblock because of it's thermal transfer stricktly across the AB plane. It's thermal insulation across the C plane would make it a little harder to design a block. On that same note, it would be an IDEAL product for sub-ambient cooling for that exact same reason. With it's AB transfer it could pull heat extremely efficiently from the CPU while insulating the sides of the block/evap which would make overall insulation of the CPU MUCH easier.

 

[lucas122478]

Maybe should pay close attention to the grade they sell at graphitestore.com - maybe its of the lowest quality. GRADE: PG-SN Prolly should get in touch with someone at minteq since they actually manufactorer the stuff, where as graphitestore just sells it. They maybe selling inferior quality P.Graphite for the cost of superior P. Graphite.

 

[lucas122478]

Okay apparently there's 3 different types of "PG" There's the SN that graphitestore offers, which is pointless to buy, because it is on scale with copper and then there's "CM" which also is on scale with copper. Then there's "HT" that Minteq manufacters and they claim it is greater than 5 times copper, and matches the performance of CVD diamond at 1700 W/mk

http://www.globalspec.com/Feat...mal_Spreaders_/36525/0


So i'd steer clear from graphitestore, I'm going to email someone at Minteq later this week and get more information and if there are any companies that sell their product.

edit: i went ahead and sent them an email stating the skepticism against their claims against the data spread for other forms of PG. Requested for data showing comparisons between their product against copper heatsinks. Maybe end up getting them to mold a heatsink and ship it to anandtech to have a review on it lol.

 

[DerwenArtos12]

Slick sh!t. too expensive for me though.

 

[lucas122478]

Well the reply to my email was brief. pretty much just replied back with what we already know. but he did reaffirm it being 1700W/mk


Thank you for visiting us at GlobalSpec

The data on our website on thermal conductivity for our material was derived through independent flash diffusivity measurements. I believe you are talking about subnucleated and continuous nucleated versions of pyrolytic graphite when you refer to SM and CM. As to your question on the comparison of heat sink materials, I will leave this question to your technical abilities to the effects of copper (400 W/mK), or aluminum (200 W/mK) against our material at 1700 W/m K.

Regards,

Richard Lemak
General Manager
MINTEQ International, Inc.
Pyrogenics Group
640 North 13th Street
Easton, PA 18042
www.pyrographite.com

Phone: 610-250-3398
Fax: 610-250-3325
Email: richard.lemak@mineralstech.com
E-mail Date/Time: 04/01/2008

 

[aigomorla]

Originally posted by: lucas122478
I will leave this question to your technical abilities to the effects of copper (400 W/mK), or aluminum (200 W/mK) against our material at 1700 W/m K.

HOLY COW! MUST BUY!

That make one hell of a waterblock.

 

[PCTC2]

Originally posted by: aigomorla

Originally posted by: lucas122478
I will leave this question to your technical abilities to the effects of copper (400 W/mK), or aluminum (200 W/mK) against our material at 1700 W/m K.

HOLY COW! MUST BUY!

That make one hell of a waterblock.

Mill one for me too?

 

[aigomorla]

Originally posted by: PCTC2

Originally posted by: aigomorla

Originally posted by: lucas122478
I will leave this question to your technical abilities to the effects of copper (400 W/mK), or aluminum (200 W/mK) against our material at 1700 W/m K.

HOLY COW! MUST BUY!

That make one hell of a waterblock.

Mill one for me too?

You know what i just realized... I dont think a normal mil could cut this. The thing might be too brittle. :\

 

[DerwenArtos12]

Originally posted by: aigomorla

Originally posted by: PCTC2

Originally posted by: aigomorla

Originally posted by: lucas122478
I will leave this question to your technical abilities to the effects of copper (400 W/mK), or aluminum (200 W/mK) against our material at 1700 W/m K.

HOLY COW! MUST BUY!

That make one hell of a waterblock.

Mill one for me too?

You know what i just realized... I dont think a normal mil could cut this. The thing might be too brittle. :\

would require a diamond/carbide bit. PLEASE read my post earlier about the issues with using this for a water block. The AB transfer is amaizing but, the thermal insulation it gives across the C would make waterblock design difficult. The only thing I can think of to try and improve on that flaw would be to coat the internal diffusion area with silver. It would hinder total heat transfer slightly but it would distribute the heat from the horizontal surface areas to the verticle surfaces of a pin array and probably be better for total heat disipation on a side to side block like the apogee series. With a block like the fuzion or supreme, the greatest ammount of heat diffusion comes where the water hits the tops of the pins so, exposed PG may work best on a dual but, with a quad or a nehalem, I'd still have to suggest coating it.

 

[Rubycon]

Originally posted by: DerwenArtos12


would require a diamond/carbide bit....

What about one of those 55kPSIG boron slurry water jet things?

BTW I read the title as prophylactic graphite. :laugh:

 

[lucas122478]

Well under their technical data for the PYRIOD it says this:

http://www.minteq.com/our-prod...id-pyrolytic-graphite/

PG has excellent thermal conductivity, approaching copper in the a-b direction, whereas it acts almost as a ceramic on the 'c' direction. In an annealed PYROID® HT state, the thermal conduction properties increase up to four to eight times that of copper and aluminum, respectively (thermal conductivity as high as 1700 W/m°K). We are able to selectively bond the planes through special fixturing to take maximum advantage of its directional conductivity. In this regard it is an excellent material in special thermal management applications, including lids and heat sinks.

So they can set it up for heat sinks supposebly? Thickness can be made up to 1 inch thick, but can stack them on top of each other for more thickness if needed, and diameters up to 18 inches, they can also manufactorer this stuff into cylindrical tubes, could have them custom make a heatsink hollow it out, and the tubes made custom fit them into it, have the tubes extend a good distance out, its density is close to 2.25 g/cc and makes it technically non-porous so water can't leak?

The chart down at the bottom shows the a plane being 1700 w/mk and then c plane being 7 w/mk , So what would happen if you have a custom block made, with say 8 cylinder pyroid tubes coming out of the side of the block leading up, into another block above that, what would be consider the C plane on a cylindrical tube? Would that eliminate the insulation problem from the c plane?

 

[aigomorla]

what there trying to say is that from top to bottom, the conductivity is simular to copper.

however in spreading the heat sideways and thoughout the panel, its awesome?

Care to correct me if im wrong derwen?

 

[lucas122478]

So what are we talking about reality wise, if this would actually work at 1700 w/k how much of an improvement would you see? if heatsink A is running on a 3.0ghz cpu, with an overclock up to 4.0ghz at temps around 50C at load, then with this Pyroid heatsink would you see it running at 10C since its 4x better than Copper? or running an overclock of 7.0 ghz at 50c? Anyway to give a round about figure on what kind of improvement you would yield?

 

[DerwenArtos12]

Originally posted by: aigomorla
what there trying to say is that from top to bottom, the conductivity is simular to copper.

however in spreading the heat sideways and thoughout the panel, its awesome?

Care to correct me if im wrong derwen?

EDIT:Cliff notes, from the chip to the top of the chunk it would transfer really well but from side to side it would insulate really well so it would take one of a couple specialized designs to get the proper surface area to actually work. It is feasable though.

EDIT2: and from side to side, I don't mean it would transfer about the same as copper or aluminum, it would transfer about 1/57th the heat from side to side as copper but from top to bottom it would transfer about 4.25x as much as copper.

Actually they've furthered my thesis of the materials inability to transfer across the C plane and added that in the specialty we're considering using they've actually anealed the PG so the molecules are better aligned in a single direction and stretched it to increase efficiency so it can transfer from one side of the material to the other but, it will not transfer heat along itself. They've gone a step further to stack the layers directionally so that it will transfer ehat more evently. Essentially think of it like a bunch of strands on the molecular level. When they first make it it's really eradic and random. They heat then cool it very quickly to line up all the strands so it's more efficient but, the strands only lay one direction and they transfer heat from one side of the strand to the other but, not along the length of the strand so when they assemble the pieces they turn each layer 90* so that the thermal insulation changes directions with each layer making it more consistant. It will still not transfer across the C plane. That means that heatpipes will not work with it because it would have to transfer across planes which it doesn't do, it just transfers from one plane to another very efficiently and sheds heat to whatever it contacts very efficiently. Were you to want to build a water block with it you wouldn't want the standard pin array layout the sides of the pins wouldn't transfer heat into the water just the surface of the base and the top of the pins. Teh most efficeint way to increase the usable surface area would be to creat pyrimid shaped pins so you have a lot of B surface area with the A surface against the chip and the C area being the sides of the block. Even a design similar to the stepped plane evaporator design used in phase cooling would be good, lots of A and B surfaces but, it would have to be a HUGE block. We're talking, 1.5" tall with 6-10 open plane chambers 1/16th to 1/8th tallwith an almost equal sized closed plane for stability and pressure retention with nozzles between the planes, not just openings like in the evaporators.

 

[DerwenArtos12]

Ok, this is where I prove I'm nuts, I'v already got a design started but, it's based around my thermalright block. If I could get some dimensions on the d-tek, the ek supreme and the apogee GTX I'll see if I can't desing a base for each to be built out of this stuff. I think the d-tek or supreme would probably be the best bets, though my thermalright would actually make a good basis for a block out of this stuff, essentially just designing a new base to use the exsisting tops of these blocks using a pyrimid shape to maximize the B surface area.

 

[shyby]

Hello ,

So according your experience , What is better ?

a. Attach a panasonic graphite pgs sheet alone to a heat source plane . ( 1700 w/m-k)

b Attach a panasonic graphite pgs sheet and on top of it attach an aluminum finned heatsink . ( 1700 w/mk graphite sheet + 200 w/m-k aluminum finned heatsink )

 

[BonzaiDuck]

Maybe something like 7 years ago when I first used nano-diamond paste, someone sent me an web-link to an article about a fem Chinese PhD doing research on various forms of carbon as a TIM solution. The news was promising. Of course, micronized diamond is just carbon.

So one has to ask what this bodes for heatpipes, since they might be even more effective with such a heatsink base and fins. Could they, for instance, make a heatpipe out of the same material and still boast a million-years MTBF?

Other news in the last year foretold of heatpipe or vapor chamber tower coolers using a different fluid of more favorable vaporization and condensation characteristics -- enhancing effectiveness.

Somebody told me that you'd get better cooling by filling your custom-water reservoir with alcohol over water. But the potential for leaks, the possible damage to pumps (propagating leaks), and flammability would make that as bad as a blind man working in his home Methamphetamine lab. That's either a folk prescription or something based on fact. Feel free to comment.

 

[aigomorla]

no... OCZ attempted to do this.. and the results...
http://www.dailytech.com/OCZ+Displays+Carbon+Nanotube+Cooler/article7571.htm

Graphite is hydrophobic... >.<
so the project was doom'd before it even began.

Also great necro... :T

7 Yr necro.... debating if i should lock it and stuff it under a rug.
I'll see how the activity goes.

 

[BonzaiDuck]

I didn't realize what an old thread it was until you mentioned it . . .

So what do you think? Will there be anything coming to the market soon which is significantly better than the units in the 2015 9-way cooler round-up at Anandtech? I don't think the list was complete, but I'm curious about the Keanu Reeves or Reeven Okeanos -- whatever it's called. One review says it performs so well mostly because of the fans, but there's a consensus that the fans are noisy.

I looked into it just a bit more. The 120mm fan tops out at around 1,800 RPM. Both fans are rated around 65 CFM +/- some margin. So the bundled fans are limp. I think the reviewer may have been wrong. But I don't see any resellers in the States who offer it.

 

[lucas122478]

no... OCZ attempted to do this.. and the results...
http://www.dailytech.com/OCZ+Displays+Carbon+Nanotube+Cooler/article7571.htm

Graphite is hydrophobic... >.<
so the project was doom'd before it even began.

Also great necro... :T

7 Yr necro.... debating if i should lock it and stuff it under a rug.
I'll see how the activity goes.

Wow, it's been almost 16 and 1/2 years since I proposed this idea. Smartphone manufacturers now use it in phones such as Apple to dissipate heat. So it has been used in some capacity since the proposal for heat dissipation beyond its previous usage back then.

Since it's hydrophobic, why hasn't anyone tried coating it with a metallic alloy to increase its strength and water resistance? I thought that maybe it would defeat the purpose of its high thermal conductivity, but maybe not. I am by far not the smartest man in the room. So hopefully, if anyone is still following this thread, you can chime in on what I bring to the table below has it been attempted yet or just no one has tried yet, and if not is it viable?

One thing we did not have 16 years ago was AI. So I went through the motions to see what would be viable or not based upon the assumption that chatGPT was trained correctly and the information it spits out was/is accurate. Here's the process below:

I started a session with chatGPT 4o and filled it into the history and then I asked it a series of questions to come up with a viable option.

First, Why hasn't anyone tried coating (PG) in a metallic coating such as copper to increase its strength and resistance to moisture/water?

Going back and forth with chatGPT, I found out it was a no-go due to thermal expansion mismatch and it would eventually cause delamination and/or cracking over time. The top three other candidates for a metallic alloy coating that had near the same thermal expansion properties of (PG) was Invar, followed by Molybdenum, and Tungsten.

Invar has a terrible thermal conductivity of 10-15W/mk. However, Molybdenum is 138 W/mk and Tungsten is 173 W/mK.

Then there was the process of applying the metallic coating to the (PG). Physical Vapor Deposition (PVD), Chemical Vapor Deposition (CVD), and Sputtering were the best options.

Sputtering ended up being the cheapest out of the three.

Finally, the conclusion was this:

Using sputtering to coat pyrolytic graphite with tungsten is a feasible and cost-effective approach. It avoids the high temperatures of CVD, reducing the risk of thermal damage to the PG. Additionally, sputtering is generally cheaper and could provide a protective and functional tungsten coating for your cooling solution. This approach should still offer superior thermal performance compared to traditional AIO systems, leveraging the high thermal conductivity of PG while addressing the challenges of mechanical robustness and moisture protection.

Going a step further, what if we used something other than water that also was non-corrosive and would not damage electronics if there was a leak, and the answer was this:


Even though glycol solutions have lower thermal transfer capabilities compared to water, the superior thermal conductivity of pyrolytic graphite should make the proposed cooling system significantly better than current AIO coolers.

The key factors are:

• PG’s High Thermal Conductivity: Will greatly enhance heat dissipation despite the glycol solution's lower thermal conductivity.
• System Benefits: Glycol solutions provide additional benefits such as freeze protection and corrosion resistance, which can improve the overall reliability of the cooling system.

Therefore, while glycol solutions might reduce the cooling effectiveness slightly compared to pure water, the advanced thermal properties of PG should ensure that the proposed system remains highly effective and potentially superior to current AIO coolers.

Thoughts? I couldn't find where any attempts on this had been done yet?