Best material for heat transfer?

[MagisD]

Lets say you were given the option of being able to have custom blocks, cpu, gpu, mb etc and rads what would you have them made of ? And why ? Seperate answers for block and rads fine. Price isn't an issue but an estimated cost would be nice.

Restrictions:

Must be comercially available.

Must be able to be worked with common technology of today.

Must last in long term setup 2-3 years.

All this fuss over different materials in upcoming products has got me wondering.

Sent from my X10a using Tapatalk

[jayhall0315]

The best heat transfer material that is commonly available and would fit your needs, is what everyone already uses: copper.

If you have deep pockets and want something exotic, go with silver.

However, I think you will find, that after many hundreds of hours of investment, CAD design, CNC milling, and custom dremeling, you will be hard pressed to better something like the Swiftech Apogee XT, XSPC RX360 rad and Swiftech MCP 655 pump. If you want to go off in a totally new direction (or at least one that many folks rarely have heard of), try a totally silent passive system. There is a worklog showing how a build which is essentially a massive copper/brass radiator, can be done.

[SuperSilent]

cpu block - single crystal man made diamond. with micro pin matrix etched using an high temerature catalytic iron etch process.

radiator - full copper with square pattern fins ultrasonically welded into position rather than soldered

cost could possibly be an issue tho :-P

[jayhall0315]

LOL Silent. A block like that would not be in Magis' house but beside the Hope Diamond in the Smithsonian.

[asura]

SuperSilent, why sell yourself short with copper for the rad, go with silver, as fine silver doesn't oxidise you can fuse the fins to the pipes with a focused hydrogen torch.

With fine silver at a little over £450/Kg as bullion it'll be a little bit steep.



Copper, copper, copper. The leap to silver isn't worth it, it's a leap in price and a baby step in gains.

[kinghong1970]

silver is quite soft, no?

[Alexandr0s]

I suggest graphene. It has a thermal conductivity that's higher than diamond. When they find a good way to mass-produce it (probably in the next decade), cooling applications might actually become viable.

[Metroid]

All made of diamond, can't go wrong with it. :sick:

[MagisD]

Just a note this is just a big what if ?

I have no intention of making any or it.

Welll maybe a silver cpu block for my birthday. Its that or a couple of 30 inch screens. Hehehe

Sent from my X10a using Tapatalk

[Metric]

I suggest graphene. It has a thermal conductivity that's higher than diamond. When they find a good way to mass-produce it (probably in the next decade), cooling applications might actually become viable.I was under the impression that graphene was hydrophobic...? :shrug:

[BlueAqua]

I was under the impression that graphene was hydrophobic...? :shrug:

I haven't seen anything where is has been hydrophobic. I've been studying some semiconductor material lately and I agree that graphene would make an excellent choice. It has essentially no band gap which enables free electron flow, more than any other substance. This translates to being the best conductor for electricity and has excellent thermal properties, better than diamond.

It's pretty interesting stuff. I doubt it will be used anytime soon in heat transfer applications but talk has been to use it as a thin sheet over solar cells to connect the cells together electrically without sacrificing the amount of surface area each cell has, since it's essentially transparent.

[SuperSilent]

admitily graphene would be awesome - if it wasnt fundamentally only 1 atomic layer thick. perhaps graphite machined with its sheet latice structure perpendicular to the heat spreadder base. that would give you a k value of 2x diamond but only inplane. but that would severly limit conduction through the aprox 1/2 of the base not directly connected to a micro fin in the water path

hmm or perhaps single walled carbon nanotubes in a silver matrix :)

if only someone had £1.5 milion to spend on a waterblock :P

[MagisD]

Hehehe at that point might as well make the case out of silver for extra passive convection cooling.

Sent from my X10a using Tapatalk

[MagisD]

Can ya tell I was bored with the inlaws over for canadian turkey day and couldn't hide in the man cave i call a den.

Sent from my X10a using Tapatalk

[mingbogo]

How about gold? Diamond an gold what else not to like?

[zanzabar]

Hehehe at that point might as well make the case out of silver for extra passive convection cooling.

Sent from my X10a using Tapatalk

silver looses conductivity when it tarnishes and in open air it will tarnish quick

[Metroid]

Perhaps pure liquid copper instead of water :rofl:

[meanmoe]

nano liquid copper with nano diamond particles

I actually read a brief article a little while back about an extremely porous ceramic block being used as a heat exchanger. The material naturally formed with 'channels/mesh' inside... They showed what I assumed to be microscopic pictures of it - I'll see if I can find it again..

[MagisD]

How about gold? Diamond an gold what else not to like?

Call that one the OG block?

Sent from my X10a using Tapatalk

[terramir]

Well, silver and multi crystalline diamond only have a 10% and 100% advantage over copper not worth it considering the costs and difficulties, graphite is too brittle and single crystalline directional graphite must have a clear axis between the cold and hot-sides for it to work. Again too much of a pain in the arse to be worth it unless your in outer space. Then again in outer space there are other ways of dealing with heat produced that are much simpler to accomplish.

What could work but it would be a pain, micro heat-pipes inside every pin, how difficult would this be?
First you make the copper water block, then from the bottom you drill holes into every pin made, then you put less than a 10th of the volume of the water ethanol mixture you will use for the heat-pipes, add a few microfine copper wires as wicks, now freeze the piece mind you right now the piece is upside down with a bunch of holes where there is liquid and a bunch of hair fine copper wires inside. Now you freeze the block and mixture that is almost at the top of the pins down to like -50C so it's rock solid and it won't evaporate, then you pump out the air from the chamber this frozen workpiece is in and at less than 6% of atmospheric pressure and with a very very cold workpiece you weld shut all the holes in the surface, now you have the almost optimal base plate for your water block :p: now you make rads that are built out of copper completely out of copper and your in business.
mind you the surface area of the radiator would have to be 10x of the surface area of what a rad is today, and well then your water will have a negligible difference to the environment so that the work that you just did with that waterblock does not go to waste, and well also one more thing that water-block has an operating temp from -10 to 50C max so if your water temp ever exceeds 50C the whole thing becomes useless, however the bottom plate of the water-block will transfer heat to the water like nothing ever used before, because of the latent heat of vaporization the temp I adjusted the air pressure for was about 50 C boiling point, of course it will start evaporating at 10C easily and well even above that it will keep moving the heat to the top of the pins like nothing you'd ever seen, and as long as the water there keeps flowing by taking that heat away you should be fine.
but to produce such water-block you'd need 100k plus in equipment, on the cheap it might be able to be done if you use a vapochill to cool down the bottom of the water block while welding the holes shut. if you were fortunate enough to find yourself a glovebox that could withstand such a vacuum, and you had a pump to continue evacuating it while you weld the holes shut. then polish the surface with a bow ever so light and your in business. this was my dream passive heat-sink design adapted for a water block.
but in the heat sink design I would have used copper pipes and threading a solid base plate and well closed the tops up with copper material threaded as well and then soldered shut, hmmm about 50 heat pipes would have done to make a heat-sink that could handle about 250W and keep it with-in 10C of environment at all times :D and well there the making is much simpler, just boil the water and let the steam push out the air. once it cools down the pressure inside the heat-pipes would be virtually non- existent. :D
my $.02
terramir

[Church]

Imho much simplier and cheaper to make would be TEC+common waterblocks of these days. Pitty those add insanely high powerdraw if you plan to cool hot cpus of today and thus are were more fit for older/cooler cpus dumping less heat.

[asura]

silver looses conductivity when it tarnishes and in open air it will tarnish quick


:down:


Sterling silver oxidises because of the copper in the alloy. Fine silver is not alloyed, what you call tarnishing is oxidising. As I already pointed out, fine silver does not oxidise.

And once more with feeling, fine silver does not oxidise. :yepp:

[terramir]

Imho much simplier and cheaper to make would be TEC+common waterblocks of these days. Pitty those add insanely high powerdraw if you plan to cool hot cpus of today and thus are were more fit for older/cooler cpus dumping less heat.
Hmmm TEC's well I used the freezone, all a matter of keeping the delta T down :D. The freezone, although not a novel concept (someone built something like it I read about in some forums before they came out with it. A modified freezone setup with a rad in the loop (which will act as a heater on idle I know) could draw less than 65W of power including fans and pump and do A very good job indeed. However how exactly to modify it so it's the most efficient I have not figured out yet :D


:down:


Sterling silver oxidises because of the copper in the alloy. Fine silver is not alloyed, what you call tarnishing is oxidising. As I already pointed out, fine silver does not oxidise.

And once more with feeling, fine silver does not oxidise. :yepp:

You know I need to try that :p fine silver as a base plate for my cpu and other blocks, wonder if the 6% is worth the price of 23 dollars an oz.
lol.
terramir

[asura]

Remember the difference between bullion price and the cost of some usable metal.

e.g. at the time or writing fine silver is at £470.044/Kg, the thickest usable sheet I can buy is 3.2mm and works out at £667.48/Kg for up to 50g, and £646.59/Kg thereafter.

That's more than a third extra of the "worth" of the metal just to get it into a usable form. And that's from the UK's biggest supplier of bullion to jewellers.

That would mean spending over £345 on sheet silver to make an Ag copy of EK's supreme HF. Or a little over £60 for a new base for a Ybris Black Sun.

[MagisD]

Somebody need to put out a limited edition silver bling bling block. LOL

Sent from my X10a using Tapatalk

[Church]

This (http://shop.aquacomputer.de/product_info.php?products_id=2509) won't do?

[MagisD]

Didn't even know that existed. But its good to know. To bad I dislike the aesthic design of it.

Sent from my X10a using Tapatalk

[asura]

Somebody need to put out a limited edition silver bling bling block. LOL

Danger Den did once upon a time.


This (http://shop.aquacomputer.de/product_info.php?products_id=2509) won't do?

Quite simply, no.

Is sterling silver not fine silver, sterling has a lower thermal conductivity than fine silver. And a lower thermal conductivity than copper. It also has a lower effusivity than copper. And a lower effusivity than fine silver.

All this theoretically means is that it will perform worse than the same block in copper and the same block in fine silver.

However, it needs testing, it could work out differently, theory is all well and good, needs testing.

[Church]

asura: I certainly never going to buy waterblock that expensive, but even if i don't know about thermal properties of silver of that purity used in it, i really doubt AC would release it if it were to perform worse in real life then it's common copper analogue. Shoggy claimed it cooling better, i don't remember though by how much.

[Gimmpy224]

a solid copper container full of mercury chilled at -30c with the mobo and what not sitting in its center?

[asura]

That's why I underlined the "theoretically" I can't imagine that it would perform worse, but the theoretical numbers say it will...

I'd love to see the exact same block design in copper, sterling silver, and fine silver tested. However, unless it's a home grown block the chances are very very low, why should someone make a sterling and a fine silver block, would just defy reason.


Re-reading the previous, my "quite simply no" made it sound like I was defiantly saying no, the tone was supposed to turn out as a balanced "I don't know" but it didn't, apologies for any confusion.

[zacrobmer]

Who needs a block? Someone needs to invent a direct water to die method of cooling, kind of a block with a small rectangular hole in its bottom just smaller then the die and an epoxy (or some other sticky water tight) seal. It would probably be cheaper then fabricating a diamond! :up:

[Church]

Hmm, actually fresh and interesting idea. I'm guessing that something still would be neeeded to enhance turbulence, but still, water directly on IHS of cpu .. interesting. No more trying to make base of waterblock thin as possible while still mechanically keep strong enough .. I wonder who'll be out there as first one with such block. Idea seems promising.

[asura]

Bigest possible drawback (other than shorts) is the lowered surface area for the thermal transfer to water... Another one of these - let's experiment - things.

[zacrobmer]

The issue would be it would probably have to be chip dependent, and would probably have to be a permanent mount. An o-ring or pressure mount would not be enough, it would have to be epoxied or some other strong sealing bond. As far as turbulence, the inlet being directly above or slightly offset with slope that "bounces" the water off the chip would probably suffice.

[Waterlogged]

I wouldn't bother trying to to pull the heat away from the source faster, instead, I'd look at ways to expel it from the loop faster. All the methods above will still rely on a rad/heat exchanger putting the heat into the air. Let's find a way to improve the amount of heat air can absorb. . .then we can worry about pulling the heat from the source faster. ;)

[MagisD]

That's why i started the thread,i had some misconceptions and wanted to see what people came up with.

Sent from my X10a using Tapatalk

[mlwood37]

So if some one created a liquid for water cooling that contained a graphite you would love it ......

http://en.wikipedia.org/wiki/Boron_nitride

http://img804.imageshack.us/img804/7543/dscf7446.jpg

White graphite in liquid form for water cooled PCS ...

[MagisD]

Lmao I'm still waiting to test your gun metal grey. So make it graphene gun metal grey.

also wifes cut my wc budget off, but promised me I could get something pretty for my birthday in april.

I also have a artist/jeweler friend who's wow'd with the modding sub-culture and wants to try making hand made blocks. Dammit she does nice engraving so an idea sorta half formed.

Debating funding that or getting my 30 inch screens.

Sent from my X10a using Tapatalk

[mlwood37]

This needs proper testing ive done some testing my self but i relay need it to go out for external testing as as my results are invalid in reality. Ive been holding back about talking about it but since no one can help might as well just show it for now and see if i can get some one interested in testing it.

[werdwerdus]

the direct die water block has been done, i'm sure i remember seeing it somewhere

[Sadasius]

Who needs a block? Someone needs to invent a direct water to die method of cooling, kind of a block with a small rectangular hole in its bottom just smaller then the die and an epoxy (or some other sticky water tight) seal. It would probably be cheaper then fabricating a diamond! :up:

Been done and performs much worse then with a waterblock to disperse the heat. There is something about it floating in the dead threads of the past.

Here it is...http://www.xtremesystems.org/forums/showthread.php?t=225354&highlight=Direct+Die+watercooling

[BeepBeep2]

Quite simply, no.

Is sterling silver not fine silver, sterling has a lower thermal conductivity than fine silver. And a lower thermal conductivity than copper. It also has a lower effusivity than copper. And a lower effusivity than fine silver.

All this theoretically means is that it will perform worse than the same block in copper and the same block in fine silver.

However, it needs testing, it could work out differently, theory is all well and good, needs testing.


Apparently you guys don't realize that baseplates are under a bunch of pressure (and actually concave so when you apply pressure it ends up even) and that fine silver would not be able to take the same pressure as copper and still keep it's original form.


Been done and performs much worse then with a waterblock to disperse the heat. There is something about it floating in the dead threads of the past.

Here it is...http://www.xtremesystems.org/forums/showthread.php?t=225354&highlight=Direct+Die+watercooling

No jet plate to increase flow (directly into) the die? ...

[Church]

I wonder if better results then with previous directdie cooling tries cannot be reached with something from both camps:
1) use not direct die, but with IHS in place, as it's much larger then die of chip;
2) to enhance turbulence and a bit heat take off area use something like "inverse" block - eg. plate with microchannels turned arround so that tips of those microchannels touch IHS
EDIT
Though on second thought removing IHS and it's TIM so that base of waterblock can act in place of it might be more effective, as probably there won't be much extra heat dissipation from not directly connected/soldered to IHS tips of those microchannels.

[terramir]

Remember the difference between bullion price and the cost of some usable metal.

e.g. at the time or writing fine silver is at £470.044/Kg, the thickest usable sheet I can buy is 3.2mm and works out at £667.48/Kg for up to 50g, and £646.59/Kg thereafter.

That's more than a third extra of the "worth" of the metal just to get it into a usable form. And that's from the UK's biggest supplier of bullion to jewellers.

That would mean spending over £345 on sheet silver to make an Ag copy of EK's supreme HF. Or a little over £60 for a new base for a Ybris Black Sun. Actually silver casts real nice and well it it's quite sane if you have the tools it would not be much of a problem. so you just take that bullion melt it down and cast your block :p:


This (http://shop.aquacomputer.de/product_info.php?products_id=2509) won't do? Well sterling silver is an umbrella for various .925 silver alloys, if it's copper it's alloyed with and uses less than .001 nickel the transfer rate should still be better than copper although not as good as pure silver. That block probably uses cheap sterling with a bunch of impurities so probably perform worse. then again most copper out there they use for blocks is not pure stuff either but 110 which has a bunch of other junk in it. so none of the cpu block's I've seen use pure .999 copper so none of them reach the theoretical maximum of 401 silver was like 426 so pure copper would be cool as well , just one catch pure copper about as soft as silver.


Apparently you guys don't realize that baseplates are under a bunch of pressure (and actually concave so when you apply pressure it ends up even) and that fine silver would not be able to take the same pressure as copper and still keep it's original form.

actually that is not quite correct, while it is true that pure silver has a hardness of 2.5 to 3 pure copper only has a hardness of 3 as well copper alloy's used in water blocks are harder, but then again that stuff doesn't transfer heat as well, so long story short non of our blocks are pure anything (except acetal or delrin tops, however the fuzion v2 uses a delrin top that has some sort of composite integrated) . Pure silver should work fine with a decent bow and some sort of o-ring or insert (like the dtek) pressing on the center to bow it.
Anyways silver blocks would be cool, however I dunno if someone actually would like to put the R&D in to really investigate that.

Another idea Add pelt's that are water-cooled and use two loops and one giant rad (to keep the water with-in 5 C or environment and well you could get below environment temps with just about 60W of pelt cooling power :p:

Second idea, forget the res and the rad, use pure distilled water and a little fountain to cool your water, catch is you'll be paying about $0.50 to $2 a day on evaporation of the distilled water and you'll need a good filter to prevent junk from the air contaminating your blocks LOL. but trust me a little mountain splashing the water the latent heat of evaporation will chill that water down real quick, need about a 2 gallon pond to make sure it dun run dry lol. (but then again no too much work):rofl:
terramir

[iddqd]

You used to be able to get silver blocks, the Storm ones come to mind. IIRC, the benefit wasn't worth the price tag, so most people ended up buying the copper version instead.

[zeroibis]

This (http://shop.aquacomputer.de/product_info.php?products_id=2509) won't do?

I would buy this purely on the basis that my computer being called the Silver Bullet 2 would be much more awesome with a silver heat sink. I just wish there was one with a clear top and a silver base at reduced cost...

[werdwerdus]

i think i remember the dangerden tdx came in silver at one point...

[meanmoe]

I wouldn't bother trying to to pull the heat away from the source faster, instead, I'd look at ways to expel it from the loop faster. All the methods above will still rely on a rad/heat exchanger putting the heat into the air. Let's find a way to improve the amount of heat air can absorb. . .then we can worry about pulling the heat from the source faster. ;)

WL, I'm with you here... The blocks aren't the choke point for me. It'd be nice if there was something smaller/better/more efficient than the current line of radiators at removing heat from the loop. And can we go ahead and get it out of the house while we're at it. Soon too, first week in Nov may be XS WCG Supercomputer week part II. edit: think feser is going to do it? you should check back with your contacts there ;)

[Waterlogged]

WL, I'm with you here... The blocks aren't the choke point for me. It'd be nice if there was something smaller/better/more efficient than the current line of radiators at removing heat from the loop. And can we go ahead and get it out of the house while we're at it. Soon too, first week in Nov may be XS WCG Supercomputer week part II. edit: think feser is going to do it? you should check back with your contacts there ;)

Hmm, my source doesn't seem to think so. :ROTF: