Liquid-Metal Cooling Loops

[elasticdog]

Just saw this on Slashdot and thought they looked intriguing...


Advantages

• * Very low thermal resistances
  * Silent performance
  * High reliability due to no moving parts and the simplicity of the solution
  * Small form factors
  * High power efficiency
  * Supports extremely high heat flux densities (no longer a limiting factor)
  * Orientation independent
  * Able to easily cool multiple heat sources
  * Scalable heat removal via current control
  * High volume solution based on commodity parts
  * High heat transport capabilities

http://www.nanocoolers.com/technology_liquid.php

[breakspirit]

Any indication of cost? I'm too cheap even for normal water cooling

[frostedflakes]

The system is so simple, I wouldn't be surprised if it was cheaper than current water cooling. Apparently because the liquid metal is so good at absorbing heat, elaborate block designs are not necessary, therefore lowering complexity and cost. Also, the pump is very simple, basically consisting of a few magnets. No moving parts = no bearings or need for lubrication, which = greater reliability.

Very interesting product.

[alglove]

Liquid metal at room temperatures? I wonder what they are using. Some sort of gallium alloy, perhaps?

[Coolin]

Pure mercury perhaps?

[StarfishChris]

It comes with a bicycle-pump attachment to pressurise the system and turn the metal liquid.
(This means it can only be prepared by strong people, so you may want to go to the gym in anticipation.)

[Tibors]

Last time I looked there where two ways to make metal liquid. Heat it till your CPU melts too. Or mix it with mercury, which isn't exactly non-toxic and environmentally friendly.

The liquid metal is non-flammable, non-toxic and environmentally friendly.

[Coolin]

Heat it till your CPU melts too.

Only Prescotts are capable of that.

[wim]

Only Prescotts are capable of that.

he (Tibors) means that molten metals (with exception of mercury) are at such high temperatures that it would be the cpu cooling the liquid instead of the other way round

but there is another way to make a metal liquid at given temperature: decrease pressure. however i'm not sure what material this would be practicable/possible with, because the melting point is relatively insensitive to pressure. you'd probably need some impossibly low pressure.. (not to mention having to avoid sublimation and the triple point of the material. but who knows, maybe they invented some alloy which is capable of this at room temperature?!)


edit: NaK sodium+potassium? cool

[Rusty075]

The fluid, according to the manu, is an alloy of gallium and indium, and has a melting point only slightly above room temp. (pure gallium melts at about 30°C) How they've dealt with some of the other nasty properties of these metals remains to be seen: they're corrosive to almost every metal out there, and they expand when they solidify, a quality that could result in a ruptured system if you leave your notebook out in the cold car overnight.

I'd have a hard time believing that this system could be cheaper or much more efficient than the heatpipe coolers in notebooks today. The developers must be banking on the continued rise of CPU heat. (I bet they're not fans of the P-M's )

[tay]

Its not likely to be sodium or mercury for obvious reasons. Galinstan is a liquid metal that could be used as is Gallium.

EDIT : duh didnt read rusty's post above.

[Tibors]

edit: NaK sodium+potassium? cool

It also reacts violently with air and water. It is apparent that this alloy is associated with several material and handling problems. Liquid sodium has nevertheless been used as a coolant for nuclear reactors, which shows that these drawbacks can be managed.

[andywww]

I'd have a hard time believing that this system could be cheaper or much more efficient than the heatpipe coolers in notebooks today. The developers must be banking on the continued rise of CPU heat. (I bet they're not fans of the P-M's Laughing )

It would be more efficient because the fluid would be actively pumped, unlike in a heatpipe. The really neat thing is that the pump would be solid state and silent b/c it'd ideally magnetically propel the metal fluid. As far as cheaper...lol... but hey, silent pump that won't malfunction, higher efficiency working fluid...add in a passive radiator... and it sounds awfully quiet. Possibly a bit out of the reach of a do-it-yourselfer though.

[nuttybing]

Looks really good on paper and has an enormous potential too. At least there'll be another option to choose from other than water, phase change etc

Still it is just another method of transferring heat away from a heat source, like heatpipes. Therefore cooling rate and noise would depend on the heat dissipating medium, I guess there'll be more passive radiators or even case side panels that act as a radiator in the future

[halcyon]

EDIT: I just noticed nuttybing is writing basically the same thing and I'm just repeating things. Oh well :)

Heat transfer from hot spots is indeed getting better.

I hope that thing becomes readity faster than the cooligy design (is it still out in any form?).

What bothers me is that with the rising cpu/gpu (still rising with dual cores) heat outputs, huget amounts of heat must also be dissipated.

It's not good enough that the heat is transferred away from the cpu.

It also needs to be dissipated efficiently and silently somewhere.

The current passive radiator designs just don't cut it.

Unless one wants to have a wall sized art piece, it's becoming increasing difficult to passively dissipate 200-300W+ of heat (up-to 150 for CPU, 75-150 for GPUs and 50W for everything else).

We need not just better heat "sinks", but also better radiator innovation (solids to air heat dissipation).

regards,
halcyon

PS Of course, I'm a firm believer in "less is more" and that CPUs should become more efficient, giving away less heat. To some extent this is happening (A64 Venice core, Pentium-M, etc), but the gains in process/design are lost in increased voltages/die size/transistor counts/switching frequencies, hence the overall temp output is still on the rise (on the short-to-medium term).

[alglove]

I did a bit of digging around, and these look like good candidates for the mystery liquid metal: http://www.indium.com/products/fusiblealloys.php

Of particular interest are Indalloy 46L (61 gallium / 25 indium / 13 tin / 1 zinc, liquid point = 7.6 degC) and Indalloy 51 (62.5 gallium / 21.5 indium / 16 tin, liquid point = 10.7 degC).

Reading elsewhere, pure gallium expands about 3% when changing phase from liquid to solid. This is manageable if the tubes are not completely filled with liquid metal. The rest of the tube can be filled with some inert gas, like nitrogen or argon. Also, pure gallium is minimally corrosive at moderate temperatures to gold and, more imporantly, nickel.

[andywww]

Its funny, the liquid metal system has the potential to change passive radiator design as well- one of the limitations of radiator design is problem with overly restrictive radiators reducing flow rate- like in a waterblock, a more restrictive radiator can often perform better per amt. of liquid flowing through it, but you get diminishing returns because of the lowering flow rate.

Anyway, with a silent magnetic pump, you could make it as powerful as you like- so your radiators could be highly restrictive - so they could be smaller and lighter too.

This post is a bit incoherent...and I'm wondering if making the radiators more restrictive would really provide that much more benefit. Blah, sorry!


A totally OT idea i was having was using TEC between radiator and waterloop- that way the radiator would be more efficient as it would be dissipating more heat. The drawback is that it couldn't be that much more efficient as TECs need to stay below 80C in current designs...but hey, maybe that thought will become more feasible and we'll see active radiators.

[Tibors]

Reading elsewhere, pure gallium expands about 3% when changing phase from liquid to solid. This is manageable if the tubes are not completely filled with liquid metal. The rest of the tube can be filled with some inert gas, like nitrogen or argon.

This won't work for two reasons. First the gas will accumulate in the systems highest point. E.g. leaving part part of the radiator non-functional. Or even more horrible, it accumulates in the CPU block. That is why people with watercooling systems "bleed" the cooling loop or use big reservoirs that are guaranteed to be the highest point. The last is not really an option with this. Or have you envisioned this in systems with several litres of liquid metal?

Secondly expansion when solidifying works differently than expansion of a fluid. There is no telling which part of the tubing will solidify first. Then acts like a plug in the tube and the liquid behind it has no expansion room left. Put an open glass bottle of water outside on winter night (If you live somewhere where it freezes in winter of course.) Big chance that you'll have a bottle shaped piece of ice and a lot of shards the next morning. The water freezes "outside in" forming a plug in the bottle opening. then the expanding ice breaks the bottle.

[andywww]

An additional issue is that pumping would be impossible until the entire loop was unblocked- relying only on convection to cool the CPU until the cpu can melt all the metal. Which, while it might work (in the ideal situation it would stay at the freezing point ) is kinda iffy neh?

[sthayashi]

Secondly expansion when solidifying works differently than expansion of a fluid. There is no telling which part of the tubing will solidify first. Then acts like a plug in the tube and the liquid behind it has no expansion room left. Put an open glass bottle of water outside on winter night (If you live somewhere where it freezes in winter of course.) Big chance that you'll have a bottle shaped piece of ice and a lot of shards the next morning. The water freezes "outside in" forming a plug in the bottle opening. then the expanding ice breaks the bottle.

Isn't water one of the only substances that expands upon solidification?

[Katharsis]

looks great for notebooks, tho i dont imagine the emf pump really has enough strength to work for a multi block full tower. now that i think about it, you could always just run individual ones to each part of the system to cool. personally im not too concerend about what the metal is, it could even be suspended in an oil solution. as long as it really works. lets just hope you never have a leak in the system, environmentally friendly or not, its probably a bit more eletrically conductive than water

edit: just read some more of the site. they actually make coolers for desktop cpus and video cards. wonder if they do single orders?

[Tzupy]

My question is - be gentle if I missed this info somehow - how flexible the pipes can be? Heatpipes are almost rigid. The best solution IMHO is to have the 'ambient exchanger' positioned exactly behind the back 120 mm fan. You can't usually do that with heatpipes, since the position of the CPU and back fan are variable from one case / motherboard to another.

[andywww]

So I took a look at the articles on their site, and there was one which mentioned a company called cooligy, which has apparently designed a silent solid state water pump. Oooh . It looks to me to be a lot like the "ionic wind" effect used in the ionic breeze. There's another company mentioned called Thorrn micro which appears to be developing computer form factor iconic wind devices.

Wow!

[Tibors]

Isn't water one of the only substances that expands upon solidification?

No, not one of the only, one of the few. See Rusty's post somewhere in the beginning. Pure Gallium expands 3% when solidifying. IIRC there are some more crystaline substances that have this behaviour. But like pure Gallium, we won't encounter them in our daily lives.

[blaze]

Technology like this has always existed, the only thing that I can think of that would have caused a slowing in its deployment would be large companies like Dell, Intel etc, lets be honest here, they profit from failing fans, as retards just go ahead and buy a whole new system. Fans are good for the economy.

[alglove]

Well, there is also a growing demand for this type of thing. Without such a demand, nobody would seriously consider making it and selling it. There would be no demand, because it would be considered too expensive, overkill, or both.

I mean really, would you put one of these on a Pentium II 266, when a passive heatsink and a 92mm case fan would suffice? I think not.

[Erssa]

http://theinquirer.net/?article=23331 Looks like sapphire is bringing something out...

[alglove]

http://theinquirer.net/?article=23331 Looks like sapphire is bringing something out...

Already under discussion here:
http://forums.silentpcreview.com/viewtopic.php?t=22173

[BrianE]

I thought water expanded when frozen mainly because of the air dissolved in it (which is forced out when it crystalizes)?

This technology looks very promising (and quiet ), but even if a version came out within this year, I would (personally) still wait a year or two for costs to come down and, more importantly, for the technology to mature more. Unless this stuff is already commonly used and well developed (I don't know, maybe for industrial use somehow) there's bound to be teething problems and some major changes for the first little while. Just my opinion....

[alglove]

Water expands because the crystalline structure at 0 degC actually forms a lattice in which the water molcules are further apart than in the liquid form, in which they are free to flow around each other. It is certainly an unusual property, but not unique. This is why ice floats on top of water... since the molecules in ice are farther apart, the solid is less dense than the liquid.

This is also what allows ice skating. With the added pressure of the skate (from the skater's weight), the ice gets compressed and some of it actually reverts back to water. The skater skates on a thin film of water, which refreezes after the skater has passed by.

As you get colder, there are actually different forms of ice that are more dense, because the molecules in the solid water get rearranged into a tighter lattice. This does not happen immediately, but it does happen over time.

This is the same phenomenon observed when white tin slowly turns to grey tin below 13.2 degC. They are two different crystalline forms of tin. The white form has many uses, but the gray form is brittle and powdery. This becomes a problem if you have an item made of pure tin and the temperature becomes too cool. This is called "tin disease" or "tin pest", and it was a problem for pipes used in church organs in nothern Europe. The pipes would actually start to disintegrate over time because of the cold weather.

OK, I am getting *way* off-topic here. I'll stop.