Holy crap! You can get a CPU cooler that moves heat by pumping a sodium/potassium mix!
And here it is: The Danamics LM10, recently reviewed by NordicHardware. It certainly makes the old TS Heatronics Zen CPU Radiator look boring.
The sodium/potassium mixture, called "NaK" from the chemical symbols of the elements, can have a melting point as low as -12.6°C. So it's not incredible that you can use it to cool a PC processor. NaK isn't even very difficult to pump, partly because it's less dense than water. It's highly electrically conductive, so the Danamics cooler shifts it with a non-contact electromagnetic pump.
But NaK is usually used as a coolant in nuclear reactors, not PCs. When you only want to move CPU heat to room-temperature air via a pumpable coolant, and how that coolant interacts with fast neutrons is not very important to you, then about all you can say for NaK is that it's better than mercury.
As the NordicHardware reviewer points out, the high specific heat capacity of water makes it the obvious choice for this application, and many others. Water is streets ahead of every other inexpensive liquid for most coolant applications.
And water also has the advantage that it will not explode if it's ever exposed to even dry air. You can't say that about NaK. (Don't even ask about NaK and moisture.)
The gripping hand, though, is that there's very little reason to make a CPU cooler that uses pumped liquid coolant, unless the coolant is being pumped to a large separate radiator. This is how normal PC water cooling systems, and automotive water cooling for that matter, work; using pumped coolant allows you to have a radiator much larger, and more conveniently located, than you could if all of the cooling fins had to be strapped straight onto the CPU, or engine.
(Sodium has a role in automotive cooling too; some engine valves are hollow, and sodium-filled.)
If all you're doing is moving heat to the fins on the top of a CPU cooler, though, you can just use solid metal and/or heat pipes, because the mountain and Muhammad are pretty much in the same place already.
Years ago, I looked at a water cooler that worked in this way. It was pretty useless. The LM10 is much better than that; NordicHardware concluded that it did at least work about as well as a high-end conventional air cooler.
But the LM10 costs 2,199 Danish kroner, which as I write this is $US375. You can get a whole good-quality CPU water-cooling rig for about a hundred bucks less than that, and top-end conventional coolers are of course far less expensive.
But conventional coolers won't make your computer explode if someone hits it with an axe.
So I suppose the choice is yours.
7 December 2008 at 2:34 pm
I see someone else loved the Motie books too.
7 December 2008 at 8:20 pm
I think this is a swell idea. But it clearly doesn't go far enough. Does anyone know of a convenient way to use dynamite for PC cooling? Or, failing that, chlorine trifluoride? Let's make computers something only real men can use!
7 December 2008 at 9:43 pm
Do you have to fill out a MSDS and hang it on the side of your puter? can this even be shipped conventionally?
I just had this image of a UPS truck catching on fire because the driver dropped a larger box on the smaller box too hard.
7 December 2008 at 9:44 pm
Yeah, ClF_3 is fun. Did you read the safety sheet? It's absolutely hilarious to handle, as would be expected for a substance that was intended to burn down concrete bunkers.
The electromagnetic pump has something going for it, since I suppose it potentially lasts pretty long, but then you'll need a fan as well. So in the end it's a pretty useless thing, at least until CPUs get well into the petahertz range.
7 December 2008 at 9:44 pm
They're just preparing for the day that CPU dies are made from uranium.
7 December 2008 at 9:45 pm
Oh, we have got to send one of these to Dan, cause I want to see him break one open to verify, that yes there is indeed a corrosive, flammable fluid inside.
7 December 2008 at 11:35 pm
I'd say a computer that refuses to die gracefully in an axe fight is a selling point to much of the cooler's target audience.
8 December 2008 at 5:30 am
If it were a fair bit cheaper, I'd totally use it. Just let someone try to screw with my computer!
8 December 2008 at 8:03 pm
Chemistry is not my strong point, but aren't Sodum and Potassium key ingredients for action potential in neuronal activity?
Why don't people explode when they think too much?
... or is that the secret of the "Chewbacca Defense"?
8 December 2008 at 8:17 pm
DBT: sodium and potassium ions are important in various biochemical processes (an Na/K pump of a wholly different kind is important in almost all cells IIRC). However the cooler works with sodium and potassium metal. The metal atoms have a electron just lightly wafting around at their extremities, and will merrily give that up to pretty much absolutely anything to form the +1 ion. So they're good reducing agents, and correspondingly, they oxidise very easily.
10 December 2008 at 4:03 am
In the early days of the US Naval nuclear reactor program, one of the first engineering problems they needed to solve was coolant choice. A reactor with a water coolant loop was installed in the first nuclear submarine "Nautilus", and one with a NaK coolant loop was installed in the second "Seawolf". I can't find the exact quote, but the director of the project Hyman Rickover, said something along the lines of "If the ocean was made of sodium and potassium, some idiot would want to put a water cooled reactor on a submarine"
16 December 2008 at 8:56 am
I don't think the use of NaK in reactors is all to do with its nuclear properties: as I recall from my Mech Eng days, it's something to do with the Prandtl Number of the substance - a dimensionless number which incorporates density, specific heat capacity, viscosity and thermal conductivity. The lower the number, I think, the better suited the fluid is to a slow-moving (ie predominantly conductive) coolant circuit - and liquid metals have much lower Prandtl numbers than water.
It's too late at night for me to do any calculations, but... Oh! Or is it the Nusselt number?
25 December 2008 at 1:48 pm
At least NaK isn't shock-sensitive, nor environmentally dangerous, nor particularly toxic to biological systems, nor is it particularly likely to react with the metal tubing. All in all the thing's actually fairly safe.
But yeah, you probably don't want to crack it open and pour the contents into the kitchen sink.
6 December 2014 at 6:12 am
"make your computer explode if someone hits it with an axe."
...A feature worth paying for, IMO.