Slashdot Mirror
Liquid Metal CPU Cooling
IceFoot writes "Bored with water cooling? Try a liquid metal cooler. It's a proven technology, used in nuclear reactors for decades because it carries heat away much better than a heat sink, heat pipe, or water cooling."
Try a liquid metal cooler. It's a proven technology, used in nuclear reactors for decades because it carries heat away much better than a heat sink, heat pipe, or water cooling. /me picks jaw up off the floor.
:-D
Liquid metal cooling is used in reactors because of the *extreme temperatures*, not just because it's more efficient. The metal (usually Sodium, but sometimes lead) is maintained in a molten state as it passes through the reactor and on back to the heat exchanger. Are they *really* saying that a CPU is going to pump enough heat to maintain a molten state inside the cooling device? If yes, that's kind of scary.
Maybe it's time to rethink the approach of driving up power usage to 300 watts just to get an extra 2 frames per second on Doom? Either that or we should start installing nuclear reactors in computers!
Javascript + Nintendo DSi = DSiCade
Liquid-Metal Cooling Loop Technology for CPU and processor cooling,
for laptops, desktops, servers, and graphics cards
The ever increasing demands put on cooling solutions for semiconductor devices have never been greater than today and there are no indications that these requirements will diminish in the future. With higher power dissipation due to higher speed processors, ever increasing leakage losses and extremely high heat flux densities due to hot spots on the chip, the demand for advanced cooling solutions continues to increase.
Until recently the demand for advanced cooling solutions was reserved for a small fraction of the ICs produced. Today these needs are becoming pervasive. New CPUs in almost every application are starting to require more than just a heat sink and a fan, and the need is not just with the CPU. In a modern portable computer or server there are several different heat sources that require advanced cooling. One can easily imagine a CPU, GPU, power supply, and other heat sources that need to be cooled.
nanoCoolers has developed a unique approach to cooling these high power heat sources. Cooling with liquid metals has been used for decades in the nuclear reactor industry, but never before have the systems been miniaturized and developed specifically for computer cooling. nanoCoolers has developed solutions to address the high heat source issues for portable computers, desktop computers, servers and other electronic applications. Within each of these categories are specialized situations that have additional needs, such as the elevated temperature requirements for ruggedized computers, or the overclocking requirements from gamers. nanoCoolers' advanced liquid metal cooling solutions address each of these concerns.
Processor Cooling and CPU Cooling for Portable Computing
Each application has issues with high heat flux densities and high power dissipation, but each also has their own unique issues that need to be addressed. nanoCoolers' solution for portable computers not only solves the power dissipation and high heat flux densities with the use of a highly thermally conductive liquid metal but also allows the system to be completely orientation independent. Since our solution is a completely filled and sealed unit, there are no gravitational effects on the thermal solution and therefore on the computer itself. Our electromagnetic pump, consisting of magnets and electrodes allows for extremely small pumps with a variety of profiles. Since the pump has no moving parts, it is inherently reliable. In the future, advanced cooling solutions will be required in portable computers for cooling CPUs, GPUs, other ICs, power supplies and even fuel cells. Our technology lets the system designer determine how many heat sources they would like to cool and at what remote location they would like to dissipate the heat. Another trend for portable computers is to make the computer thinner. nanoCoolers' heat exchangers can be made extremely thin to allow for these design challenges. The heat can be efficiently removed from the heat source and then transported to a remote location where it can be rejected to ambient air. Finally, one of the most important issues with a portable computer is the battery life of the unit. nanoCoolers' thermal solution is not only very power efficient, it could also be designed to vary based on the amount of cooling needed. If the system is idling, the current to the pump could be reduced or even shut off. However, if the CPU is running at 100%, the pump current could be increased for maximum cooling. These attributes allow for system designers to be able to design the very best portable solution available.
CPU Cooling and Graphics Card Cooling for Desktop Computing
Desktop computers have many of the same issues as all CPU driven devices; high power dissipation and high heat densities. Our desktop solution solves the most demanding thermal requirements. Desktop solutions might not be as concerned about power efficiencies,
This actually was first used at Los Alamos in part of the bomb project in WW II - see John Mcfees book "the curve of binding energy".
However, I very much doubt that sodium will be the metal of choice for CPU cooling, no matter how popular it is in submarines. The obvious candidates are mercury and gallium. Mercury is rapidly falling out of favor because it is so toxic and, if you spill it and it gets under the floorboards it is floor removal time. Gallium is a little expensive.
Panurge has posted for the last time. Thanks for the positive moderations.
Best of luck trying to get rid of the heat. Remember, convection won't work, only radiation.
Ydco co
Graphite is not a metal; its a form of carbon. Chernobyl was a bad Russian design, based on graphite as the moderator and IIRC gas as the coolant, not based on liquid metal at all.
Many American reactors do use pressurized water, not liquid sodium, for cooling. The primary (really "hot" in both senses) loop runs at several hundred degrees, but pressure keeps it from boiling. There's also the Boiling Water reactor design, which does indeed let the primary water boil and generate steam, which condenses in the heat exchanger and is returned as a liquid.
Gallium melts around room temperature.
It's most likely Galinstan, a metal alloy developed by Geratherm to replace mercury in medical thermometers.
In the case of a cooling system, the heat flux will be higher than with water or alcohol (heatpipe...). The specific heat's waaay lower, but the thermal conductivity (as in the rate the heat's absorbed or dissipated...) is much, much higher. So, if you have a decent convective flow via thermosiphon or by way of pumping, it becomes this very extended air-cooled heatsink.
You won't be overclocking with this stuff unless you couple it with something like Peltiers or Vapor-phase, but you CAN make a decent quiet PC with it.
I am not merely a "consumer" or a "taxpayer". I am a Citizen of the State of Texas
A Bismuth alloy of
Bismuth 49%
Lead 18%
Tin 12%
Indium 21%
has a melting point of 58%C
it could be used safely and is widely available
no need for liquid sodium pity.
Graphite is a) not a liquid and b) not a metal.
.info domain used for a legit site? Who knew?)
Graphite was present in Chernobyl, but it was used as a moderator. The coolant was our good friend h2o.
http://www.chernobyl.info/ has great info (The
In the early days of commercial radio these alloys were used as a conductive form to secure galena or other semiconductors for use as the detector in 'crystal' radio sets. Low melting point avoided damage to the mineral.
They are also used in making for many low temp(so as not to damage the mold) casting of patterns from a single rubber(latex) mold for use in making mold 'trees'.
I'm sure there are some /. readers who know of other uses for 'Woods' metal?
Let us know...
In addition to the other (correct) points you make, the specific heat (in J/(g*K)) may be lower, but the heat capacity (in J/K) isn't. Since most of these systems are probably volume/area limited instead of mass limited, I'd consider heat capacity more relevant, and reveals an even greater advantage for metals. I'd use J/(g*L) or something like that if you want to use specific heat, that way you consider the higher density of metals.
You're right. Water has the HIGHEST specific heat of all substances in the liquid phase. The only compund that has a higher specific heat than water is ammonia but only in the gas phase and in a certain temperature range.
-The Chemist
It depends on what type of metal you use. Different metals have differing specific heat capacity.
get your specific heat calculation on!
Sodium (and sometimes potassium) is used inside high-performance automobile engine pistons and valves to transfer heat from the surface of the piston to the skirts (or the valve face to the stem), where the heat can be shed to the engine block. Porsche and Mercedes Benz have been doing this for thirty years or more.
Chernobyl was a water-cooled graphite-moderated reactor.
There were a few bad things about this design:
1) If the reactor loses all of its coolant, it does not lose its moderator. Thus, losing coolant does not slow the reaction down. In fact, I believe that the Chernobyl reactor had a number of operating regimes where increases in temperature would increase the output power.
2) Graphite is very combustible. Highly flammable materials in an extremely high-temperature environment such as a nuclear reactor is a Bad Idea. Especially in a facility with no containment building whatsoever.
U.S. reactors are very different. Like Chernobyl, they are water-cooled, BUT they are also water-moderated. If they begin losing coolant, the reaction will begin to slow down. There are no highly combustible substances in the reactor core, and even if there were, U.S. reactors have very strong containment buildings so that if something goes horribly wrong, it will not likely ever escape containment.
Liquid-metal reactors have the disadvantage that their coolants are in some cases very reactive, but that's not much of a problem with a strong containment building, especially since some of the liquid-metal reactors are FAR more efficient as far as making use of their fuel and also produce waste that has a much shorter half-life than the waste from pressurized water reactors, making disposal much easier.
retrorocket.o not found, launch anyway?
Correct, but that is by weight. With a CPU, you want as much heat-absorbing capacity on as little space as possible, so it makes more sense to calculate the heat capacity per unit volume, which is the heat capacity times the density. The density of most metals is between 3 and 10 times more than that of water, so there you have your factor 4 back. Plus the advantage of a much better heat conductivity.
Avantslash: low-bandwidth mobile slashdot.
Wait, you want to replace mercury with a metal that reacts violently with oxygen and water vapor in an explosion and which reacts with water vapor to form the strongest base known. CsOH is caustic enough to go through glass and will go through metals. IIRC, the only safe way to store cesium is to keep it in a glass ampule under a vaccuum or an argon atmoshphere.
I would stick to the mercury, at least with mercury you can use EDTA or some other chelating agent to sequester it and counteract mercury poisoning.
"When you sit with a nice girl for two hours, it seems like two minutes. When you sit on a hot stove for two minutes, it
Radiative heat transfer isnt as bad as you think, its a function of T^4, whereas convection is a function of T^1.
The background of space is aproximately 4 degrees kelvin. So running your computer at room temperature (~304 kelvin, lets make numbers easy). 300 to the fourth power is a big number. And we've been doing radiative heat transfer for a long time. In fact, on some missions (for example, Voyager) they had to install resistive heaters to keep the compters warm enough to keep them running because it was so cold.
IAAAE. (I Am An Aerospace Engineer).
-Philski-
the specific heat capacity of water may be higher, but it does not transfer heat well at all. put your water bottle upside down buried in snow, bottom will freeze, top will stay melted. you'll also note that metal heats up a lot faster than water. why? because it transfers heat a lot faster. that's the property you want for cooling. quickly picks up heat from the cpu, and quickly gets rid of it at the radiator or fan.
It explodes on contact with water. Or almost anything else for that matter.
The sodium binds with the -OH in water to produce NaOH, a powerfully corrosive base. This happens to releases a hell of a lot of heat. It also happens to release a hell of a lot of gaseous hydrogen (H2O minus OH leaves H). Hydrogen+heat, kaboom.
If you take a lump of solid sodium metal and toss it into a lake it will sink for a moment, reacting as described. In a split second it explodes throwning the chunk of metal back up into the air - possibly at a random angle. It falls back into the water (assuming it didn't hit you in the face) and repeats. I've heard it can easily go on for 15 minutes or a half hour.
Splash-BOOM.... up it goes... down it comes...
Splash-BOOM.... up it goes... down it comes...
Splash-BOOM....
Metallic potassium is even better if you can get it. It is even more intensely reactive than sodium, plus the explosion should have a cool purplish color to it.
Hmmm, I just realized somthing... if we're talking about liquid sodium, well tossing that in a lake would damn near detonate all at once rather than a series of blasts. The initial contact and heat and explosion would blast the sodium into a near mist and into the water. Don't try this one at home kids.
-
- - You can't take something off the Internet! That's like trying to take pee out of a swimming pool.
but the article said if you ever get bored with water cooling
What could me more fun than proven technology - LN2 cooling, that has been used for supercomputers for decades.
The beast : http://www20.tomshardware.com/cpu/20031230/
They don't say what metal it is -- but mercury is out of the question -- too hazardous.
An alloy of Indium and Gallium is liquid at room temperature -- I have used it in laboratories for some other application. I think it is safe to assume that they plan to use some alloy metal involving In, Ga, Sn etc. All these are by the way, considered moderately hazardous. They will definitely have to come up with a solid plan about what to do if it leaks out during installation/assembly/disassembly. Also, since all these are conducting, they will short any part of a circuit board they contact -- and they can not be simply wiped away. BTW. In-Ga eutectic also eats up aluminum.
Pumping is an easy problem really. For pumping any conducting metal (mercury), people always use a type of induction pump. The path of the fluid is lined with magnet and a current is flowed through the mercury and the force exerted on the current carrying liquid by the magnetic field propels the liquid in its path.
Another promising Austin startup, NanoCoolers, says it is nearly ready to offer evaluation samples of its processor-cooling modules, based on a liquid form of gallium and indium.
An alloy of gallium and indium. It is liquid well below room temperature, with a boiling point in the ballpark of 2000 C.
Another neat trick is that the system has no moving parts. The tubing passes through a magnetic feild. A pair of electrodes stick into the liquid metal and introduce a DC electric current, effectively creating a liquid electromagnet. The electric current through the magnetic feild is exactly the same as single winding of an electric motor - except the motor force is directly on the liquid metal itself. This force pumps the liquid around the cooling loop.
Silent, and no failure prone moving parts.
-
- - You can't take something off the Internet! That's like trying to take pee out of a swimming pool.
'How toxic is this "Galinstan" compared to mercury?'
Not very toxic at all. That is kind of THE POINT in replacing mercury in thermometers, after all.
From the MSDS
Inhalation: The extremely low vapor pressure of Galinstan makes absorption through inhalation negligible.
Ingestion: No adverse health effect has been observed or reported. Galinstan passes through the digestive system without effect.
Skin: Skin oils may be reduced through continuous contact.
Eyes: Direct contact with the surface of the eye may cause irritation. Eye protection is recommended when potential direct eye contact is possible.
So don't take a bath in it or anything.
-Mark
MHD pumps work via the Lorentz Force and therefore need electrode contact with the fluid at right angles to the magnetic field.
Keep in mind, that most MHD pumps are at best 30% efficent, so you'll need a little more juice to move something like Galinstan. The only problem with using traditional pumping with something like that is that it wets every surface except things coated with Gallium Oxide and it alloys with most all metals to some small or large extent (You flatly do not want to expose Aluminum or Magnesium to this stuff, it'll rot it like Mercury does...). The only decent pump is going to be a MHD pump or an inductively driven Tesla style pump to begin with. Anything else will get contaminated with the pumped liquid or expose it to eventual oxidization...
I am not merely a "consumer" or a "taxpayer". I am a Citizen of the State of Texas
See their technology page.
The Web is like Usenet, but
the elephants are untrained.
Actually in the cold war era, it was believed that the USSR had submarines that had this type of propulsion system. The fact that it has no propellor makes it pretty silent, which is a bonus for submarines.
As a former submarine chaser from the cold war era, I can say that "it" was not believed so. But Tom Clancy made a lot of money from the idea.
The magnetostrictive effect didn't scale nearly well enough to do more than power a few small scale demos. Definitely not a sub. And, for what it's worth, the mechanism from the slashvertisement is not the same as a magnetostrictive drive.
It wets the surfaces of anything, including glass and plastic, but stuff like Gallium Oxide. A thin coating of GaO2 is present in the new non-mercury thermometers so you can actually read them.
All in all, it's obnoxious, but it's not anywhere near as bad as NaK alloys or liquid Na- there's a good reason why they abandoned that stuff as it'd attack almost anything in existence in short term. Same goes for Mercury- save that it's pretty damn toxic in addition to being an aggressive metal.
I am not merely a "consumer" or a "taxpayer". I am a Citizen of the State of Texas