[The Young Ion]

I am trying to design a simple, fairly low budget, full submersion computer cooling system.

From a chemical perspective, I need some liquid that does not ionize (which to my knowledge means does not conduct electricity) and (preferably) conducts heat well.

Here's an illustration of the problem: an experiment was conducted in which a computer was run while submerged in distilled water. The computer ran for about five minutes then short circuited, sustaining no permanent damage however.

I am working on the idea that this was because the distilled water ionized some of the metals it was in contact with, which led to a spark being conducted.

So again, I need a liquid that will not ionize and conducts heat well.

[enahs]

Your theory is generally correct.

This has been looked at many times before, the problem is there are not many chemicals in the fluid phase at typical atmosphere pressure that do not ionize (or ionize other impurities) that have a specific heat higher then air.
There are lots of chemicals that would work, just not at the temperature range and atmosphere pressure a computer system operates.

So really, chemically your only option without spending thousands on building special apparatuses and chemicals would be submerging all the non mechanical parts in something like olive oil.
Its heat capacity is less then half of water, over 100 times more viscous (will not flow as easy) and will go rancid though.


Water could work, if you get really really pure water (I mean, really pure, make your own pure) in a super clean container, and really clean off the computer components first, multiple times with your excess pure water. You could then constantly circulate the water through a deionizer. They are not cheap though, and you have to replace them regularly/the cartridges. You also have to seal the container excellently, as just the dust in the room floating around landing in the water is enough to make it conduct electricity.


Water is some amazing stuff.

[Borek]

Plus you have to separate water form air to avoid carbon dioxide dissolution.

[The Young Ion]

Would it be possible to dissolve something in the water (clay??) which would block the electrical conduction?

Failing that, is there any substance (something like melted plastic but more user friendly) that I could coat the components with and which would provide a thin electrical barrier between water and silicon while still carrying heat through.

Just out of curiosity, could you explain to me the mechanics of electrical and heat conduction in chemicals? I have a general idea about electricity, but really none about heat.

[Borek]

Whatever you dissolve it will most likely increase water condutivity. Purest of the purest of the purest (Sir!) water is what have very high resistance, most substances you may add will dissociate decreasing water resistance. Substances that don't dissociate (many organics) have much lower heat capacity than water.

[enahs]

Just out of curiosity, could you explain to me the mechanics of electrical and heat conduction in chemicals? I have a general idea about electricity, but really none about heat.

That is more complicated then you know! But a quick, simplified description.

Heat is just a form of energy, more specifically (and simplified) the kinetic energy of the molecules).

Water has a high heat capacity because it is polar and forms very strong hydrogen bonds. This means it takes a lot of energy to make the bonds between each distinct water molecule to twist, vibrate and separate. This is what gives water is very high specific heat, hydrogen bonding (very important chemical concept). Hydrogen bonding is a form of intermolecular attraction, and there are many other types as well.

Like I said, that is a very simplified version, and I am not sure you will even understand that, as it requires you to know a lot of other stuff chemistry related.


As far as a substance to put over the contacts, it might be possible. It would have to be something that it’s self obviously does not conduct electricity, and is not soluble in water.

The first thing that comes to my mind would be a paraffin wax. It is an extremely good electrical insulator and not soluble in water. Depending on what kind you have the melting range is from ~47-65^oC (~116-150^oF). So you would have to get some with the upper range, and monitor the system carefully because if you hit the melting point then it will wash off in the water. It has a specific heat of about half that of water, so it should transfer heat fairly well.

It would be easy to melt then dip the components in, however it would be very hard to get a very thin layer. If the layer is thick it will just trap the heat in and have the wrong effect.


It however burns very readily and very hotly (it is used in some bipropellant rockets).

The biggest problem is though, that it might melt, as a CPU processor will have some very high heat spots and it so it might just be impossible to use it for this application or to get a thin enough layer



Your best and safest and cheapest bet would probably actually be to build an air tight chamber around the parts (make sure you have plenty of volume). Remove all the air from it, and fill the chamber up with Helium gas. Helium gas has a specific heat of over 5 times that of air, it has a higher specific heat then water actually. It is non reactive so it would not harm the components (and it is readily available and cheap). This would allow you to pull the heat away from the computer components really fast and efficiently. You would then need to remove the heat from the helium. You could pump the helium through a radiator to remove the heat.

[Borek]

Helium gas has a specific heat of over 5 times that of air, it has a higher specific heat then water actually.

Are you sure? Heat capacity, or heat conductance? I know He has high heat conductance, but somehow I doubt its heat capacity - its monoatomic and light, heat capacity is higher for more massive molecules with many degrees of freedom. Not to mention the fact that He has very low density (second lowest to be exact) for a gas - so the mass of coolant will be very low in 'normal' conditions.

[enahs]

My 2005 edition of the CRC Handbook of Chemistry and Physics lists:

List Helium as 5.193 J/g*K

Where as Oxygen (as O₂) has 0.918 J/g*k
And Nitrogen (as N₂) has 1.040 J/g*K

All at 25^oC and 1 bar.

It lists air as 1.0035 J/g*K at 0^oC

[enahs]

Wait, I just noticed after typing that up, that it was referring to the per gram, which is kind of pointless in this case as we need per mol. Doh!
I am a tard. To many finals today and yesterday, my bad.

Ignore what I said about the gas stuff then, I am just being a tard today. There are gases that would work, but they are toxic and expensive.

I went back and checked my data about Paraffin wax to make sure I was not screwing up their too, and it was correct.

Long day, sorry about that.

[The Young Ion]

So the two major problems with paraffin are that it will start melting somewhere between 47 and 65 celsius and that in order to get it to conduct heat and not trap it you need a very thin layer.
The heat does not seem like too much of a problem to me because I wouldn't want to be running my computer at anywhere near 65 degrees celsius anyway. For instance, right now my cpu temperature is a  puny 33 degrees. The highest I've seen it go up to is around 45. My system temp is of course much lower. So paraffin might actually be of service by warning me if my system overheats.
I should imagine that one could get a very thin covering of wax by heating it until it becomes very thin, then dipping, then withdrawing and letting the components drip dry. The thinness could be controlled by heating the components gently while they dry to let even more wax come off.

Using gases also interests me. Do you have an idea as to how much power and gas would be needed to expand the gas into the tower (main computer box in case you didn't know) and then recompress it and bring it around again thus chilling the interior of the case? I should think that would be very effective assuming it could be done without the need for huge pumping equpiment.

Finally, if all else fails, I had also thought about the combination of water and some kind of oil, using a small layer of oil to insulate the components from electrical conduction while using a current of chilled water to carry away the heat from the oil. The only thing I didn't like about this idea was that most (or all?? ) oils float ontop of water and not beneath is, thus putting the components at the top of the case where all the heat would collect as opposed to radiating the heat up from the bottom into the water current.

Correct me if I'm wrong, but you said that thermal conductivity is based on how much energy there is naturally binding the molecule together so that the more energy there binding the atoms, the more heat is required to break it apart, or to put it another way, the more heat the molecule can 'hold' or withstand.

Thanks much for your help, I feel deeply humbled in the presence of such knowledgeable peoples. 

[Borek]

Correct me if I'm wrong, but you said that thermal conductivity is based on how much energy there is naturally binding the molecule together so that the more energy there binding the atoms, the more heat is required to break it apart, or to put it another way, the more heat the molecule can 'hold' or withstand.

Not exactly. Energy (for your purposes) is stored almost entirely as kinetic - in different forms of movements. Imagine small monoatomic molecule (say He) - it can only move in three directions. It has almost no moment of inertia. Now imagine linear biatomic molecule - it can move in three directions, but as it has moment of inertia it can also rotate - so it can store more energy. It can also vibrate as if both atoms were connected with a spring. The more complicated molecule, the more energy it can store. Usually more complicated molecules contains also more bonds so more binding energy - but there is no simple relation between binding energy and heat capacity. If any, there is a relatively simple relation between molecule geometry and the amount of energy it can store. But there are other limits that came here - complicated molecules are heavier (they contain more atoms) and heavier molecules are rarely gaseous for example.