# Computer cooling with dry ice, ideas and question; thermodynamics

I am designing a cooling system for my computer and had a few questions.

So I have a computer water cooling radiator and I want to cool it as much as I can.

My first two ideas were an old window A/C unit and blowing air over the cold radiator in there onto my radiator. Or put the whole radiator again in water and fill the water with ice indefinetely...

My next idea, which I am going to pursue, is to set dry ice on the radiator. Or to submerge the radiator in some sort of liquid with a low freezing point (acetone?) and get that liquid as cold as I can by sublimation the dry ice in it.

Using the acetone is a good idea because it can grab all the cold from the dry ice (surface area is covered by liquid; opposed to it just sitting on the rad and 3/4 of the dry ice is incontact with the air doing nothing) But I am not sure how well my radiator would stand up to being submerged in acetone (not to mention explosions...)

So what would be my best bet to using the dry ice to chill the water? How do I calculate how much I need for the amount of watts in the watter?

I know my water cooling loop must remove about 625watts at max computing power, so how much dry ice do I need to remove that amount of heat?

I think setting dry ice on it works well (have done it before) but I am scared of local freezing inside the radiator, and that ice hitting my pump impellor then (no-no). Also if I get the water too cold I will start to get condensation on some parts (no-no).

• Does your computer produce more heat than the water cooling can handle? – OSE Apr 12 '13 at 14:00
• No, its currently at very acceptable ranges. – KDecker Apr 12 '13 at 14:07
• What is the objective of your design? Is it to optimize energy efficiency, keep the processor very cold, or something else? If efficiency is your goal, the A/C unit is probably not a good bet because it requires a good amount of electricity to run. For health reasons I would probably avoid using acetone. You could always just use regular water and ice and change often though. – OSE Apr 12 '13 at 14:18
• Ehh, just for fun, hobbyist. I guess the colder the better is what I am going for now. But basically a practical way to use the dry ice to cool the radiator – KDecker Apr 12 '13 at 14:20
• You are going to need a supply of dry ice (some quantity per hour of operation). This may slowly fill your room with CO2 and start suffocating your pet hamster, cat, dog siblings/children/parents etc. Dry ice doesn't seem a good choice for a material to transport heat from one place (your radiator) to another (your room). – RedGrittyBrick Apr 12 '13 at 14:42

A safer solution - buy everclear.

1) Mix one part ethanol to three parts water. The mixture freezes at around -15C. Make ice cubes out of most of this mixture, cool the rest. Your freezer should be able to just freeze this ethanol+water mixture, if not add some water.

2) Replace the water in your computer cooler with a 30% ethanol - 70% water mixture. This won't freeze and jam your impeller, and it wont burn (the rubber might dissolve though)

3) Immerse your radiator into a bucket of full of chilled 25% ethanol and pour the ice cubes into it.

4) As your ice cubes melt, replenish them. As the ethanol evaporates out occasionally add a shot of ever-clear.

Now you switched the unpleasantness of CO2 intoxication, for the pleasure of ethanol intoxication!

(As noted in the comments, with such low temperatures you'll have condensation problems. So buy a bunch of silica packets and dump them into your case and seal it up. Or blow dry hot air on exposed electronic parts. Or whatever)

• Question is, do I want to pass out from taking shots or CO2 inhalation? – KDecker Apr 12 '13 at 18:07
• CO2 is what's responsible for that chocking feeling. N2, CO, CH4, they'll asphyxiate or intoxicate you gracefully. – Lenzuola Apr 12 '13 at 18:20
• Does your CPU actually produce 625 watts of heat, or is that the computer ratting? Freeze the ethanol+water mixture in a small bladder, zip-lock bag or condom. Then the % mixtures will not get changed in step 3 above. Since you are up for experimentation the "latent heat of vaporization" from phase change, takes considerable heat, as compared to liquid conductance. – Optionparty Jan 3 '14 at 22:52
• @Optionparty Vaporization? Are we boiling ice cubes here? – Aron Dec 2 '14 at 9:08
• @Lenzuola This system would cause dangerous condensation in the rig. You would need to insulate all cooling equipment to the extreme to prevent the rig shorting out (that or you dehumidify the air completely). – Aron Dec 2 '14 at 9:09

I don't know about the practical issues regarding transferring the heat to the dry ice without encasing the radiator in a block of ice. To be honest, the whole idea sounds completely insane to me. (This does not automatically mean I disapprove.) I can, however, answer the question about how much dry ice you will need.

The enthalpy of sublimation of dry ice is $571\,\mathrm{kJ/kg}$. (source: Wikipedia.) This means that every kilogram of $\mathrm{CO_2}$ that sublimes will absorb $571000\,\mathrm{J}$. If we assume that you'll be dumping all of the heat from the CPU into the dry ice, this means you will need $625/571000\,\mathrm{kg} \approx 1 \,\mathrm{g}$ of dry ice every second, or about a kilo every 15 minutes or so.

Of course, it might not be quite so bad if you're also using the radiator as a radiator and dissipating some of the heat into the air as well. But still, I think a dry-ice-cooled processor is going to require rather a lot of dry ice. (As some people commented the amount of $\mathrm{CO_2}$ gas this would give off would be dangerous in a confined space. I wouldn't attempt anything like it indoors.)

• It's also very difficult ,mechanically, to cool something with a solid that sublimes. Firstly it's hard to get a good thermal contact between the solid cryogen and a heat exchanger and then the boiling gas creates a thermal barrier. Solid LN2 used to be used to cool instruments below 77k before mechanical coolers and it was a royal pain – Martin Beckett May 12 '13 at 18:13