Immersion cooling in vacuum. Is it possible?

[Kamikaze_Raven]

Hello. I'm a new guy to this forum, but I've been watching linus tech tips for years. I am planning to build a high end PC by the end of this year or early next year. I am trying to read every news about PC buildings.

 

Recently I've found some videos of the so called immersion cooling solutions that really attracted my attention.

 

First is the 3M Novec liquid. Which seems like the best performing one among all. They have a very low boiling temperature, and can transfer heat very efficiently. Essentially it is a phase change cooling solution. The drawback is that this liquid is extremely expensive. A gallon of those cost hundreds. And the whole system must be sealed in an air tight case (because they boils). Plus, 3M does not sell those liquid to consumers. Only companies are allowed to purchase them for commercial use. I also read from somewhere that those liquid are harmful when inhaled because they will condense in human lung, causing all kinds of issues.

 

 

 

Then there is this mysterious closed loop phase change liquid cooling that Linus talked about in one of his videos. The creator did not disclose what liquid is in it.

 

 

 

And then there is this guy from Engineered Fluids with immersion single-phase cooling solution. I read from their website. This fluid sounds much safer than the 3M Novec. It does not boil under normal temperature. It does not transfer electric. It essentially acting like mineral oil immersion cooling. But it is cleaner, and looks like it does not degrade any parts of a computer (mineral oil degrades rubber). Plus, it is much cheaper than 3M Novec, and consumers can buy them with no restrictions. 20 liters (about 5.3 US gallon) of those liquid cost around 300 to 400 dollars.

 

 

 

 

Then here comes my idea. Immerse the PC in a low pressure chamber filled with the single-phase liquid from Engineered Fluids. We all knew that if the air pressure is low, the boiling point of liquid will be lower. We can manipulate the pressure so that this fluid will boil at around 35C. I suspect that mysterious phase change solution in Linus' video used this technique. Their CPU cooling block looks very heavy duty. And they have a pressure sensor sticking out. They could be using a safe liquid in a low pressure chamber so that the liquid will start to boil under low temperature.

 

What do you guys think? Does this sounds viable? Anybody know how to get all the cords out in a sealed air tight chamber? As the liquid boils, pressure will build up again. Do you think this thing will need a safe valve to prevent the whole thing from exploding? Please let me know your opinion.

 

There is another thread that talked about putting PC in a vacuum chamber. But it doesn't seem like people discussed about the boiling point.

 

[Madgemade]

4 hours ago, Kamikaze_Raven said:

Their CPU cooling block looks very heavy duty. And they have a pressure sensor sticking out. They could be using a safe liquid in a low pressure chamber so that the liquid will start to boil under low temperature.

Seeing as this was just a CPU block, the safe liquid could just be water. According to this, if you reduced the pressure to 1 psi or less, then water would boil at 30C or less. This is not much better temp wise than chilled water, but due to boiling it would probably transfer heat quicker.

The main problem with this sort of system would be keeping the pressure down and removing the heat from the evaporated water as traditional radiators couldn't be used. That video seems to have a beefed up radiator which can probably withstand the higher external pressure with a normal water radiator couldn't.

 

Edit: After thinking about this a bit more, you wouldn't be able to go below room temp very easily because whatever cooling plate you used would need to be kept at a temperature below the boiling point.

I think this is quite doable though, I just don't think it's worth the effort because the same results are achievable by setting up a custom loop with a chiller (or just going outside in winter).

Edited May 20, 2019 by Madgemade
add more

[Kamikaze_Raven]

16 hours ago, Madgemade said:

Seeing as this was just a CPU block, the safe liquid could just be water. According to this, if you reduced the pressure to 1 psi or less, then water would boil at 30C or less. This is not much better temp wise than chilled water, but due to boiling it would probably transfer heat quicker.

The main problem with this sort of system would be keeping the pressure down and removing the heat from the evaporated water as traditional radiators couldn't be used. That video seems to have a beefed up radiator which can probably withstand the higher external pressure with a normal water radiator couldn't.

 

Edit: After thinking about this a bit more, you wouldn't be able to go below room temp very easily because whatever cooling plate you used would need to be kept at a temperature below the boiling point.

 I think this is quite doable though, I just don't think it's worth the effort because the same results are achievable by setting up a custom loop with a chiller (or just going outside in winter).

 

Thanks for the input. I think the goal is not to go below ambient temperature but to keep it as close to ambient as possible.

 

This video made me think that regardless how huge the radiator is, we still might not be able to remove heat from CPU as efficient as we needed. If you watch from 6:59, you will see that the CPU still gets to 80C even with this insane loop where the water temp stays stable at 20s. I suspect low pressure phase change might help? I'm not so bright at physics......

 

[Madgemade]

16 hours ago, Kamikaze_Raven said:

you will see that the CPU still gets to 80C

This was related to the CPU not being de-lidded as he said in the video. The GPU never got above 26C so it certainly works. The problem moves from getting heat out of the loop to getting heat from the silicon and into the water itself.

[akio123008]

I'm not quite a refrigeration expert, but I'm not completely sure this is really a phase-change cooling solution at all.

 

In actual phase change cooling, your transporting a refrigerant between areas with different pressure levels. First the refrigerant evaporates in a low pressure area to absorb heat, then you pump the vapor into a high pressure area, where it's forced back to liquid, and therefore releases energy it absorbed when evaporating in the low pressure area. 

 

In this system however, the pressure is low throughout the entire machine. Sure the boiling point is lower, so it will evaporate, but the condensation happens simply due to the vapor being cooled in a radiator of some kind; you're not compressing it back into a fluid. This basically means the fluid will never get colder than ambient temperature. The key to real phase change cooling (refrigeration) is forcing the fluid to evaporate, mostly using pressure differences. Here, the fluid only naturally evaporates because of the heat from the cpu, and condenses naturally by the cool radiator.

 

Your idea would probably work, but it's just a very complicated and expensive way of doing something that can also be done with regular cooling. In fact, you might even just buy a phase change cooler, as that's cheaper and can achieve sub-ambient, and even sub-zero temps.

[Madgemade]

32 minutes ago, akio123008 said:

In fact, you might even just buy a phase change cooler, as that's cheaper and can achieve sub-ambient, and even sub-zero temps.

Indeed and that's probably what I would recommend for the OP. The idea of reduced pressure loop is interesting but it's only use is shown in that video. Main problem is not being able to achieve lower temps than ambient and only advantage is not needing a pump and potentially better efficiency at the block but that's only in theory.

[akio123008]

33 minutes ago, Madgemade said:

only advantage is not needing a pump and potentially better efficiency at the block but that's only in theory.

Yes that's right. You could even see it as reverse-refrigeration; normally pressure differences are used to force evaporation/condensation for cooling/heating, whereas in this system natural evaporation/condensation causes pressure differences, moving the fluid/gas so that you don't need a pump (kind of like how a heat-pipe works actually)

[Kamikaze_Raven]

9 hours ago, akio123008 said:

In actual phase change cooling, your transporting a refrigerant between areas with different pressure levels. First the refrigerant evaporates in a low pressure area to absorb heat, then you pump the vapor into a high pressure area, where it's forced back to liquid, and therefore releases energy it absorbed when evaporating in the low pressure area. 

 

I thought phase change means liquid to gas and gas to liquid change. Pressure difference should be irrelevant? Thank you for all the input. I now agree it's not worth the hassle. I probably will go with AMD cpu so insane cooling is not necessary.

[akio123008]

4 hours ago, Kamikaze_Raven said:

I thought phase change means liquid to gas and gas to liquid change. Pressure difference should be irrelevant?

Well I guess you could call any setup where something evaporates "phase change" but the kind of phase change cooling that can achieve below-ambient temperatures needs a pressure difference between areas to work. 

 

I believe the physics principle involved is the fact that a refrigeration/phase change cooling system does work (work as in W equivalent to E - energy) to move heat between places, regardless of, or even against, the natural heat transfer. That means a real phase change cooler could in theory even be used to heat up the cpu and cool the room, even though the room is much colder than the cpu. This is also the reason why it cools below ambient temperature; normally below ambient temperature the heat transfer from the cpu to the outside would stop, because the temps are the same. However, phase change cooling can use work to move heat regardless so it can keep transferring heat to the outside, and keep cooling the cpu to below ambient temperature.

 

The kind of setup you're talking about only utilises natural heat transfer just like normal water or air cooling. There's no work done to transfer the heat. That's also why a normal water cooler for example, will never cool below ambient temperature, because natural heat transfer stops when the temperature of the cpu is the same as that of the room. It will also never move heat from the room to the cpu, as the cpu is waaay hotter, and natural heat transfer will always go from high to low temp areas. 

 

But yes, you could call it phase change anyway as the fluid evaporates, it's just not what most people would expect from a phase change cooling system.

[Kamikaze_Raven]

15 hours ago, akio123008 said:

Well I guess you could call any setup where something evaporates "phase change" but the kind of phase change cooling that can achieve below-ambient temperatures needs a pressure difference between areas to work. 

 

I believe the physics principle involved is the fact that a refrigeration/phase change cooling system does work (work as in W equivalent to E - energy) to move heat between places, regardless of, or even against, the natural heat transfer. That means a real phase change cooler could in theory even be used to heat up the cpu and cool the room, even though the room is much colder than the cpu. This is also the reason why it cools below ambient temperature; normally below ambient temperature the heat transfer from the cpu to the outside would stop, because the temps are the same. However, phase change cooling can use work to move heat regardless so it can keep transferring heat to the outside, and keep cooling the cpu to below ambient temperature.

 

The kind of setup you're talking about only utilises natural heat transfer just like normal water or air cooling. There's no work done to transfer the heat. That's also why a normal water cooler for example, will never cool below ambient temperature, because natural heat transfer stops when the temperature of the cpu is the same as that of the room. It will also never move heat from the room to the cpu, as the cpu is waaay hotter, and natural heat transfer will always go from high to low temp areas. 

 

But yes, you could call it phase change anyway as the fluid evaporates, it's just not what most people would expect from a phase change cooling system.

 

I am fully aware the phase change cooling solution I talked about will not cool below ambient. It is not my goal to cool below ambient. Below ambient cooling introduces a whole lot more trouble.  I merely believed that phase change cooling can transfer heat more efficiently.