Engineers have developed a passive evaporative cooling membrane that dramatically improves heat removal, it managed 800 watts of heat per square centimeter

Posted by nohup_me@reddit | hardware | View on Reddit | 28 comments

[-]

csory@reddit

Still, 800W/cm2 sounds waaay too much. That would be 3.2kW on a typical 4cm2 die…

[-]

TDYDave2@reddit

That just means the cooling bottleneck will be elsewhere in the system.

[-]

Sevastous-of-Caria@reddit

Stack them untilyou put it outside a laptop. Make the laptop out of copper plating and boom. You made yourself a bed warmer.

[-]

JesusIsMyLord666@reddit

They didn’t mention the temperature delta at which this was achieved. Could be at 100C or something ridiculous.

[-]

Dpek1234@reddit

In theory

So assume at least a 4x reduction due to everything from theory not being perfoect to useing worse materal

[-]

account312@reddit

No, that's what they actually achieved (admittedly in the lab). The theoretical limit is higher.

[-]

From reading the article, it sounds like this is just a proof of concept to incorporate evaporative cooling into a normal heat exchanger so as to take advantage of the large amount of energy needed to cause the phase change from liquid to vapor.

The big problem I see is that the boiling point of water at 1 atm pressure is 100C, higher than GPUs and CPUs should operate at for long duration. Meaning to incorporate this concept would require either a closed primary loop using another coolant that can boil at a lower temperature or a closed primary loop that operates under a vacuum to reduce the boiling point of normal water. Either way seems costly and a lot more complex when expanded to hundreds of thousands of processors in a server facility, when compared to existing cooling methods.

[-]

blisteringbarnacles7@reddit

Humans use evaporative cooling, and we operate at a lower temp than most silicon!

[-]

KaskaMatej@reddit

but under normal circumstances dont produce 800 watts of heat per square centimetre.

[-]

Explosivpotato@reddit

Humans generally produce less than 800 watts per human.

[-]

mrbeehive@reddit

800W is a lot for a human.

https://www.youtube.com/watch?v=S4O5voOCqAQ

[-]

puffz0r@reddit

Bro clearly hasn't had a ChatGPT girlfriend yet

[-]

Morningst4r@reddit

Exactly. People who think water only evaporates at 100C must be really confused as to why wet floors don't stay wet forever after being mopped.

[-]

rddman@reddit

The big problem I see is that the boiling point of water at 1 atm pressure is 100C

Water does not need to be 100c to evaporate. Extreme case: in vacuum water ice directly "sublimates" to vapor.

[-]

waxwayne@reddit

Alcohol or some other liquid would work.

[-]

Visible-Advice-5109@reddit

Normal heat sinks already use evaporative cooling. The chip is cooled by heat pipes which use water under vacuum to evaporatively cool the chip and then yhe water condenses back higher up the heat sink.

https://gamersnexus.net/guides/981-how-cpu-coolers-work

[-]

AtatS-aPutut@reddit

Oh wow I thought those pipes were just for looks

[-]

Qweasdy@reddit

A closed primary loop under vacuum with water would be a nightmare

Wait till you find out that the device you typed this comment out on probably already has a partial vacuum inside the vapour chamber/heat pipes in it.

[-]

the_corruption@reddit

I mean, the article mentions that evaporative cooling in heatpipes are already used for cooling, so not sure why you think the article author was unaware of that.

[-]

New_Enthusiasm9053@reddit

Ethanol is cheap and boils at 80C which is fine for a processor. Methanol is also cheap and boiling point is 65C and not really a risk in small quantities for the primary loop, it only needs to move the heat far enough that the secondary loop of water can absorb the heat fast enough.

[-]

AspectSpiritual9143@reddit

hello, why IT billed me 3 bottles of vodka today?

[-]

account312@reddit

Water has a high boiling point for the same reason that it moves a lot of heat when boiled off: It takes a lot of energy to break the intermolecular interactions. Water has something like 250% the enthalpy of vaporization per gram of methanol and is about 25% denser. You'd have to boil off a significantly higher volume of methanol to move the same amount of heat. That may not be practical.

[-]