3 ms·

So I looked around for real figures of merit on these things - efficiency is but one. One thing I noticed is that every press release failed to mention the heat

So I looked around for real figures of merit on these things - efficiency is but one. One thing I noticed is that every press release failed to mention the heat flux of these devices. Gscholar to the rescue. \DeltaT = 5 K and a max Q of 135mW/cm^2 for a research device from MIT. Note, max Q implies \DeltaT of 0K [1] On digikey I pulled up the first data sheet I could find. [2] https://www.cuidevices.com/product/resource/cp60.pdf https://www.cuidevices.com/product/resource/cp60.pdf The CO60133 can pump 10 W @ 10K. Max Q is 12W from an Sa of 2.5 cm. Or a Q of 4W/cm^2 @ \DeltaT of 10K. Or 30 times better for the Peltier device. These devices are cool, but they're not even remotely powerful enough to replace Peltier devices, never mind traditional refrigerants. [1] https://fisherp.mit.edu/wp-content/uploads/2020/10/129.full_.pdf https://fisherp.mit.edu/wp-content/uploads/2020/10/129.full_... [2] https://www.cuidevices.com/product/resource/cp60.pdf https://www.cuidevices.com/product/resource/cp60.pdf

K0balt 3y ago
They are really unpractical at the moment, but so were peltiers and solar cells when they first started out. Now those technologies are robust, effective, and practical. With enough research, perhaps these will also be practical. I think a MEMS implementation with chip scale devices might fit the bill if suitable materials can be discovered. Since these are basically capacitors, If we say Dt is 1c and the single cycle density is low, we might be able to use MEMS to get a MHz cycle rate and layer them 100 thick like NAND. Then you’ve got an extremely high thermal flow (millions of cycles per second) and 100c Delta—T in a device that can be produced at chip-scale. Peltier devices can’t unlock this process, since their base efficiency is so dismal that you would just be creating a heating element. I don’t know if any of this is possible with -this- device, but it is meant to be an example of how modern technologies can create unexpected outcomes from humble processes, when the efficiency of the granular process is high enough. If this could leverage MEMS in a similar way to how massive parallelism revolutionised power MOSFETs, it would be a fantastic tool for humanity.
crote 3y ago
The problem is that your linked peltier consumes 22.8W of power to pump those 10W. That is a horrible efficiency, and completely useless for most applications. At that point you're probably better off with plain old resistive elements when trying to heat, or literally anything else if you are trying to cool. Being limited to a small temperature difference and power capacity per area isn't that big of a deal when you stuff a whole bunch of them together, especially considering the goal is to replace traditional building-sized heat pumps. Needing tens of thousands of them only matters due to cost - and that's way less of an issue when it is more efficient than the alternatives.
Dylan16807 3y ago
There are use cases where a 30x increase in surface area would be fine. For example, 10W of heat pulled out of an insulated cooler would do a good job, and you could easily dedicate 100 cm^2 to the device. Of course it would have to do that at a much larger delta T than "near zero".

K0balt 3y ago

They are really unpractical at the moment, but so were peltiers and solar cells when they first started out. Now those technologies are robust, effective, and practical. With enough research, perhaps these will also be practical. I think a MEMS implementation with chip scale devices might fit the bill if suitable materials can be discovered. Since these are basically capacitors, If we say Dt is 1c and the single cycle density is low, we might be able to use MEMS to get a MHz cycle rate and layer them 100 thick like NAND. Then you’ve got an extremely high thermal flow (millions of cycles per second) and 100c Delta—T in a device that can be produced at chip-scale. Peltier devices can’t unlock this process, since their base efficiency is so dismal that you would just be creating a heating element. I don’t know if any of this is possible with -this- device, but it is meant to be an example of how modern technologies can create unexpected outcomes from humble processes, when the efficiency of the granular process is high enough. If this could leverage MEMS in a similar way to how massive parallelism revolutionised power MOSFETs, it would be a fantastic tool for humanity.

crote 3y ago

The problem is that your linked peltier consumes 22.8W of power to pump those 10W. That is a horrible efficiency, and completely useless for most applications. At that point you're probably better off with plain old resistive elements when trying to heat, or literally anything else if you are trying to cool. Being limited to a small temperature difference and power capacity per area isn't that big of a deal when you stuff a whole bunch of them together, especially considering the goal is to replace traditional building-sized heat pumps. Needing tens of thousands of them only matters due to cost - and that's way less of an issue when it is more efficient than the alternatives.

Dylan16807 3y ago

There are use cases where a 30x increase in surface area would be fine. For example, 10W of heat pulled out of an insulated cooler would do a good job, and you could easily dedicate 100 cm^2 to the device. Of course it would have to do that at a much larger delta T than "near zero".