Peltier effect generates spatially-separated hot and cold sides. Electrocaloric effect generates temporally-separated hot and cold periods. The Peltier effect is simpler to harness into a refrigeration unit (put the cold stuff on the cold side, dissipate heat from the hot side), but has lower potential efficiency.
Now we just need to combine it with thermal transistors* on the front and back sides to gate and pump the heat in one direction. Conduct -> Cool -> Insulate -> Heat -> Conduct -> Cool... (while doing the opposite on the heat-sinking side, of course)
Thank you. That's a clear description. I read through that whole article and still was not sure why it was not a Peltier cooler with oil flowing around.
> Temperature delta can be increased by just putting devices in series.
Yes, but depending on efficiency this can scale quite horribly. Multi-stage Peltier coolers, for example, used to be absolutely horrible in yhis regard - a two stage device would need to dissipate 100W at the high temperature side to provide just 1W at the cold side. (I remember doing the math for a project around 2016, don't know how much better it is these days.)
> A syringe pump pushes the silicone oil one way through the stack
This is better than the refrigerant cycle we're using now, sure, but I don't see how this is "solid state"
> “We can scale it because those elements we are using are already commercialized for other purposes.”
> For one thing, none of the present ceramics’ key elements are appealing for mass production. Lead is toxic; scandium is prohibitively expensive; tantalum is a conflict material in Central Africa and, Defay says, best avoided.
So it's not available with the current materials they need?
I think the idea is that there are actually two jobs going on, and one of them has been solid-state-ified: (1) cycling something between hot and cold and (2) ensuring emit-heat-to-environment happens separately from the absorb-heat-from-contents part.
In a regular refrigerator, refrigerant is pumped around doing both things at once, however we could imagine a system where there's two loops with a heat-exchanger: One small liquid+gas loop for refrigerant, and another silicone-oil loop.
Not sure if that was intention, but fridge system can be split into two parts producing temperature change, and moving the heat. In a typical compressor fridge the gas does both, change the temperature by expanding/ compressing and move the heat by pumping it around. I guess the electrocaloric effect does the first half in solid state. With oil being pumped doing second half in a non solid state way. Isn't system capable of moving heat automatically a fridge? No- you can have a heat mover which is capable to only move heat from hot to cold like a water cooler in a PC.
I wonder what the lifespan of these refrigerators would be. You might think that solid state devices would have longer lifespans, because there's no mechanical wear. But in fact SSDs die faster than spinning disks. Batteries swell and electrodes corrode. It seems like solid state electronics actually tend to be short-lived.
Which of SSDs or HDDs die faster depends strongly on their workload.
For a write-intensive workload, SSDs certainly die much faster than HDDs.
Otherwise, HDDs die faster, but modern SSDs die much faster than flash memories made with old technologies, which can have a lifetime of 10 to 20 years, while few SSDs can be expected to have a lifetime much longer than 5 years.
All but one of the many SSDs I've worked with that failed did so by refusing to respond to any commands from that point forward. The other one was actually probably ok, but did a big reallocation and performance was trash during the process, so we replaced it.
Some of my hard drives that failed when that way, sudden unexpected disappearance from the bus, but most of them provided signs of upcoming disaster, which could likely have been averted if the signs were considered in a timely way.
The failure rate of the SSDs was much lower than the HDDs, but the means of failure was more troublesome.
I wonder what brand of SSD did you bought ? I only have a single SSD that "died": the read/write performance became miserable for some reason, but it had no issue other than that.
hard drives store bits magnetically , ssd's store bits electrically. You could call both solid state if the SSD wasn't specifically named due to its close relation to dram and cache lines that need active power. I would also consider hard drives more solid than ssd's if you tested for how long each could resist entrophy if there was no source of low entrophy.
uh...no, “solid state” doesn’t have competing popular definitions. Solid state means the state can be read or written without moving parts or gaseous components. SSDs have this property, hard drives clearly do not.
> Solid state means the state can be read or written without moving parts
Ahh so the data on this cooler can be read without moving parts, interesting.
OP was talking about the definition of solid state in a broader sense. In that broader definition (useful outside of computer hardware) "solid state" could refer to the parts that don't require constant power to maintain state.
OP was demonstrating how the term could be reasonably used to mean different things in different ways meaning that "solid state" is not a very useful term in it's own, it needs context to be meaningful.
> Ahh so the data on this cooler can be read without moving parts, interesting.
I never said data, I said state. If you're getting hung up where I said “written” just think “changed” instead.
> In that broader definition (useful outside of computer hardware) "solid state" could refer to the parts that don't require constant power to maintain state.
This particular “broader definition” is not used in any domain, nor is it at all useful.
Nor was OP talking about any domain outside of computer hardware. They were talking specifically about hard drives and SSDs.
> OP was demonstrating how the term could be reasonably used to mean different things in different ways meaning that "solid state" is not a very useful term in it's own, it needs context to be meaningful.
They were demonstrating that if you make up definitions arbitrarily then nothing means anything. Regardless, “solid state” is a well-understood term with a specific widely understood meaning in the domain we’re all obviously discussing.
Would it theoretically be possible to create a nano-scale (or pico-scale?) electric generator that converts heat (atomic or molecular motion) into electric current?
It sounds too good to be true to have a refrigerator that generates energy, but I lack the education in quantum physics to understand why it would be unreasonable or impossible.
The Thermoelectric effect can produce electric current from a temperature difference (i.e. potential), but heat by itself can't do any work unless it has something colder to heat up. Refrigerators just do the inverse, they use an electrical potential to create a difference of temperature between the outside and inside of a box.
There are very small batteries that use the changes in its environments, but it use in general vibration, or light, or sugar (see https://en.wikipedia.org/wiki/Sugar_battery).
But the power is very small, used for very low power devices. One could think to use the sun cycle (become a little bit warm, then a little bit colder during the night), but at this point solar panel do the trick.
To extract work, you need to move energy across a gradient. You can only generate electricity by moving heat from an area of high thermal energy to low thermal energy.
The peltier or Seebeck effect does just this. By heating one side of the element and cooling the other, you get a small amount of electricity. It's impractical and very inefficient, though.
Because you must cool one side, you have to have some sort of system to remove heat as quickly as you put it in. Active cooling with a fan requires too much energy, so you're left with passive cooling. As well, the amount of power you get goes down as the cold side gets hotter.
There are plenty of good uses, though. Old gas water heaters actually used an electronic circuit to control the gas. A thermocouple is placed above the pilot light and creates enough energy to trigger the solenoid valve. Many types of temperature sensors work in the same way.
I'm not finding a great ELI5 of the electrocaloric or magnetocaloric effects. Specifically, the IEEE article doesn't explain how shuffling the generated heat away from the electrocaloric material results in a net cooling effect. If one tried that with a thermoelectric element, it wouldn't work, because when it was shut off, it would just cool down to ambient temperature.
Is it correct to say that they're analogous to gases changing temperature as the pressure of the gas changes? i.e. not a continuous generation of temperature change, but a one-shot change between two temperatures that's dependent on the outside force (pressure/electrical/magnetic/etc.) and some inherent capacity of the medium? That feels like the only way one could take away the resulting heat, remove the outside force, and have the element end up cooling down to a lower temperature than it started out, but I am not a physicist.
I always wonder what the COP efficiency of my fridge is. I mean a minisplit AC is about 3 to 1. A fridge is basically a small version of the same heat pump. So you would think solid state might actually be less efficient, though quieter and more reliable.
Capacitors are known to add distortion to signals that pass through them, especially those that use a high-k dielectric such as the ceramic type. It's why capacitor selection in ultra low distortion audio amplifiers is an important consideration.
Leaving capacitive effects aside and concentrating on secondary effects, it's been my understanding until now the electric field in a bulk dielectric is never 100% in sync with the applied signal as the field itself physically distorts the dielectric, as such nonlineararities are introduced into the signal as it's transferred across the capacitor and that they vary to some degree with the signal's (field's) frequency and amplitude. Nevertheless, it's never been very clear to me what happens at the molecular level that contributes to this nonlineararity.
Essentially, losses in capacitors are due to I²R in its electrodes, bulk resistive leakage within the dielectric as well as resistive impurities in the dielectric and that signal distortion is principally due to the intrinsic nature of the dielectric itself. That is, signal distortion is caused by the action of the field on the dielectric which physically distorts it, and although this action is essentially elastic the effective energy through the capacitor is never a 100% truly accurate analog of the applied signal in that group delay and related effects are introduced, and these contribute distortion (that is, energy essentially remains the same but the signal is changed albeit to a minuscule extent). This new work seems to indicate there are additional reasons for why capacitors contribute distortion to signals.
After reading this article it's occurred to me to question how much heat is actually generated within the dielectric by, say, nonelastic strain or deformation of its crystal structure (through molecular movement, phonon coupling losses and or whatever). Heat generated through actual dielectric deformation is energy and by conservation of energy it has to come from the input signal. Signal distortion will then occur by virtue of some energy changing into heat which then becomes subject to thermal inertia. In effect, energy through the capacitor can be lost and if reabsorbed [by adiabatic process?] into the dielectric’s field then the transferred signal will not precisely mirror the input, thus additional distortion is incurred. Moreover, I'd then contend that distortion harmonics introduced into the signal by the dielectric and its related processes are not only likely to be complex but also frequency and amplitude dependent, and would only be predictable with a full understanding of both the dielectric's electrical properties and the heat flow within and around the dielectric.
This mechanism may also account for why high-k ceramic capacitors are more lossy at higher frequencies (it also seems the only true distortionless capacitor would be the vacuum type where e0 is the effective 'dielectric').
I'd appreciate it if someone with a better understanding of these issues could enlighten me.
Since posting I've reread this and I'm sorry I posted it. Even with my limited knowledge of the subject there are too many complex issues involved here for me to have summarized them in the simplistic way that I have done.
reply