Nano-engineered thermoelectrics enable scalable, compressor-free cooling

So it's a better Peltier element? The article only seems to compare it to existing thermoelectric devices and not standard refrigeration units so I'm going to assume they haven't gotten even close to that efficiency. If they had I would assume they wouldn't shut up about the fact.

Also one of the biggest if not the biggest downside of these chips is, unlike a split refrigeration circuit, the front gets cold while the back gets hot which means you can't move the heat very far.

> the APL team achieved nearly 100% improvement in efficiency over traditional thermoelectric materials at room temperature

Peltier effect refrigeration has very low efficiencies (5%) so while this is an amazing accomplishment it will not replace other more mechanical cooling methods.

For sure this doesn't replace mechanical cooling.

But efficiency is extremely important in this context, not just for saving energy, but because the inefficiency manifests as heat generated, which undermines the intended refrigeration. So as far as Peltiers go, a doubling of efficiency is like a 3x ~ 4x improvement in effectiveness. Peltiers are already used for cooling in some contexts (eg. cooling CCDs) and this greatly grows the envelope for where they can be effectively employed.

IIRC there is an application in solar panels where thermo electric cooling could play a role if we could get the efficiency just slightly higher.

Thermoelectric cooling needs as much research as possible. Mechanical cooling is extraordinarily space consuming. CHESS has the potential over the next 10 years to largely replace vapor compression in most systems other than the most extreme gradients or scales. They are small enough to incorporate into most devices and would allow smaller devices more thermal load. In some ways I think efficient TEC like CHESS could be more useful than room temperature super conductors.

One step at a time... it would be astonishing if any thermoelectric device can leapfrog mechanical compressors.

Most IR cameras use Peltier cooling so better coolers should lead to better cameras.

> can't move the heat very far.

Heat pipes (as in CPU heatsinks) can passively move the heat up to a feet away. Far enough to allow effective insulation between cold and hot side. From there you can move the heat further away with a fan.

The paper[1] has some actual details, like this:

Under low-heat-pumping, with minimal role of parasitics, TFTEC modules offer four times the Coefficient of Performance (CoP) advantage over bulk devices. As an example, system-level CoP with a 16-couple TFTEC module is ~ 15 for small temperature differentials of 2 °C, pumping about 1.2 W heat load using 80 mW of electric power. Such small-scale high-CoP cooling is relevant for distributed refrigeration or compartmentalized refrigeration as well as for use in future electronic thermal management

They also note that the maximum cooling power density depends inversely on thickness, and this is where the thin-film TECs like this gets most of their improvements from, compared to millimeter thick regular TECs.

Just a quick scan before going to bed, but looks interesting for certain applications.

[1]: https://www.nature.com/articles/s41467-025-59698-y

From the abstract:

>system-level coefficient-of-performance is ~15 for temperature differentials of 1.3 °C.

There's a long way to go. As far as I know, the leader in condensed-phase refrigeration cycles is still the sodium iodide ionocaloric method, which blew past all of the competing methods (magnetocaloric, elastocaloric, thermoelectric) when it was announced in 2022:

https://www.science.org/doi/10.1126/science.ade1696

...but the temperature drop of 25 C is just barely practical for air conditioning in warm (but not desert) climates.

So, these devices.... turn heat into electricity? Where does the electricity go, back into the system it's powering?

> turn heat into electricity?

No, they turn a temperature gradient into electricity. If one side is heated and the other cooled, you can get current flow on the two leads. And as with many electrical devices, it can also be run in reverse: if you put a voltage across the leads then one side will get hot and the other side will get cold.

Obligatory video on the inefficiencies of thermoelectric cooling/Peltier elements from Techonology Connections: https://www.youtube.com/watch?v=CnMRePtHMZY

Also in the article, it is implied that there is no chance to replace mechanical compressors for great thermal powers, but for small thermal powers, from a few watt to a few hundred watt, thermoelectric devices may become preferable, due to small size, simplicity and reliability

Heat pipes only reduce the thermal resistance between 2 points. They cannot cool something below ambient temperature.

Thermoelectric coolers do not compete with heat pipes. They are useful only when you want to obtain a temperature lower than the ambient temperature. Otherwise, heat pipes or liquid flow cooling are the right solutions.

just noting that household fridge/freezers are in that power range...

I need this for my compact compost freezer [1]

[1] https://www.envirofreezely.com/

Nah. Heat pumps are ~10-100x more efficient than thermoelectric. Thermoelectric is just inefficient mechanism and is inherently difficult to scale up as the more electricity gets generated so does more heat which inhibits the temperature gradient you’re trying to utilize. There’s a reason water cooling is preferred instead of peltier to ferry heat away from electronic.

Magnetocaloric is super interesting though as an alternative to heat pumps. Likely the next big revolution in this space.

Thermoelectric cooling is pretty inefficient, because the materials need to balance competing requirements:

- Good thermal insulator - Good electrical conductor - Good semiconductor

This is because the hot & cold sides are sandwiched closely together as a PN junction, so once you move heat from one side to the other, it just leaks right back. Mechanical cooling doesn't have this problem, because the hot & cold sides are separated by thin bits of tubing. This makes the thermal leakage a "minor annoyance" in a mechanical system as opposed to "literally the whole problem we're trying to solve" as it is with thermoelectrics.

One work-around is to stack lots & lots of thermoelectric coolers on top of each other. That reduces the temperature difference at each individual PN junction, which in turn lowers the leakage. That's what this team is doing, but using layers that are only a few nanometers thick, so they can fit dozens or hundreds of junctions in a single package.

>Heat pumps are ~10-100x more efficient than thermoelectric.

Peltier junctions are a type of heat pump.

> Mechanical cooling is extraordinarily space consuming.

You'd wind up taking up even more space with a TEC solution at these efficiencies. To replace a 5-ton condensing unit you'd have to reject on the order of 50-100kW of heat.

Mechanical heat pumps are 10x-100x more effective than peltier heat pumps.

Twenty years ago there was a company trying to commercialise thermoelectric cooling based on a vacuum gap: https://web.archive.org/web/20031213235132/http://www.coolch...

They claimed 55% Carnot efficiency based on a 30-100 angstrom gap maintained by piezoelectric controllers, and a method to construct large electrodes with matched surfaces so that the gap could be maintained over a large area. It all sounded plausible but never went anywhere as far as I know.

Incidentally that means all their patents will have expired...

You could have heat pipe filled with liquid that evaporates at 5 degrees. This way it would draw heat from ambient level temperature and lead it to peltier device that would cool it below 5 deg and liquefy it back again. This way you could have peltier in the middle of your thick insulation layer with heat pipes drawing the heat into it from the cooled space and drawing the heat from the other side of it outside (using traditional heat pipes this time).

Presumably night vision on drones and missiles and things suddenly gets a lot smaller and deadlier?

Further to that, have a look at the refrigeration units on chest type portable fridges. They’re really not very big, compressor smaller than a roast chicken, small low speed fan similar to an auxiliary cooling fan in a PC, a controller board, and a few meters of metal tubing.

They typically consume around the 50 to 80 watts while the compressor and fan are running, and generate two to four times that in cooling capacity.

Surely people have adapted these in to PC cooling units?

I’ll watch your video later when I get home. Mostly leaving a comment so I can find it easier later.

Freezing food waste prior to composting it results in much faster breakdown in the compost.

Yeah but, household freezers are typically capable of freezing many tens of kilograms of material down to -18 to -24 degrees C / 0 to -10 F

Peltier coolers aren’t anywhere near this.

I think you may have just reinvented absorption refrigeration, previously invented by Ferdinand Carré in 1858.

https://en.wikipedia.org/wiki/Absorption_refrigerator

Down the heat pump rant rabbit hole we go.

It’s really, really weird to comment on the efficiency of these devices on an article like this, without actually checking the paper being referenced. Like, we know traditional thermoelectric are inefficient. But that’s the whole point of this research. To improve it.

It seems like they achieve a CoP of 1.3-6.8 (depending on heat transfer load) versus e.g. - CoP of 2-4 which is common for a household refrigerator. So we are already in similar territory.

The article also references a Samsung refrigerator already in the market using a hybrid system with thermoelectric to achieve higher efficiency. So clearly commercial thermoelectics are already efficient enough to have a role in efficient cooling.

https://news.samsung.com/global/samsung-unveils-new-refriger...

The article has the CoP numbers for the thermoelectric element used in that Samsung refrigerator as well, if you’re interested.

It’s really, really weird to comment on the efficiency of these devices on an article like this, without actually checking the paper being referenced. Like, we know traditional thermoelectric are inefficient. But that’s the whole point of this research. To improve it.

It seems like they achieve a CoP of 1.3-6.8 (depending on heat transfer load) versus e.g. - CoP of 2-4 which is common for a household refrigerator. So we are already in similar territory.

The article also references a Samsung refrigerator already in the market using a hybrid system with thermoelectric to achieve higher efficiency. So clearly commercial thermoelectics are already efficient enough to have a role in efficient cooling.

https://news.samsung.com/global/samsung-unveils-new-refriger...

I think the role of the peltier is to allow them to design the compressor to be more efficient in a temperature maintenance mode.. so their peltier is probably not more efficient than the compressor in low heat transfer mode. That’s exactly the mode where the CHESS device is making massive improvements, so clearly it unlocks the potential for a thermoelectric-only refrigerator that’s more efficient than one using a compressor

The article has the CoP numbers for the thermoelectric element used in that Samsung refrigerator as well, if you’re interested.

That doesn't change the fact that Peltier junctions are a type of heat pump.

I’m not sure what they mean by “traditional materials” here, but from their paper the COP is in the range of 1.3 - 6.8 depending on heat load which is similar to a modern household refrigerator. So yeah, it really does seem to be good enough to replace mechanical cooling.

The paper also references a Samsung hybrid refrigerator that already uses a thermoelectric device to improve efficiency (probably by letting the compressor operate in a more efficient mode most of the time).

According to the paper Samsung uses a bulk TEC device with a COP of 1.2 - 3 depending on heat load. That’s already fairly close to mechanical cooling. If it wasn’t it wouldn’t have made sense for Samsung to use it in a refrigerator whose whole selling point is efficiency.

I mean, clearly the 100% improvement is for the high heat load COP relative to that Samsung device, right? From 3 to 6.. and I think 6 is better than most commercially viable mechanical cooling solutions, no?

But isn't condensation based cooling like 500% efficient?

The Carnot limit is the theoretical upper limit of the efficiency of a heat pump, so the stated number is presumably with respect to that, not heat moved per unit energy input like you're quoting.

there are other alternatives for refrigeration

https://en.wikipedia.org/wiki/Thermoacoustic_heat_engine

it should be pointed out that thermoelectric cooling that was able to outperform mechanical pumps, would still be mostly useless for on device cooling as it cant move heat any distance, with it's own heat stuck in the same box or package, making design pivot around that limitation.

What if you use the technology in places where you actually want to maximize heat conductivity?

I'm thinking of the separation walls in counterflow heat exchangers (only useful at the end where the incoming stream is closer to its end temperature than the delta offered by thermoelectrics I guess). Can it do whatever it does across a temperature gradient?

Not really. The role of heatpipes is not to cool, just to transfer heat to and away from peltier device so it's easier to insulate hot and cold side.

A lot of people throw around 5% efficiency for Peltiers, and it's just not true - it depends heavily on the temperature differential and current vs. IMax. You can (with care) drive them >2.0 COP.

This isn't anything like a compressor or heatpump system, but Peltiers get a bad rap... they move heat really well if you're not pushing them to the edge.

Here's a nice chart. At 10k difference and 0.1 current max, you're over 2.5 COP. https://www.meerstetter.ch/customer-center/compendium/71-pel...

It’s astonishing how here, in an ostensibly technical forum, the vast majority of comments are left by people who did not even skim the article. At least read through the informative comments left by those who did read it before giving your venerable opinion based on the information that was already in your head? I mean, isn’t the point of an article or paper to present new, novel information?

The abstract of the paper mentions temperature differential of 1.3°C and 2°C - not really inspiring for use in a refrigerator.

Neither the article nor the abstract go out of their way to compare the efficiency of the new system to traditional heat pumps. Makes it kinda hard for a lay person to really assess the situation.

It’s the day before a holiday in the US. No one is thinking, we’re all just knee-jerk existing for another handful of hours and then… Then we engage in that once a year ritual of lighting up explosives and making a pilgrimage to the ER, where no one can afford proper treatment. /s

Happy 4th all!

(With apologies for the wildly off topic comment)

I thought Magnetocaloric cooling [1] was what will kill "traditional" heat pumps, but if we will have something with even fewer moving parts that is awesome.

[1] https://www.youtube.com/watch?v=qcscUoP8FNk

I didn’t know this, and it jumps out to me because 10k is pretty much the exact difference between room temperature and wine fridge tenperature - I wonder if this is actually not a horrible application for peltiers?

I was suggesting combining Peltier element _and_ heat pipes.

Wait, won't thermoelectric cooling need a LOT more surface area to have any comparable cooling performance to compressor-based systems?

Found a wine fridge at the thrift store last month, pulled the model number, realized it was basically a Peltier cooler and a fan, and thus likely to be still operational. Powered up just fine, so...

$10 and an hour of deep cleaning later, and now we have a wine cooler in our basement. I don't recall the specs or power consumption offhand, but it does keep my beverage-of-choice a few degrees cooler than ambient. :)

that's an idea but would you not have to power each one which then destroys efficiency anyway

I'm not sure why we would want to replace mechanical systems. A portable AC unit with 10,000 BTUs uses ~650wh these days, and it'll turn a 50m^2 apartment into an icebox. That's not much more than a flagship GPU pulls. This is also about the least efficient class of aircon you can get.

Sure less energy usage is always better, and if we could get the same out of mere single digit wh power draws that would be cool. But I don't think thermoelectrics are ever going to get us there.

You can get 4 COP with a regular air cooled chiller and up to 7 COP if you add an evaporative cooling tower.

Variable frequency drives have made running pumps and fans a lot more efficient, even residential HVAC equipment is starting to get EC motors or VFD driven A/C motors. I can run my 10,000 BTU (1kW) window unit at 68F for an entire month in the summer and it only costs $50 due to the variable speed fans and compressor pump.

What’s the use case for peltier coolers, wearable cooling?

power consumption is worst than compressor system

I think the gotcha is that you need a beefy heatsink and fan (and power for the fan) to keep the hot side anywhere close to room temperature.

I'm not sure of the exact reasons but you don't really see vapor-compression heat pumps in the tens- to low-hundreds-of-watts range. So I suspect there are some scaling factors where the reduced size/noise/complexity of a solid-state device starts to be more important than the extra energy that it uses.

I think most of the commercial bed cooling systems are thermoelectric (ChiliPad, Eight Sleep) and they seem to work fine, but by the time you get to the scale of a small fridge or dehumidifiers the products are generally awful.

A Carnot heat pump maintains the temperature in a house at 20C on a day when the temperature outside is 5C. What is the coefficient of performance of the heat pump?

The coefficient of performance (COP) of the Carnot heat pump is 19.5.

The coefficient of performance of a typical heat pump in a british home is around 4.

There is obviously a huge difference between 4 and 19.5 - although a good chunk of this is explained by large temperature differentials in the condenser and evaporator, and a british desire to use a water heating loop.

machining metal parts to tight tolerances when they're only a millimeter wide is awfully expensive.

But making tiny things with lithography is really cheap (in volume).

The middle ground is what one needs for a 10 watt vapor compression pump. And to my knowledge nobody has built a 1 watt pump with lithography - although an array of electrostatic scroll compressors does look like it could work.

HN hasn't been a good place for technical discussion in at least 5-6 years, maybe longer.

X is much better for this kind of stuff these days (just follow the right people and stick to the following tab)

Thanks for the link. One of the projects I'll probably never get around to is a thermoelectric-augmented fan coil unit for air-to-water heat pump retrofits. The existing emitters (and crucially the in-wall distribution piping that's expensive to insulate and vapor seal after the fact) would stay just above the dew point, and then the augmented fan coil would work to remove latent heat (humidity) by dropping a bit below the dew point via thermoelectric coolers that reject heat into the return piping.

It's a relatively small delta-T and in most climates a relatively small fraction of the overall cooling load, so it might just barely pencil out.

Good point. Cheerio.

X? But then you have to follow people, and inherently that backs you into a syncopation bubble. I value HN for the diversity of well considered opinions and ideas… but sadly that is dying.