Hacker News new | comments | show | ask | jobs | submit login
High-power thermoelectric generator utilises thermal difference of only 5°C (newelectronics.co.uk)
165 points by rbanffy 5 months ago | hide | past | web | favorite | 90 comments

The Carnot efficiency of a heat engine is 1 - T_low/T_high. When operating between 305K and 300K (room temperature(ish)), for example, this is 1 - 300/305, or about 1.6%. In other words, the absolute most amount of energy a device operating over a 5 K temperature difference at room temperature is pretty dismal. Just because their band gap (for lack of a better term) is small does not make this device "useful."

For all stories about heat engines running between smaller temperature differences, the smaller the temperature difference, the less excited you should be. For example, I have a device that reaches the Carnot efficiency across temperature gaps of 0K! HINT: It's a rock. Or a paper clip. Or a wet tissue. Or...

Generating useful energy from large temperature differences is old hat, though. 1.6% efficiency could be plenty for some tiny device that needs very little power.

Err, it's 1.6% of heat flux gets turned to electricity in a perfect device. This device is not perfect, and you are likely lucky to turn 1% of heat flux into electricity.

Sure. That could still be useful. There are many applications for a device that uses a tiny amount of power and needs to harvest it all from its environment.

But at that point if you're near a population center, you would probably be better off scraping RF energy

It could still be really useful for remote sensors and who knows what else. A lot of places a little power is all you need.

One of the batteries that Dad worked on back in the 1950's was for the Union Pacific railroad. It was intended to be loaded into an underground vault along the tracks in remote areas, and would provide power to run signals through most of a winter. If modern LED power draw is low enough, and the new thermocouple is efficient enough, they could run the signals off the temperature difference between the mountain air and the temperature of the earth about 10 meters down.

Yeah -- interstellar space, we're talking 67.5% Carnot efficiency.

I don't think heat engines would work in space generally due to the low mass of stuff around you that can absorb heat.

They work great as long as you have a source of heat and a good radiator. You don't need mass as a heat sink.

Interesting, I wasn't thinking about that. Thanks

At a practical level, your initial instinct was right. In space, nobody can feel you convect. Essentially all of the waste heat has to be disposed of via radiation, and this is orders of magnitude less effective than convection at the temperatures that any known material can withstand. Worse, you don't have to radiate "just" the waste heat. Eventually, all of your electric energy becomes heat, too. In other words, all of the energy you extract from your decay source will end up as heat that you have to get rid of via radiation. (Well, you could wind a giant spring, I suppose. But, a spring that large would be a better radiator than energy storage device.)

All that being said (and true!), you also have a very cold sink to reject waste heat to, whether you want to or not. The end result is that heating is as much an issue as cooling. Thus spacecraft usually wind up with blankets and radiators, both. :)

Here's a detailed look at the thermal control system for the ISS, for the curious: https://www.nasa.gov/pdf/473486main_iss_atcs_overview.pdf

>In space, nobody can feel you convect.

I know I'm on HN, as opposed to one of the political sites I read, when I run across that sort of remark.

From there:

“RTGs use thermoelectric generators to convert heat from the radioactive material into electricity. Thermoelectric modules, though very reliable and long-lasting, are very inefficient; efficiencies above 10% have never been achieved and most RTGs have efficiencies between 3–7%”

Also interesting, the price of RTGs:

"Plutonium 238, which is used to power Radioactive ThermoGenerators such as the ones that power the Voyagers, Galileo, and Cassini probes has not been produced in the US for 25 years, but NASA and DOE have budgeted $50 million to restart production of about 2 Kg per year for 5 years…an average cost of about $5 million per Kg. That is expensive stuff!"

One SNAP-27 unit, the type used in Apollo missions, providing only around 70 W, had around 4 Kg, which means its Pu price alone would be at least around 20 million, if it would be produced now.

So, a kind of thermal solar panel that could possibly scale down/cheap better than a solar panel?

Wet tissue usually has a greater temperature delta due to evaporative cooling... But that's just pedantic; everything else was spot on. ;)

This is not a heat engine, as there is no cyclical nature and it's not turning the heat into mechanical energy, therefore the Carnot efficiency need not apply.

Interestingly, while the Carnot Limit is derived using heat engines, it applies to any "device" that extracts energy from a temperature difference. If it were not so, then one could place, for example, a thermoelectric device between two energy reserves at different temperatures and use it to power a Carnot heat pump to drive the temperature difference higher, or to sustain the temperature difference and extract work. This is a perpetual motion machine, violating the thermodynamic laws of which the Carnot Limit is a manifestation. (It is a general result, obtained from hypothetical devices.)

One can sort of "feel" this result by imagining the voltage pushing a cloud of electrons around a circuit, producing shaft work in an electric motor, and/or heat if there is any electrical resistance.

Carnot efficiency still applies, even if the thermoelectric effect doesn't have a mechanically cyclical nature. http://www.pnas.org/content/112/27/8205

The article you linked doesn't follow your statement, in fact it suggests a different theoretical limit, and only suggests a engineering formula for high temp TEGs to mesh up with the observed results due to the Thomson effect on heat transfer.

They suggest a new formula because the "the conventional efficiency formula (Eq. 1) often misleads and gives rise to an impractically high efficiency prediction".

> The [thermoelectric] conversion efficiency by Eq. 1 is the product of the Carnot efficiency (ΔT/Th) and a reduction factor as a function of the material’s figure of merit Z = S^2 ρ^−1^κ−1, where S, ρ, and κ are the Seebeck coefficient, electrical resistivity, and thermal conductivity, respectively.

There are many devices for which power consumption is measured in the microwatts. This could certainly be useful for them.

Requiring a 10C difference to generate power isn't useful when you only have a 5C difference.

parent was implying that a 5C difference is practically useless, if technically possible, so don't bother popping open the champagne when it's announced.

It's not useless. There are not that few IOT style applications where one could extract significant energy from e.g. atmospheric temperature cycling during the day relative to a thermal mass. This would not require sunshine, only that the temperature over the day changes about 10K between minimum and maximum. Or with a large enough thermal mass it could use this cycling over multiple days, which might be more reliable. It also has the great benefit of not requiring something fragile exposed to the sun. This makes the feasible reliability potentially much higher, depending on how good this thermoelectric generator lives.

Can anyone explain the importance of this to someone with no understanding of physics?

As I see it, this is good for low power iOT devices that are installed in remote / hard to reach areas (say underground etc. ..). A small temperature difference could produce power enough to run the device without any battery.

Heart Pacemaker require around 10uWatt and this new power source seems to provide 12uWatt for 1 cm2. If temperature difference could be maintained, then could provide a low weight, low power source that do not require changing battery.

Is there a 5C difference anywhere in the body? Skin temperature to blood temperature, maybe?

That's what I was thinking. 37°C inside vs. up to 32°C (90°F) outside and you still have enough temp gradient to produce the power. All you need to do is install this heatsink on your chest :)

And to satisfy precedent, pipe any leftover power after running the pacemaker to a ring of white LEDs.

Testicles versus main body is the classical example. Sperm production and conservation just works better around 32°C.

On a more practical note, I wonder how feasible a thermoelectric powered earpiece would be...

How does that works? For a period of time before you try to get someone pregnant you should try to keep your testicles at that temperature to get the best quality possible?

Yes. Boxers. Loose pants. No hot baths or saunas. Good luck!

There are likely papers on the exact effect over to be found on google scholar.

Thanks. If that energy is stored in a battery or capacitor, how could it bleed off any excess charge (in your example 10 v 12uW). Is this a thing that’s is problem anywhere?

Getting rid of a few microwatts is never a problem. If these things behave like Peltier elements, they'll simply float their voltage up to a design maximum then stop extracting energy as the electric field balances the energy of the electrons they're trying to extract from.

Heat engines generate work (power) through the movement of heat from hot to cold. Conventional systems (steam turbines, Stirling engines, regular thermoelectric generators) require a temperature difference of much higher than 5C between hot and cold, which limits the situations they can be used in. This apparently can work at much lower temperature differences than other systems while generating 'useful' amount of work.

I wonder if this could be used to power an electric recovery generator in a Internal Combustion Engine in hybrid applications. If my back of the envelope calculations are correct a typical 2 litre engine exhausts over 1000 litres of exhaust gasses at ~300c per minute. A fair amount of energy is lost to the atmosphere this way.

In combined cycle power plants they use the exhaust gasses of a turbine engine to power a secondary steam power plant. This obviously would work in a car if it weren't for the huge complexity and cost.

Sure,but you are judt bolting one heat engine to another (the motor). I'm assuming the efficiency of the motor is better than the TEG (generally the case), so why not just draw power from the motor instead?

This technology is probably more applicable to situations where you have a source of low quality heat and you want to extract a tiny bit of power.

An internal combustion engine relies more on the heat from the chemical reaction than the heat of the engine block. And the hot gasses are just blown out the exhaust pipe. It isn't a closed cycle.

Bolting a heat engine to the internal combustion means that the excess heat from the engine block and the exhaust gases can do some work before convecting, diffusing, or radiating away. The usual problem is not the additional weight of the heat engine part, but the enormous radiator you would need to maintain a proper cold well. This would likely be a ribbed (finned) aluminum plate covering the entire underside of the car, with scoops and fans to ensure sufficient airflow across it.

The combustion engine part could then be redesigned to produce higher temperatures, as the heat engine portion can be actively driven if necessary to cool the engine block--or to warm it, as might be needed for diesel startup.

The typical way to extract power from exhaust gas heat is to add a turbo.

And in Formula 1 now this is done "completely" with the waste heat recovery units. Basically, a normal turbo frequently generates more boost pressure than the engine can use, so the recovered power is just wasted again through a wastegate. But in F1, they've managed to build very long turbo shafts, so they can be coupled to an electric motor, and thus remove excess boost by generating electricity.

Combustion engines don't rely on heat so much as pressure (can be viewed as the same thing in some models).

There's no need to have "excess" heat in an engine.

PV = nRT

In a fixed volume (the cylinder, on the time-scale of ignition), pressure is proportional to temperature.

Higher temperature in the cylinder bleeds more heat into the engine block, but also produces more force on the pistons.

Nitrous oxide systems do this, at risk of overheating the engine. If you were to actively drive a Stirling integrated into that engine, it would actively cool the engine, forcing its heat into the cold well. You would overheat your oversized radiator, instead of your engine.

Yeah, checking wikipedia the efficiency of TEGs are ~5-8% so I guess that is why this idea isn't used. It would be hard to justify the cost / weight at that limit.

There's got to be more to it than that. I would suspect that energy recovered over the service life isn't enough to more than break even.

If up front cost and/or weight/packaging were the issue you'd see them employed at least occasionally in marine/rail/off highway/stationary industrial applications where physical trade-offs aren't as big of an issue and the up front cost is can be more easily amortized over the long service life of the equipment it's tacked on to.

Many applications don’t have a weight limit, things like generators powering remote living locations.

Formula 1 cars use a much simpler method to recover energy from exhaust gases - a generator attached to the turbocharger.

Aircraft engines had an "simpler solution" to this problem, they geared the output of the turbo charger down to the crankshaft speed and directly recovered the power into mechanical energy.

This was known as a turbo compound engine: https://en.wikipedia.org/wiki/Turbo-compound_engine

These improved fuel efficiency by 15-35%. It's actually the reason my mind jumped to recovering energy from exhaust gasses. Doing this directly from heat would be elegant if it could ever be made cost effective.

Well as you know current power generator technology for the most part uses a heat source (burning coal, nuclear reactor, etc.) to produce steam from water, this pressure difference is used to drive a generator which turns the rotational energy of the turbine into electrical energy. Thermo-electric devices like thermocouples and thermopiles convert heat energy directly into electrical energy.

I didn't see in the article a T_low or T_high where this device was operational. Normally TEGs work well with high temperature differences (attached to the side of a hot wire pipe with a heat sink on the opposite side), and they work poorly at lower temps (room temp). Anyone know what temperature range this device is made for?

Misleading title:

> high-power density of 12 microwatts per 1cm2

That actually is a fairly high power density for this small of a temperature difference. The main application is probably remote sensors that run on a milliwatt or less.

Still I agree it sounds misleading.

The article says:

> shortened the silicon nanowires to 0.25nm

That's a very short wire. Isn't that about the diameter of a silicon atom? I thought the smallest wires we could make on silicon were about 20nm _wide_.

Yes, that's almost certainly wrong (or, at least, should no longer be called a nanowire). A length of 0.25 nm is of the order of the interatomic distance between a single pair of Si atoms.

The atoms aren’t that small it’s just difficult/impossible to reliably manufacture devices with features smaller than that using visible light photolithography.

A silicon atom is about 220 picometers across (subject to change depending on how you add, subtract, or squeeze the electrons).

12 microwatts per 1 cm2

0.12W per sqM

for an intuitive sense, it would take 125 hours to charge a standard phone with 1 sqM device

Or in other words it would be possible to continually power an iPhone with the temperature difference and area available in a tent. I think the context is important to whether this is presented as something negilable or something useful. Though there is absolutely no doubt that we aren’t going to see extreme amounts of power from this in tiny devices.

It also depends hugely on the nature of the generator - if it's a flexible sheet which can be used to make a tent, then awesome. If it's a rigid sheet then you'd be far better off making your tent out of solar panels even if they only saw an hour a day of sunlight.

They are not mutually exclusive though. And your tent would tend to have a higher temperature gradient during the night when there’s no sun.

A 2 cubic meter tent has a surface area on the order of 10 square meter. The temperature differential can easily exceed 10 kelvin during a cold night, else your face will freeze off.

So the potential for the application you cite is whole watts of power, enough to recharge a phone or a lantern overnight. (though, at current silicon fabrication costs, the price would be prohibitive)

So photovoltaics from the sun + waste heat is used by this on the backside. If cheap enough.

For the lowest-possible power wireless transmitter and camera (black and white, VGA), how big an area would you require then?

Right. But (I am an ignorant, sorry), how much wifi power can you get for how long?


The total power of the signal of a wifi router is around 2W. (Should I multiply it by x2 or x3 to power the internal system, the blinking leds, heat loss, ...?)

For 2W you need ~20m² (~200sqft). That is the size of two rooms. Probably using the surface of your whole roof you can get 4W or 5W.

Yes, but that’s for a 5°C difference. It isn’t clear if that changes with larger differences.

Thanks a lot. No IoT with this yet, then...

https://www.thingsquare.com/blog/articles/sensortag-power/ You would be suprised what you can pull off with wireless sensors. These people pulled off 1 year with a 2032 battery.

This could potentially power a low power beacon every so often. It doesn't seem practical for IOT wearables or for anything industrial at that power density. I'd be curious about the cost, in particular what manufacturing process size can be used for this. 1cm² of wafer can still be quite pricey at .000012 watts.

5°C Delta T generation is pretty impressive though. There are some commercial systems that can operate with 60°C temperature Delta (organic rankine cycle) and recover about 12% of that energy.

They'd need to adapt to something like aluminium foil or UHMWPE film (made like paper from the fiber-spinning solution) as a base and coat that with poly-Si. Even if it is only half as efficient, it should be a lot cheaper and possibly strong enough to be laminated into fabric like e.g. Gore-Tex membranes (extruded microporous PTFE film). A reverse electric blanket would be fancy...

By the sounds of it this is intended for use with things like Zigbee or LoRAWAN.

or 1000x less than a conventional photovoltaic technology

Could this make radioactive batteries more feasible?

By bringing the temperature differential down, we could use a fairly weak heat source to charge up a device.

Would we be able to use a tiny amount of shielded radioactive material to provide a constant heat source and use that heat for the temperature differential?

12 uW per cm2 is still poor, you'd need gigantic sizes for just a few watts. See the thread down here for some more numbers.


True but that's at 5c. What happens if we go to 10c differential? Is the gain linear or does it become more productive as the differential gets higher?

A radioactive device that gives off that much heat wouldn't make sense for something like your TV remote control. However, maybe it could help power your off-the-grid house.

Does a radioactive material exist that could give off a few Celsius degrees of heat for years?

Would it be feasible to have homes use that material (properly shielded) without being a threat to national security and safety?

> Would it be feasible to have homes use that material (properly shielded) without being a threat to national security and safety?

No - the "hotness" is pretty much intrinsically unsafe, people can break into the shielding, and if you limit it to easily containable alpha emission it has a short life.

It's a whole subtopic with uses in satellites too far away from the sun (Voyager), or rovers on surfaces where they don't see the sun constantly / to much atmospheric interference (Curiosity). But I guess there are many more, just not for your wristwatch or your home.


Not used for wristwatches, but still in use powering pacemakers! How many people are there still walking around with bits of Plutonium in them, I wonder?

Radioactive materials are used in wristwatches.


> Would it be feasible to have homes use that material (properly shielded)

(Technically,) Yes.

> without being a threat to national security and safety?


Wonder how useful this might be for space travel.

Spaceships are essentially thermos bottles (metal tube surrounded by vacuum).

So lots of surface with large temperature differences.

A temp difference doesn't really exist when one of the substances is vacuum. Without an actual material for the energy to flow into, all heat loss happens radiatively. Which definitely can be captured, just not with this sort of technique.

The external temperature is the 3K level of the cosmic background radiation. The external wall emits radiation but also absorbs radiation.

Probably the problem of this application is that it's much better to have very good insulation in the walls than to try to capture the escaping heat and try to use it for something useful as heating.

>> For instance, it may be possible to charge your smartwatch during your morning jog someday.

Already possible using 1970s mechanical tech. Or solar. Im sure there are uses, but dont go with wristwatches for your explanations.

> Already possible using 1970s mechanical tech.

Or 1980s: https://en.wikipedia.org/wiki/Automatic_quartz

smartwatch, now just watch.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact