SoylentNews is people
SoylentNews
from the getting-a-charge-out-of-painting dept.
NewAtlas has a story about a new thermoelectric paint which can turn any surface into an electricity generator.
Thermoelectric generators convert heat or cold to electricity (and vice-versa). Normally solid-state devices, they can be used in such things as power plants to convert waste heat into additional electrical power, or in small cooling systems that do not need compressors or liquid coolant. However the rigid construction of these devices generally limits their use to flat, even surfaces. In an effort to apply thermal generation capabilities to almost any shape, scientists at the Ulsan National Institute of Science and Technology (UNIST) in Korea claim to have created a thermoelectric coating that can be directly painted onto most surfaces.
Variously known as the Peltier, Seebeck, or Thomson effect, the thermoelectric effect is seen in semiconductor devices that create a voltage when a different temperature is present on each side or, when a voltage is applied to the device, it creates a temperature difference between the two sides. In this instance, the new paint created by the UNIST researchers is used specifically to heat a surface when a voltage is applied.
The specially-formulated inorganic thermoelectric paint was created using Bi2Te3 (bismuth telluride) and Sb2Te3 (antimony telluride) particles to create two types of semiconducting material. To test the resultant mixture, the researchers applied alternate p-type (positive) and n-type (negative) layers of the thermoelectric semiconductor paint on a metal dome with electrodes at the top and the base of the dome.
Applying a voltage across the electrodes, the researchers were able to measure a temperature gradient from the hot top of the dome to the cooler bottom. According to the researchers, the entire device generated an average power output of 4 mW per square centimeter.
Original Paper (Complete text)
(Score: 2) by opinionated_science on Monday December 12 2016, @12:44PM
4 mW / sq. cm.
10^4 sq.cm = 1 sq.m.
So 40W/ sq.m.
Sun puts out an average of 1kW/sq.m. at the Earths surface.
Seems quite effective , no?
(Score: 2, Disagree) by VLM on Monday December 12 2016, @01:05PM
Some more math fun... outputs 4 mW per sq cm. The paper implies it is 500 uM thick. So 3e-11 cubic inches of Te per sq cm (love mixing imperial and metric, feels like automotive work). All metals are half a pound per cubic inch plus or minus half a pound. So a pound of Te would make about tens of billions of cubic centimeters of stuff. There's no point using more than one sig fig here. Continuing, Te costs are all over the place but with increased demand of a somewhat inelastic supply the price might bounce back to historical hundred bucks per pound. Working the other side lets say after application it works for 2000 hours after which its a total loss and isn't recycled, which might be pessimistic for industrial use or optimistic for consumer applications. 2000 hours times 4 mW is a whopping 8 watt-hours per sq cm total lifetime energy production (An AA battery is like "eh about 4 Wh" and is also maybe a cm across so imagine a lifetime output of four stacked AA on every sq inch...) Now the economic value of a watt-hour in the form of bulk electrical power is maybe 5 cents per 1000 watt hours so figure the electricity produced is maybe a six hundredth of a cent. Meanwhile the Te used in the paint layers costs maybe hundreds of millionths of a cent.
To a first approximation the paint layers have a net capital gain but the labor cost is staggering and the production rate is practically zero. The labor cost of the layers looks very expensive.
It would not make a useful primary source of energy.
It might be useful for measurement. Why you wouldn't use a simple COTS thermocouple is mysterious. I suppose painted on has good thermal conductivity.
For non-primary energy applications you can do fun stuff with milliwatts of electricity. Strange microcontrollers can run down there. A modest size "thing" could light an indicator LED (perhaps a temperature warning?)
(Score: 0) by Anonymous Coward on Monday December 12 2016, @01:24PM
Aren't you just a little ray of sunshine ...?
(Score: 3, Interesting) by VLM on Monday December 12 2016, @01:32PM
Another fun math is I'm way to lazy to look it up but a stereotypical TO-220 power transistor case has a surface area somewhat larger than a sq cm but not ridiculously so, and in bare air the thermal resistance is about sixty or so deg/watt so guessing a hundred degrees/watt is reasonable per sq cm. Meanwhile the paper claims peak power output of 4 mW per sq cm at 100C hot side and presumably room temp cold side so we'll call it "somewhat less than a watt thermal" to generate enough temperature to get about four thousandths of a watt out. So we're in no danger of perpetual motion here, figure half a percent efficient.
This shows the paradox of how thermal electric elements can lower efficiency of a plant or whatever. See half a percent of the output of a nuclear power plant is a lot of power and unpressurized condensers operate at boiling water 100C temps so superficially this sounds like free energy. However the TE modules don't run "at" 100C they run across a drop of 100C... Now you're gaining half a percent in the TE modules however running the condensers at 200C instead of 100C means you're losing, oh, I donno, maybe 25% of the power from the turbines for a net loss of 24.5% when you add back in the TE modules. Or you have the immense capex problem of expanding the condenser to cool as much with the TE elements in the way as without, but again it turns out that using the worlds most amazing condenser without the TE modules in the way would result in really cold condensate and more power generated in the turbines than you'd ever get from the TE modules.
Likewise for cooling if it takes a TE module 20 watts to refrigerate 1 watt (which isn't all that far out of line) then if you have a 100 watt CPU and 100 watt heatsink then putting a TE module in between and dumping 2000 watts into the TE module means your poor little 100 watt heatsink has to dump 2100 watts of heat, which it can't, meaning you'll roast the CPU. Or if you bought and installed a 2100 watt heatsink you'd just break even, however, if you bolted that 2100 watt heatsink to the CPU directly, it would run so incredibly cool you'd have to worry about condensation, plus you wouldn't have to pay for the TE modules.
TE modules only "work" for really weird applications, like motion and vibration free cooling of telescope cameras or in general places you don't care about cost or energy efficiency you just want a certain controllable delta-T.
(Score: 0) by Anonymous Coward on Monday December 12 2016, @07:38PM
20:1 IS pretty far out of line for conventional devices -- the one in TFA is just really bad for geometric reasons. Assuming you've chosen a suitable device for the task, rather than having found one that's really too small in your parts box and "making do", COP will be between 0.3 and 1.0; in fact you can get well over 1.0 if you're designing for efficiency above all (you'll need to oversize the TEC at least 3x, and run at a modest ΔT).
See this helpful graph [tetech.com] to follow the numbers in the following paragraph:
You're describing a COP of 0.05, which typically happens when you're running the TEC right at its limits. Assuming that's the case, you should reduce the current a bit, the ΔT would drop just a little, but the COP would improve to 0.1, and you'd need half the power. Conversely, you need to dump half the power, so the gain from using your 2100W heat sink with "only" 1100W would likely outweigh the ΔT reduction. Or, if you really need the high ΔT, use a two-stage system, with each stage seeing half the ΔT, they'll each be operating at much higher COP, and need far less than half the power.
TECs are really quite practical, as long as you can afford to buy the right size for the job, instead of trying to make a smaller one work out in the corners of its capabilities.
(Score: 2) by TheRaven on Monday December 12 2016, @04:58PM
sudo mod me up
(Score: -1, Offtopic) by Anonymous Coward on Monday December 12 2016, @01:49PM
Instead let's talk about release 100 of a libre vidya game:
http://soylentnews.org/article.pl?sid=16/12/11/1758223 [soylentnews.org]
Free Video Game "ChaosEsque Anthology" Reaches Release 100
(Score: 2, Informative) by Anonymous Coward on Monday December 12 2016, @02:05PM
These devices don't convert heat to electricity. They convert a temperature difference to electricity. The heat doesn't disappear, it just gets transferred to the other (colder) side. Once the heat difference is gone, no more electricity.
If you can find a way to actually convert heat to electricity, you are going to be rich (assuming nobody patents the method before you can use it). Heat is what happens to waste energy. Heat is where the energy lost in a perpetual motion machines goes. Find a way to turn heat back into electricity, and you have solved both the perpetual motion machine and global warming.
When you build a refrigerator that has an electric socket rather than an electric plug, and the answer to "my beer isn't cold enough" is to plug another PC into the refrigerator, that's when you are converting heat to electricity.
And before anyone says "steam locomotive": The temperature difference between the boiler and the water tank is what allows to build steam pressure. Cold water in -> heat -> expands into steam.
(Score: 1) by Scruffy Beard 2 on Monday December 12 2016, @02:23PM
it is called solar power.
Efficiency hovers around 20-30%.
Cooler objects also have back body radiation, but there is less energy to work with.
(Score: 0) by Anonymous Coward on Monday December 12 2016, @02:42PM
wrong. that's the photoelectric effect, very different from heat.
read it up. einstein got the fucking nobel prize for explaining that it's not heat...
(Score: 2) by Scruffy Beard 2 on Monday December 12 2016, @05:41PM
It is possible to solar array [solarreserve.com] that uses heat to turn a turbine, rather than the photoelectric effect.
(Score: 0) by Anonymous Coward on Monday December 12 2016, @02:53PM
Heat is where the energy lost in a perpetual motion machines goes.
Well it ain't really a perpetual motion machine then is it?
(Score: 4, Interesting) by GreatAuntAnesthesia on Monday December 12 2016, @03:53PM
Find a way to turn heat back into electricity, and you have solved entropy
You're not thinking big enough. [multivax.com]
(Score: 1, Interesting) by Anonymous Coward on Monday December 12 2016, @06:53PM
In any thermoelectric device, you have two ends. In operation, either can be the cold end, and the other the hot end, depending on electrical polarity. A conventional TE device is a thin plate, with one end on each face; you sandwich the device between a hot object on one side, and a heatsink & fan (or watercooling block, or whatever) on the other side, and it works (either generating power by slowing down heat transfer, or increasing heat transfer at the expense of power).
In this painted-on TE device, however, the hot end is one region on the surface it's painted onto, and the cold end is another region on the same surface. So it won't work well at all on a surface with decent thermal conductivity (the experimenters used glass substrates), as most of the heat flux would go through the underlying surface rather than the TE layer. In fact, it's hard to think of a situation where it's likely to work well for generating electricity -- the paper depicts a hemisphere with a hot pole and a cold equator, but I can't think where one would see a heat distribution like that on an existing surface.
And if you have to add a glass substrate connecting, say, a hot metal cylinder head to a cooling block, and paint TE stripes on that substrate, the ability to shape that glass arbitrarily and paint over the resulting curves doesn't seem so great -- why not just shove a planar TE device between the head and the cooling block? Even if you have to add an aluminum spacer to adapt a curved surface on the head to a flat surface for the TE device, that doesn't seem worse than having to add a glass substrate connecting them.
And the worst part is, the extended length from the hot end to the cold end increases resistance, which is largely responsible for the very poor performance. Conventional planar configurations have many short elements because it works better that way.
If I try to think where this could make sense, the big potential of their hemispherical demonstration unit seems to not be conforming to arbitrary surfaces, but the fact that the hot and cold ends are not the same size. In fact, with the hot end at the center of a circular disc, and the cold end at the outside, you significantly increase the surface area -- this could benefit convection. Consider the back of your refrigerator -- it has a long, serpentine metal tube to cool and condense the refrigerant. We'd like to keep this cooled by natural convection, as it already is, but also generator electric power from it. Take a series of plastic discs with a hole sized to push onto that tubing, and paint/print TE elements on both sides -- push dozens (hundreds?) of these on each horizontal run of tubing, with insulating spacers between. The idea is that increase in surface area from the discs balances out the increased thermal resistance and the partial or complete insulation of the tubing between fins -- same heat flow as before, but now it (mostly) flows through the TE elements. (However, I haven't run any numbers on this, and even if it works, it's not clear this sinterable paint is actually better than stacking discrete semiconductor blocks around the tube, with an aluminum or copper heatsink fin, since we're disregarding the purported benefit of conforming to arbitrariy curved surfaces. In either case, I expect it will prove ludicrously expensive.)
In fact, it seems like the most interesting part in the paper may be the one not being talked about -- using this sinterable semiconductor "paint" to fill molds, and producing round discs (or arbitrary shapes) of p-type and n-type material to be used in building more conventional planar TE devices. I'm not sure how the semiconductor blocks in conventional devices are produced commercially -- it's not clear to me that molding to arbitrary shapes is (or is not) actually an interesting step forward for commercial devices, but it's interesting (even if not commercially, it could be good for hobbyists, though Te is toxic), and actually produces devices with reasonable performance.