Slashdot Mirror
LED's Efficiency Exceeds 100%
New submitter Paul Fernhout writes "Physicists from MIT claim to have demonstrated that an LED can emit more optical power than the electrical power it consumes. Researchers suggest this LED acts like a heat pump somehow (abstract). Is it true that 230% efficient LEDs seem to violate first law of thermodynamics?"
They must have used the wrong cable, causing the light to go faster than C and mess with their readings.
Exceeds 100% ELECTRICAL efficiency is the key here. The conservation of energy is still intact because it supposedly uses heat energy to supplement.
So if I get the article right - LED cooling?
Really puts a whole new perspective on LED clad 'gaming'-machines, which as you know - should have blue LEDs for cooling, and red LEDs for superior overclocking.
Or you could read the damn links and find out. But I guess easier to make guesses.
"In this house we obey the laws of thermodynamics!"
My sci-fi novel, Ghost Thief, is now available from Amazon.com.
From the article: "The researchers didn’t try to increase this probability, as some previous research has focused on, but instead took advantage of small amounts of excess heat to emit more power than consumed. This heat arises from vibrations in the device’s atomic lattice, which occur due to entropy." The other thing to note is that these LEDs are being run at REALLY low power.
For those wondering about conservation of energy, it's intact. The extra energy comes from heat / vibration in the system.
For those concerned about the second law of thermodynamics, it's not specifically addressed in the article, but the smart money's on entropy increasing in this experiment. The second "law" is really just statistics though (law of large numbers anyone?), and as with most statistics people are still arguing about what it really means. See http://en.wikipedia.org/wiki/Second_law_of_thermodynamics#Controversies and http://en.wikipedia.org/wiki/Fluctuation_theorem
"30 picowatts and measured an output of 69 picowatts of light - an efficiency of 230%. The physical mechanisms worked the same as with any LED: when excited by the applied voltage, electrons and holes have a certain probability of generating photons. The researchers didn’t try to increase this probability, as some previous research has focused on, but instead took advantage of small amounts of excess heat to emit more power than consumed. This heat arises from vibrations in the device’s atomic lattice, which occur due to entropy."
They are not claiming more than 100% efficiency in total terms.
See my journal for slashdot ID's by year. Mine created in 2005. http://slashdot.org/journal/289875/slashdot-ids-by-year
These guys must have been hanging out with Martin Fleischmann and Stanley Pons .
11 comments down and finally someone has actually understood enough of the summary to know that they aren't claiming that conservation of energy is dead.
Slashdot - News for Nerds, Stuff that Matters, in ISO-8859-1 Has just realised that beta makes this signature redundant
It says in the summary (and in the article) that the LED at very low electrical input levels, acts as a heat pump. It absorbs local heat energy and converts into photons.
So you get more light out than electricity in, because you're stealing heat and converting it to light. It's not more than 100% efficient, it's multiple energy sources being used. No breaking the laws of thermodynamics.
I'm out of my mind right now, but feel free to leave a message.....
Definitely GPS timing error.
Never say never. Ah!! I did it again!
From the article: "In their experiments, the researchers reduced the LED’s input power to just 30 picowatts and measured an output of 69 picowatts of light - an efficiency of 230%." It only would violate the conservation of energy if it converted the electricity to more electricity than came in. It's just converting the electricity into light very efficiently with a great ratio.
Nothing is more dangerous than a programmer with a screwdriver.
The LED is "consuming" external heat to produce the additional light. The article is pretty clear and an enjoyable read.
Global warming to the rescue!
To offset political mods, replace Flamebait with Insightful.
It's a good example. The hub of a wagon wheel will be warm to the touch. That heat comes from the motion of the wheel. A sympathist can make the energy go the other way, from heat into motion. I pointed to the lamp. Or from heat into light.
There was an art to choosing your projects in the Fishery. It didn't matter if you made the brightest sympathy lamp or the most efficient heat-funnel in the history of Artificing. Until someone bought it, you wouldn't make a bent penny of commission.
I didn't follow through to the abstract, but the article didn't claim to be creating net energy. There could be other causes for more net energy emitted than applied, such as the device being on fire.
Gently reply
Thank you for that, TFS made it look like they somehow managed to overcome the laws of thermodynamics. I was going to ask you to explain farther but RTFA instead.
It seems to me that it's just fiddling with the numbers; it isn'r really ">100%", it's "better than 100% when compared to higher power levels."
I did expect a better article from Psysorg. I'm still not completely clear on it, but it seems to have no practical value, only an academic excersize.
Free Martian Whores!
Wow I'm totally shocked, what's the world coming to? :) All you have to do is actually read the linked article to see there's no sort of thermodynamic violation of any sort implied, not that most of the people posting here will bother to RTFA.
According to TFA, they are actually taking advantage of other sources of energy in addition to the electricity provided by the wall plug. So it's not really the LED getting "greater than 100% efficiency", it's really "producing more light than you would get if you only took advantage of the electricity from the wall plug".
And they're talking in the range of 69 picowatts of light output, using only 30 picowatts of "wall plug" energy input. So it's quite believable.
as most people think Light Emitting Diode when they hear LED.
But in this experiment they are referring to a Large Entropic Dilemma.
So the results make perfect sense.
intellectual property law is philosophically incoherent. it is your moral duty to ignore it or sabotage it
Now, all we need is a solar cell with 100% efficiency and we're in business.
That was my first thought. Even the explanation of the extra energy coming from heat is nice and doesn't stop solar cell from working. Might actually work well together: the cell is going to leak heat, this will soak it up.
Surely if this is true the "light" is not the big story.
If you can take "heat" and convert it into another form of energy that is HUUUUUUUGE NEWS- yes I know, steam engines, etc, but they require a large difference in temperature.
Imagine if your fridge/freezer- GENERATED power- by taking heat energy and converted it into electricity?
"That's the way to do it" - Punch
Yes, to say 230% efficient is really a false statement. There's no violation of thermodynamics, it's just that the LED has more energy sources than the electrons it's drawing down the wire.
The world's burning. Moped Jesus spotted on I50. Details at 11.
Interesting to see the number of posts saying that this is absolutely not possible - reading through the article, it seems possible and maybe there is enough here to study the phenomena enough to warrant more investigations.
The LED seems to be emitting 69 picowatts (pico = 10^-12) when only 30 picowatts of electricity is being pumped in with a measurable decrease in the temperature of the LED. This implies that the LED is acting as a heat pump, converting heat energy into light. If you've ever seen a Peltier cooler in action (or worked through the operation), it seems like to me this is possible.
Note that the power level this phenomenon is observed at is extremely low - the result is maybe good enough for cooling a few molecules of beer - but I think there is something here that should be investigated to see if any usable applications could come out of it.
myke
Mimetics Inc. Twitter
Yeah, but man, a it's a completely solid state heat pump that dumps waste heat as usable light - now that's something. Just imagine: every server, instead of needing cooling, can have this stuck to the heatsink and mounted on a tall pole. No more datacenter, we'll have datapoles, and our streets will be full of them :)
A successful API design takes a mixture of software design and pedagogy.
A third option is that you didn't read the article.
You must be new here...
-Arthur
Cave ne ante ullas catapultas ambules
Given that we can cool stuff down to less than a nanokelvin IIRC, and it took some ingenuity (Nobel prize for Bose-Einstein condensate), I doubt this LED will come anywhere close.
A successful API design takes a mixture of software design and pedagogy.
Read the article? Heck, I didn't finish the headline. As soon as I realized it didn't mention iPads I went straight to the comments to argue we should instead discuss iPads.
Why don't we have iPad 4 speculation yet?
1. I for one welcome our new iPad 4 overlords and their app that allows you to put hot grits on Natalie Portman and disguise it in a bad car analogy.
2. Ask if it runs Linux, and then cite another failed year of Linux on the desktop.
3. ???
4. Profit.
What were we talking about again?
http://blindscribblings.com - Tasty pop-culture in conceptual fashion.
I can see a practical application right off the bat - completely silent cooling for computers and satellites. As I understand, particularly the latter is a huge pain in the ass for engineers.
The fun trick will be to point this at a 45% efficient photovoltaic panel to generate the electricity.
"Who is the Journal of Quantum Physics going to believe?" --Stephen Hawking
So the lights sucking the heat out of the air and feel physically cold to the touch?
Does this 230% conversion ration only work in really high heat location or is this in room temperature?
Would this technology not really work in -40 degree winter environments?
Troll is not a replacement for I disagree.
No it isn't fiddling with numbers. You are missing the heat pump bit.
The device is taking X amount of energy from the electricity supply and X * 1.3 of energy as thermal and converting this to X * 2.3 as light. i.e. it is 230% efficient when comparing light output to electrical input. Equally, it is 100% efficient when comparing light output to electrical and heat energy input combined.
This does take a little bit of thinking to get your head around but I have a more common example in the shed outside. It contains a heat pump which is 350% efficient. It takes 2kW from the electricity supply and outputs 7kW of heat energy to heat my house. The missing 5kW comes from the pipes in the garden as heat energy. The result being that the garden is slowly being cooled. http://en.wikipedia.org/wiki/Heat_pump
wot no sig
In theory.
I'm out of my mind right now, but feel free to leave a message.....
Now, some people might still be bothered by this, because the idea of using ambient heat to do useful work is another one of those "perpetual motion machine" kind of claims. Heat represents a disordered (high-entropy) state, from which you cannot extract useful work. The relevant thought experiment here is the Brownian ratchet: the idea being that you have a ratchet that gets bombarded by random molecular collisions (in water or air, say). The ratchet will turn foreward when a random collision is strong enough, and so over time you can use this turning motion to wind a spring and thus convert random thermal motion into stored energy. The reason this doesn't work in real life is because if random thermal motion is enough to overcome the pawl on the ratchet, then the pawl will be 'hot' enough that it will randomly and spontaneously lift up, turning the wheel backwards. The only way to avoid this is to have the pawl at a lower temperature than the rest of the mechanism: this works, but it's well-known that you can extract useful work from a thermal gradient, so the laws of thermodynamics remain intact.
Coming back to this present result, how does this device use ambient heat to generate useful photons? Sure, it acts as a thermoelectric cooler, establishing a local thermal gradient, but this sounds like 'cheating' in that it's a way to extract energy from the entropy of the surroundings! The very first sentence of the scientific paper addresses this:
Basically, the device is converting high-entropy thermal energy into even higher entropy incoherent electromagnetic radiation (light output). So, the second law of thermodynamics is not violated. Essentially, this device is acting as a way to connect thermal degrees of freedom to E&M degrees of freedom. The system, wanting to increase entropy as much as possible, tries to spread energy through all these degrees of freedom, which means creating some photons at the expense of some of the heat in the material.
It's a neat bit of physics, and will probably have implications for device efficiency and other applications.
The semiconductor PN junction is amazing. That's what's fundamentally inside LEDs. When appropriately tuned, PN junctions (a) permit electron flow in only one direction, demonstrating their diode nature, (b) convert current into light, like an LED, (c) convert current into a heat differential, like a Peltier junction cooler, (d) convert light into current, like a photo cell, (e) convert heat differential into current, like a solid-state thermionic energy converter, (f) act like a voltage-tunable capacitor, like a varactor, and more. In fact, to a very coarse first approximation, all PN junctions exhibit each of these characteristics to a greater or lesser degree.
So what's this group done? Shown that an appropriately tuned PN junction (or stack of them, I'd imagine) can be used to simultaneously act as a solid-state thermionic energy converter *and* an LED. Thus, it converts applied electricity to photons, but also converts a heat differential to electricity, which gets converted to photons as well, meaning it's sucking heat out of its immediate evironment. Cool stuff, if you'll pardon the pun.
Put my fist through my alarm clock with its ding-dong death inside my ear. - The Blackjacks.
No, it is not. The linked article is quite clear: the LEDs are geting colder, so the extra power output comes from the environment.
Now, 230% efficiency suggests that it is operating as a >100% efficiency heat pump, and that's also impossible. It might be decomposing itself in an endothermic(sp?) chemical reaction, or something.
It's not impossible. Heat pumps are routinely of greater than 100 percent efficiency, because they don't measure the input heat, just the output heat and the input electricity. Sure, it's a marketing scam, but what-r-ya gonna do?
When our name is on the back of your car, we're behind you all the way!
I once observed a low threshold LED (has a much less than 1.4V on-voltage) that was only attached by one lead, with the other lead hanging freely in space. The LED was quite clearly "on". When you put your finger closer to the free hanging lead (but not touch) it got brighter. It was just acting as an antenna in a room with lots of EM radiation around, and the induced current was enough to light it up.
"I have never let my schooling interfere with my education." - Mark Twain
This is not as incredible as it sounds. To explain how it works, it is perhaps easiest to start with a simpler device. I could take a brick, connect a battery to it and say "Look! This brick is only consuming one milliwatt of electric power, yet it is emitting one Watt of infrared radiation. That is 100 000 % efficiency!" If I did the same thing at 1 000 degrees Celcius, the brick would even be emitting visible light (wether connected to a battery or not.)
What the people at MIT do is a little more complicated. They don't use the black body radiation directly. Instead they take electrons that would have emitted infrared photons, add some more energy to them, and get visible light. For this to work, they only have to add the difference between the energy of an infrared photon and a visible photon, yet they get the light output of a visible photon. At a temperature of 135 degrees Celcius (that is 275 degrees Farenheit if you happen to live in Belize or the United States) the difference between the black body radiation and visible light was small enough that they managed to get over 100 % efficiency. No laws of thermodynamics were violated.
I've always wondered how glowing metal aligns with the second law of thermodynamics. It seems to directly convert heat (lower order energy) to light.
Heat is not an inescapable sink of energy. It's just that at a large scale, entropy (including heat) must increase.
When metal glows when hot is it consuming anything or utilizing the difference in temperature in some way?
It's losing heat. If a block of metal emits 1 J of light, it cools by 1 J.
Or said another way, if you put a piece of metal in a perfectly insulated hot box would it still glow forever?
A piece of metal in an insulated box, as one would normally think of it, still has two items interacting thermodynamically: the block of metal and the inside of the box. If the metal is magically suspended in the box and the box otherwise contains a vacuum, then there is no conduction or convection between the box and the block. There is still, however, radiative heat transfer -- the light the block is emitting. If the inside of your box is a perfect mirror, finally you will have an insulating box that is not heated by the block. In this case, the block will glow forever -- but that's because the light it emits due to heat reflects off of the box and is reabsorbed by the block, heating it. (In other words, if it emits 1 J of light, it cools by 1 J, but then absorbs the same 1 J of light, heating by 1 J.)
For example, is the LED getting colder? Could it be converting heat to electricity?
Wow, yeah they should look into that...Oh wait, they already did and the LED is indeed getting colder and they postulate that it is converting heat into light.
The truth is that all men having power ought to be mistrusted. James Madison
*NEWS FLASH* The internal combustion engine used worldwide for over a century produces a jillion percent more output power than the electrical input power used to fire the spark plugs.
Yeah, but man, a it's a completely solid state heat pump that dumps waste heat as usable light - now that's something. Just imagine: every server, instead of needing cooling, can have this stuck to the heatsink and mounted on a tall pole. No more datacenter, we'll have datapoles, and our streets will be full of them :)
You're not thinking like an evil genius. You've got a 10MW data center. You have a way to convert the heat load into light. And now you want to distribute it all over and make street lights out of it?
Whatever. I want my huge frackin' laser.
Yeah it's not the first law it violates but the second.
When you visit a hardware store, and see a pack of incandescent bulbs that advertise 100% efficiency, you should thank the law of thermodynamics for obfuscating yet another specification.
It's a neat bit of physics, and will probably have implications for device efficiency and other applications.
It's the solution for global warming.
Take a large bank of these over-efficient LEDs. Shine them on a solar panel. Power the LEDs from the solar panel output. Everything in the vicinity of the LEDs gets cold. Make lots of these. Problem solved.
If it seems like a perpetual motion system, it probably is. If you've got a 230% efficient LED, then you can have a 50% efficient solar panel and still come out ahead.
The only problem is what to do with all the excess electricity these things will produce.
> The fun trick will be to point this at a 45% efficient photovoltaic panel to generate the electricity.
No chance, at 2.5um that is even theoretically impossible. Higher efficiency requires much shorter wavelength.
While my knowledge and understanding is limited I think that the extra power in the light output comes from heat. So light power out is greater than electrical power in but if you consider thermal power AND electrical power then total efficiency is under 100%. Thus the first law of thermal dynamics is safe.
Only temporarily. If this device is converting heat into light as a byproduct of converting electricity into light, it's still LESS THAN 100% efficient, it just means it's getting the energy from the environment around it, NOT making it itself, OR it is using its own internal heat, meaning the longer it does this trick, the colder it gets. This provides a fundamental limit, namely 0 Kelvins, assuming it can even operate at all anywhere near that cold. Once it reaches this temp., it should start operating at below 100% efficiency, per the laws of thermodynamics.
Don't get me wrong, if they have figured out a reverse amplifier transistor, one that uses a small current to bias a PN junction, and then receives energy from the background, they've effectively done what Tesla was trying to do around the time he died, just in a completely different way, making a device that can receive the sun's power indirectly, like an antenna, only it would use the environment itself as the antenna, the very air around it.
Of course, at the picowatt level, unless they can use 22nm or smaller silicon printing tech to make HUGE numbers of these, I would think they'd be prohibitively expensive. People concerned with what color light they output are not seeing the big picture. This allows us to finally, (and don't mod this funny, I'm not joking) use solar power 24 hours a day, since the sun bakes the earth, and the heat remains even at night, (unless you're some poor SOB who lives in Northern Montana, at high altitude where the temp in the dead of winter, at night, gets real close to absolute zero, they say...)
I once had an idea like this myself, but never pursued it. My version used conventional refrigeration... It feels good to be vindicated. Guess I should have thrown caution to the wind and built that prototype after all.
Or, just maybe... they cocked up the experiment, which seems vastly more likely, or they're making it up. Is a bit odd for the usual Slashdot April 1 bullshit. But just to throw this out there, remember everyone, grown men are prone to playing idiotic pranks and getting a cheap thrill from believing others believe them when they make shit up, and feel that at the close of March, each year, that some archaic change to a calendar is adequate excuse to give in to those urges, and act like children. I do not look forward to this particular time of year.
I wouldn't say it is a scam but it is misleading if you don't understand where the extra energy comes from, but then who cares as heat pumps are much more efficient at heating and cooling a house than a furnace, resistive heating elements, and air conditioners.
Time to offend someone
Not only silent, but you can beam the light to an external collector that produces electricity. It's like having a peltier element where the hot side can be in a different building.
Comment removed based on user account deletion
run it through the faster than light cables too
This is not violating the first law of thermo (energy conservation). It is getting the energy it needs from it's environment.
However it might possibly be violating the second law of thermo. Turning heat into light at high efficiency should not be sustainable. energy in the form of light has more less entropy than energy in the form of heat.
I could imagine that, in burst mode, that some energy is somehow being stored so that it can when triggered temporarily emit more or seemingly defy entropy. For example perhaps the crystal lattice is disorganizing during emission and then self healing to an organized state over time. This would be taking energy from the environment and shedding entropy to the environment and not neccessarily viloating any laws.
So some game is being played and I'm surprised anyone would publish the findings without an explanation for this.
Some drink at the fountain of knowledge. Others just gargle.
So, the LED converts free heat along with the electrical energy into light. So I suppose, yes, if you had a closed, dark room, and you made an array of these lights, and you plugged them into a power source, and you spotted them on a PVC with over 50% efficiency, and you routed that PVC's output back into the LEDs, you would have yourself an ongoing power source that would actually increase in power until you tapped some of it out for safety. Which you could then use! Furthermore, the meanwhile, the LEDs thermocoupling the heat energy along with the joules of electrical energy and transforming it all into light are steadily making the entire environment cooler.
#1:) higher efficiency
#2:) cooler house
That means you are also cooling your house while you are generating perpetual motion! This is good because you will have to keep your house heated in order to continue to supply energy to the system you have created. The sun is the most obvious source! If you switch from tinfoil to copper foil and cook the outside of it twice before wearing it, your foil hat can act as a sun magnet!
At any rate, this all seems like an argument for further complicating a field of study I have luckily not yet spent college money on. They will end up having to either re-do their study with a mind toward clarifying where energies are coming from and how they're measurable, or physics is going to have to come up with a yet even more complex unit of measurement (oh, great) in order to avoid further claims such as these.
"Stratigraphically the origin of agriculture and thermonuclear destruction will appear essentially simultaneous" -- Lee
I would like to see the entire experiment rather than an incomplete summary. here are a few questions.
1. How long was this effect present?
2. What was the temperature of the LED as the power was decreased?
3. Was the same effect there if the power was started at 0V and slowly increased?
If it only was present for short periods while the power was decreased the effect might even be a capacitance release of power stored in the LED. In the lower efficiency phase electrons may be stored in the LED and released as the power gets lower.
... that adding LEDs to stuff made it cooler.
Collector's Edition
To be fair, you pretty much had to miss about an entire week of news around the internet to have missed out on the loose cable story. People tend to take their light-speed limits pretty seriously these days.
Yes, to say 230% efficient is really a false statement.
Depends on your perspective. If you are selling these as LED light bulbs that output twice as much energy then they take from the wall plug, then yes, they are 200% efficient. They don't output more energy than what is put into the system, but they do output more energy than you put into the system and since we are all (as a species) self-centered egomaniacs, that is all that matters and the terminology is correct (for the audience).
Monster Cable TOSLink?
In the book Sundiver by Robert L. Forward, a research ship traveling inside the Sun gets its drive sabotaged, and they escape by using the cooling laser as a drive, freezing everyone aboard.
Lasers are today used to cool to a few milli Kelvins, and below, very close to absolute zero temperature. The reflected colour of the laser is a little bit less pure, as the thermal vibrations are removed by increasing the entropy of the laser light.
The same principle is going on in this light diode.
SO tired fof the crapped 'Science fiction' prophet crap.
You have hundreds of people doing 'scientific' stuff, and some happen to turn out to be the way it's now. Big deal.
Setting aside the fact that most 'prophetic' example are really no more then a bad attempt to give an author more 'cred'.
The Kruger Dunning explains most post on
A) you would just need 40% efficient panels.
B) taking heat out of the atmosphere would be fine.
Of course, if we could get 40% efficient panels, that would be a world class break through.
The Kruger Dunning explains most post on
It sounds like it violates the 2nd and 3rd laws of thermodynamics if not the first.
The first law of thermodynamic is...you do not talk about thermodynamics :)
I see a lot of joke comments talking about hooking this LED, that is getting a significant amount of high entropy heat and turning it into higher entropy light, to a photoelectric cell. I can understand the heat to light in terms of thermodynamics, but assuming an efficient enough photoelectric cell isn't turning high entropy energy into low entropy usable electrons "cheating?" Could someone tell me why this wouldn't work in the real world (I just assume it can't)? Do photovolts not work at that energy level? Would the energy be so low that the photovolt would simply loose the electrons to signal loss before generation enough of a voltage gap (though this would be fixable with well insulated micro circuitry one would think)? The explanation of turning heat into light thermodynamically seems acceptable to me, but I do not understand where the loss would come when converting said light back into electrons.
If this result could occur at useful energy levels, it would be revolutionary. But very likely it's just an interesting anomaly that occurs at very low energy levels. My guess is that at such low energy levels, the disordered nature of the energy in the ambient heat doesn't come into play, but at higher energy levels it would. There simply wouldn't be enough heat energy in the surrounding environment to create the same effect at higher energy levels. Please let me know if there's a flaw in my undertsanding of the science.
Is this some form of Maxwell's demon, having the same effect but in a way not so far envisaged ? It seems to me that it takes the heat energy, tops it up with some electrical energy and before that process can reverse it radiates the energy away as light. The radiating away has the effect of the trap door - preventing the reversal.
This device may not work well if there are many of them that can shine on each other, an incoming photon could knock an electron up into the conducting band leaving a hole behind and generating some heat. Thus to be useful the light that they generate would have to be directed away with little reflection.
I've just realized something remarkable: the slashdot comments are less stupid than those in the surrounding environment, where people are having trouble grasping that the 2nd Law of Thermodynamics is the issue, not the 1st.
Now, if only we could harness this gradient of stupidity.
"That's a bright idea, he said coldly."
Also, heat-activated lighting. Also, if you can suck heat from the environment to make light, and then pump the light to solar cells to make electricity, you have a heat-to-electric converter.
Maxwell's demon must be rolling over in his randomly displaced bed right about now.
I've fallen off your lawn, and I can't get up.
What an intriguing idea. A light that actually consumes heat and emits it as light. One energy source is as good as another -- I have to wonder if there are many other areas of research where we've focused on electricity as the power for a device when there might be alternative transforms from other sources available.
In short, no, it doesn't sound like they're breaking any laws of thermodynamics or energy balance equations. Instead, they're just using an unusual source to boost the inputs: heat.
I do not fail; I succeed at finding out what does not work.
Don't you know, the first law of Thermodynamics is: We don't talk about thermodynamics!
Although were not talking a lot of power it might be something if you could get the components small and efficient enough such that for each combo led+photocell you could power at least 1.5 other units. Not base 2 but still significant if you can get a billion such micro units working.
A think there are more efficient ways to harvest power from heat. I'm still waiting for process with micro rectifiers for converting heat vibrations directly to electricity.