[ars]

Conservation of energy is not the issue. It's not enough just to have energy in = energy out. In order to turn heat into any other kind of energy you MUST have a cold sink.

If you want to have black body radiation it's not enough to just have a dark area - the recipient of the light must be colder than the source.

So this would only work if the LED was heated more than the room around it. I guess if they put it in an oven, and then viewed it from a window in the colder room.

[Tuna-Fish]

> In order to turn heat into any other kind of energy you MUST have a cold sink.

No, in small enough scales this is not necessary. This is not "heat moving from a hotter object to a colder object and doing work on the way", this is "heat being directly converted to photons". When you view a patch of hot gas as individual molecules bouncing about you can transform the heat of a molecule into momentum of your target at 100% efficiency (heat is kinetic energy!). In quantum environments, the second and third law do not work like they do in the macroscopic world. The second and third are still not broken in the large because to do this kind of trickery you need to have a lot of exact information, and that has a cost in entropy. (So you'd need a maxwell's demon to scale this up.)

[ars]

No, this is just not true.

You can only convert the kinetic heat motion of a single molecule to kinetic energy in your target if your target is colder than the source! And there's your cold sink.

The efficiency is not 100% because the target sends energy back to the source since the target is moving (from heat kinetic energy).

If the target was standing still efficiency would be 100% - but that's the same as saying the target is at absolute zero and we already know that Carnot efficiency is 100% if the sink is at absolute zero, so it makes no difference that you are dealing with a single molecule.

To your second point that this is "heat being directly converted to photons" - that's exactly the definition of blackbody radiation. But the blackbody also absorbs radiation from the environment it is in, so it's not perfectly efficient either.

[pbhjpbhj]

It's simply not a full system analysis, > unity efficiency WRT input electrical energy. There's just a corresponding draw of energy from the environment that is also producing photons.

>"You can only convert the kinetic heat motion of a single molecule to kinetic energy in your target if your target is colder than the source! And there's your cold sink." //

This is counter logical. You're saying that a small fast moving body can't impart energy to a large slow moving body. Can you explain further how this works at the single molecule level?

[ars]

> There's just a corresponding draw of energy from the environment that is also producing photons.

It is only possible to do that if the photons are sent somewhere that is colder than the device.

> You're saying that a small fast moving body can't impart energy to a large slow moving body.

No, I'm saying it can't impart all of its energy, and the more energy the other body has the less energy it can impart. i.e. it's exactly Carnot efficiency.

Perhaps what you don't realize is that Carnot efficiency doesn't represent lost energy. It represents heat energy than can not be converted to another form, and remains in the source.

[pbhjpbhj]

>It is only possible to do that if the photons are sent somewhere that is colder than the device. //

Where are you deriving this from. What force prevents me shining a light towards a hot object.

[DougBTX]

There needs to be a cold sink since heat is just a variation on kinetic energy (small vibrations), and momentum needs to be conserved. The released photons have momentum too, and it looks like they get that from the heat (=vibrations) in the LED.