It does NOT cool things. To use it you have to heat it hotter than wherever it is shining into.
It uses that heat to make light, but it is not a cooler (i.e. a heat pump).
It uses that heat to make light, but it is not a cooler (i.e. a heat pump).
The paper[1] behind a pay wall show has a graph showing its cooling power vs current.
I think where the light is shining into is open space so the source temperature has to exceed what ever temperature the local background radiation is at; a common local minimum being the cosmic background radiation.
The cosmic background radiation is not exactly common. You'd have to be in space.
BTW, I don't think we are actually arguing.
Unless you claim that the LED stops producing the >100% efficiency levels at room temperature, I don't see why you're claiming it can't cool anything down.
edit: I don't mind a downvote now and then, but I would really prefer a discussion or explanation to a fire-and-forget downvote.
Light is a more highly ordered form of energy than thermal noise; it can be used to do work. If you could convert thermal noise into light and beam it away, you could use that as the cool side for a heat engine and have a perpetual motion machine.
By "beam it away", I assume you mean into space or somewhere else you don't care about heat being. In this case, the cold side of the device isn't the real cold sink, it's just a proxy for the several million cubic lightyears of background-temperature near vacuum which will eventually wind up absorbing the light. Aliens on a distant planet may be able to extract power from the light; bully for them, but I already had to pay extra for shipping, so there's no funny business going on.
What makes you think that violates thermodynamics?
(For what it's worth, "perpetual motion machines" are not outlawed by classical thermodynamics, only those which do additional work. There's nothing in theory which prevents you hooking up a 100% efficient emitter and a 100% efficient receptor up with superconducting wire and letting it run forever, so long as you don't extract any energy from the cycle, for example by observing that it is taking place. There may be something in quantum physics which states that such a system would have to leak information about itself somehow, but I don't know enough to say.)
Are you asking something more specific than that?
I'm thinking of high-frequency transistors / ICs with heatsinks and a bank of these lights being attached.
It wouldn't be a primary cooling method, but if you needed a light on your device anyways then you could make use of this mechanism as well. Imagine backlights for LCDs in mobile devices. An iPad or Android tablet can easily reach 100 F when playing games, and for most cases that's warmer than ambient air temperature.
I don't have access to the original paper (haven't checked arxiv yet). Do the lights need to be at a certain temperature for this to work at all? I know that in similar experiments (condensates) the lattice vibrations only propagate at temperatures much lower than 70 F. At room temperature there is too much chaotic motion in the structure. I'd see that as a major limitation on scale and real-world application.
Where are you getting that?
A Peltier-effect refrigerator can certainly cool things lower than the ambient temperature. It does that by moving the heat from one place to another.
I'm not aware of anything fundamental that says that the heat energy has to be radiated as heat - it could just as easily be radiated as light. So why couldn't this cool things?
(Your argument that it could somehow work if you shone it on something that was cooler than the device implies time travel - the photons would need to now the destination is cooler before they are emitted.
From the description given in the article, it seems that the energy input is simply being used to move energy from one location to another – it just so happens that the energy is sourced as heat, and is deposited as light. This is really not much different from a heat pump, so I see no reason it can't move heat up a gradient.
When used as a heat pump, the device might be useful for solid-state cooling applications or even power generation.
http://phys.org/news/2012-03-efficiency.html
A heated semiconductor light-emitting diode at low forward bias voltage V<kBT/q is shown to use electrical work to pump heat from the lattice to the photon field.
http://www.science20.com/cool-links/optical_heat_pump_100_pe...
When an electrical power input as low as 30 pico-watts is applied, the researchers consistently measured a 69 pico-watt optical power output, seemingly defying the law of conservation of energy. However, upon further investigation, the researchers showed that the system was able to derive extra energy from the thermal energy present in the environment. More specifically, the LED was able to convert the heat generated by the vibrations in the LED’s silicon lattice structure into additional emitted photons, increasing the LED efficiency to over twice unity.
Achieves greater than unity efficiency by simultaneously converting thermal and electrical energy into photons....Cools the surrounding environment as it draws thermal energy from it.