Dave,
This is not exactly a Maxwell’s demon, since there is no discretionary filtering. I like your suggestion about thermal peaks being involved and there is undoubtedly some kind of resonance. The LED was notably in the IR range to begin with. Possibly the Boltzmann tail of the thermal distribution was trimmed and captured. I have been unable to find important update details on this LED. Here is the website of the Russian Institute which made the LED: http://www.ibsg-st-petersburg.com/index.html It is only ~70 picowatts light equivalent. Perhaps no more was said about it – due to either finding an error in measurement– or the project becoming “black” due to military significance. But otherwise it would be crazy not to pursue this with massive effort. It is worth noting that LEDs are simple diodes and presumably quite a few can be put on a chip. A picowatt is one trillionth of a watt, so if one could etch one trillion diodes on a semiconductor, then it would be rather impressive – with about 40 watts of net gain which is in the form of cooling. How many discrete devices can be put on a chip these days? My guess is that it is in excess of a trillion. From: David Roberson This discussion is interesting. Perhaps the existing thermodynamic laws apply mainly to black body types of interactions when radiation is associated. Clearly the light emitted by an LED is not of that nature. It is narrow band radiation at a level that is much higher in these bands than would be expected according to the temperature of the device. Also, the DC input power contributes a significant portion of the net radiation output in a direct conversion process. This behavior is very unlike most of the systems used to derive the thermodynamic laws. Perhaps there really does exist at least this one loophole that can be breached. A clear understanding of exactly how the random thermal motion within the LED can be converted into light at this level of efficiency would be desirable. Could it be that the random peaks in thermal energy that follow a Gaussian distribution are the key? Near the thermal peak one might find that a little help from the DC source is sufficient to cause electrons to jump into higher orbitals. If enough of these occur in a short period of time a population inversion may come into existance which would then drain the excess energy by positive feedback and subsequent radiation pulses. The excess energy would have to come from that random thermal motion that was tapped leading to cooling of the device. Is this an example of an atomic Maxwell's demon? There may be some interesting concepts hidden within this effect. Dave -----Original Message----- http://www.wired.co.uk/news/archive/2012-03/09/230-percent-efficient-leds Notice that this LED has a COP of 2.3… or 230 percent overunity. That implies “perpetual motion”. “However, while MIT's diode puts out more than twice as much energy in photons as it's fed in electrons, it doesn't violate the conservation of energy because it appears to draw in heat energy from its surroundings instead.” When it gets more than 100 percent efficient, it begins to cool down, stealing energy from ambient, which is exactly what must happen in any OU device, unless there is nuclear reaction pathway or another “supra-chemical” way to convert mass into energy. BTW - If photocells could be obtained which are ~70% efficient, then in principle, yjey could be mated to the LED for the proverbial “eternal light” … but the output is so low you would need a few million of them to be useful… but you get free air conditioning as a fringe benefit J
