A light-emitting diode (LED) that emits more energy than it consumes has been published by researchers from MIT.

That violates the second law of thermodynamics, right? Isn't that specifically listed as a no-no in the Science 2.0 FAQ? Not the violating, you are welcome to go ahead and do that, but claiming to do so in an article, that is.

Lead researcher Parthiban Santhanam told Institute of Physics writer Tim Wogan it isn't.  "The most counterintuitive aspect of this result is that we don't typically think of light as being a form of heat. Usually we ignore the entropy and think of light as work. If the photons didn't have entropy (i.e. if they were a form of work, rather than heat), this would break the second law. Instead, the entropy shows up in the outgoing photons, so the second law is satisfied."

In this instance, they were able to emit more light energy than they consumed in electrical energy. The idea of this kind of optical 'heat pump' - as they term it - that converts lattice vibrations into infrared photons which cool the area was first predicted in 1957, and who wouldn't want LED efficiency of 200%? Why wasn't this done already? The band gap of the semiconductor, the energy that will make an electron–hole pair, is the rub here (BYU has a good utility for comparing different kinds of semiconductors), just like it is in solar panels and those got $44 billion in subsidies over a two-year period for nothing, so we can forgive science for taking its time in this case. Individual photons can have energies different from the band gap, that is sort of the point - but while some pass through unnoticed most electron–hole recombinations actually result in heat waste, absorbed by the semiconductor in the form of quantized lattice vibrations called phonons. Jan Tauc at the Institute of Technical Physics in Prague postulated in the 1950s that since this idea provided a way for radiation to remove heat from a semiconductor lattice, there was no barrier in principle to an LED being more than 100% efficient. Technology and materials just had to catch up. Easier said than done, and no one had done it for 55 years, but now these MIT folks say they have. 

Can we bring on perpetual motion? What's the catch? Efficiency is a colloquial concept to people and of its own does not mean much but it goes without saying efficiency alone can not solve problems.  This is really low power so you aren't going to see a 100% efficient refrigerator. But you might see a faint super-efficient LED used in small medical applications.

Citation: Parthiban Santhanam, Dodd Joseph Gray, Jr., and Rajeev J. Ram, 'Thermoelectrically Pumped Light-Emitting Diodes Operating above Unity Efficiency', Phys. Rev. Lett. 108, 097403 (2012) 10.1103/PhysRevLett.108.097403