A student of mine has drawn my attention to this article. (Full article in Physical Review Letters here, a more accessible summary of it in Physics World here.)  It describes a light emitting diode (LED) that has a greater than 100% efficiency at converting electrical power to light. That is, put 30 picowatts of electrical power in, and we can get 70 picowatts of light out – an efficiency of more than 200%.  For those that don’t know, ‘pico’ means ‘times ten to the power of minus twelve’, or ‘a millionth of a millionth’. A picowatt of power isn’t much.

Uhh? What happened to conservation of energy? How can you get more power out than you put in? No, this doesn’t violate the laws of physics. The point is that 30 picowatts of ELECTRICAL power input (not total power input)  goes in, 70 picowatts of light comes out. There must be another power source as well, to make up the missing 40 picowatts.

It’s heat, from the surroundings. The LED, when operating, is cooling its surroundings. So we have 30 pW of electrical power and 40 pW of heat energy going into the device, and 70 nW of light coming out. It does make sense. It’s like a mini heat-pump. The heat pump in your house takes energy from the air outside and pulls it into the house. To do this requires energy input from the electricity supply – but the total energy (heat) you get inside your house is greater than the energy put in from electricity, so we can say the pump is more than 100% efficient.

Does this mean that we have a wonderful new source of lighting that beats the ‘energy-saving’ fluorescent bulbs hands-down?  Well, if you had enough of them, maybe. However,  to get such an efficiency, the LEDs run at very small currents – if you turn up the current (and make them brighter) the LEDs efficiency drops. So we need a huge number of these things, running at small currents, to make up a significant light source. But, as the Physicsworld article explains, there may be other niche applications, such as communications, where one can needs less bright light sources than for lighting your office.

So what about using these devices to cool? With enough of them, couldn’t we cool the atmosphere and solve global warming? Alas, no. The devices are producing radiation in the form of light, and this can be considered a form of heat. When the light strikes something and is absorbed, it heats up what is doing the absorbing. The obvious example is sunlight heating the earth’s surface. So, for 30 pW extracted from the local region, we get 70 nW elsewhere. It doesn’t help.

Just possibly the device could be used as a fridge for something very, very small, giving localized cooling. But for violating the second law of thermodynamics, no.