I've always been intrigued that electrical energy can just disappear out of a circuit 'into the aether' as RF.

Of course thinking about it, some energy 'vanishes' as EM radiation even in a filament lamp or infra-red bar heater but not that much.

Recently, with LED efficiencies in the order of ~30%, it's interesting to think that for every 10W or so dissipated in LEDs, only say 7W or so needs to be considered as heat !

Question: I'm familiar with the photoelectric effect in solar panels etc but is the light emission mechanism in LEDs related to this or somewhat different ?

  • 1
    \$\begingroup\$ The emission of light from an LED is indeed related to the absorption of light in a solar panel; it's the same process happening in reverse. Also, 30% efficiency is really quite low for an LED! \$\endgroup\$ – Hearth Oct 14 '20 at 1:56
  • \$\begingroup\$ 30% seems to be fairly typical of LEDS manufactured for lighting currently. Optimised solar cells can apparently approach 30% conversion efficiency (without going multi-junction) too. I was actually looking at some data only today and the fluorescent tube isn't that far behind ! Fluorescents age worse of course. \$\endgroup\$ – Graham Stevenson Oct 14 '20 at 2:26
  • 1
    \$\begingroup\$ High end blue LEDs are over 70% efficient (watts in per watts out). White lighting is somewhat less efficient because the phosphor has to waste some energy as heat to turn a blue photon into a green one, and even more to turn a blue photon into a red one. \$\endgroup\$ – user1850479 Oct 14 '20 at 2:30
  • \$\begingroup\$ The fluorescents I use are rated for 50k triphosphor 5000’’K true daylight with better spectral balance than LEDs and 88 lpw. Although if you start them twice day only rated for 30kh yet longer if left on all the time. Mine use a quad ballast single to each end of a 4ftx 1” (T8) each independent of the other, so any number of tubes may be used or switched off with a HV remote switch \$\endgroup\$ – Tony Stewart EE75 Oct 14 '20 at 2:31
  • \$\begingroup\$ 88 Lumens per Watt is certainly good. Looking today at some relatively inexpensive 3030 Cree LEDs, their efficiency was ~115 Lumens per Watt. \$\endgroup\$ – Graham Stevenson Oct 14 '20 at 2:34

As others have mentioned, photoelectric effect (solar cells) and photoemission (LEDs) are pretty much inverse processes.

My high-school-physics mental model is like this: Electrons orbit a nucleus in shells. The physical size of those shells dictate how much energy the electrons have (and somewhat depend on the type of atom). Thus, to move from one shell to another the electron needs to gain or lose a specific amount of energy. A "free" electron can be thought of as being in the "next", unfilled shell of the atom.

The actual lingo, then, moves to energy "bands". The outermost electrons, in the valence shell, have an energy corresponding to the "valence band" and free electrons are in the unfilled shell corresponding to the "conduction band". The amount of energy between the two is the "band gap".

So, in the photoelectric effect we have a photon whose energy corresponds to the band gap (or maybe more) coming in and transferring its energy to an electron in the valence band. That gives the electron enough energy to jump to the conduction band, and if we put an electric field around the thing that field will carry away the electron, giving us current.

In photoemission we have some extra electrons provided by an external current. When they come across an atom with a vacancy (because we have p-doped material...) they are captured by the atom, moving from the conduction band to the valence band, losing an amount of energy corresponding to the band gap in the process. That lost energy comes out as a photon*, and the frequency (i.e. color) is inversely proportional to the amount of energy, which is fixed.

As to questions of efficiency: A big factor that is missing here is that with an LED all of the energy you put in is being converted to a single wavelength, and it can be very efficient. With a solar cell a single wavelength can be converted with good efficiency but the light you're putting in is a very broad spectrum of wavelengths, and not all of these have enough energy to move an electron across the band gap. Of interest may be multi-junction solar cells, which combine multiple materials to have band gaps capturing different parts of sunlight's spectrum.

*The lost energy can actually be split between the photon and a phonon, which is some vibration of the crystal latice. This has something to do with quantum spin and is way, way above my head, but the important take-away is that practical LEDs are made with "direct band gap" materials where all of the energy goes into the photon and there is none "lost" as a phonon. (Lost as in doesn't produce light.)


A quick google search suggests it's the electrons jumping into holes that give off photons. Looks like there's plenty of info out there on it.

At a shallow level, it maybe the exact opposite of the solar cell effect. In fact, LEDs also work as solar cells.

  • 1
    \$\begingroup\$ Presumably there's some optimisation for purpose. I can't imagine much light being emitted by a solar panel if I pumped power into it. \$\endgroup\$ – Graham Stevenson Oct 14 '20 at 2:21
  • \$\begingroup\$ @GrahamStevenson Correct - solar cells are not optimized for emission. But a quick google suggests they do work that way. \$\endgroup\$ – Jesse Gordon Oct 14 '20 at 2:34
  • \$\begingroup\$ I recall discussing the photoelectric effect 'in class' some 49 years ago. Photons of sufficient energy liberate electrons. Planck's constant came into it regadring the energy required. Yes, it sounds like a similar effect in reverse AIUI. \$\endgroup\$ – Graham Stevenson Oct 14 '20 at 2:41

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.