On a somewhat divergent topic but hopefully relevant enough to fit here, I'm wondering if anyone has a good accurate qualitative description of the photonic light emitted by common LEDs anyway.
Specifically I refer to the cases without use of phosphors though comments about that interaction may be interesting also.
Obviously from the data sheets you end up with a sort of gaussianish spectral distribution that is often some dozens-ish of nanometers wide say FWHM.
And at ~2eV / photon for a milliwatt of optical emission you end up with around 10**15 ish photons per second if my quick calculation is right.
So actually inside common sorts of LEDs what is happening on time scales of N * 10**-15 seconds or so as these photons are emitted?
Given a very rough quantum efficiency of ~50% that'd be around 2 electrons dropping over a 2V bandgap to create one photon and I guess some phonons or whatever due to the other one getting lost as heat.
So are multiple spatially distinct regions over the active area of the LED emitting totally independently of each other over even short time scales?
Does light "bounce around" to any significant extent inside the LED and form a "cavity field" so that one photon emission influences others to any appreciable degree? Clearly the answer is "not much" since it isn't a LASER with lots of cavity passes through a resonator but, still, I wonder to what extent any correlation exists spatio-temporally over the die on at least short (few photons) time scales?
So at such "low" photon emission rates of maybe 10**-15/s is each emission basically temporally and spatially independent?
How is the breadth of the spectrum produced? Do you tend to have physically different fixed spatial regions over the die where there's a spread of bandgap so that particular points "like" to produce certain wavelenghts while others produce others in a continuum?
Or due to the overall emission process and geometry and so on do you just get a bunch of similar emissions of "wave packets" that are all produced from "characteristically similar" junction areas over the die and the envelope of these "wave packets" just intrinsically has the xx nm wide "gaussian" bandwidth of the macroscopic spectral distribution that is observed? So the breadth of the spectrum would be actually truly reflected even looking only at a small number of 'photon wave packets' emitted over any isolated region of the die?
Does anything significantly change about the optoelectronic character of the emissions even over the span of very low currents (e.g. femtoamperes) all the way up through XX mA nominal currents? Of course the lattice warps some with increasing temperatures which shifts the color a little but otherwise is that about it as far as OE behavior difference from single electrons to full intensity or even in high current pulses?
Does the color spectrum produced by a single specific LED "age" much for whatever reason of slow diffusion or migration or whatever when run at full intensity over X months / years?
What causes the "rise time" and "fall time" of optical radiation in a non-phosphor based LED anyway? Basically just die resistance, die capacitance, package inductance and nothing really else?
For that matter what DOES a common 5mm or 3mm packaged LED or SMD PLCC / 0805 / whatever LED look like on a VNA in conditions of various forward or reverse bias? Has anyone made any decent SPICE models of ordinary LEDs and not just ones intended for high performance telecom / instrumentation?
Besides "exotic" purpose-made devices (e.g. laser diodes, superluminescent diodes, ...) are there no LEDs that can produce a significantly narrower spectral line by internal semiconductor processes and without much external influence (not talking about phosphor etc. nor giant external optical filters)?