How is light emitted by an incandescent lamp? I am looking for better understanding of how light is produced in an incadescent lamp. More specifically: how is the kinetic energy of electrons converted to light?

*

*Are we dealing with interband transitions or with intraband relaxation involving photons? Is this Bremsstrahlung (electrons lose their energy as light when colliding with crystal impurities/defects)? Or is this a thetmal radiation resulting from Joule heating?

*How is the emission affected by presence of impurities and imperfections of the crystal lattice? Do phonons play a role?

*What properties make a material more suitable for use as a filament: should it be a metal? Should it have a crystalline structure? Will any metal produce light, if a high current is passed through it in vacuum?

Update
The term describing the processes in the incadescent lamp is thermal bremsstrahlung, see the posts on this subject here and here.
 A: The only requirement for radiation to occur, in insulators or conductors, is acceleration of charges or magnetic fields. Bound electrons surrounding a nucleus can be stimulated to radiate by thermal agitation of the nucleus. Rotations,  vibrations etc. All atoms have either dipole or multipole magnetic moments, these will also radiate when agitated by heat. This is thermal spectrum radiation, under certain conditions it can have a "black body spectrum"  Obviously an electron transitioning from one, non radiating stable atomic state, to another non radiating stable atomic state, will also radiate briefly, this form of radiation is the source of line spectra.
A: I would like to address something the other answers do not mention, that is, how the electric current heats up the filament on the molecular level and why it can store this energy and why it keeps glowing after you turn it off (no current).

A proportion of the collisions result in excitation of the metallic electrons to higher energy levels, which may produce light emission upon returning to the lower stable energy level. Continuous collisions between electrons produce a resistance to the flow of the mobile electrons, and atoms of the filament are induced to vibrate by the interaction with the moving electrons. The vibrational energy results in the production of a significant amount of heat, and a characteristic of resistive filament lamps is that only about ten percent of their energy input is turned into light, most of the remainder being emitted as heat (infrared electromagnetic radiation).

https://micro.magnet.fsu.edu/primer/java/lightsources/filament/index.html
As you can see, we are dealing with thermal radiation, and the electrons' (in the current, as they scatter off/collide with the molecules) kinetic energy is transferred to the molecules:

*

*vibrational (this is the most important in your case)


*rotational


*translational energies.
Please note that there are electronic transitions too but I am not mentioning those.

For a molecule with N atoms, the positions of all N nuclei depend on a total of 3N coordinates, so that the molecule has 3N degrees of freedom including translation, rotation and vibration.

https://en.wikipedia.org/wiki/Molecular_vibration
Now when you turn the the lamp off (no current), the molecules are still storing the extra energy that was transferred to them by the current's electrons' (by way of scattering/ collisions), and as the filament tries to reach thermal equilibrium with the environment (cool off), this extra energy causes the  molecules in the filament to keep relaxing by emitting photons (the molecules relax to a lower energy level by emitting photons, including the visible range).
A: Supplementary answer to the OP's clarifying comment:

The main point that still needs clarification, in my opinion, is whether we are really dealing with thermal radiation here, i.e., whether the role of electric current is only to heat the material (since the current flow itself is not a thermal state

Good question, and...
Yes, we are really dealing with thermal radiation here!
The heating produced by the flow of conduction electrons in the bulk of the filament is not related to the thermal radiation coming from the few tens of atoms near the metal surface that are producing the photons that we see.
We know this because we can do some experiments:

*

*When we turn on or off the current, the light produced ramps up or down with a radiative time scale of milliseconds, which is how much time it takes the filament to heat up or cool down from its previous equilibrium "OFF" or "ON" temperature.

*We can try to measure the 100/120 Hz flicker of the bulb, and see that it's perhaps a percent. An incandescent bulb does not turn off 100 or 120 times per second. Its light stays relatively constant. We can make light trails with our eyes or cameras with pulsing light sources like some brands of LED lights (e.g. cheap battery operated portables) or florescent lights or some kinds of mercury or sodium vapor street lights, but we can't reproduce those effects with incandescent lights.

Now, this does not mean that electron collisions in metals can't make visible light, but the chances that a conduction electron can get 2 or 3 eV of kinetic energy before hitting another electron and that that also happens within tens of angstroms of the surface so that the light gets out is extremely small.
Basically the tungsten does two totally separate jobs at the same time:

*

*acts as a suitable temperature-dependent resistor such that it reaches thermal equilibrium and radiates 100 watts or whatever power it's supposed to

*acts as a thermal radiator, producing light when heated


update: @Ruslan's comment links to two excellent videos!

*

*Incandescent light bulb turning on at 1000 fps - High-Speed Entertainment shows some low amplitude pulsations of the brightness of the light. I don't know how much they've slowed it down for display though.


*Incandescent light burning out at 1000 fps - High Speed Entertainment shows a filament burning in air and then breaking, and we can see the decay of the light emitted from each piece as it cools down.

Then it breaks, no current flows, and the light continues but starts to dim:

When it touches another part of the bulb, that part cools more quickly by conduction than by radiation, so it turns dark. But the bit at the right can't cool easily along the filament because it's thermal conductivity is low along the wire, so it's still glowing fairly brightly:

A: https://en.wikipedia.org/wiki/Thermal_radiation
Depending on the theoretical frame used it may as well be called Bremsstrahlung (free(-ish) electrons in the metal scattering into each other)
As far as I can imagine, impurities and crystal lattice defects will affect electrical properties in the first place.
Do phonons play a role? Not sure, I think electrons dominate the heat exchange in metals. One needs heat exchange in order to bring heat to the surface of the filament. You may as well think about phonons scattering electrons (in other words, the crystal lattice and the electron gas exchanging heat).
Should it be metal? Not really. But it should be at least somewhat conductive for the electricity.
Yes, any metal (and any solid substance in general) will produce light when heated in vacuum (or in any transparent media) as long as it stays solid. One needs some 750K in order to produce some faint visible light or ~3000 to look like a normal incandescent bulb.
The best material for a incandescent lamp filament will be:

*

*more or less conductive so it can be heated by electricity

*be absolutely reflective or transparent for non-visible electromagnetic waves and black for the visible spectrum.

*stable against decomposition, melting or evaporation at the desired temperature (equal to the desired color temperature of the lamp, for most practical purposes 3000..6000K)

Since we don't have the ideal material, we use tungsten and we try hard to make it evaporate slower, using inert gases and halogens. On the other hand, gases make bulbs less effective because they carry away some of the heat from the filament. That's why we have better or worse gases for filling the bulbs.
Other materials were used or considered in the past, like tantalum or carbonized natural filaments like cotton and wool.
