Mobile Charge effects on Band Diagrams Hi I have a question which I think is related more to definitions than anything else. 
Consider a band diagram.  PN junction, heterostructure, whatever it might be. It's a very useful tool to determine where there is field, where there is space charge by slopes, band bending etc.  
However, as I recall, the strictest definition is that this is the potential an electron sees due to the hybridization of the atomic lattice states.  Any charges, i.e. defect/traps, surfaces, polarization charges, etc. are all fixed.  Does this mean that mobile charge cannot contribute to the band diagram in the strictest definition?
For instance, the Kirk effect in HBT/BJTs essentially "undoes" some of the reverse bias band bending when a large electron current is passing through such that there is a condition J less than qns for the band diagram to be plausible.  OR what about strongly correlated systems, like perovskites, where alot of the physics is due to electron electron interactions, which is NOT including in BDs? Alot of work goes into Mott-Hubbard transistions, which if I recall correctly cannot be inferred from BDs.  
So are band diagrams technically only the potential seen in a fictional "1 electron" system, which is a valid approximation assuming an electron gas?  Or could one in principle, create BD's which do involve mobile carrier potential (Assuming one could solve those difficult Hamiltonians!)
Thanks! I'd love to hear some opinions!
 A: Atomic orbitals are a solution for an electron configuration around an isolated atom.  Outer orbitals turn into bands when many atoms are close together in a solid (or liquid, for that matter).
An electron in a metal is in a part-filled band, and while it is free to wander inside the conductor, it is not 'free'; if it were, cathodes in vacuum tubes
wouldn't require heaters.
As to defects and mismatched atoms, the knowledge we have of bands is from calculations on repetitive structures (like a crystal lattice); defects and impurities
are local, and bands are global (apply to a large lattice) not
local.   The local electronic state near an impurity isn't part of the band picture.   For working semiconductors, the impurities are very few and
far between, so they don't change the bands much...
In perovskite superconductors, the actual electron collective couplings
are weak (interesting, but modest temperature is all it takes to scramble
their effects).  Similarly, ferromagnetism is a collective electron
effect that is too subtle to show up in a band diagram.
