I've been reading this question and the answers it received, and I still have some unresolved questions.
It is common to use the semiclassical approximation in describing charge carrier transport in semiconductors (for example, see Ashcroft & Mermin chap. 29). Wavepackets are combinations of momentum eigenstates that give some localization to a particle (electron or hole) within the limits of the HUP. This is helpful in view of the importance of position and distance in semiconductor device design (e.g., depletion regions, base layer widths in BJTs, FET channels, etc.). It is also noteworthy that many simplified diagrams of p-n junctions have an x-axis.
Basically, my question is this: When we discuss occupied states in a semiclassical context, does this only refer to energy/momentum eigenstates (delocalized, without reference to position), or does it mean instead that the energy states in a particular x,y,z region are occupied?
In the case of depletion regions, no (or very few) electrons are available in the conduction band, and no (or very few) holes are available in the valence band because of recombination. To me that seems like states that are occupied (or unoccupied) locally.