Besides of photon, what else can excite electron from valence band to conduction band, then to form exciton? The title is my question. Besides of photon, what else can excite electron from valence band to conduction band, then to form exciton? thank you.
 A: In principle, any of the processes that lead to exciton relaxation (i.e., finite time of exciton and electron-hole pairs) can also happen in reverse direction and create an exciton. However, most of these processes are too difficult to use in practice.

*

*One could cite here Auger scattering, that is collision with other electrons, notably an electron already excited to a higher energy band, in which case it could transfer to a valence electron an energy close to that of the gap, i.e., sufficient for creating an exciton.

*Another option is a multiphonon process, where several (likely optical) phonons may have sufficient energy for creating an electron-hole pair.

Another possibility is using a semiconductor device, in a geometry similar to a semiconductor laser/photodiode - in this case electrons and hole are injected in the active region from adjacent regions enriched in only one type of carriers (p-/n-regions). The difference is that in lasers and photodiodes the electron-hole pairs are strongly coupled to the electromagnetic field - we do not want them to become excitons, since this would broaden the line, i.e., reduce the quality of the laser. However, one can envision opposite conditions, where the radiative transitions are suppressed, and the gas of electron-hole gradually cools down to form excitons.
The literature on the Bose-Einstein condensation of excitons (a popular subject about a decade and a half ago) might contain many ideas along these lines.
A: A very simple answer might be that any particle which carries energy and has a nonzero cross-section for electrons (which I believe is theoretically true for any particle) could possibly collide/interact with an electron and transfer the energy required to raise the orbital level.   Photons turn out to be the easiest to generate and control.
I see a few comments related to applying RF fields, which after all are just a flux of very long-wavelength photons.  While NMR, "nuclear magnetic resonance," scanners function by finding resonant frequencies within atoms [note: I still hate the anti-panic renaming as "Magnetic Resonance Imaging"] ,  I'm not sure that attempting resonant absorption in electrons is possible. Certainly it'd take a very large-multi-photon absorption event to bump up the orbital.
