Dear exciting community and developers,
when calculating absorption spectra and the energy loss function with the Bethe-Salpeter approach for finite momentum transfer (q-dependent BSE calculations) I noticed strange, distinct peaks and I I’ve been struggling to see where these are coming from.
The peaks have the following properties:
- the peaks appear ‘at the end of the spectrum’, so e.g. when including states such that there is absorption up to 30 eV the peak will appear slightly below 30eV
- the appearance and position is momentum dependent; for the momentum being equal to one or two reciprocal lattice vectors the peaks disappear
- in a system with semi-core states, the peaks disappear as soon as the semi-core states are included in the BSE hamiltonian (e.g. GaAs: peak is there for 4 valence bands and 4 conduction bands at some momentum k but disappears for the same calculation with Ga d-states or Ga and As d-states additionally included in the BSE)
- peak position doesn’t change with the number of conduction bands but peak height decreases with increasing number of conduction bands (testes for Silicon with 4 valence bands and varying the conduction bands between 8 and 30)
Is this perhaps a know effect when a BSE spectrum is calculated for a larger energy range than covered by the included conduction bands? Or has anyone seen the same thing and has some explanation?
Unfortunately I cannot upload the figures that would show what I described above as the upload functionality doesn’t work for new users…
I would appreciate any comments on this - thank you.