gw and bse calculations for N-doped semiconductors

[waynebeibei]

Dear All

I want to perform a GW and BSE calculation for a metallic system (actually, an N-doped semiconductor). Pls see the attached files. My questions are two:

1) in o.qp file, DFT results show it indeed is an N-doped semiconductor. For GW results, how can I defined the fermi level? If I just shift the position of CBM to compare with DFT results as the usual semiconductor case, the GW results also show the same conclusion: it is an N-doped semiconductor. Is it correct?
2) after BSE calcultions, the absorption spectrum with e-h effects is strange: it has a lots of negative values? Why?

Best Regards,
Jiaxu Yan

[Daniele Varsano]

Dear Jiaxu,

1) Yes, in GW you calculate corrections to the DFT band structures, so should take the DFT calculation as a referenve (I think...).
2) The negative excitation energies, I think that come from the fact you have fractional occupation numbers. This is a metal, so you have some occupied states above the Fermi energy and holes below. When you build your excitonic Hamiltonian, you are considering both v->c transitions and c (now occupied)-->v transitions (now empty) near the Fermi level. Such negative energies should disappear reducing you electronic temperature (smearing), but I'm not sure you can totally eliminate them, as for 0 smearing your DFT calculation will have difficulties to converge. Moreover consider to include the coupling part in your Bethe Salpeter calculations. Anyway I'm not familiar with calculations in metals, so may be someone in the forum can give you better advises.

Cheers,

Daniele