the binding energy is negative for the excitation

[tomtom007]

Dear all

I want to use ‘ypp’ to analyze the optical transitions after a BSE calculation. Firstly, I got the files named o.exc_E_sorted using ‘ypp –e s’. In o.exc_E_sorted, I can find such infos:
# E [ev] Strength Index
3.503 0.3076E-6 1.0000
3.50 0.292E-10 2.00
3.503 0.1735E-8 3.000
3.50910 0.01435 4.00000
3.510 0.3280E-6 5.000
3.510 0.5677E-7 6.000
3.512 0.2804E-6 7.000 …….
Then, I use ‘ypp –e a’ to see the E-H pairs that contributes to Excitonic State 4. In o.exc_weights_at_4 file, I got such data:
# Electron-Hole pairs that contributes to Excitonic State 4 more than 5%
#
# K-point [iku] Weight
# 0.000000 0.333333 0.000000 0.525244
# 0.000000 0.333333 0.200000 0.999929
# 0.000000 -0.333333 0.200000 1.000000
#
# Band_V Band_C K ibz Symm. Weight Energy
#
24.00000 25.00000 10.00000 2.00000 0.50260 2.16800
24.00000 25.00000 5.00000 2.00000 0.50242 2.16800
24.00000 25.00000 10.00000 5.00000 0.50239 2.16800
24.00000 25.00000 5.00000 5.00000 0.50231 2.16800
24.00000 25.00000 10.00000 6.00000 0.50226 2.16800
24.00000 25.00000 5.00000 3.00000 0.50216 2.16800
….. …..

My question is why the energy gap between Band_V24 to Band_V25 (2.168 eV) is smaller than the excitation energy (3.5091 eV)? If I am not wrong, the energy of bands transition minus the excitation energy is the binding energy of the exciton, so the binding energy is negative for this excitation?



Best Regards
Jiaxu Yan
Division of Physics and Applied Physics
Nanyang Technological University
Singapore

[Daniele Varsano]

Dear Jiaxu,
please note that the "gap" reported in o.exc_weghts is the Kohn-Sham gap, and the binding energy is the Quasi-particle gap minus the excitation energy. In your BSE calculations, I presume you used QP energies (by a previous GW calculation or modeling it via a scissor operator), so if you want to calculate the binding energy you should consider the QP gap previously calculated and not the KS gap.

Hope it helps,

Daniele