incorrect GW gap for LiCl

[zsjan]

Dear all:
I cannot reproduce the gap for LiCl after lots of tests. So I am now seeking for your help. The literature value of LiCl is 5.91 eV for LDA and 8.99 eV for GW. I can get the LDA value but my GW calculations always give me value about 8.76eV, 0.23eV smaller than published value. Here is my input file for GW calculations:

em1d # [R Xd] Dynamical Inverse Dielectric Matrix
gw0 # [R GW] GoWo Quasiparticle energy levels
ppa # [R Xp] Plasmon Pole Approximation
HF_and_locXC # [R XX] Hartree-Fock Self-energy and Vxc
EXXRLvcs= 1471 RL # [XX] Exchange RL components
Chimod= "Hartree" # [X] IP/Hartree/ALDA/LRC/BSfxc
% BndsRnXp
1 | 500 | # [Xp] Polarization function bands
%
NGsBlkXp= 1000 RL # [Xp] Response block size
% LongDrXp
1.000000 | 0.000000 | 0.000000 | # [Xp] [cc] Electric Field
%
PPAPntXp= 27.21138 eV # [Xp] PPA imaginary energy
% GbndRnge
1 | 500 | # [GW] G[W] bands range
%
GDamping= 0.10000 eV # [GW] G[W] damping
dScStep= 0.10000 eV # [GW] Energy step to evalute Z factors
DysSolver= "n" # [GW] Dyson Equation solver (`n`,`s`,`g`)
%QPkrange # [GW] QP generalized Kpoint/Band indices
1| 1| 1|100|
%
%QPerange # [GW] QP generalized Kpoint/Energy indices
1| 8| 0.0|-1.0|
%

Please give me some suggestions! Thank you very much!

[Daniele Varsano]

Dear Shijun,
it is not easy to help you with the few information you provide. Discrepancy with literature or experiments can be due to several reason.
Some suggestion:
1) Check carefully all your convergences: g-vectors, bands, k-points etc. (Usually the number of Gvec in the exchange is much larger that the one needed to converge the screening.)
2) In order to check the convergences parameter, I suggest you to calculate the GW corrections for the top and bottom valence/conduction bands, this it will speed up your calculations a lot, now you are calculating the corrections for 100 bands, and I do not think it is useful.
3) When comparing with literature, try to understand the differences of implementations, I do not know with which results you are comparing, but there could be many source giving a discrepancy of ~0.2 eV, among them: basis set (see e.g. https://www.jim.or.jp/journal/e/pdf3/46/06/1100.pdf) , different plasmon-pole models, pseudopotentials etc. etc.

Best,

Daniele

[zsjan]

Dear Shijun,
it is not easy to help you with the few information you provide. Discrepancy with literature or experiments can be due to several reason.
Some suggestion:
1) Check carefully all your convergences: g-vectors, bands, k-points etc. (Usually the number of Gvec in the exchange is much larger that the one needed to converge the screening.)
2) In order to check the convergences parameter, I suggest you to calculate the GW corrections for the top and bottom valence/conduction bands, this it will speed up your calculations a lot, now you are calculating the corrections for 100 bands, and I do not think it is useful.
3) When comparing with literature, try to understand the differences of implementations, I do not know with which results you are comparing, but there could be many source giving a discrepancy of ~0.2 eV, among them: basis set (see e.g. https://www.jim.or.jp/journal/e/pdf3/46/06/1100.pdf) , different plasmon-pole models, pseudopotentials etc. etc.

Best,

Daniele

The literature I am compared is PHYSICAL REVIEW B 82, 195108 (2010). I have used the same TM potentials, the same cutoff, the same code (quantum espresso) for LDA. However, the results are still cannot converge to their value even I increase bands up to 500. The value of EXXRLvcs and NGsBlkXp I have also tested but still cannot obtain the value in this paper.

[Daniele Varsano]

Dear Shijun,
in the work of Kang and Hybertsen a different plasmon pole model (HL) is used for the frequency dependency of the dielectric matrix *Ref.11 of the paper you mentioned". Yambo uses the Godby Needs model. Depending on the material, differences can be even bigger than 0.2 eV, see e.g.
PHYSICAL REVIEW B 84, 241201(R) (2011)
This could be a reason for the discrepancy. Which model is better to use it is hard to say.
Best,
Daniele

[zsjan]

I see. Thank you!