hBN_scf.in effection

[xjxiao]

Dear all,
I noticed that after performing a self-consistent calculation using Quantum ESPRESSO with the `hBN_scf.in

hBN_scf.in.txt

` file provided on the Yambo website, I was unable to read the Fermi energy from the resulting `hBN_scf.out

hBN_scf.out.txt

` file. The output only showed something like this:
```
highest occupied, lowest unoccupied level (ev): 5.0633 9.3098
```
However, after adding the parameters `occupations = 'smearing', smearing='gaussian', degauss=0.02`, I was able to obtain the following in `hBN_scf_smear.out

hBN_scf_smear.out.txt

`:
```
the Fermi energy is 6.6238 ev
```
Why is this the case? And will adding these parameters affect subsequent Yambo calculations?

Yours,
Xiao

[Daniele Varsano]

Dear Xiao,

Why is this the case?

This is reported in the QE FAQ: https://www.quantum-espresso.org/faq/fa ... tency/#6.7

As you are dealing with a semiconductor, you can set the Fermi energy at the top of the valence band.

And will adding these parameters affect subsequent Yambo calculations?

Absolutely not, Yambo in any case rescale the energy setting the zero at the top of the valence band.

Best,

Daniele

[xjxiao]

Dear Daniele,

Absolutely not, Yambo in any case rescale the energy setting the zero at the top of the valence band.

In my system, under PBE calculations, it is an indirect bandgap semimetal, meaning there is a small overlap between the valence band and the conduction band. I found that the bottom of the conduction band is set to E0=0.
# K-point Band Eo [eV] E-Eo [eV] Sc|Eo [eV]
1 60 0.053582 -0.687920 -1.259927
4 61 0.000000 -0.480911 -1.137575
Is this problematic?

Additionally, under GW correction, the bandgap opens, with the valence band maximum located at a high-symmetry point, but the conduction band minimum lies between two high-symmetry points, making it difficult for me to determine its exact location. However, to measure the convergence of input parameters, I need to determine the indirect bandgap value. How should I determine this? Is the indirect bandgap value related to the density of the k-point grid at this point?

Furthermore, for the parameters BndsRnXp and GbndRnge, how many times the number of empty bands is generally required compared to the number of valence bands? I have seen in the literature that it is more than 6 times, but it seems that my system converges at just 1 times. Is this reasonable?

Thanks!
Yours,
Xiao

[Daniele Varsano]

Dear Xiao,

In my system, under PBE calculations, it is an indirect bandgap semimetal,

I do not exactly what you are calculating, but I suggest you to check carefully your DFT ground state. hBN is a semiconductor and also, this statement is not compatible with you found previously:

Code: Select all

highest occupied, lowest unoccupied level (ev): 5.0633 9.3098

Furthermore, for the parameters BndsRnXp and GbndRnge, how many times the number of empty bands is generally required compared to the number of valence bands? I have seen in the literature that it is more than 6 times, but it seems that my system converges at just 1 times. Is this reasonable?

This is system dependent, anyway a number of conduction band equal to the number of valence band is quite unusual. Please note that this parameter is dependent to the size of the dielectric matrix (NGsBlkXp), the two parameters should be both converged. Please check if you have set NGsBlkXp to an uncoverged low value.

Best,
Daniele

[xjxiao]

Dear Daniele,

I do not exactly what you are calculating, but I suggest you to check carefully your DFT ground state. hBN is a semiconductor and also, this statement is not compatible with you found previously:

Code: Select all

highest occupied, lowest unoccupied level (ev): 5.0633 9.3098

Thank you for your response. The material I mentioned in the subsequent question is not hBN. I brought it up here only because I noticed that the calculation results differ from what you mentioned, "Yambo in any case rescales the energy setting the zero at the top of the valence band." In this semimetal, the bottom of the conduction band is set to E=0 instead of the top of the valence band.
I apologize for any confusion caused. I would still appreciate your help in answering this question.

In my system, under PBE calculations, it is an indirect bandgap semimetal, meaning there is a small overlap between the valence band and the conduction band. I found that the bottom of the conduction band is set to E0=0.
# K-point Band Eo [eV] E-Eo [eV] Sc|Eo [eV]
1 60 0.053582 -0.687920 -1.259927
4 61 0.000000 -0.480911 -1.137575
Is this problematic?

Additionally, under GW correction, the bandgap opens, with the valence band maximum located at a high-symmetry point, but the conduction band minimum lies between two high-symmetry points, making it difficult for me to determine its exact location. However, to measure the convergence of input parameters, I need to determine the indirect bandgap value. How should I determine this? Is the indirect bandgap value related to the density of the k-point grid at this point?

Thank you again.

Yours,
Xiao

[Daniele Varsano]

Dear Xiao,

In this semimetal, the bottom of the conduction band is set to E=0 instead of the top of the valence band.
I apologize for any confusion caused. I would still appreciate your help in answering this question.

So it is correct, as the bottom of the conduction band is at lower energy than the top of the valence band.

Best,

Daniele