BN band gap value

[Fabiof]

Dear all,


I want to do a BSE calculation for h-BN. But i can't get the right band gap with PPA.

For what i see in the literature, the band gap is indirect (Gamma-K) and ranges from 6 to 8 eV.

In the report file i have the right information about the DFT energies, direct band gap (K-K) point around 4.6 eV:

Code: Select all

 States summary         : Full        Metallic    Empty
                           0001-0004               0005-1000
  Indirect Gaps      [ev]: 4.592870  7.710691
  Direct Gaps        [ev]:  4.59287  10.41483

*X* K [1] : 0.000000  0.000000  0.000000 ( cc) * Comp.s 16761 * weight    0.0069
              0.000000  0.000000  0.000000 (rlu)
  E -17.52856  -5.13060  -1.35442  -1.35442   4.71876   5.59385   5.86439   6.30399

*X* K [19] :-.436E-16  0.876131  0.000000 ( cc) * Comp.s 16662 * weight   0.01389
              -.287E-16  0.577350  0.000000 (rlu)
  E -13.95238  -7.81559  -7.17581   0.00000   4.59287  12.39409  13.17556  14.49928

But when i do PPA calculations, i get an indirect band gap (Gamma to K point) which is what i want, but with a small value.

Code: Select all

QP [eV] @ K [1] (iku): 0.00      0.00      0.00
   B=3 Eo= -1.35 E= -0.61 E-Eo=  0.74 Re(Z)=0.88 Im(Z)=-.2828E-3 nlXC=-17.65 lXC=-17.36 So= 1.133
   B=4 Eo= -1.35 E= -0.58 E-Eo=  0.78 Re(Z)=0.88 Im(Z)=-.2853E-3 nlXC=-17.56 lXC=-17.36 So= 1.081
   B=5 Eo=  4.72 E=  5.51 E-Eo=  0.79 Re(Z)=0.96 Im(Z)=-.1727E-3 nlXC=-1.171 lXC=-3.652 So=-1.658

 QP [eV] @ K [19] (iku):-.287E-16 0.577      0.00
   B=3 Eo= -7.18 E= -6.16 E-Eo=  1.01 Re(Z)=0.90 Im(Z)=-.6240E-2 nlXC=-13.70 lXC=-14.71 So=0.1153
   B=4 Eo=  0.00 E=  1.62 E-Eo=  1.62 Re(Z)=0.89 Im(Z)=-.3035E-3 nlXC=-14.53 lXC=-16.56 So=-.2205
   B=5 Eo=  4.59 E=  7.24 E-Eo=  2.65 Re(Z)=0.95 Im(Z)=-.5899E-4 nlXC=-2.988 lXC=-8.259 So=-2.489

As you can see the K[19] point includes the the maximum of valence band, and K[1] point includes the minimum of conduction band.
I have a gap around 3.8 eV.


This calculations was done with an DFT ecut of 100 Ry, 1000 BndsRnXp, 1000 GbndRnge, NGsBlkXp= 13 Ry
also i included

Code: Select all

  RandQpts= 100000000          # [RIM] Number of random q-points in the BZ
 | RandGvec= 1            RL    # [RIM] Coulomb interaction RS components
 | CUTGeo= "box z"              # [CUT] Coulomb Cutoff geometry: box/cylinder/sphere X/Y/Z/XY..
 | % CUTBox
 |   0.00000 |  0.00000 | 45.00000 |        # [CUT] [au] Box sides
 | %

My supercell has a vacuum space of 50 bohr.

Without the truncation i get a gap around 3.6 eV.

Do you know what am i doing wrong, considering that the band gap ranges from 6 to 8 eV?

I have attached the yambo input, report file and also a script to plot the interpolated band structure.

[myrta gruning]

Ola Fabio
I looked at the report. I noticed that already the HF indirect band gap you get is 5.17 eV which is way too small.
I ran some old job on 2D hBN and got a HF indirect band gap > 10 eV using 12x12 grid and 80 Ry not touching yambo default.
However I could not spot anything in the inputs that can justify the difference, but I think that you should try to get first a reasonable HF band gap.
Since one can run HF in few secs on a PC, you should be able to run several tests in relative short time. Furthermore HF does not have so many parameters to adjust as in the PPA/GW part.

[Fabiof]

Ola Myrta,

Thank you for your answer.

The only thing i can change with a yambo -x (HF) calculations is the flags EXXRLvcs and the number of k_points, right?
By changing these flags i still get a small HF indirect gap.

I also tried to do calculations with different pseudopotentials and my HF indirect gap is always around 5 eV.
But this gap is not result from the Gamma to K transition. The HF indirect gap from Gamma to K is > 10 eV.

Fabio

[myrta gruning]

Ola

The other thing may be something in the DFT input, even though from the yambo report everything seems OK on that side.
I re-run my old calculations with the most recent gpl version of the code just to be sure, but I just get the same results for the HF as before.
If you wish, you can post a small version of the runs, I can run it and see if I can spot what is wrong.

[Fabiof]

Ola Myrta,

I attached the DFT input and output files with the pseudopotentials.
Maybe you can check the output files to see if something is wrong.

Thank you,

Fabio

[myrta gruning]

I modified the pwscf input (see attachment) and it works (reasonable HF bandgap).
I would have to investigate the reason for this difference as apparently results should be the same.
In the meantime I hope this helps you to advance with your calculations.
Best,
Myrta

[Fabiof]

Now I'm getting a reasonable band gap with PPA calculation.
Thanks for the help.

Fabio