Hydrid functional

[magorzata_makowska-janusik1]

Hello,
We calculate the electron structure of the CsGeCl3 and CsPbCl3 using PBE and HSE06. To calculate the properties using HSE06 we proceed in 3 different ways:
1. DOS is calculated by PBE + DOS is calculated by HSE06 + BANDS are calculated by HSE06
2. DOS is calculated by HSE06 + BANDS are calculated by HSE06
3. directly BANDS are calculated by HSE06.
In the first approach, after the PBE calculations, we fix the charge and use calculated waves to calculate DOS using HSE06.
My question is – why we obtain the same Eg value using the 1 and 3 approach? Why in approach 2 the E g is much higher than in the 1 and 3 case?
Please could someone answer why is this difference and how should be a methodology to use hybrid functional to calculate the electron properties of crystals?
Regards
Malgorzata

[alexey.tal]

Dear Malgorzata,

Thank you for your question.

Could you please provide the files for these three cases that you are describing so that I can better understand what exactly you are trying to compare?
Please see the forum guidelines.

Best wishes,
Alexey Tal

[magorzata_makowska-janusik1]

Dear Alexey,
Thank you very much for your answer.
I have prepared file when I put all 3 approaches descriptions and INCARs for each one.
Please, say me why the 1 approach give the same results as approach 3. The approach 2 give Eg vale much higher.

How is the method (approach) to calculate Eg by hybrid functional?
Best regards
Malgorzata

[magorzata_makowska-janusik1]

Dear Alexey
I will add that I am not surprised that in approach 2 the Eg is the largest. Here DOS and BAND are calculated with a hybrid functional. I am surprised by the small value of Eg in approach 3. Here BANDs are immediately calculated with hybrid functional. Logically, Egfrom 3 should be the same as Eg calculated in approach 2.
Why is this? What is done wrong and what does Vasp calculate in these different approaches.
Best regards
Malgorzata

[alexey.tal]

Thank you for providing the INCAR files, but it would also be very helpful if you could provide all other input and output files (see guidelines).

As I don't see the OUTCARs I don't know what band gap you find in these different cases. However, one mistake that I see in your INCAR files is that you are trying to restart your hybrid calculation from the charge density (ICHARG=11) and you should see this warning:

Code: Select all

 -----------------------------------------------------------------------------
|                                                                             |
|     EEEEEEE  RRRRRR   RRRRRR   OOOOOOO  RRRRRR      ###     ###     ###     |
|     E        R     R  R     R  O     O  R     R     ###     ###     ###     |
|     E        R     R  R     R  O     O  R     R     ###     ###     ###     |
|     EEEEE    RRRRRR   RRRRRR   O     O  RRRRRR       #       #       #      |
|     E        R   R    R   R    O     O  R   R                               |
|     E        R    R   R    R   O     O  R    R      ###     ###     ###     |
|     EEEEEEE  R     R  R     R  OOOOOOO  R     R     ###     ###     ###     |
|                                                                             |
|     Calculations where the Fock exchange is evaluated cannot be run         |
|     with ICHARG>10. This includes hybrid-functional calculations,           |
|     Hartree-Fock calculations, etc. Mind: To restart such a                 |
|     calculation, you need to provide the WAVECAR file.                      |
|                                                                             |
|       ---->  I REFUSE TO CONTINUE WITH THIS SICK JOB ... BYE!!! <----       |
|                                                                             |
 -----------------------------------------------------------------------------

So I suspect that you are actually comparing PBE and HSE band gaps. If I understand correctly, you are performing a bandstructure calculation. We have a very detailed guide for that.

[magorzata_makowska-janusik1]

Dear Alexey,
I am sending INCAR and OUTCAR files for two different systems calculated by 3 different aproaches.
Calculating CsGeCl3 and CsPbCl3 the Eg was taken from VASPkit. We obtain values

CsGeCl3
1. DOS_PBE+DOS_HSE06+BAND_HSE06 E=3,6 eV 2. DOS_HSE06+BAND_HSE06 E=3,6 eV 3. BAND_HSE06 E=1,65 eV

CsPbCl3
1. DOS_PBE+DOS_HSE06+BAND_HSE06 E= 2,98 eV 2. DOS_HSE06+BAND_HSE06 E=5,09 eV 3. BAND_HSE06 E=2,9 eV

For CsGeCl3 the 1 and 2 approaches give the same value - it is maybe normal, but for CsPCl3 the values are completelly different.
Inthe atachment the files are described as following
1 aproach - files 1.1, 1.2 and 1.3
2 aproach - files 2.2 and 2.3
3 aproach - files 3.3.
Please could you help me to solve my problem
Best regards
Malgorzata

[alexey.tal]

The problem with this calculation is ICHARG=11 which means that the hybrid functional calculation should restart from the charge density. However, the exchange-correlation hybrid functionals depend on orbitals, so the charge density is not enough to restart the calculation. If you want to restart your hybrid functional calculation from a preceding PBE calculation you should set LWAVE=.TRUE. in the PBE step and don't set ICHARG in your hybrid functional calculation.

Also, I see that you are using VASP 6.2.0, which is quite old now. A newer version of VASP would have shown the warning that I posted above and not allowed a user to perform such a calculation.