Keywords for LEXCH tag in POTCAR when use different functional

[jingyun_ye4]

Hi

I have a question regarding LEXCH tag of POTCAR. For example, if I want to compare functional (e.g., PBE, PBEsol, RPBE, revPBE, B3LYP) for the same system. I will use POT_GGA_PAW_PBE POTCAR for all the elements. My question is should I change the LEXCH tag in POTCAR to the name of the functional.

Such as:
LEXCH = PE (for PBE)
LEXCH = PS (for PBEsol)
LEXCH = RP (for RPBE)
.....

I read several posts on VASP forum and research gate. I could not found a solid answer to this question. If anyone knows, I think this will help many users.

The second question: If we need to change LEXCH tag, why? What's the impact when I specify different keyword for LEXCH tag? Since all the parameters in POTCAR are the same, but LEXCH tag. I don't what will be changed in the calculations.

We found if I use LEXCH = PE for RPBE functional, the calculations also complete.

Thanks!

on whether should modify the LEXCH tag in POTCAR. I tried with the example provided by vasp website, https://cms.mpi.univie.ac.at/wiki/index ... O_LSDA%2BU. At first I tried with PBE, and then with GGA=B3 without any change in POTCAR. The energy from PBE is about -10.80eV, and from B3 is about -11.79ev. As we can see that the energy are very similar only except that there will be a warning when running GGA=B3 without change LEXCH to B3. Next, I changed LEXCH=B3, but the energy is totally different, it is -23.86eV. So I am wondering which is correct for using B3 functional?

[henrique_miranda]

You should not change the LEXCH tag in the POTCAR file.
The POTCAR file is intended to be used as read-only by the user.
The tag LEXCH indicates what exchange-correlation functional was used to generate the POTCAR.
We distribute POTCAR's for PBE and LDA.

Using the PAW POTCARs provided with VASP, you can change the exchange correlation functional used by simply changing the GGA INCAR tag.

[jingyun_ye4]

Thank you for your response.

If this comment "The tag LEXCH indicates what exchange-correlation functional was used to generate the POTCAR." is true, why the POTCAR in OUTCAR are the same when I specify GGA=PE, PBEsol, RPBE, RevPBE. I think this means that no matter what I specified in GGA, the calculation using PBE POTCAR.

There are only three type of POTCAR provided PW91 (LEXCH=91), LDA (LEXCH=CA) and PBE (LEXCH=PE).

[henrique_miranda]

Yes, no matter what you set in the GGA INCAR tag you will use the POTCAR you specified.
Note that VASP will re-calculate the energies of the reference atom and other PAW quantities with the exchange-correlation functional you specify in the INCAR with the GGA tag if it is different from the one that was used to generate the POTCAR.
This is what allows one to use the PBE POTCAR even if afterwards you change the exchange-correlation functional.
When you modify the LEXCH tag in the POTCAR you trick VASP into thinking that the POTCAR was generated with a different exchange-correlation functional and the energies of the reference atoms are not correctly updated.

[jingyun_ye4]

Thank you for your response.

I found on vaspwiki website wiki/index.php/Specific_hybrid_functionals. For PBE0, HSE functional, vaspwiki suggest to use PBE POTCAR. However, for B3LYP functional, no suggestion on POTCAR.

We would like to use B3LYP functional, so I found two discussion on Researchgate and I included below for other researchers who face the similar problem. Some people suggest to replace LEXCH tag with B3. If this is wrong, it will be helpful to update the information on vaspwiki (wiki/index.php/Specific_hybrid_functionals) to let the user know which POTCAR is suggested to use.

Or add some information (wiki/index.php/Available_PAW_potentials) to discuss the POTCAR usage when different functional specified.

https://www.researchgate.net/post/How-c ... lculations
https://www.researchgate.net/post/How_t ... al_in_VASP

Thank you for your kind help!

[henrique_miranda]

Thank you for bringing this to our attention.

We modified a few wiki pages to clarify this issue for other users:
wiki/index.php/GGA
wiki/index.php/POTCAR
wiki/index.php/Specific_hybrid_functionals

As for the questions/answers on researcher gate, I see that in the first one that you have posted a colleague of ours already warned that one should not change LEXCH in the POTCAR file.
I also answered the second question pointing out the same.
I did not find a way to downvote the original answer and we have no way to remove it.