Geometry relaxation with external electric field

[anton]

Dear all,
I want to relax the geometry (atoms and, eventually, also cell vectos) of a crystalline periodic system in the presence of an external electric field. It looks like that the EFIELD_PEAD tag does not affect the geometry at all. Digging into the manual and the web, it looks like that the appropriate tags are EFIELD, LDIPOL, IDIPOL and DIPOL, that must be combined to achieve the convergence, but in the manual it's written that they are for slab calculations; moreover, the question was, in some way, already posed in this forum but no clear answer was given so far. If for the EFIELD_PEAD tag there is a reference, unfortunately, there is no reference on how the EFIELD, LDIPOL, IDIPOL and DIPOL are implemented in VASP.

Please, could you tell me:
1) which tags I must use to perform geometric relaxation of 3D crystalline periodic systems in the presence of an external static electric field (please, provide explicit examples of typical values for the tags),
2) which are the published manuscripts where it is explicitly said how the EFIELD, LDIPOL, IDIPOL and DIPOL tags are implemented in VASP.

Thanks a lot in advance for your kind help.

Antonio Cammarata
Czech Technical University in Prague

[admin]

The implementation of EFIELD_PEAD in VASP evaluates the influence of the external electric field
on the electronic structure only. The contribution to forces acting on atoms is not impelemented.
Therefore, the study of the influence of the external field on geometries is not possible.

[anton]

I understand that the setting the EFIELD_PEAD tag does not affect the geometry due to the implementation.

I set distinct values for the EFIELD, LDIPOL, IDIPOL and DIPOL tags and I obtain differenct geometries and the electronic distribution. Please, let me know which are the published papers where the implementation of the EFIELD, LDIPOL, IDIPOL and DIPOL tags is documented and tested.

Thanks a lot in advance for your kind reply.

All the best

Antonio Cammarata
Czech Technical University in Prague

[admin]

The VASP implementation closely follows the "PEAD" approach of Nunet and Gonze
(Phys. Rev. B 63 (2001) 155107).
The example of the calculation on NaF is given in manual:
https://cms.mpi.univie.ac.at/vasp/vasp/ ... .html#6173