Question about electric field without dipole correction.

[yujia_teng]

My system has dipole, so usually we need to add dipole correction. If we don't use dipole correction, there will be electric field in the vacuum and the potential profile has slope in vacuum. Now I want to see if I apply the external electric field with EFIELD tag, can I cancel the field in vacuum when there's no dipole correction, or make the potential flat in vacuum region. Unfortunately, from my tests, it doesn't work. So I'm wondering is it possible to do this in VASP?

In my tests, I don't see the effect of electric field in the potential profile and total energy if LDIPOL=.FALSE. The tag EFIELD is even not in OUTCAR. The slope remains same as the original case: no E-field and no dipole correction. It only shows up when LDIPOL=.True. So looks like E-field must be used together with dipole correction on (EFIELD=xx with LDIPOL=.True.)?

Some input/output files and figures are attached.

[jonathan_lahnsteiner2]

Dear Yujia Teng,

I was checking your input files. If you want to apply an external electric field and not compute dipole corrections you have to set LMONO=True.
See here the wiki page for further information:
https://vasp.at/wiki/LMONO
If you do not set LMONO the tag EFIELD from your INCAR file will be ignored. You can verify this in the OUTCAR files you supplied by the missing EFIELD tag in the INCAR file tag list at the beginning of the OUTCAR file.

All the Best Jonathan

[yujia_teng]

Dear Jonathan,
I added the LMONO tag, but looks like EFIELD tag is still ignored. In the OUTCAR, under the section 'dipole corrections', there's still no EFIELD tag. So the resulting potential profile and total energy is still same as original case. Is that because my system is a 2d charge-neutral slab so the tag doesn't work, or I did something wrong here?

Best,
Yujia

[jonathan_lahnsteiner2]

Dear Yujia Teng,

I was checking your input files. I think the tag IDIPOL is missing. As described in the vasp wiki the EFIELD is applied along the direction set by IDIPOL
https://vasp.at/wiki/EFIELD
Please do a careful check if this resolves your problem.

All the Best Jonathan

[yujia_teng]

Dear Jonathan,
The IDIPOL tag is added in my INCAR. It's in the directory 'IDIPOL-3' and is also in the OUTCAR. But EFIELD is still ignored.

relevant part in INCAR:

# Dipole correction
#LDIPOL = .TRUE.
IDIPOL = 3
#DIPOL = 0.5 0.5 0.5

LVHAR = .True.
EFIELD = 0.1
LMONO = .True.

OUTCAR:

Dipole corrections
LMONO = T monopole corrections only (constant potential shift)
LDIPOL = F correct potential (dipole corrections)
IDIPOL = 3 1-x, 2-y, 3-z, 4-all directions
EPSILON = 1.0000000 bulk dielectric constant
LVACPOTAV = F vacuum potentials using an averaging scheme for the charge density
VACPOTFLAT = 0.1000000 required flatness to determine vacuum potential

Best,
Yujia

[jonathan_lahnsteiner2]

Dear Yujia Teng,

I was now reconsidering your original question:

Now I want to see if I apply the external electric field with EFIELD tag, can I cancel the field in vacuum when there's no dipole correction, or make the potential flat in vacuum region?

In principle, your logic is correct: since the potential slope in the vacuum is caused by an artificial electric field, applying an external electric field of the exact same magnitude in the opposite direction could mathematically "cancel" it out and flatten the potential. However, in practice for VASP, there are three major reasons why you should not use EFIELD for this purpose and instead use LDIPOL.

1. Without dipole correction, the slope of the potential in the vacuum depends on the dipole moment of your slab and the total length of your simulation cell.
   • If you change your vacuum thickness (cell size), the artificial field strength changes.
   • This means you would have to manually re-calculate and tune the EFIELD value every time you adjusted your cell or moved an atom, which is highly inefficient.
2. When you apply an external electric field (EFIELD) in VASP, the code applies a sawtooth potential (because of periodic boundary condition). This creates a jump (discontinuity) in the potential at a specific point in the cell.
   • If you use EFIELD without LDIPOL, you still have the uncorrected dipole interaction between periodic images.
   • Even if you managed to flatten the potential in one part of the vacuum, you would likely end up with a huge, unphysical potential jump elsewhere that complicates the convergence of your total energy.
3. The LDIPOL = .TRUE. tag is specifically designed to do exactly what you are describing, but automatically.
   • It calculates the internal dipole moment.
   • It determines the necessary "counter-field" (the dipole layer).
   • It applies it precisely so that the potential is flat in the vacuum and the periodic images are decoupled.

So there is no need to do the counter field calculation by hand. For more information on the EFIELD (if still needed) tag I would recommend to take a look at the VASP wiki https://www.vasp.at/wiki/index.php/EFIELD.

All the Best Jonathan

[yujia_teng]

Dear Jonathan,
Thanks for the detailed explanation. It's very clear to me. I was thinking a simple relation \(P=\varepsilon_0 E\), where E-filed just depends on the slab dipole moment. I didn't realize that electric field value in vacuum here is not fixed, because it depends on vacuum region. And there're some other facts you mentioned here makes it not a good idea to apply E-field only. The situation here is different from the just ideal simple electrostatics.

Best,
Yujia