Issues with EFIELD_PEAD, LPEAD_RELAX, and dielectric response in VASP 6.4.3

[askhetan]

Dear VASP team,

I am using VASP 6.4.3 to study the effect of an external electric field on the total energy and forces of an insulator using the EFIELD_PEAD tag. Additionally, I aim to compute the static dielectric response, i.e., the dielectric matrix, to determine the polarizability.

According to this VASP forum post: “https://www.vasp.at/forum/viewtopic.php?t=19899”, this should be possible, and the forces should include contributions from the external electric field. However, I have encountered several issues during my calculations and would appreciate clarification on the following points:

1. LPEAD_RELAX not recognized in OUTCAR
The site I referred states that to apply EFIELD_PEAD to forces, LPEAD_RELAX = .TRUE. must be set. I have included this in my INCAR file, but:
• This setting does not appear in the OUTCAR file.
• I could not find LPEAD_RELAX in the official VASP wiki documentation.
Is this setting still valid, or does VASP handle it automatically without explicitly writing it to OUTCAR?

2. EFIELD_PEAD value printing issue (First relaxation step INCAR)
I observed inconsistencies in how EFIELD_PEAD values are printed in OUTCAR:
• EFIELD_PEAD = 0.0 0.0 0.00001 → OUTCAR reports:
EFIELD_PEAD = 0.0000 0.0000 0.0000
• EFIELD_PEAD = 0.0 0.0 0.0001 → OUTCAR correctly reports:
EFIELD_PEAD = 0.0000 0.0000 0.0001
• EFIELD_PEAD = 0.0 0.0 0.000001 → PEAD-related settings do not appear in OUTCAR at all.

Does this indicate that VASP ignores electric field values below a certain threshold? If so, what is the minimum effective EFIELD_PEAD value? Should it be ≥ 1E-4 for the field to be applied correctly?

3. Inconsistency between relaxation and LCALCEPS electric field settings (Second step INCAR)
I perform structural relaxation under an external electric field by setting EFIELD_PEAD = 0.0 0.0 0.0001 and use the resulting CONTCAR for a subsequent LCALCEPS calculation. However, when I apply the same field setting (EFIELD_PEAD = 0.0 0.0 0.0001) in the LCALCEPS step, the OUTCAR file reports: EFIELD_PEAD = 0.01 0.01 0.0001

This suggests that VASP modifies the applied electric field during the LCALCEPS calculation.

• Why does VASP adjust the EFIELD_PEAD values in LCALCEPS?
• Is there a way to ensure that the field strength remains consistent between the relaxation and dielectric response calculations?

4. Polarizability Tensor Calculation
To calculate the polarizability tensor from the dielectric tensor, I am using the following equation, as given in VASP wiki documentation:

ϵij∞ = δij + 4π/ϵ0 ( ∂Pi / ∂εj ) ; i, j=x, y, z

Here, the term "4π/ϵ0 ( ∂Pi / ∂εj )" represents the polarizability tensor ,αij. Can you please confirm if this interpretation is correct?

Below are POSCAR and my INCAR's settings (for both steps):
Ce O
1.00000000000000
11.6302339033276603 0.0000000001217020 0.0000000000860563
5.8151162812246247 10.0720783993327281 0.0000000000860508
5.8151162812246247 3.3573592084659345 9.4960466737075215
Ce O
27 54
Direct
0.0000000000000000 0.0000000000000000 0.0000000000000000
0.9999999999904219 0.9999999999907772 0.3333333432481851
0.0000000000668123 0.9999999999815472 0.6666666864963631
0.9999999999549800 0.3333333433098460 0.0000000000000000
0.0000000000313705 0.3333333433006160 0.3333333432481851
0.0000000000217923 0.3333333432913932 0.6666666864963631
0.0000000000455600 0.6666666865204220 0.0000000000000000
0.0000000000359819 0.6666666865111921 0.3333333432481851
0.0000000000264038 0.6666666865019693 0.6666666864963631
0.3333333432971699 0.0000000000000000 0.0000000000000000
0.3333333432875989 0.9999999999907772 0.3333333432481851
0.3333333432780208 0.9999999999815472 0.6666666864963631
0.3333333432521499 0.3333333433098460 0.0000000000000000
0.3333333432425718 0.3333333433006160 0.3333333432481851
0.3333333432329937 0.3333333432913932 0.6666666864963631
0.3333333432567613 0.6666666865204220 0.0000000000000000
0.3333333432471903 0.6666666865111921 0.3333333432481851
0.3333333433235808 0.6666666865019693 0.6666666864963631
0.6666666865083783 0.0000000000000000 0.0000000000000000
0.6666666865847688 0.9999999999907772 0.3333333432481851
0.6666666865751907 0.9999999999815472 0.6666666864963631
0.6666666865493269 0.3333333433098460 0.0000000000000000
0.6666666865397488 0.3333333433006160 0.3333333432481851
0.6666666865301707 0.3333333432913932 0.6666666864963631
0.6666666865539383 0.6666666865204220 0.0000000000000000
0.6666666865443602 0.6666666865111921 0.3333333432481851
0.6666666865347821 0.6666666865019693 0.6666666864963631
0.2500000000006395 0.2499999999893205 0.2500000000394849
0.2500000000142109 0.2499999999856115 0.5833333134904777
0.2499999999814904 0.2499999999708677 0.9166666865358479
0.2500000000228866 0.5833333134192671 0.2500000000394849
0.2500000000364579 0.5833333134155581 0.5833333134904777
0.2499999999540918 0.5833333135000842 0.9166666865358479
0.2499999999602309 0.9166666865097426 0.2500000000394849
0.2499999999738023 0.9166666865060336 0.5833333134904777
0.2500000000270504 0.9166666864912898 0.9166666865358479
0.5833333134664755 0.2499999999893205 0.2500000000394849
0.5833333134800469 0.2499999999856115 0.5833333134904777
0.5833333134473193 0.2499999999708677 0.9166666865358479
0.5833333134887155 0.5833333134192671 0.2500000000394849
0.5833333135022869 0.5833333134155581 0.5833333134904777
0.5833333134199279 0.5833333135000842 0.9166666865358479
0.5833333134260670 0.9166666865097426 0.2500000000394849
0.5833333134396383 0.9166666865060336 0.5833333134904777
0.5833333134928793 0.9166666864912898 0.9166666865358479
0.9166666865090178 0.2499999999893205 0.2500000000394849
0.9166666865225892 0.2499999999856115 0.5833333134904777
0.9166666865758373 0.2499999999708677 0.9166666865358479
0.9166666865312578 0.5833333134192671 0.2500000000394849
0.9166666865448292 0.5833333134155581 0.5833333134904777
0.9166666865484387 0.5833333135000842 0.9166666865358479
0.9166666865545778 0.9166666865097426 0.2500000000394849
0.9166666865681492 0.9166666865060336 0.5833333134904777
0.9166666865354287 0.9166666864912898 0.9166666865358479
0.0833333357894617 0.0833333358411963 0.0833333358383683
0.0833333357669801 0.0833333358273336 0.4166666567649813
0.0833333358433705 0.0833333358181108 0.7500000000131593
0.0833333357858024 0.4166666567163020 0.0833333358383683
0.0833333358492894 0.4166666567024393 0.4166666567649813
0.0833333358397113 0.4166666566932165 0.7500000000131593
0.0833333358267510 0.7500000000261480 0.0833333358383683
0.0833333358042694 0.7500000000122853 0.4166666567649813
0.0833333357946913 0.7500000000030553 0.7500000000131593
0.4166666567346198 0.0833333358411963 0.0833333358383683
0.4166666567121382 0.0833333358273336 0.4166666567649813
0.4166666567025601 0.0833333358181108 0.7500000000131593
0.4166666567309605 0.4166666567163020 0.0833333358383683
0.4166666567084789 0.4166666567024393 0.4166666567649813
0.4166666567848694 0.4166666566932165 0.7500000000131593
0.4166666566859405 0.7500000000261480 0.0833333358383683
0.4166666567494275 0.7500000000122853 0.4166666567649813
0.4166666567398494 0.7500000000030553 0.7500000000131593
0.7499999999458211 0.0833333358411963 0.0833333358383683
0.7500000000093152 0.0833333358273336 0.4166666567649813
0.7499999999997371 0.0833333358181108 0.7500000000131593
0.7500000000281375 0.4166666567163020 0.0833333358383683
0.7500000000056559 0.4166666567024393 0.4166666567649813
0.7499999999960778 0.4166666566932165 0.7500000000131593
0.7499999999831175 0.7500000000261480 0.0833333358383683
0.7499999999606288 0.7500000000122853 0.4166666567649813
0.7499999999510578 0.7500000000030553 0.7500000000131593

First relaxation step (with applied electric field):
EDIFF = 1E-5
EDIFFG = -0.05
IBRION = 2
ISPIN = 2
ISMEAR = 0
SIGMA = 0.05
ISIF = 3
NSW = 1
NELM = 800
ENCUT = 800
ISTART = 0
PREC = Single
LREAL = Auto
LCHARG = .FALSE.
LWAVE = .FALSE.
ALGO = All
LASPH = .TRUE.
LMAXMIX = 6
METAGGA = R2SCAN
LPEAD_RELAX = .TRUE.
EFIELD_PEAD = 0.0 0.0 0.0001
LCALCEPS = .FALSE.
#ISYM = 0

Second step (LCALCEPS calculation using relaxed structure):
EDIFF = 1E-5
EDIFFG = -0.05
IBRION = -1
ISPIN = 2
ISMEAR = 0
SIGMA = 0.05
ISIF = 3
NSW = 0
NELM = 800
ENCUT = 800
ISTART = 0
PREC = Single
LREAL = Auto
LCHARG = .FALSE.
LWAVE = .FALSE.
ALGO = All
LASPH = .TRUE.
LMAXMIX = 6
METAGGA = R2SCAN
EFIELD_PEAD = 0.0 0.0 0.0001
LCALCEPS = .TRUE.
#ISYM = 0


I would greatly appreciate any insights or clarifications on these points.

Thank you in advance for your help!

Best regards,
Abhishek Khetan

[ahampel]

Dear Abhishek,

thank you for reaching out to us on the official VASP forum.

Let's try to take these questions one by one.

1. LPEAD_RELAX is not written to the OUTCAR, but it is recognized by VASP and used to calculate forces. You should be able to see the difference in the relaxed structure. The flag is not documented as it is an experimental flag. So please use it with caution, but it is still valid.

2. The printing of EFIELD_PEAD is just limited to 1e-5 .

3. If LCALCEPS=True and you set EFIELD_PEAD = 0.0 0.0 0.0001 then all elements smaller than 1e-5 will be reset to their default value, which is 0.01 with LCALCEPS=True, which is written on the wiki page of the EFIELD_PEAD tag: "Mind: If EFIELD_PEAD is used in combination with LCALCEPS=.TRUE., electric field values below 1E-5 will be reset to the default value." So yes, if you want consistent field strength between the relaxation step and the dielectric response calculation you have to choose for each component a value larger or equal to 1E-5!

4. Yes I think you are correct. However, note that this static dielectric response here with LCALCEPS is not the full dielectric tensor. See for reference here: https://www.vasp.at/wiki/index.php/Cate ... properties and maybe also here https://www.vasp.at/mmars/day2.pdf . Another option to obtain this is for example ALGO=CHIO although much more expensive.

Best,
Alex

[askhetan]

Dear Alex
Thanks for your very quick and very helpful response.
At present I am looking only for the static contribution.
Best regards
Abhishek