Inconsistencies in results when applying electric fields

[nuha_sahir]

Hello.

I am applying an external electric field to a system consisting of graphene with a dissociatively adsorbing H₂ molecule, in order to study the effect of the electric field on the catalytic activity of the 2D sheet. I am monitoring the chemisorption energy (Ec = E(slab+2H)-E(slab+H2)) and the dissociation barrier (Eb = E(transition state)-E(slab+H2)).

I have a few concerns:
1) No noticeable effect of the electric field - The changes in Ec and Eb after applying the electric field are very small compared to the zero-field case. They appear essentially the same as without the field. However, the total energies reported in the OUTCAR after convergence are clearly different from the zero-field case, which suggests that the field is indeed being applied. It is surprising that the relative energies (and therefore the adsorption barrier and chemisorption energy) remain unchanged. Based on experimental observations and also on published work using DMol3 [Z. M. Ao and F. M. Peeters, Appl. Phys. Lett. 96, 253106 (2010)], I would have expected a change in trend, even if not quantitatively identical.

1. Attempts to resolve the issue - I have tried the commonly recommended fixes (from the forum and wiki), including turning off symmetry (ISYM=0), enabling dipole corrections (IDIPOL, LDIPOL, DIPOL), and increasing vacuum thickness. None of these changes affected the outcome.

2. Possible code-specific differences? I have also seen similar questions from other users (https://www.vasp.at/forum/viewtopic.php?p=31677#p31677) reporting the same issue. Could the difference arise from the way VASP applies the electric field compared to DMol3, leading to the lack of sensitivity in relative adsorption energies/barriers?

Any guidance on whether I am applying the electric field correctly in VASP, or if there are known limitations in how field effects are treated in such slab calculations, would be very much appreciated. I have attached the sample input (converged) files for a slab+H2 system for reference and this is my INCAR:

Code: Select all

SYSTEM = GRAPHENE

!!SYSTEM INITIALIZATION!!
ISTART = 0
ICHARG = 2

!!ELECTRONIC STEPS!!
ALGO = Normal
!!GGA = PE
ENCUT = 500
ISMEAR = 0 
SIGMA = 0.05
EDIFF = 1E-6
NSW = 1000

!!IONIC STEPS!!
PREC = Accurate
IBRION = 2
!!NFREE = 2
POTIM = 0.15
EDIFFG = -0.01
NELM = 500
NELMIN = 5
!!SMASS = 1.0

!!CELL RELATED!!
ISIF = 2
ISYM = 0
LREAL = Auto
SYMPREC = 1E-4
!!ADDGRID = .TRUE.

!!CORRECTIONS!!
IVDW = 11
IDIPOL = 3
LDIPOL = .TRUE.
DIPOL = 0.5 0.5 0.5

!!MAGNETIZATION!!
ISPIN = 2
!!NUPDOWN = 0
MAGMOM = 162*0.7 0.7 -0.7
!!NELECT = 448
EFIELD = -0.5

!!GENERAL/CALCULATION OPTIMIZATION!!
LWAVE = .FALSE.
LCHARG = .FALSE.
NCORE = 4
!!NPAR = 4

!!MOLECULAR DYNAMICS!!
!!MDALGO = 2
!!TEBEG = 300
!!TEEND = 300
!!MAXMIX = 40	

!!NEB CALCULATION!!
!!NWRITE = 2
!!IMAGES = 16
!!SPRING = -3
!!ICHAIN = 0
!!LCLIMB = TRUE
!!IOPT = 3
!!MAXMOVE = 0.2
!!TIMESTEP = 0.1

Any help would be much appreciated. Thanks in advance!

Nuha

[manuel_engel1]

Hello Nuha,

Thanks for posting on the VASP forum. First off, this is a rather complex question. The issues you encounter could be related to many different things (as you are already aware), so it is impossible to discuss it in full detail here.

Here are some general hints and ideas that could help shed light on this matter:

• Check your dipole correction placement and direction of the E-field. Since I only have the INCAR file, I cannot verify it myself. The rest of the settings look very reasonable.

• A perpendicular external field in graphene may be strongly screened or cause nearly identical energy shifts for all states you compare, so adsorption/dissociation energy differences may be very small even though the total energy changes.

• H₂ is non-polar; dissociation to two H atoms changes local charge but may produce only a small net dipole change on graphene. If the field couples primarily to dipole changes, the effect can be tiny.

I will discuss this matter with my colleagues, taking into account also the previously asked question on the forum. In the meantime, could you please upload the remaining input and output files following our forum posting guidelines? I will also move this topic over to Using VASP.

[nuha_sahir]

Hi Manuel!

Thank you so much for replying. I have attached a zip folder with the inputs and OUTCAR for the initial state and final state (energy(FS) - energy(IS) gives the chemisorption energy). Let me know if anything else is required.

Regarding your suggestions: I believe the dipole placement is correct in my setup, as the sheet is centered in the simulation cell and the center of mass is at ~0.5 along z, where the dipole correction is applied. The other two points you raised are certainly possible. I was hesitant because the DMOL3 study (mentioned in my first post) reported a significant difference under similar conditions, but I will revisit these aspects in light of your advice.

I look forward to your feedback after your discussion with your colleagues.

Thanks again!

Nuha