UT theoretical chemistry code forum

Dear Professor:
I have been working on the dissociated path of a charged adsorbates on SWNT. I constructed a finite length carbon nanotube with charged peroxodisulfate ions (with +2e) adsorbed on it. The transition state calculations for the reaction path have been converged under vacuum, after which I have added implicit solvation parameters (EB_k = 80, LSOL = .TRUE.) to calculate the single point energy of each reaction state to get the solvation effect on this reaction. However, the energy of the transition state obtained from solvation calculation is lower than the initial state, which is obviously unreasonable (the reaction energy barrier is around 1 eV under vacuum).
1. Is it possible to qualitatively reflect the trend of the reaction path by only calculating the single point energy under implicit solvation? Is it necessary to include solvation parameters for transition state calculations? This requires too much computational resources to be consumed.
2. Is it available for vasp-vtst to calculate the reaction paths of charged adsorbates, and is it necessary to add other parameters besides NELECT?
3. Is the implicit solvation accurate for the energy calculation of the ions? Although this is not related to vtst's code, I will be glad to hear your opinion since I haven't got a reply in vaspsol's forum yet.

Here are my INCAR parameters for the single point calculation with solvation:

PREC = Accurate
ENCUT = 520.000
NPAR=4
ISYM=0
IDIPOL=4
LSOL = .TRUE.
EB_K = 80
IBRION = -1
NSW = 1

ISIF = 2
NELMIN = 2
EDIFF = 1.0e-05
EDIFFG = -0.03
VOSKOWN = 1
NBLOCK = 1
NWRITE = 1
NELECT = 690.0
ISYM = 0
NELM = 200
ALGO = Fast
IVDW = 12
VDW_S6 = 1.000
VDW_S8 = 0.7875
VDW_A1 = 0.4289
VDW_A2 = 4.4407
INIWAV = 1
ISTART = 1
ICHARG = 0
LWAVE = T
LCHARG = .FALSE.
ADDGRID = .FALSE.
ISMEAR = 0
SIGMA = 0.05
LREAL = Auto
LSCALAPACK = .FALSE.
RWIGS = 0.77 0.32 0.73 1.02 0.99

Well, as you say, many of these questions are scientific in nature rather than technical. I can, however, try to answer some of the technical points.

First, I would have no expectation of the energy going up or down when you include the implicit solvation model. It is certainly strange to have a transition state become lower in energy than the initial state, but this could be due to the fact that you are doing a single point calculation for each image. My recommendation is to relax your initial and final states with the implicit solvation model and then run another NEB calculation, again, with the implicit solvation model. I don't understand your comment that the implicit solvation model makes the NEB calculation too expensive - it should not be the most expensive part of the calculation.

The question of whether modeling solvation is necessary is entirely dependent upon the system. There will be cases where implicit solvation has a negligible effect and others where (for example) screening of charges is critical to an accurate description.

You can use vasp-vtst to model charged species - just remember that vasp will add a uniform background charge to compensate and make the cell neutral. The solvation model is helpful in these calculations because it will screen charges locally, rather than have them interact with the background charge over the entire cell.

The accuracy of the implicit solvation model is again related to your system. If you have strong interactions between the solvent and your system, it can be a good idea to include some explicit solvation and then implicit as well. I have seen results from the Hennig group, who wrote the solvation model in vasp, where there is fairly rapid convergence of the solvation energy with just a single layer of solvation molecules around a species of interest.

Thank you so much for your opinion, this helps me a lot!