VASP Forum

Dear VASP support team,

in two recent publications, Ryosuke Jinnouchi et al. calculated the redox potentials of transition metal ions such as Fe2+ and Ag+ in an aqueous environment. All the used atomic models have a net charge, which was probably set by the directive NELECT. The net charge was compensated with a background charge. I wonder which correction was used to prevent the spurious long range Coulomb interactions between charged species and their periodic copies. Unfortunately, nothing related to the latter is mentioned in both publications.

Nonetheless, which charge correction would make sense in case of atomic models for fluids? The Freysoldt–Neugebauer–Van de Walle (FNV) correction, the Kumagai–Oba correction, the Makov–Payne correction, or the Coulomb kernel truncation approach?

Thanks.

René

Kernel truncation is not suited for 3D models.

FNV, Kumagai–Oba and Makov–Payne (LMONO=.TRUE. in VASP) are all not well suited for impurity models such as the ions solvated in water.
For example Makov–Payne would assume a constant dielectric constant in the medium, however an ion dissolved in water would have a spatially varying dielectric environment around the impurity.

Fortunately water has a high dielectric constant (80) and long-range interactions are effectively screened out since the dielectric constant enters in the denominator.
To show the effet of the cell size was investigated in https://www.nature.com/articles/s41524-024-01295-6

The Kumagai–Oba method aka eFNV method might be suitable if the image charge correction is determined as an ensemble average from 100 or more individual DFT calculations. In each DFT calculation the eFNV correction is calculated with a charged cell of an ion embedded in differently oriented solvent molecules of a liquid. The different structures of the ion surrounded by solvent molecules are frames of MD simulations in which an MLFF is used. Does this sound reasonable?

Since the discussion diverges from the intention of this board "Using VASP", feel free to move this discussion to another board index which suits better.

Thanks.

René