Re: [lammps-users] kspace cutoff when separating an atom from the surface - limitation or error?

Hi all,

Final check in on this. Hoping someone can confirm why when an atom is separated from the surface even well beyond the cutoff (as per my files in message 1, this is defined as 12A), I still see a reduction in KE. When doing this using a different potential it behaves as expected (reaxFF for example). I have tried using the kspace slab modifier but still see this reduction. I expect this modification would remove periodic interactions, making me feel it is simply a limitation of the kspace method (infinite reductions in energy due to interactions).

Is there perhaps a way to remove any of these interactions beyond the cutoff or is my goal just not attainable using these methods?



your understanding of the coulomb interaction with long-range is incorrect as well as the impact and purpose of the slab correction.
the cutoff applies only to part of the interaction, the remainder is handled by the kspace style. the slab correction doesn’t “shorten” that interaction with the slab, but just subtracts the interactions with periodic images of the slab.
since the coulomb interactions decay with 1/r there are still significant contributions at a very long distance from a surface.

reaxff uses an approximation to coulomb interactions (Wolf sum, IIRC), that is not a good choice for a surface, it is best for bulk crystals. it differs from a straight cutoff by having a smooth truncation of the interaction, but it is effectively a short range approximation that assumes that interactions will be screened. but in your setup they are not.