Avoiding solution clustering

Hello, I am seeing if there is any established method in LAMMPS to effectively regulate the “dispersion” of species to avoid unwanted clustering.


  • Unwanted clustering in aqueous metal salt solution

  • TIP4P water

  • LJ+Coulomb cations/anions (potentials taken from literature)

  • Periodic boundary conditions are applied

  • Spacing/count matches experimentally-measured density/concentration at room temperature

What was attempted:

  • fix langevin at room temperature, hoping the random forces would help --> result: clustering still seen
  • “Electronic continuum correction” method in literature, where the aqueous species charges are reduced to effectively model water screening --> result: clustering still seen
  • Adding charge transfer and polarization behavior among solution species, a more complex option seen in literature --> before going this route, I just wanted to check if LAMMPS already had some kind of “effective” method established to avoid unwanted clustering. The most I could find from a search was compute cluster/atom.

Thank you,

two questions:

  • what kind of cations/anions are you using?
  • have the potentials for those ions been parameterized for TIP4P water?


Hi Axel, thanks for your reply earlier. After more extensive review, the cations/anions (trying a wide variety of mixtures) are parameterized to the water model as individual species, but not as mixtures. Found some workarounds in LAMMPS (setforce, addforce, set velocity), but after more review, this looks like a deeper research question on the parameterization itself than a LAMMPS question.
Thank you,


indeed, and neither of the options you mention are addressing the core issue, which is the balance of the LJ parameters for the cross-ion interactions. For mixed salt solutions, they usually cannot be inferred from mixing rules. the CHARMM force field has for that purpose an NBFIX section which contains specific overrides for the mixed terms, which is mostly applied to ions.
the situation gets even worse with multivalent ions (like Ca^2+ or Mg^2+ or Fe^2+ or Al^3+) where there are significant many-bondy contributions that would need to be added and/or the charges need to be redistributed/equilibrated to avoid overbinding.