Parameters for ions are typically based on reproduction of the experimental
free energy of solvation. In Charmm, the ion is assigned a formal charge
and then the LJ parameters are optimized to reproduce the free energy
of solvation. There is a Beglov and Roux paper in which they did this
for sodium. Finally, it should be noted that one set of LJ parameters for
an ion may not yield proper interactions with all types of functional
groups. A trick to get around this in charmm is the use of NBFIX for
to assign specific ion-atom LJ parameters that balance such
interactions.
alex
