Probably dont matter that much. You only need 3 parameters, charge, 'ion radius' and 'ion hardness'
hardness... the smaller the ion the less the results will depend upon this
radius.... i would just lift out the value out of crystal tables.
Not as arbitary as it might seem. Experience of looking at considerable numbers of ions, the MD hydrations they produce, and how these compare to experimental structural measurements shows that big ions are basically impossible to do... (Cl for instance) here polarisability, and the other non-physical aspect of the model are comming into play. With the likes of Beryllium, its almost impossible to get wrong.