>First of all, 3-21G basis set is too small for thermochemistry. Triple or, better, quadruple-zeta basis sets augmented with diffuse and polarization functions are commonly used.
>Second, RHF is also inappropriate for thermochemistry. You should use correlated methods (DFT or MP2 geometry optimization with single-point MP4 correction).
>Third, and most important is the solvation model. PCM cannot account for specific solvation of the cation and functional groups of the anion and the parent ligand. You should explicitly solvate the cation, anion, and complex with sufficient amount of water molecules (5-6 for Na+, at least 1-2 molecules for each O atom of the ligand and complex, 1 water molecule for N atom) and embed this cluster in PCM. Obviously, you should take into account hydration/dehydration of Na+ during complexation (excess water molecules should also be embedded in PCM; probably, they should be combined in a cluster, because water always forms hydrogen bonds with its neighbors).
>Finally, vibrational calculations (gas phase, harmonic oscillator, rigid rotor) are irrelevant for solution thermochemistry. PCM calculations directly give free energy of the solute in solution (TOTAL FREE ENERGY IN SOLVENT). To calculate the complexation free energy, you should consider the free energy of the following equilibrium (L is the ligand, m + n = p + q, (H2O)q is a cluster of q water molecules):
>[Na(H2O)n]+ + [L(H2O)m]- = NaL(H2O)p + (H2O)q