SAPT electrostatics concerns

we are testing the SAPT electrostatic interaction between Na and Cl compared to a ESP computed at different methods (aug-cc-pvtz for all). The HF ESP of the Cl- interacting with a positive point charge probe at the corresponding distance yields -588.7 kj/mol for HF ( 590 for MP2 and is also comparable for CC), while SAPT0 electrostatics gives -608, which should be comparable to HF. The argument was that the interaction point charge model should give stronger interaction, but we are getting SAPT stronger by a margin.
Do you have any idea what could be the source of the discrepancy?


The first thing I notice is the basis set, which for SAPT0 calculations should ALWAYS be jun-cc-pvdz. The performance of SAPT0 relies heavily on cancellation of errors, so jun-cc-pvdz must be used. Secondly, more information on your tests are required to examine this issue any further; for instance, what is the distance between the Cl- and the point charge? Can you post your input files?