@michal you are exactly correct that you should be wary of breaking aromatic bonds within F-SAPT. Unfortunately, A-SAPT isn’t currently available within the public release of Psi4 – our apologies. This may change in the future, but for the moment I would like to caution you with respect to utilizing atomic fragments within F-/ISAPT.
As a general rule, I recommend strongly against using atomic fragments in either the F-SAPT or ISAPT partitioning schemes. The reason for this is that atomic fragments can be dicey at best within the F-SAPT post-analysis phase of the computations (i.e., running fsapt.py within your fsapt/ directory); the charge assignment for link bonds between atomic fragments (either by charge or 50-50) can
- yield drastically different results between link assignment scheme, and
- both kinds of link assignments can lead to spurious dipoles on atomic fragments,
either of which can influence the overall F-SAPT analysis in unexpected ways. However, since utilizing atomic fragments seems to be your only current way forward, just be very careful to check that your results make chemical sense. Also, be sure to thoroughly check the Orbital Check section of the fsapt/frag_A.dat and fsapt/frag_B.dat files, as values more than 0.05-0.1 electrons lost from doubly occupied orbitals in a fragment can indicate that the F-SAPT analysis of a particular fragmentation scheme may no longer be valid.