I am interested in parameterizing dodecaborate cage [B12H12]2- and its halide derivatives [B12X12]2-, X=Cl,Br,I. I read the original paper on procedure, and did some literature survey. I have some idea how to go about it, but would like some sanity check on the methodology from more experienced members of the community.

The molecules are quite rigid and I think that for the most part it should be straightforward. This is what I planned:

a) LJ params for start can be adopted initially from CgenFF; I noticed the B201 atomtype available in the latest 2020 version. Luckily, bonded halides are well represented thanks to more recent efforts (although bonded to carbon, but I hope we can get away with this). So even in this case atom types are there.

b) Bonds, angles and torsions look simple as the molecule is rigid. So just setting parameters to match the QM optimized geometry should work.

c) Partial charges also do not look very tricky. But I am guessing this would require the most effort; dodecaborate has a negative charge on the hydrogens and forms dihydrogen bonds with water (B-H---H-O-H). [B12Cl12]2- (and Br derivatives) have a negative charge on the halide atoms, whereas Iodine is largely neutral. There is also the issue of sigma hole on halogen atoms and I am not sure if it is relevant in my case.

d) I The water interactions need to be modeled correctly. Maybe fitting the LJ and charges to fit the QM interaction energies would suffice.

Some suggestions and improvements on the aforesaid procedure would be highly appreciated.