Dear all,

I am simulating a ionic liquid (TFSI-/Li+) diluted in diglyme with classical MD

https://pubchem.ncbi.nlm.nih.gov/compound/3816071

https://pubchem.ncbi.nlm.nih.gov/compound/Diglyme

I have tried long time in calculating the Li diffusivity using parameters from LigParGen and MSD/Green Kubo as implemented in Gromacs, basically following the MDTutorials approach (EM minimization, NVT+NPT equilibration and a MD production run). I have tried several combinations of parameters, rcoulomb, rvdw, nstvout, tau_t and tau_p, etc... but MSD is always too low and GK is not really converging, so one could choose the set of parameters according to the goodness of the results.

I then tried to use a more adequate force field, specific for ionic liquids, ie fftool (https://github.com/paduagroup/fftool).

How can I properly choose the atom types from the attached oplsaa.ff file?

Some are evident, such as C in the C3F groups of TFSI, other are less, such as the middle Cs in dyglyme which are connected to H2 and O and C, some are just wrong, eg for the SO2 group there is nothing (the only available S in from DMSO).

Unfortunately I get some pretty horrible results, with the density which is less then half the experimental one.

could anyone suggest how to get fftool right or an alternative solution?

Thank you

BR

Marco