Force constant for S-N-S angle in NPF2/NTF2

[mattiafelice-palermo]

Dear Prof. Acevedo,

I'm currently working on a system containing FSI anion and was exploring the force field proposed in this work and validating the bonded contributions against the PES from QM calculations.

In particular, I noticed a large discrepancy between the PES from QM (B97-3c) for the S-N-S bond force constant and the PES from the MM force field derived in this work:

(angle units are in degrees, apologies for the inaccuracy in the figure)

In comparison, fitting the bond force constant to the QM PES yields a force constant of about 170 kJ/mol/rad, against the 671 kJ/mol/rad from this work.

I wanted to ask if there's a trivial explanation for this mismatch! Maybe I'm doing something wrong in my calculations.

Thank you for your help in understanding this!

Mattia

[orlandoacevedo]

Hi Mattia,

I am unsure where the discrepancy comes from. Perhaps from fitting to B97-3c? This S-N-S force constant was originally developed by Canongia Lopes and Pádua in https://doi.org/10.1021/jp0476545

Best,
Orlando

[mattiafelice-palermo]

Hello,

thank you very much for the prompt reply! I will go through the publication and try reproducing the PES with HF/6-31G(d) + resampling at MP2/cc-pVTZ(-f) as the authors did. Although I find it unlikely that the change in level of theory could result in such large discrepancy.

If I get more insights I will send an update!

Thank you again and kind regards,

Mattia

[mattiafelice-palermo]

I might have found the culprit for why this force constant is quite off compared to QM reference values.

From the paper of Canongia Lopes and Pádua, they mention they were not able to derive the angle constants directly from QM calculations for bistriflylimide, due to diagonalization issues:

Tracing back to the mentioned paper from Halgren (https://pubs.acs.org/doi/10.1021/ja00168a015) , I see that the angle constant is calculated through an empirical formula:

Going through the paper would require a lot of effort (it's one of those good old extremely difficult papers...). My guess is that this kind of empirical formula might hold well for more "standard" species. Applying them to peculiar moieties like bistriflylimide, which features a negatively charge nitrogen connected to two disulfone groups, might be a bit too much of a stretch.

It is just a hunch though, and I might be off track.

I will perform an energy analysis of a QM tailored force field (via the Q-Force library) versus these emprical parameters to assess how much they affect the PES of the force field when compared to a reference QM PES.

Thank you again in the meantime for the feedback!

Mattia

[orlandoacevedo]

Hi Mattia,
Thank you for the update! During my time in the Jorgensen group, one of the major philosophies that we followed for OPLS-AA force field development is to focus on nonbonded parameters and torsion rotations (particularly those of longer alkyl chains, etc.) given their strong influence on solvent properties. We didn't emphasize trying to reproduce vibrational frequencies - bonds or angles (although of course it is important). I am glad that you are investigating it and looking forward to hearing about any advances you make.
Best,
Orlando