Zero OPLS Lennard-Jones parameters for H atoms and overlapping atoms

GROMACS version: 2019.6
GROMACS modification: Yes/No

Hi Gromacs users.

I am using Gromacs-2019.6 and the OPLS-AA force field for my simulations.
I have noticed that the combination rule is 3 (geometric average for both sigma and epsilon). However, the Lennard-Jones parameters for hydrogens in hydroxyl (OH), or amine (NH), are zero.
Does that make sense? Is there any physical reason for that?

As well, I am modelling the molecular packing of a phosphoric acid-based molecule, i.e., with O=P(OH)2 functional group. However, after geometry optimization, I see in some cases that one of the hydrogens (of P-O-H group) is overlapping with the doubly bonded oxygen (P=O) or the oxygen of the second P-O-H group.
I am not sure if the forcefield is the reason for that.

Have you ever faced such a situation? How did you fix that?
Thanks in advance for your help.


H atoms rarely have their own LJ parameters, because their radii are too small compared to the heavy atoms to which they are bonded. Therefore, most force fields assume zero LJ interactions of H atoms, and the rest of the atom types are parametrized in conjunction with this assumption.

This sounds more like a problem with bonded parameters being problematic such that you get a distorted internal geometry.

Hi Justin,

Thanks a lot for your reply!

Unfortunately, I have checked my forcefield.itp and files but have not found anything strange.

As a test, I have tried giving non-zero LJ parameters to the hydrogen atoms of the phosphate (using parameters for hydrocarbon). I don’t have those overlapping anymore. Although I agree with your first comment, I am now confused with my own simulations. It seems that adding the LJ parameter solved the issue but I am not sure if I am creating artificial interactions between the hydrogens and the rest of the system.
What should I do? Keep the LJ parameters or set them at zero?

Thank you very much.


Adjusting LJ parameters requires reparametrization of relevant bonded interactions (namely dihedrals, which are scaled in a specific way to be a correction factor for short-range inaccuracies in LJ and Coulomb potentials). If your initial molecule is distorted, this suggests either poor internal (bonded) parameters or assignment of charges, which lead to overly strong intramolecular attraction.

Hi Jian
I have a similar problem, how were you able to solve this?