With very large eps and sigma potential is suddenly repulsive

GROMACS version: 2018-2021
GROMACS modification: No
Dear all,

When having a large epsilon and sigma value (eps > 10, sigma > 2.0 nm) the distance where the potential is not repulsive any more is increased. A pbc molecule (a uniform polymer block in vacuum crosslinked through pbc) needs to be at least 4xsigma away from a “wall” (a frozen uniform block of beds in vacuum) to not lead to a too high potential in mdrun’s first step.

What also happens is, that the potential is not attractive anymore. I tested this by pulling the polymer block towards the surface, and the LJ potential is never getting negative.

I am using normal vdw treatment, md integrator in NVT. Periodic only in xy. Using a 12-6 LJ wall at z=0 and z=zmax which has sig,eps=0.0. Very basic setup, tested very different setups too. changed the cutoff, cutoff modifier, rlist, verlet-buffer.

What am I missing? How can this setup lead to a repulsive potential, where the LJ energies never go below zero?

The forcefield.itp is:

[ defaults ]
; nbfunc comb-rule gen-pairs fudgeLJ fudgeQQ
1 2 no 1.0 1.0

[ atomtypes ]
;name at.num mass charge ptype eps sig
P 1 0.0 0.000 A 10.0 2.0
W 1 0.0 0.000 A 10.0 2.0

[ nonbond_params ]
P W 1 10.0 2.0

[ moleculetype ]
; Name nrexcl
Surface 0
[ atoms ]
1 W 1 W W 1 0.0 10000.0
2 W 1 W W 1 0.0 10000.0

[ moleculetype ]
; Name nrexcl
Surface 3
[ atoms ]
1 P 1 P P 1 0.0 10000.0
2 P 1 P P 1 0.0 10000.0


A issue may be in the synthax directive [atomtypes], the value of sigma should come before epsilon.

the syntax of is [atomtypes]
atom type; m (u); q (e); particle type; V; W;
where the non-bonded parameters consist of the van der Waals parameters V ( c6 or σ, depending on the combination rule) and W ( c12 or ϵ)

I hope it helps

w o w… yes… indeed. Thank you very much Alessandra for taking a closer look!