Modeling interlayer LJ interactions with full periodic boundary conditions for single CG sheet

LAMMPS LAMMPS General Discussion
1

Hello all,

I am working on a coarse-grained (CG) model of single-layer materials (similar to graphene) and computing the layer and multi-layer mechanics. In these models, it is standard to:

  • exclude nonbonded interactions within the same sheet
  • include Lennard-Jones (LJ) interactions only between different sheets

This is consistent with literature I refer to (e.g. Ghazanfari et al. 2022, Ruiz et al. 2015)

1. My goal

I want to simulate a single CG sheet consisting of 1 type bead under fully periodic boundary conditions (PBC in x, y, z), while enforcing:

  1. No LJ interactions between CG beads within the same sheet, but governed by bonded interactions
  2. LJ interactions between periodic images along z (i.e., mimicking interlayer interactions)

Or in other words, intra-sheet interactions are only boned interactions, out-of-plane interactions are only LJ interactions. Specifically, for single layer model under PBCs, the out-of-plane interactions cross boundaries in Z direction.

2. Problem

I tried to define different layers as different types so LJ interactions among the types (i.e. layer) can be specified. BUT The single-layer CG model only has 1 type bead.

As far as I know, in LAMMPS, pair interactions do not distinguish between:

  • same-sheet neighbors
  • periodic images of the same sheet

So there are technical problems:

  • If I disable LJ for same-type pairs → no interlayer interactions exist → box collapses along z under NPT
  • If I enable LJ → unwanted intra-sheet LJ interactions appear

And subsequently:

3. Current workaround

For multilayer systems (e.g., 10-layer tactoid) under full PBCs, I assign different “types” to each layer and define LJ only among different types. This works and show good structure as Ghazanfari et al. 2022. It means the structure, pair- and bond- coefficients should be correct.

10-layer model relaxation
10-layer model relaxation871×733 62.8 KB

In fact, even a bilayer model also works well, only if the LJ can exist.

However, for a single layer with full PBC, this approach does not apply. In my work, full PBCs are necessary to be applied.

4. Question

Is there a recommended or standard way in LAMMPS to:

  • simulate a single layer system with 1 type CG bead
  • use full periodic boundary conditions
  • enforce “no intra-layer LJ but interlayer LJ across periodic images”?

Or is the correct approach to instead construct a minimal multi-layer unit (e.g., two layers with different atom types) to represent the periodic system?

5. About Attachment

I can provide minimal input script and force field file if needed. But I think the issue appears to be conceptual rather than numerical.

Thanks for your help!

2

You cannot do meaningful dynamics with just one periodically replicated layer. The atoms will interact with themselves along the short dimension. Please also note that there is the INTERLAYER package with some potentials that can be used for adjusting potentials with a spherical cutoff to model layered materials. Most of them are tuned for specific systems, so you have to read the corresponding publications to find one that is generic and can be parameterized for your specific needs.

3

Hi Axel, thank you a lot for your reply. It is very helpful for my later modelling. I will review more about the periodic boundary conditions and modelling constrains.