[lammps-users] Seeking for help regarding the dielectric command

Dear Sir

I was trying to study a charged surface and polyelectrolyte adsorption interaction. I am trying to incorporate the polarization effects by introducing different dielectric constants for the surface and the polyelectrolytes. It seems that in LAMMPS we can incorporate only one dielectric constant. Is there way to get around this problem?

Samyabrata Chatterjee

It looks to me as if you are completely misunderstanding what the purpose and effect of the dielectric command is as well as are lacking understanding in how you would go about modeling what you want to model. For both there is lots of published knowledge available and thus you should make a better effort researching this.

First problem is that the dielectric command does NOT set the “dielectric behavior” of any material, but rather the relative permittivity of the background medium. In your case it is obviously vacuum and thus coulomb interactions must be unscaled and the dielectric constant should remain on the default value.
For all-atom simulation the dielectric constant is not an input parameter but a property that can be computed.

The value for the dielectric command would be different from default only if you have some coarse grain model which have charges (many coarse grained models are uncharged include approximated coulomb interactions) and where solvent effects are not included in the particle-particle interactions.

Second problem is your plan to model the polarization of a charged interface. This is non-trivial to begin with but to also have the model represent polarization of the surface, additional calculations would need to be done. In the simplest case (and ideal metal), this would require using a so-called image charge model (not available in LAMMPS since it is notoriously difficult to implement that in parallel. It is available as an externally maintained package, which is restricted to serial calculations). There are multiple, more complex models available, e.g. the DIELECTRIC package that was recently added to LAMMPS, but also multiple, externally maintained packages, each with slightly different approaches to the same family of problems. I strongly suggest spending time studying the relevant published literature since you first need to gain a sufficient understanding of the physics before it is even worth trying to set up a simulation with any of those add-on packages. From my personal experience as a graduate student studying electrified interfaces I can assure you that there are MANY ways to make subtle errors in setting up simulations and their boundary conditions resulting in significant errors in the results, and it can be quite challenging to get everything correct.