gloriatab/lammps-gg

An implementation of the Mahadevan-Garofalini dissociative water potential compatible with LAMMPS

Mahadevan-Garofalini Water Potential Implementation for LAMMPS

This is the code that implements the Mahadevan & Garofalini potential in LAMMPS. This code is not very thoroughly tested and is not the code used to obtain the results in any publications put forth by Garofalini or others at the Interfacial Molecular Science Laboratory at Rutgers.

Although the numerics and overall dynamics of the LAMMPS code are consistent with the original implementation by Mahadevan and Garofalini, the density of water at 1 atm, 298 K comes out consistently higher in this LAMMPS implementation by a small amount. Using a tabulated 2-body potential generated by the original Mahadevan code with LAMMPS (pair_style table) shows the same behavior. This suggests that the difference in density is caused either by the 3-body potential in LAMMPS or some difference in the thermostat or barostat, but the exact reason has never been determined.

Nevertheless, this repository contains all of the files one would need to add to LAMMPS. The following files are included:

• pair_gg_coul_wolf.cpp and pair_gg_coul_wolf.h - the code to calculate the 2-body potential with Wolf summation
• in.ggh2o - sample input that illustrates how to combine the custom 2-body with LAMMPS' own implementation of the Stillinger Weber potential to get the 3-body interactions
• gg.sw.kcal - the Stillinger Weber parameters to get the correct 3-body interactions for the dissociative water potential. The two-body parameters are all zeroed out.
• data.knite9.0792 and data.knite9.4000 - sample water configurations to run

Inputs that exercise the SiO2 or water-silica interactions have not been created, but there should be a way to use LAMMPS's Stillinger-Weber potential to achieve the same 3-body cross-terms for Si-O-H described in the Mahadevan paper.

This code was last tested with a February 2016 (16Feb16) release of LAMMPS and installation should be a simple matter of

1. copying pair_gg_coul_wolf.cpp and pair_gg_coul_wolf.h to the src directory within the LAMMPS source distribution
2. adding #include "pair_gg_coul_wolf.h" to src/style_pair.h
3. make my_system as usual

This code has not thoroughly tested, so there may be bugs. It is intended to be a helpful starting point for implementing the Mahadevan-Garofalini Dissociative Water Potential, but it comes with no guarantees of correctness.