[lammps-users] Out of range atoms-cannot compute PPPM (src/KSPACE/pppm.cpp:1912) when equilibrated Na2S system using Buckingham potential

Der LAMMPS users,

The LAMMPS what I am using is development version (8 April 2021) and the msi2lmp is v3.9.9. The molecular geometry of Na2S and 1,4-Dichlorobenzene were optimized by Gaussian 09.

I met some questions about bond atoms missing when equilibrate the mixture of Na2S and 1,4-Dichlorobenzene. So, I step back to equilibrate Na2S and 1,4-Dichlorobenzene respectively. Both data were prepared by msi2lmp. The equilibrium process of 1,4-Dichlorobenzene is ok but met the error “Out of range atoms-cannot compute PPPM (src/KSPACE/pppm.cpp:1912)” when equilibrate Na2S system.

The length of Na-S calculated by Gaussian was 2.4 Å which is 2.48 Å in the initial structure (as the picture attached). After energy minimization, the bond length of some Na-S was expanded to 4.25 Å and even to 5.19 Å. The temperature of system raised from 330 to 17471.53. I know it is unnormal. The Buckingham potential parameters of Na2S were cited from web (https://www.ucl.ac.uk/klmc/Potentials/Library/diller.lib) and the Mulliken charges were extruded from Gaussian calculation, was set as the partial charges.

The input as follows and the data, and visualized figure (before and after) as attached.

I know I made some obvious and foolish mistake but I do know how to address this problem. Help me.

Yuanqi Jiao

The input file:

Na2S-pure-5.data (2.57 KB)

Your problems are primarily due to your (let’s be polite and call it) “cavalier” approach to force field parameters and force field parameterization.

You have taken parameters that were apparently derived for core-shell model simulations, grafted on your own partial charges and then combined it with some completely different parameterization. the fact that you have optimized initial structures mean nothing.

Force field parameterization is far more complex, and in most cases you need one consistent set of parameters that can be applied to the entire system and have been parameterized to represent as many thermodynamic or material properties as possible as well as the interactions between multiple components.

Before continuing I suggest you carefully study the publications of several force fields, so you get a better sense of what I am referring to.


Dear Dr. Axel,

I made such stupid mistake that mixing the partial charges, Buckingham potential parameters for core-shell models and I am really sorry for that. I find another paper (DOI: 10.1021/acs.jpcb.8b04353) kindly provided the Buckingham potential parameters that seems to suit my system. Another approach is the Dreiding force field well cover the parameters that I need.

Thanks for your time.