Hello,

I am attempting to use LAMMPS to calculate the chemical potential of a charged system based upon the Widom Insertion method. LAMMPS is particularly useful since it can easily calculate the long range electrostatics for this application… I hope.

Essentially, I take snapshots of an equilibrated and electro-neutral sytem, randomly insert one anion (or one cation), calculate the change in energy after 1 time step, remove the particle, and repeat. The average change in electrostatic energy upon random insertion of one particle gives me the Coulomb chemical potential.

I get the expected “System is not charge neutral…” warning. However, LAMMPS is not crashing (also expected) and is calculating what appears to be reasonable changes in electrostatic energy.

Is the resulting E_coul only short-ranged since Ewald sums aren’t valid for net charge systems? What is being calculated? Is the new E_coul reliable?

Thanks for any information you may have,

Brian

Ecoul is always just the short-range part. Elong may be slightly

inaccurate due to non-zero net charge, but will include the

extra charge for N+1 atoms. Paul may want to comment

on the accuracy of Ewald or PPPM for this scenario.

Steve

Hi Brian. First off, I agree with Steve's comment.

LAMMPS adds an extra energy term for systems that are not net neutral. See line 286 of ewald.cpp or line 696 of pppm.cpp. This is referred to as the "finite volume correction term" and "inclusion of this term leads to independence of the results from the choice of the screening charge parameter [g_ewald] for net-charged systems" (see JCP vol 108, No 17, page 7072). This term is included in elong.

Widom's method may not work right for a charged particle insertion. I haven't seen this applied correctly, and I doubt that there is an established method to do what you are trying to do here. But someone else on the list might be able to point you to a literature reference where the method has been established. David Kofke's work on the subject is a good place to start: Fluid Phase Equilibria 228-229 (2005) 41-48, Molecular Physics, vol 102, No 4, 405-420. If I recall, he only dealt with net-neutral systems and insertions.

Paul