[gmx-users] Pseudo 2-D Ewald Summation and Slab Geometry

[Andrew DeYoung]

Hi, 

I am simulating a capacitor (in which the dielectric is liquid between the
plates), with charged electrodes parallel to the xy plane.  Because the
system is effectively two-dimensional, I am using ewald_geometry = 3dc,
which applies a force and potential correction in the z dimension, giving a
pseudo 2-D Ewald summation.  

However, I observe a drift in the electric potential between the plates.
Specifically, I use g_potential to solve the Poisson equation and determine
the electric potential as a function of the z coordinate.  The liquid has a
structure such that an electric double layer is formed, and this cancels off
much of the electric field due to the charged electrodes.  So, I would
expect that the potential between the plates is constant (i.e., a plot of
the electric potential Phi versus z is expected to be flat).  This is in
fact what I observe when I simulate the same system using another MD
package, using exactly the same force field parameters and the analogous
(same correction equations, just different code) force and potential
correction in the z dimension to produce a pseudo 2-D Ewald summation: in
that other MD package, the electric potential is constant (flat) between the
capacitor plates.  But in Gromacs (4.5.5), the potential is not constant
(flat) between the plates; rather, it is linear with a non-zero slope.

Has anyone else experienced such a "drift" in the electric potential when
using a slab geometry and ewald_geometry = 3dc?  Do you have any ideas of
what might cause this and how I can fix this?  

Thank you very much.

Andrew DeYoung
Carnegie Mellon University