I’m simulating a system of toy surfactants which form micelles in an electric double layer (EDL). The simulation is 2D periodic and charged, so kspace_modify slab is used. At the bottom of the box is a charged wall. At the top is a fix wall/harmonic boundary condition. When I do my analysis, I notice a substantial increase in concentration of micelles adjacent to the wall. It looks like the particles are just slowing down and kind of sticking to the wall. There’s no clear physical explanation for this; I’m wondering if this is some artifact of the simulation. Since the solvent is implicit, I’m using fix langevin.
Has anyone experienced anything like this before? I’m including a simplified version of my script, and a picture of my simulation to make things a little more clear.
If kspace_modify slab is used two-thirds of your simulation cell (in z-direction) should be empty and particles only in the center of that range. The Poisson solver used to decouple the dipoles between periodic boxes in z-direction will give wrong results otherwise.
Bottom line, if your picture is representative then your entire simulation is bogus.
If the pink wall is negatively charged and the yellow counterions are positively charged, then shouldn’t you expect the yellow counterions to accumulate near the wall simply by Coulombic attraction? This seems like a classic electric double layer (specifically a Gouy-Chapman model with an exponential density profile).
By the way - you should use extra counterions to compensate the micellar charge so that you have an overall neutral simulation. A charged simulation box with long-range interactions is really one with a continuum countercharge “smeared” across the box with uniform density. This is fine in a homogeneous bulk system, but in an inhomogeneous system like yours the effective permittivity is different throughout the solvent. (Because of the double layer!)
So you should expect compensatory charge to be inhomogeneously spread through the system as well, and the simplest thing to do is just add the extra charges in, since you would expect any region of the real-life system to be electrically neutral and have stoichiometrically matched numbers of oppositely-charged species on average anyway.
@akohlmey I don’t think the extra “void space” gets added into the output zlo and zhi - it’s just added on internally in the kspace calculations (which is fine since no particles should ever get into the void anyway)
Thanks so much. What you’ve said is exactly what I’ve done; there are more counterions than surfactants to compensate for the charges at the wall, so the system is neutral. This is exactly an electric double layer. As you can see, counterions collect near the charged surface while micelles (coions) are excluded from the region near the wall.
There’s the possibility that this concentration bump is entirely a consequence of the harmonic wall b.c. If this is the case, it shouldn’t have a meaningful effect on the results of the simulation if the harmonic wall is far enough away. Figuring this out would be a science problem, not a LAMMPS problem, so I’ll handle it on my own. I just wanted to make sure this wasn’t some obscure consequence of an improperly utilized kspace_modify slab function.
I was a little confused by the wording in the documentation. It wasn’t clear whether the extended dimension mentioned was vacuum.
I need to clarify the observation you are concerned about. Is it the case that: the micelles are less concentrated near the low-z charged plate, and sticking to the high-z boundary? (I have avoided “wall” because it’s being used to describe both low- and high-z surfaces.)
It’s worth considering – the high-z boundary is also planar and hence fundamentally different from a bulk environment for all constituents. There may be some kind of surface effect that’s beyond my pay grade.
What I do know well is how the slab correction algorithm works: it calculates the total box dipole and subtracts the dipole field from neighbouring z-images, approximating the field as a dipolar slab (hence “slab correction”). Your box is, to be completely technical, not very slabby. So it is possible that a low kspace accuracy has resulted in an incomplete slab correction, leading in turn to neighouring images still “feeling” a residual dipolar alignment, which would manifest as micelles being attracted to the overcharged ionic later “one box up”.
To eliminate that possibility you should try a higher kspace accuracy (I’d recommend starting at 1e-5 and working your way down). If you’re still unsure, increase the slab correction from 3.0. And you may also want to consider making a “nanocapacitor” structure with negative and positive plate charges on either end - if you’re going to throw the top away anyway, at least make it a definite and understandable subsystem rather than something you’re less confident about.
