[gmx-users] Dispersion correction in a heterogeneous

[Mikhail Stukan]

Berk,

Thank you very much for the answer. It was not clear from my post (sorry for that) but my main concern was indeed related to the inhomogeneity near the surfaces. I am not targeting complete quantitative agreement, but I would like to be save from "hidden" undesired qualitative effects. As far as I know, the dispersion correction contributes something like 2-5% to the surface tension value (depending on the cut-off). If I consider different salt concentrations then the inhomogeneity near the surface will be of different order (compare to the bulk). So taking into account your recommendation, long cut-off with no dispersion correction looks like a better option.

Thanks and regards,
Mikhail


>Message: 5
>Date: Mon, 4 Oct 2010 10:16:35 +0200
>From: Berk Hess <gmx3 at hotmail.com>
>Subject: RE: [gmx-users] Dispersion correction in a heterogeneous
        system
>To: Discussion list for GROMACS users <gmx-users at gromacs.org>
>Message-ID: <COL113-W10EA30F423207830F450CE8E6C0 at phx.gbl>
>Content-Type: text/plain; charset="iso-8859-1"


>Hi,

>I think you are looking at the wrong issue.
>Unless your concentration is ridiculously high, the dispersion heterogeneity will be irrelevant.
>Furthermore, at distances were the correction works, the distribution will be close to homogenous.
>But you do have an issue with dispersion correction because of strong spatial inhomogeneity because
>of the interface. The dispersion correction will have no direct effect on the surface tension.

>What is the best way to proceed depends very much on the questions you want to answer.
>The surface tension of many water models is off by 30%, so you probably won't get quantitatively
>correct results anyhow. (note that also many "standard" combinations of Na and Cl parameters are very bad)
>If you want an accurate number, given the force field limitations, you need to use a long LJ cut-off,
>possibly with twin-range settings for efficiency. But force fields often have not been parametrized with this.
>And finally, ion C6 parameters do not give realistic dispersion strengths in most cases, but as  I said,
>this effect will be negligible in normal concentration ranges.

>Berk

>>From: mstukan at slb.com
>>To: gmx-users at gromacs.org
>>Date: Sat, 2 Oct 2010 12:40:23 +0000
>>Subject: [gmx-users] Dispersion correction in a heterogeneous system


>>Dear gmx-users,



>>Although  this task has been already discussed few
>>years ago (http://lists.gromacs.org/pipermail/gmx-users/2007-January/025668.html)
>>the full summary is not clear to me. So I would really appreciate if somebody
>>could give me advice on the following subject.



>>I am trying to simulate an aqueous solution of NaCl. One of
>>the properties I am interested in is surface tension, which means that
>>correction to the pressure could be quite important.  But the system is
>>heterogeneous: values of C_6 differs from 0.0003 (for Na-Na) to 0.0180 (for
>>Cl-Cl), so they are not comparable. Is it worth to use dispersion correction in
>>such a system? Or such results will have no real meaning and the proper way would
>>be to switch off the dispersion correction and just increase cut-off (or
>>perform rerun with higher cut-off)?



>>Thanks you very much in advance,

>>Mikhail