[gmx-users] Diffusion of Ammonia Gas

[Mark Abraham]

Darrell Koskinen wrote:
> Dear GROMACS-ites,
> I am trying to determine whether OPLS can be used to model gas-phase 
> ammonia. To this end, I created a 30 nm^3 box with 729 ammonia molecules 
> in it, performed energy minimization, and ran a simulation at 298K & 
> 1.01325 bar for 20 ps followed by another simulation of 20 ps duration 
> without temperature or pressure specified. I then ran g_msd on the 
> output .trr and .tpr files. The output from g_msd was a diffusion 
> constant of:
> 
> D[       NH3] 1755.8405 (+/- 1386.3837) 1e-5 cm^2/s
> 
> I am first wondering if the simulation length was too short as the error 
> seems to be quite large if the figure following the +/- is the error.

I'd guess 20ps is at least one order of magnitude too short. Some 
literature searching might be useful to see how long other people have 
run for making similar observations. A necessary but technically 
insufficient criterion for convergence is that sufficiently long chunks 
of your run agree with themselves, so you may wish to look at the output 
of g_msd on 1ps fragments towards the end of your existing 20ps, or such.

Mark

> Secondly, the value of 1755*1e-5 cm^2 /s is an order of magnitude 
> smaller than the value of 0.2*1e-4 m^2 /s, reported in 
> /"Landolt-Börnstein IV/15A, Chapter 3: Diffusion in Pure Gases" & 
> "TEMPERATURE DEPENDENCE OF SELF-DIFFUSION COEFFICIENTS FOR GASEOUS 
> AMMONIA by Charles E. Baker"/. Therefore, I am also wondering if I am 
> performing the simulation correctly and running g_msd correctly.
> 
> Please let me know if there is any additional infomation required for 
> you to make a determination of whether I am running the simulation 
> correctly and thus showing that ammonia as represented in OPLS cannot be 
> used for a gas phase simulation or whether I have made a mistake 
> somewhere in my runs.
> 
> Much thanks.
> 
> Darrell