CHARMM Development Project Forums User Discussion & Questions General Chemistry Discussions self-diffusion of a solute

Dear All,

I'm hoping someone can answer this problem.

I'm busy calculating the translational diffusion coefficients of several solutes using MSD. Having obtained the MSD I have to fit a straight line and obtain the gradient, from which the diffusion coefficient is calculated.

In the fitting of the straight line, do I make the fit through the initial collision region of the MSD or do I ignore this region and only fit from the steady state region?

Regards
Rich

Ignore the inertial region in the beginning.

Dear Prof.

Thank you for the quick response.

Regards
Rich

For comparison to Dt from experiment, there are two major additional factors: a finite size correction (1), and for many water models, a viscosity correction. We've done both for disaccharides in TIP3P water recently (2).

(1) Yeh, I. C.; Hummer, G. J. Phys. Chem. B 2004, 108, 15873–15879.

(2) Venable, R.M.; Hatcher, E.; Guvench, O.; Alexander D. MacKerell, A.D. Jr.; and Pastor, R.W. J. Phys. Chem. B 2010, 114, 12501-12507.

An additional note-- one cannot use COOR ANAL for this; you must first use MERGE COOR UNFOLD, then get the center-of-mass time series for each molecule via the ATOM time series type in CORREL, followed by CORFUN with the DIFF keyword.

For a rectangular box I believe COOR ANAL handles PBC for the MSD calculation. Shouldn't COOR ANAL work on UNFOlded trajectories?

COOR ANAL does not seem to provide a means of getting the center-of-mass (COM) motion for larger, more flexible molecules. It appears to be limited to small, rigid solvent molecules, where a single atom is a useful surrogate for the COM, so it's okay for water and ions. However, I could not get reasonable results (identical to rigorous COM results) for neat heptane and pentadecane using COOR ANAL.

Yes, this could be a problem; if this is a common type of analysis, and the use of a single atom as a surrogate for the COM is really bad we should add the COM option to COOR ANAL (and extend its PBC awareness).

Dear Prof Nilson and Dr Venable

Thank you both for your comments. I will look at these and the references before continuing with my Dt calculations.

Regards
Rich

An additional note-- in order to apply the finite size correction of Yeh & Hummer, the simulation viscosity is required, a non-trivial task. Changes to c35b1 and later versions have made that somewhat easier; in brief, the pressure tensor from every integration is required, from either fairly long or replicate NVT simulations. The correction becomes small for larger systems, such as a cube of over 100 A edge, but I don't know the size where it becomes negligible.

Contact me via email if you are interested in a copy of ref (2) above, or copies of the scripts I used.

Dear Dr Venable,

Thank you for the additional note. Please can you inform me on where to obtain your email address to enquire about the input scripts.

Regards
Rich