[DFTB-Plus-User] self-diffusion coefficient

[. Juita]

Dear Jan,

Thank you so much for your explanations.
I calculated the diffusion coefficient byplotting the MSD from the trajectories files with VMD software and then the slope of MSD versus time plot is used to calculate the diffusion coefficient. 

If the atom coordinates move out of the cell (the coordinates are higher than the box size), do you think that the image does not cause the mistake in the self-diffusion coefficient calculation as the particles go from one side of the periodic boundary to the other? 

I am using Andersen thermostat to maintain temperature at 300 K, as can be seen in the attached input file.  I am not sure if my dynamics are right. My system consists of H, N and O. It would be great if you can give some advise on the issue.
Your assistance would be greatly appreciated.

Thank you so much for your assistance.
Best regards,
Juita

________________________________________
From: DFTB-Plus-User <dftb-plus-user-bounces at dftb-plus.info> on behalf of Jan M. Knaup <janknaup at gmail.com>
Sent: 10 February 2014 20:41
To: User list for DFTB+ related questions
Subject: Re: [DFTB-Plus-User] self-diffusion coefficient

Dear Juita,

first of all, if you want to calculate the self-diffusion coefficient
from the MSD, which is probably what you want, you must not enforce the
atoms to stay within the central periodic cell. The coordinate
modification of "wrapping the atoms around" creates an upper bound to
the MSD, making it useless for calulcating diffusion. The standard
behavior of DFTB+, i.e. let the atom coordinates wander out of the
central cell, is implemented exactly to avoid this problem.

Concerning your specific problem, it is difficult to analyze since you
do not describe the procedure how you calculated the diffusion
coefficient at all. I can make a few general remarks though:
The MD restart frequency of 100 seems extremely high. You might be
encountering aliasing effects
Are you sure your dynamics are right? In systems containing H, 1fs is
definitely the upper limit.
Are you using a thermostat? If yes, then that alone may make the
difference you observe, depending on your settings. Cf.
Alister James Page, Tetsushi Isomoto, Jan Michael Knaup, Stephan
Irle and Keiji Morokuma
Effects of Molecular Dynamics Thermostats on Descriptions of Chemical
Non-Equilibrium
J. Chem. Theory Comput., 8,4019 (2012)
DOI: 10.1021/ct3004639 <http://dx.doi.org/10.1021/ct3004639>

For diffusion, I find it best to equilibrate the system to the target
temperature using a thermostat, then continue in NVE mode to get the
actual dynamics.
Finally, depending on the length of your simulation, you need to make
sure you exclude any heat-up phases etc. and also use the positions at
the beginning of the trajectory segment you analyze as the reference for
calculating the MSD.

I hope that helps,
Jan


On 02/10/2014 12:02 AM, . Juita wrote:
>
> Dear DFTB+ users,
>
>
> I am calculating the self-diffusion coefficient from the trajectories
> obtained from the MD simulation with SCC-DFTB method. However, I get
> the value 2 orders of magnitude lower compared to the literature
> value. I checked the space and time unit, unit conversion and it seems
> that everything is right.  The system is periodic with 10 A box size.
> In the input file, I specified timestep as 1 fs, MD restart frequency
> of 100,  and the parameter sets from "3ob" as I have H, N and O in my
> system. To check the problem, I also carried out simulation of water
> as the self-diffusion coefficient of water is reported well in the
> literature, but I still have the same problem.
>
>
> The issue might be caused by the wrapping or image effect. I tried to
> modify the coordinate geometry so that all atoms were inside the box
> (within 10A box size) and re-run the optimisation. The coordinate
> after convergence, however, shows that some atoms were outside the
> box, eventhough I have specified the box size in the input file.  May
> I know how to avoid this wrapping or image effect?
>
>
> I am wondering if there are any other reasons for the discrepancy in
> the self-diffusion coefficient. I would appreciate
> if anyone can assist me with this problem.
>
>
> Thank you so much for your assistance.
>
> Best regards,
>
> Juita
>
>
>
> _______________________________________________
> DFTB-Plus-User mailing list
> DFTB-Plus-User at dftb-plus.info
> http://www.dftb-plus.info/mailman/listinfo/dftb-plus-user


--
Jan M. Knaup                      | Fon +49-(0)421-218-62351
Dipl. Phys. Dr. rer. nat.         | Fax +49-(0)421-218-62770
Universität Bremen - BCCMS        |
Am Fallturm 1                     | Jan.Knaup at bccms.uni-bremen.de
28359 Bremen                      | JanKnaup at gmail.com
Germany                           | www.bccms.uni-bremen.de

_______________________________________________
DFTB-Plus-User mailing list
DFTB-Plus-User at dftb-plus.info
http://www.dftb-plus.info/mailman/listinfo/dftb-plus-user
-------------- next part --------------
A non-text attachment was scrubbed...
Name: dftb_in-330ps.hsd
Type: application/octet-stream
Size: 6722 bytes
Desc: dftb_in-330ps.hsd
URL: <http://www.dftb-plus.info/pipermail/dftb-plus-user/attachments/20140211/c6f9a89d/attachment.obj>