calculating migration barrier energy using selective dynamics

Message

hyon-jee · #1 Post by **hyon-jee** » Thu Jan 19, 2006 11:25 pm

Hi,

I'm calculating a diffusion barrier of an atom to a vacant site.
Since I know the migration path, I move the target atoms along this path and do atomic relaxation of other surrounding atoms.

To get the energy of such transition state, I use selective dynamics, making the target atom non-movable (F F F) but other atoms movable (T T T) so that they can relax around the target atom.

What I realized is that the center of mass is not preserved in this selective dynamics case, thus all other surrounding atoms shift in some cases, making migration of target atom meaningless.

How do I get around that problem?
Thanks!

tjf · #2 Post by **tjf** » Fri Jan 20, 2006 1:11 pm

You need to fix another atom as well. If that vacancy were an atom, that's what you'd fix.

Is there another atom on the other side of the vacancy that may act as an anchor point?

Note that doing such a scan (with a few fixed points) will probably give different energy profiles depending on your anchors. Bearing that in mind, I think you should look very closely at using NEB instead.

<span class='smallblacktext'>[ Edited Fri Jan 20 2006, 02:52PM ]</span>

#3 Post by **admin** » Fri Jan 20, 2006 2:07 pm

is there any objection to use the nudged elastic band method?
I fully agree with tjf that this is the method of choice for problems like yours, because 'doing it by hand' may easily lead to a diffusion path which deviates from the energy minimum path.

hyon-jee · #4 Post by **hyon-jee** » Mon Jan 23, 2006 7:59 pm

When the diffusion path is well defined, NEB seems like a waste of time.

Then it is true that the Center of Mass is not conserved when one is using a selective dynamics?

Is there any way that I can force the conservation of center of mass?

tjf · #5 Post by **tjf** » Mon Jan 23, 2006 9:30 pm

[quote="hyon-jee"]When the diffusion path is well defined, NEB seems like a waste of time.[/quote]not[/i] well defined.

NEB is your friend here. It will do the interpolation in a very well defined and physically relevant way. What you're asking for with the fixed CoM is an amazingly specific type of constrained dynamics which is (almost) completely meaningless in the context of a 3D periodic calculation. You're asking for invariance of energy with translation not to apply!