Reaction coordinates in NEB

[kambiz]

Hi dear all,

I m checking how my NEB (compiled with vtst codes) is running. Using "nebef.pl" I got this:

0 0.00000000 -240.87926600 0.00000000
1 0.65641500 -240.49116500 0.38810100
2 3.31633700 -239.57548200 1.30378400
3 6.88612400 -239.14187900 1.73738700
4 5.65225700 -237.74746800 3.13179800
5 0.00000000 -242.46581600 -1.58655000

Does the second column represent the reaction coordinates? Should I increase the number of images?

Thanks
Kambiz

[graeme]

The second column is the force projected along the path -- it is not the reaction coordinate.

Check the neb.dat file for more information including the distance between images. This distance along the path is suitable for a reaction coordinate.

You should also check the force on image #4. If the force on that image is going to zero, you have a good saddle. If you only care about the saddle, then you don't need to add images. On the other hand, if there is a problem with convergence of that image, you should add more images to better define the tangent along the path. But unless you have a complex path, using 4 images is a good start. That said, a barrier of 3 eV is pretty large. So (without seeing your calculation) it is either not converged or there is something unusual going on. Check the movie along the neb to make sure that it's reasonable.

[kambiz]

Dear graeme,

Thanks for your answer.

My NEB hasnt converged yet, but it seems that there is a problem in Images #3 and #4.

This is neb.dat results:

0 0.000000 0.000000 0.000000 0
1 0.000000 0.380322 -0.667939 1
2 0.000000 1.100327 -2.421587 2
3 0.000000 1.846962 -2.881634 3
4 0.000000 5.433248 -3.710341 4
5 0.000000 -1.586550 0.000000 5

This is the NEB part of my INCAR:

ICHAIN = 0
LCLIMB = .FALSE.
IMAGES = 4
SPRING = -5
LSCALAPACK = .FALSE.
EFIRST = -240.8744
ELAST = -242.4604
IBRION = 3
POTIM = 0.2

Can it be because of "POTIM" or "EFIRST/ELAST"?

I really appreciate your help.

Thanks
Kambiz

[cenwanglai]

have you compiled the vtstool code to the vasp?

[rgarcia87]

The problem comes from your POTIM parameter, which has a different physical meaning for IBRION = 3 than for IBRION 1 or 2.

For the decomposition of organic molecules on metal surfaces and IBRION=3 flag, either for relaxations or NEB calculations, values between 0.007 and 0.025 are usually safe and robust, and also converges faster than IBRION=1. That is, POTIM should be more than one order of magnitude lower than for IBRION 1 or 2. Do your own tests to find the optimal value for the system you are working with.