Problems with CNEB calculatios

[dayharrisv]

Hello everybody,

I'm learning how to do neb calculations and I decided to start with a reaction of water adsorbed in mordenite. The issue is that when I do the calculations, I obtain 6 extremas (I used 4 images from nebmake.pl). I've seen in other posts differents files of mep.eps and they look so much better than mine (I left the file attached in this post).

I Wonder if someone could help me in this topic.

The incar I use is this:

System = neb agua

LCHARG = .FALSE.
LWAVE = .FALSE.

Electronic minimization
PREC = NORMAL
GGA = PE
EDIFF = 1E-6
ALGO = Fast
LREAL = Auto
ENCUT = 400
NELM = 40

NEB (set IBRION = 3 and POTIM = 0)
LCLIMB = .TRUE.
IMAGES = 4


Ionic relaxation
NSW = 400
IBRION = 3
SPRING = -5
POTIM = 0
ISIF = 0

DOS related values
ISMEAR = 0
SIGMA = 0.1

Parallelization etc.
NPAR = 16



PD: I apologize for my english, I'm not a native.

[hakuna]

it seems ok, just a bit over fitting,

my opinion only

[dayharrisv]

Thanks for answering. I'm trying to find a transition state but in this case I haven't have success. I calculated the vibrations frequencies of image 3 (that I supposed could be close to the TS, but I couldn't find any negative vibrational mode). So, right now I'm optimising the structure of image 3 looking forward for the right TS structure.

I wonder if you have any recommendations.


Also, I would like to know what that changes in curvature mean.

[graeme]

My guess is that you have not linked in our vtstcode into vasp. The graph that you show appears to have zero projected forces at each image, which is why the spline is so wiggly and you are getting extra extrema along the path. Start by greping your OUTCAR file for VTST - if you don't see a line saying the version, this is the problem.