CI-NEB problem

Vasp transition state theory tools

Moderator: moderators

Post Reply
Ali
Posts: 2
Joined: Sat May 14, 2016 12:54 am

CI-NEB problem

Post by Ali »

Dear Professor Henkelman,

I'm studying the diffusion of a Li atom on the GeP monolayer using CI-NEB method, but I have found a strange problem. I'm using the vtstcode and I can see the related parts in OUTCAR file for that but I can not see any difference between the results obtained from NEB and CI-NEB calculations. In other words, it seems that CI-NEB is not working in my calculations. I have no idea if the problems is coming form vtstcode installation or I'm doing something wrong. Below you can see the part for vtstcode in my OUTCAR file.

VTST: version 3.1, (03/28/14)

CHAIN: initializing optimizer

OPT: Using VASP QUASI-newton optimizer
CHAIN: Read ICHAIN 0
CHAIN: Running the NEB
NEB: SPRING -5.000000
NEB: LCLIMB T
NEB: LTANGENTOLD F
NEB: LDNEB F
NEB: LDNEBORG F
NEB: LNEBCELL F
NEB: EFIRST 0.000000
NEB: ELAST 0.000000


I have also attached the MEP profile created by nebspline.pl, for a CI-NEB calculation that shows the Image 2 is not located at the expected position.
I run CI-NEB calculation ans follows, Please correct me if I am doing something wrong. First I run a regular NEB calculation. After convergence, I move CONTCAR file of each 0N directories to POSCAR file. Finally I add LCLIMB = .TRUE. to the INCAR and rerun the job.
I will be grateful if you help me solve this problem.
Attachments
mep.jpg
mep.jpg (588.45 KiB) Viewed 8401 times
graeme
Site Admin
Posts: 1999
Joined: Tue Apr 26, 2005 4:25 am
Contact:

Re: CI-NEB problem

Post by graeme »

The energy path looks reasonable.

One comment: you can run the climbing image right from the start. This will be more efficient than converging a regular NEB and then turning on the climbing image.

One thing that I can not tell from what you have posted is how well converged the band is and what your convergence criterion is. If you use EDIFFG=-0.002, the highest energy image should converge very close to the saddle (top of the MEP). Check the residual force, to see if the climbing image is still converging or if you need to tighten your convergence criterion.
Ali
Posts: 2
Joined: Sat May 14, 2016 12:54 am

Re: CI-NEB problem

Post by Ali »

So many thanks for your prompt answer. I used EDIFFG=-0.02 in my calculations. Just one more question: Does CI-NEB work well with PBE-D2 or PBE-D3?
graeme
Site Admin
Posts: 1999
Joined: Tue Apr 26, 2005 4:25 am
Contact:

Re: CI-NEB problem

Post by graeme »

it should be fine.
Post Reply