NEB convergence Problem

[sungeng]

hello
I am studying the CINEB method to find the reaction barriers. I meet a convergence problem.
I use 8 images to find a reaction barrier of CO and CH to COCH, on Rh111 surface. The ends images are well converged below 0.01eV/A. But the NEB reaction which can't converge. In the Images 03~05, force oscillates at level 0.06~0.1 eV/A, I tried these methods but they all do not work.
1. incrising the ecut from 600 to 800
2. add the paramater enaug=700
3. change ibrion = 1,3 (and iopt=1)
4. change the parameter SPRING = -2 from SPRING=-5, i find two parameter give the same result. (in the attachments)
5. add the paramater addgrid= true
6. change ediff from 1E-4 to 1E-5

The attachment is my working files,which with the SPRING=-2. Same convergence situation appears when SPRING =-5

thank you for any suggestions!

[graeme]

It looks ok to me in general. The problem is likely just that you are using an aggressive optimizer (IBRION=1) with a large time step (0.3) so that it can be unstable. Switch to something like IBRION=3 and POTIM=0.1 to make sure that it converges. If it does, and you want better efficiency, try using our NEB-specific optimizers (such as IOPT=1, or reduce POTIM with IBRION=1):
http://theory.cm.utexas.edu/vtsttools/optimizers/

[sungeng]

[quote="graeme"]It looks ok to me in general. The problem is likely just that you are using an aggressive optimizer (IBRION=1) with a large time step (0.3) so that it can be unstable. Switch to something like IBRION=3 and POTIM=0.1 to make sure that it converges. If it does, and you want better efficiency, try using our NEB-specific optimizers (such as IOPT=1, or reduce POTIM with IBRION=1):
http://theory.cm.utexas.edu/vtsttools/optimizers/[/quote]

thank you for your suggestions!,I will try it