EDIFFG = -0.1 is enough for CNEB to find the barrier?

[cenwanglai]

I have read the paper Theoretical Calculations of Dissociative Adsorption of CH4 on an Ir(111) Surface authored by Graeme in PRL. In this paper, dimer method is applied basing on NEB calculation.

So, why not CNEB was employed to find the saddle point? As a starting point, what is about the covergent criterion? A coarse (EDIFFG = -0.1 for example) NEB run is enough for further dimer calculation?

My CNEB has been converged by LBFGS at EDIFFG = -0.1.

I just need energy barrier.

So, do I need to move on the NEB with lower EDIFFG such as -0.05?

For a typical geo optimization,the EDIFFG is set to be -0.03 with a single K point.

When the sadle point is found, any further geo optimization is really needed?

[graeme]

The calculations for the PRL were done before the climbing image NEB was implemented in vasp (despite the order of the publications). It would be appropriate to have used the CI-NEB method for those calculations.

A force convergence criterion of 0.1-0.01 eV/Ang is generally suitable for saddle point calculations. If you want to use the NEB for a rough calculation of the path, the force on the climbing image should be on the order of 1-0.1 eV/Ang before starting a dimer search.

When the saddle is found to within your required force criterion (with whatever method), there is no need to do subsequent calculations. Well, perhaps with more k-points or a higher energy cut-off.

[kai]

i have to add some here regarding this since i was puzzled at this for a while;

the simple anwser is: it depends;

i did comparable calculations using CINEB for C2H5Br and C4H9Br on Silicon and found that the TS geometry converged for C2 at ediffg=0.1, but not for C4. C4 converges when ediffg=0.02 is reached.

I would argue it is a good pratise to concentrate on the saddle point a bit more, and perhaps step-by-step.