NEB calculations

[valeria_butera]

Hello,
I am dealing with a system of which consists of a CO2 molecule adsorbed on a Ru-MoS2 surface. I am interested in calculating the energy barrier involved in the breaking of a C-O bond (with consequent formation of O and CO, both adsorbed on Ru atoms deposited on MoS2 surface monolayer).
I generated 5 images using the nebmake.pl script implemented by the Henkelman group. However, starting from the generated POSCARs, during the relaxation one oxygen atom was moved very far away from the surface while the C was located at the bottom of the surface between Mo and S atoms. Therefore, I modified manually the initial POSCARs and run again the job.
After 150 NSW, the calculation did not reach the selected convergence criteria, so I copied the CONTARs into POSCARs and re-run. I did that twice, but still the convergence was not met. Then I increased the EDIFGG to -0.1 (in the initial INCARs it was set to -0.01).
It seems to me that the "problem" is the convergence of the forces on the last four atoms of the POSCARs (one Ru and the CO2 molecule).
I have attached the OUTCAR file of one of the images (the image 2 which is supposed to be the closest to the TS), all the POSCARs and the INCAR file.
Any suggestion would be really appreciated.
thanks

OUTCAR.tar.gz

1.tar.gz

2.tar.gz

[andreas.singraber]

Hello Valeria Butera!

Looking at the files you sent I found two potential origins of the convergence problem:

1.) I would strongly recommend to change IBRION to 1 or 3. The CG algorithm (IBRION=2 in your INCAR) is not suitable for the NEB method as described on the Wiki page for the IMAGES tag.
2.) The same Wiki entry also mentions that starting with a low number of images is beneficial to get better convergence.

Could you please retry your calculations with IBRION=1 and a lower number of images (only 1 or 2 maybe) and see if that avoids the convergence issues?

All the best,

Andreas Singraber