NEB Energy linearly extrapolates between initial and final

[Madeline_928]

Hello,

I've been attempting to run an NEB calculation using the attached files. Below are some of the suggestions that I've taken from here to attempt to get relative energies that are higher than the initial and final relative energies (the expected trend according to literature for H2S dissociation on a Cu2O surface):

1) I've attempted the NEB method by both the standard VASP package and the energies trend towards or converge to energies that fall between the initial and final energies...I can see that it's linearly decreasing from the initial to the final for each image.

2) I've attempted the NEB method by using the IOPT 2 and 3 optimizers and it does not converge. I've also looked at the forces that correspond with the energies. The lower the force (i.e. the closer it gets to the EDIFFG setting), the TS energy again falls below the initial.

3) I've kept my ENCUT and KPOINTS grid consistent with the initial and final configuration settings.

4) I've added ISYM to initial, final, and transition state calculations - same issue occurs.

5) It seems as if I'm having a system migration, so I fixed elements that were not involved in the dissociation of the hydrogen element - same issue occurs.

6) I've attempted to add more intermediate images between - same issue occurs.

I know there are different force based optimizers available with the vtst package, but shouldn't the overall trend still give a maximum saddle point regardless of the the force based optimizer used?

I've attached in progress calculations for a standard VASP package NEB run and a vtst package NEB run with IOPT 2. Any and all feedback/help is greatly appreciate.

Also a side note: I terminated the jobs once I saw that the energy would pass the initial energy as I did not want to keep using up resources when I knew the energy would end up converging to a lower energy value.

[graeme]

I would change a number of things:

1. The POSCAR is 00 is not oriented in the same direction as the other images. Start with new initial and final configuration which have the same cell (not allowed to relax).

2. Reduce the vacuum gap to 8 Ang.

3. Start with a rectangular cell.

4. Check that the initial set of images that you generate for the NEB view as a smooth movie of the reaction you are interested in.

5. Use regular potentials; Cu instead of Cu_pv and O_s instead of O.

6. Reduce your cutoff to 280 eV.

7. Reduce from a 3x3x1 to a 2x2x1 kpoint mesh.

8. Use LREAL=Auto

9. Check NCORE; a value of 66 seems too large.

[Madeline_928]

Hello,

Thank you so much for your suggestions! I made some adjustments (per your recommendations) and I was able to get it to converge. I also did more reading on previous forums where you've helped with convergence issues and I switched to IOPT3 and it converged!

[graeme]

Awesome, thanks for the feedback!