Issues with minima and low energy barrier in NEB

[tanmoy]

Dear Prof. Hekelman,

I am doing some vacancy migration calculations in pure Ta system using NEB. The literature values of which are around 0.75 eV. I tested with 3x3x3 and 4x4x4 supercells and the results are varying too much as 0.61 and 0.44 eV resp. I have attached here the full calculations for 4x4x4 supercell.

First question is I am not getting why the results change too much depending upon the supercell size?
Second, I found a weird minima during spline fitting (which is unexpected, the endpoints are fully relaxed) for 4x4x4 supercell (see the mep.eps). Could you please help me a bit regaridng this. I might be doing something really silly.

I got pretty good results with pure Ti system though with the same INCAR etc...

[graeme]

I don't know about the variation in the barrier with system size, although a lower barrier in a larger cell is generally expected. That said, a change in over 0.15 eV from your 3x to 4x cells is a little surprising.

The minima in the spline are due to the fact that you have no frozen atoms in the system. Small translational shifts in the position of the atoms leads to an increase in the distance between images. It is the fact that the force times the distance along the path no longer gives the correct energy difference between images that the spline is overshooting the energy of the minima. Aside from the interpolation and the plot, everything looks fine to me -- this issue does not effect the saddle energy or barrier.

[tanmoy]

Dear Prof. Henkelman,

Thank you so much for your reply and nice explanation. Yes, I did not fix the largest neighbour atom. I overlooked it because I didn't get such minima while doing the NEB calculations for 3x3x3 supercell. I have now also discussed with Jutta regarding this one. Let me do it and I will let you know in case of further discrepancies.

Thank you once again.