UT theoretical chemistry code forum

Hello,

I am confused about how CI-NEB works when atoms' coordinates are fixed on POSCAR using F F F. For example, a few examples in the forum relaxed the initial and final structures using ISIF=3, but the transition states made by nebmake.pl have all fixed atoms. Though ISIF=2 is used (degree of freedom for atomic coordinates), if atoms' coordinates are fixed, how NEB moves the atoms using springs?

Please help me to understand this,

I'm not sure that I entirely understand the question. There is no sense to an NEB calculation when all the atoms are fixed (with a very small caveat that it is possible to do a solid state optimization of the cell with atoms fixed in relative coordinates). You certainly can have fixed atoms in an NEB calculation and the most straight forward approach is to keep the cell fixed (ISIF=2) and any fixed atoms should be in the same position in every image. The fixed atoms will not feel any spring forces, since they are in the same position at each image, and any forces due to the potential are zeroed out - simply put, they are not allowed to move. I hope this helps, but if I missed the point, please let me know.

Hello Professor,

Thanks for your reply. To clarify, let's say I have a crystal structure of NVPF with 278 atoms, and I want to study the migration of Na ions from a stable structure to a state where this Na ion may leave the cell. I relaxed the initial structure with ISIF = 3 up to 0.01 ev/A and the final structure with ISIF =3 (though it is not a stable state). Then for transition states, I fixed 277 atoms, and only the migration Na is free. I used ISIF = 2 to relax up to 0.1 ev/A (not relaxing below this for a large structure), spring constant -0.5 ev/A^2, timestep 0.01, IOPT = 3.

Is this process correct?

Also, can you please explain what you mean by "solid state optimization" with all fixed atoms in a cell?

Let's forget about the solid state optimization for now - it is not important.

So I'm not sure that we are on the same page in terms of how the calculation is done. My suggestion is to find a cell geometry that you are happy with and then fix it - just the cell parameters, not the atoms. Now you can move Na from one site to another. Note that there is no "outside the box"; these are periodic calculations and there is nothing special about the location of the box boundaries. But anyway, keep the cell fixed and do an NEB calculation to find the migration barrier. Don't freeze all the atoms - in fact, you can keep them all relaxed. You need to know how the lattice atoms respond to the Na diffusion, so you can't freeze them.