Hi, I would like to perform a CI-NEB calculation, in order to simulate the diffusion of Na atoms in layered oxides. I try to optimize both initial (Na atom on a given site, let's say site 1) and final (Na on another site, site 2) structures before performing the CI-NEB calculation (with the same volume kept constant). However, the final state is not stable and the atom tends to move back to its initial position (site 1). The objective is to show that the diffusion is not favourable along a given path. How can I optimize the geometric structure with the Na atom located on site 2 (final state).

I tried a selective dynamics of the final state with the specific Na atom kept fixed. But the optimization results in high forces applied on the Na atom, higher than the requested tolerance, which is -0.03 eV/A. Thanks