NEB results in a well instead of a barrier

Dear Lammps users,

I am running an NEB simulation in order to find the barrier (or saddle point) between two structure configurations. I am using conjugate gradient minimized in order to make sure that my initial and final structures are well minimized. However the result of the NEB run shows a downward evolution of the energy instead of a barrier. I have changed many parameters ( kspring, timestep, neb group, fixed boundary conditions …) but the results have not changed. I keep having a well instead of a barrier.
Could you please guide me to solve this issue.

Thank you,

There isn’t really any useful details in your email
about your system.

But Aidan can possibly comment in a general sense.


I will assume that you have initialized the system with two well converged and distinct local minima. You can verify this by making sure that the energies under PE1 and PEN in the log.neb file do not change much during the NEB calculation. If the NEB calculation relaxes to a path with no barrier, or a path with a well, then this is a property of the potential surface that you have defined. A likely explanation is that one or both of the end states are not physically-relevant low energy structures, but rather are high energy local minima that do not play a role in the material behavior.