Conclusions about neb: only suitable for rate-limiting steps?

Hi all,

According to the neb example, diffusion from an initial state to a final state involving 2 atomic “steps” required two independent simulations for analysis. However, if one simply input the initial and final atomic states (i.e. combined the process into a single hop), the energy barrier doesn’t converge, and increases essentially to infinity (if the number of iterations is increased…). However, I am also somewhat disconcerted by the thought that for each step, it is only the converged maximum energy barrier that one can utilise. SO, if you could obtain a converged outcome for a series of steps, you might incorrectly assume there is a single step-activation barrier… When in fact, there may be several?

I am wondering if anyone who has carried out neb analyses in the past can indicate whether they have successfully obtained convergence between highly complex transitions from initial and end states?

Also, if someone could possibly provide verify my results (the maximum energy barriers for hop1 and hop2 are exactly equal to the example log values:

for “hop 1”:

0.0071048579 eV.

For “hop 2”:
0.0016018849 eV…

However hop1+2 for tolerances:

$minimize 1.0e-15 1.0e-15 50000 500000


$neb 0.0 0.01 5000 5000 100 final final.hop2

does not converge but provides this output:

for “hop1+2”:
0.07122142 eV

or for :

$neb 0.0 0.01 50000 50000 100 final final.hop2

for “hop1+2”:

17.742657 eV !!!

Any comments or queries which are well-considered and helpful would also be appreciated.

Kind regards,