Dear all.
I want to treat a reaction involved some big molecules on (100) surface of fcc metal with NEB. I choose 4x4 cell in my calculation. In the NEB
calculation, I use only Gamma point and find a MEP path. If increase the k-points, the calculation is time-consuming heavily. My question is whether I can use the TS's structures obtained with Gamma point calculation to recalculate the energy barrier with more k-points.
thanks !
question about NEB calculation
Moderator: moderators
It would be best to reconverge the saddle point using more kpoints. You can use your initial gamma point band to do this efficiently. One method is to start a dimer or lanczos run from the gamma point saddle. This requires only 1 or 2 images, so it should be less computationally expensive. You can use the neb2dim or neb2lan scripts to generate the initial files for these runs.
If you don't want to try a new method, you can also start a 1 or 3 image NEB, using the climbing image method. If your path is fairly straight, you might be able to keep the same endpoints, but otherwise, you can use images from the original NEB on either side of the saddle. As long as your gamma point band captures the reaction coordinate at the converged k-point saddle, the climbing image will reconverge to the more accurate saddle point.
If you don't want to try a new method, you can also start a 1 or 3 image NEB, using the climbing image method. If your path is fairly straight, you might be able to keep the same endpoints, but otherwise, you can use images from the original NEB on either side of the saddle. As long as your gamma point band captures the reaction coordinate at the converged k-point saddle, the climbing image will reconverge to the more accurate saddle point.