UT theoretical chemistry code forum

how to acquire accurate POSCAR in saddle point from CL_NEB?

I have used the CL_NEB method to obtain the barrier of migration and then i want to calculate the vibrational frequencies at initial and transition satate.
therefore, I must first get the optimized POSCAR at saddle point. So my question is how to acquire accurate POSCAR in saddle point from CL_NEB?
CL_NEB method can only get energy at saddle point at specified image, but i want to know how to get the POSCAR corresponding to the saddle point image.

another question is if i get that POSCAR, whther i need to optimize it before calculate the vibrational frequencies.


thanks!

You can get the climbing image number from neb.dat. The CONTCAR file in the folder with the name of that number is what you want.
As along as the EDIFFG is small enough, this saddle point structure is accurate. Do not optimize when you calculate the frequency.

[quote="xph"]You can get the climbing image number from neb.dat. The CONTCAR file in the folder with the name of that number is what you want.
As along as the EDIFFG is small enough, this saddle point structure is accurate. Do not optimize when you calculate the frequency.[/quote]

Thank you for your reply.
According to the CL_NEB method, the highest energy image is just exactly driven to the saddle point,but my calculated picture mep.eps shows that the highest energy point dose not belong to any imgae.Besides, the highest energy listed in the file exts.dat does not correspond to any image I have made by nebmake.pl. So I think my EDIFFG is not small enough according to your reply. Do I can believe the image after CL_NEB calculation is exactly located to saddle point as long as the EDIFFG is set small enough and result is well converged?

The spline mep curve can be misleading sometimes. As long as the forces converged (usually 0.01 is enough), the climbing image is at the saddle point. The higher energy section on mep.eps is just the fault of spline.

[quote="xph"]The spline mep curve can be misleading sometimes. As long as the forces converged (usually 0.01 is enough), the climbing image is at the saddle point. The higher energy section on mep.eps is just the fault of spline.[/quote]

Thank you for your help.