solid state CI-NEB, not converge

Vasp transition state theory tools

Moderator: moderators

xph
Posts: 39
Joined: Tue Mar 13, 2012 9:33 pm

Re: solid state CI-NEB, not converge

Post by xph »

Two real negative modes? Then it is a monkey saddle point, though NEB is not designed to find such kind of saddle points. Did you watch the two modes to see what they look like?

I have been using SSNEB in TSASE which is also on our website under "code". I have calculated phase transformations from graphite to diamond using forces and enthalpies from VASP. The paths are always smooth even under 45 GPa. There could be bugs in VTSTCODE or VTSTSCRIPT that we have not noticed. I am working on high pressure tests these days.
rlzhou
Posts: 20
Joined: Sat Aug 06, 2011 11:02 am

Re: solid state CI-NEB, not converge

Post by rlzhou »

Did you use the enthalpies from VASP in your calculation using TSASE? In VTSTscripts, the nebbarrier.pl reads the free energies while not enthalpies of the images. Do you think that it is a reason that the highest energy image can not sit on the saddle point after spline under high pressure? Under high pressure, I thine the energy term of pV has significant influence on the relative stability of phases. We usually estimate the relative stability of different phases by comparing their enthalpies while not free energies.

I hope you can find the bugs in VTSTCODE or VTSTSCRIPT soon. Please let me know as soon as you find and fix them. Thank you.
xph
Posts: 39
Joined: Tue Mar 13, 2012 9:33 pm

Re: solid state CI-NEB, not converge

Post by xph »

Thanks to your testing, we do find a bug in nebbarrier.pl. It should read enthalpy rather than energy. We are fixing it now.

All the SSNEB calculations are fine, and you just need to run nebresults.pl again when we update the VTSTSCRIPTS.
xph
Posts: 39
Joined: Tue Mar 13, 2012 9:33 pm

Re: solid state CI-NEB, not converge

Post by xph »

Yes, it is enthalpy in TSASE.
I just compared the results between TSASE and VTST. Replacing the energies in nebss.dat with enthalpies from TSASE smoothes the path. I think this will solve your problem.
You can do it by hand if your are in hurry. Just read enthalpy of each image from OSZICAR and divide it by the number of atoms, then put the value in nebss.dat.
rlzhou
Posts: 20
Joined: Sat Aug 06, 2011 11:02 am

Re: solid state CI-NEB, not converge

Post by rlzhou »

Yes, I got the smooth curves when I use the enthalpies. Please let me know, when you update the vstpscripts. Thank you.
Post Reply