Climbing NEB calculations

1. Download VTST scripts from http://theory.cm.utexas.edu/vtsttools/index.html
2. Upload the vtstscripts.tgz file to your desired directory (e.g. 'home/username/scripts') and untar the file
3. Relax the perfect structure (ISIF=3)
4. Remove/add atoms and relax with ISIF =3 (initial and final structures)
   • Note: make sure the ordering of the atoms remains the same, i.e. if atom no. 3 moves then change the coordinates for that specific atom and don’t just delete and add a new atom in a different order
5. Use the CONTCARs from the initial and final structures and the nebmake.pl package from VTST scripts to create the images between the initial and final structures

$ /home/username/scripts/vtstscripts/nebmake.py POSCAR_i POSCAR_f <number of images>

6. Run the cNEB with executable vasp_std_vtst and the following flags in the INCAR file:

ISIF = 2

IBRION = 1 or 3

ISYM = 0 

POTIM < default or 0 

NEB

ICHAIN = 0 

IMAGES = #images 

SPRING = -5 

LCLIMB = .TRUE. 

LSCALAPACK = .FALSE. 

Note: The number of processors must be divisible by the number of images

7. Move results of initial and final structures into the 00 and 0# folders once the cNEB simulation is complete

8. Use the nebef.pl package to output the energies and the energy barrier for each image

$ /home/username/scripts/vtstscripts/nebef.pl

IMPORTANT NOTES:

VASP must be complied with VTST

IMAGES tag should equal the number of images you generated with VTST scripts

ATOM ORDER must be the same between the initial and final structure

SPRING tag = -5 is for the climbing NEB method

IBRION tag for congugate gradient method (2) generally fails (results in an error in reading IMAGES from the INCAR file). Use 1 or 3

LCLIMB tag = TRUE uses the climbing images modification (see http://theory.cm.utexas.edu/vtsttools/neb.html)

LSCALAPACK tag must be FALSE or VASP will crash running NEB calculations

HiPerGator submission script example:

#!/bin/sh

#SBATCH --job-name=<jop name>            # Job name

#SBATCH --mail-type=NONE                 # Mail events (NONE, BEGIN, END, FAIL)

#SBATCH --mail-user=<email address>      # Email address to send events to 

#SBATCH --ntasks=42                      # Number of MPI ranks

#SBATCH --cpus-per-task=1                # Number of cores per MPI rank 

#SBATCH --distribution=cyclic:cyclic     # Distribute tasks cyclically on nodes 

#SBATCH --mem-per-cpu=3500mb             # Memory per processor

#SBATCH --time=96:00:00                  # Time limit hrs:min:sec

#SBATCH --output=<output>.out            # Standard output and error log

#SBATCH --qos=<queue name>               # Queue you are submitting to 

pwd; hostname; date

module load intel/2016.0.109 openmpi/1.10.2 vasp/5.4.1

VASP=vasp_std_vtst

srun --mpi=pmi2 $VASP > vasp.log 2>&1