ssNEB or the solid state NEB can be done with VASP, but it does require recompiling VASP. The basic theory is laid out in the following paper by the Henkelman group at the University of Texas:

http://theory.cm.utexas.edu/henkelman/pubs/sheppard12_074103.pdf

Essentially, the technique maps the coordinate vector (in the NEB this is just the atomic positions) to also include the engineering strain and the force vector (in the NEB just the atomic forces) to also include the stress tensor by using a suitably defined Jacobian.  What this allows for is a change in the lattice vectors between the start and ending cell (as in the normal NEB, there still needs to be a one-to-one mapping of the atoms between the initial and final cell). In general, it can be used to study phase transformations in which cell degree of freedom and not just internal degrees of freedom are necessary. In the paper above, it is used to explore the wurtzite to rock salt phase transition in CdSe and I've used it to explore the phase transition in NbO2.

Henkelman's group has developed a code that you can patch into VASP and then recompile VASP to use.  Some supercomputing centers already provide a variant of VASP including this code.  It is called VTST or VASP Transition State Tools and can be obtained here:

http://theory.cm.utexas.edu/vtsttools/

The relevant flag is "LNEBCELL."

Also, from speaking with one of the developers, the simulation cell used for the calculation should be oriented such that the lattice vectors when listed as rows in a matrix form a lower diagonal matrix (eq. 6 in their paper) for optimal results.

Another benefit of the VTST code is that it adds extra flexibility even for regular NEB calculations over the default in VASP (for example climbing image and additional force based optimizers) as well as other transition state methods.  Additionally, they provide many scripts for setting up and analyzing transition state theory calculations.

ssNEB or the solid state NEB can be done with VASP, but it does require recompiling VASP. The basic theory is laid out in the following paper by the Henkelman group at the University of Texas:

http://theory.cm.utexas.edu/henkelman/pubs/sheppard12_074103.pdf

Essentially, the technique maps the coordinate vector (in the NEB this is just the atomic positions) to also include the engineering strain and the force vector (in the NEB just the atomic forces) to also include the stress tensor by using a suitably defined Jacobian.  What this allows for is a change in the lattice vectors between the start and ending cell (as in the normal NEB, there still needs to be a one-to-one mapping of the atoms between the initial and final cell). In general, it can be used to study phase transformations in which cell degree of freedom and not just internal degrees of freedom are necessary. In the paper above, it is used to explore the wurtzite to rock salt phase transition in CdSe and I've used it to explore the phase transition in NbO2.

Henkelman's group has developed a code that you can patch into VASP and then recompile VASP to use.  Some supercomputing centers already provide a variant of VASP including this code.  It is called VTST or VASP Transition State Tools and can be obtained here:

http://theory.cm.utexas.edu/vtsttools/

The relevant flag is "LNEBCELL."

Also, from speaking with one of the developers, the simulation cell used for the calculation should be oriented such that the lattice vectors when listed as rows in a matrix form a lower diagonal matrix (eq. 6 in their paper) for optimal results.

Another benefit of the VTST code is that it adds extra flexibility even for regular NEB calculations over the default in VASP (for example climbing image and additional force based optimizers) as well as other transition state methods.  Additionally, they provide many scripts for setting up and analyzing transition state theory calculations.

Thank You Dear Dr. Andrew O'Hara for your help and support always on researchgate.

Shilendra

maybe these reference will help you:

[1]D. Sheppard, P. H. Xiao, W. Chemelewski, D. D. Johnson, and

G. Henkelman, J. Chem. Phys. 136, 074103 (2012).

[2] N. A. Zarkevich and D. D. Johnson, J. Chem. Phys. 142, 024106

(2015).

Dear Dr. Andrew O'Hara, I am using VASP 5.3.3 which was complied by our computer center engineer. Now I want to do some ssNEB calculation, as you told in your previous post it is needed new compilation.

How do check installed VASP is suitable for ssNEB calculation or it requires fresh installation and what changes requires in new installation.

My other query is that without Henkelman's code we can do ssNEB in VASP ?

I asked my colleague he was saying we can generate image file (POSCAR file) from vtst code and can execute in VASP separately but he is not sure. Please clarify installation queries and what is right way to do ssNEB calculation.

Thank You

Shilendra

As far as I am aware from looking at the source code, VASP itself only implements the standard NEB method.  In order to use other force optimizers, other techniques (climbing image), and the ssNEB, you have to recompile the code with the instructions provided by the Henkelman group website (their website provides both the code that gets added to VASP as well as some helpful scripts).

The code they provide is freely available and can be linked into VASP during compilation rather straightforwardly. If you are using a copy provided by your computing center, they may be able to recompile VASP with the VTST code added in (I know that some centers already provide such a version).

The scripts they provide for setting up and analyzing NEB calculations can be used with the regular VASP implementation, but you still won't have access to the additional functionality that the VTST code provides.

Dear Dr. Andrew O'Hara, I installed VASP with vtst tools for ssNEB calculation. Without ssNEB tag it is executing results but when I am doing calculation with ssNEB tags job is suddenly terminating and showing no error. My INCAR

System=Phase_transformation_of_Au(hexagonal)_to_Au(fcc)

ISTART=0

ICHARG=2

ISYM=2                                                                                                           ENCUT=345

ISMEAR=0

SIGMA=0.05

ISIF=3

IBRION=1

NSW=1000

EDIFF=1E-5

EDIFFG=-0.02

LREAL=False

PREC=Accurate

IALGO=38

NPAR=8

IMAGES=4

SPRING=-5

ICHAIN=0

LCLIMB=.True.

POTIM=0

LNEBCELL=.TRUE.

IOPT=3

Where I have done mistake, I tried but could not resolve this problem.

Please suggest me.

Thank You

Shilendra

Dear Dr. Andrew O'Hara, problem has been resolved, I was giving  number of images wrong. First I was including initial and final images also in IMAGES tag, now I am giving only images as number of intermediate images. Now job is running well but some convergence problem is coming.

Thank You

Shilendra