We are working on kink nucleation and migration using NEB method in LAMMPS. To provide intermediate replicas we need to make kink pairs. Can anyone suggest a method or tool that can be used to make double kink.
It's a bit tricky as the kinked configuration is inherently unstable. The way I've done it before is to
Create configurations of straight dislocations in positions A and B corresponding to two neighboring Peierls valleys. This can be done by computing the Eshelby elastic solution (i.e. the Stroh method) and displacing atoms from a perfect bulk structure. Make certain you use the same initial perfect structure as the atom IDs need to match. If you are doing a monopole configuration with fixed boundaries, I like to set the fixed atom positions to be for a dislocation halfway between A and B.
Create a configuration that is half at A and half at B by cutting the above in half and pasting them together. Check that the cut is coherent - all atom ids not in the A part are in the B part. This gives a rough initial kink structure to start with.
Relax the AB system and dump configurations often. As it relaxes, the kink structure will fix itself, and the system will eventually migrate fully to either A or B.
Set up your NEB with one end point being one of the dump configurations from #3 just before the kinks annihilate each other, and the full A or B that #3 did not relax into. When you run NEB, the kinked end point will relax just like #3 into either A or B and hopefully the intermediate configurations will be tricked into following the kink path.
I did lots of manual manipulations of configurations when I did it before. Nowadays, I would use the atomman Python package https://www.ctcms.nist.gov/potentials/atomman/ (which I manage).
There is a configuration builder for dislocations that builds monopole, periodic array, and small-scale dipole systems. It is very easy to manipulate configurations in other ways, such as slicing and pasting the two end configurations together. Also, APIs to access potentials from the NIST Potentials Repository, and tools for setting up, running, and analyzing LAMMPS simulations.
With a little bit of time, I should be able to put together a script/notebook that builds the configurations as described.
I've attached a Python script using atomman to create the initial unrelaxed kink configuration and the two compatible end configurations A and B. This just builds the configurations, so it is still up to you to use LAMMPS to relax them as described above, and then do NEB.