This python package helps you create orthogonal grain boundary supercells for atomistic calculations. The code is based on the
coincident site lattice (CSL) formulations for cubic materials (sc, bcc, fcc, diamond). I intend to extend it to hcp structures soon.
This code produces a final structure to be read in [LAMMPS](https://lammps.sandia.gov/).
# Structure
There are two main scripts: [_csl_generator.py_](./csl_generator.py) and [_gb_generator.py_](./csl_generator.py) which you need to use in this order to produce the final gb structure.
@@ -145,14 +146,15 @@ The following is a one of these 50 GBs visualized by [Ovito](https://ovito.org/i
<imgsrc="./exGB.png"width="50%">
- _**A note on microscopic degrees of freedom:**_
In the absence of a consensus on how to find the global minimum energy GB structure I used atom removal + rigid body translations according to the description above to find the minimized structure. For rigid body translations, the smallest translation vector to guarantee the minimum energy structure is not well defined, therefore you can make the mesh as dense as you wish by choosing larger a and b values. By trial and error in fcc elemental cases (such as Al and Cu) I have come to the rule of thumb conclusion of 50 initial structure.
In the absence of a consensus on how to find the global minimum energy GB structure I used atom removal + rigid body translations according to the description above to find the minimized structure. For rigid body translations, the smallest translation vector to guarantee the minimum energy structure is not well defined, therefore you can make the mesh as dense as you wish by choosing larger a
and b values. By trial and error in fcc elemental cases (such as Al and Cu) I have come to the rule of thumb conclusion of 50 to 100 initial structures.
If you are a more rigorous user you can just create the GB structure and run a more involved minimum energy search routine.
- _**A note on the minimization procedure:**_
I often do a three stage minimization at 0K followed by an MD annealing simulation in [LAMMPS](https://lammps.sandia.gov/).
The 0K miminimization is composed of: A congugate gradient minimization of the energy of atoms, the simulation box and then atoms again
similar to a procedure explained [here](https://icme.hpc.msstate.edu/mediawiki/index.php/LAMMPS_Input_Deck_for_Grain_boundary_generation).
For the annealing simulations I use an _nvt_ ensemble followed by damped dynamics. Depending on the GB structure and your final purpose you can run annealing simulations for different time spans.
