How to transform the disordered structure to ordered structure?

Hello, pymatgen users! Now I have a disordered structures with fractional occupancy, i.e. metastable bixbyite Y6WO12.


Among this, the two Wyckoff sites 8a and 24d were occupied simultaneously with 75%Y,12.5%W and 12.5% vacancies.
x y z Occ. U Site Sym.
1 O O1 0.08333 0.37087 0.14708 1.000 0.000 48e 1
2 Y Y2 0.28508 0.00000 0.25000 0.750 0.000 24d 2…
3 W W3 0.28508 0.00000 0.25000 0.125 0.000 24d 2…
4 Y Y4 0.00000 0.00000 0.00000 0.750 0.000 8a .-3.
5 W W5 0.00000 0.00000 0.00000 0.125 0.000 8a .-3.
I want to construct the ordered structure model and calculate the formation energy or band structure by VASP calculation. Can i do this by Class EnumerateStructureTransformation in pymatgen.transformations.advanced_transformations module? Is it a feasible method to handle this problem?

Hey @Caodazhuang, there are a couple of methods you can use:

These work in different ways - SQS attempts to create a supercell of a given size that represents the most random arrangement at that size; OrderDisorderedStructureTransformation tries to minimize the electrostatic energy, which is not a direct proxy for the DFT energy (they are very different)

For generating supercells ≥25 atoms, you likely want to use one of the Monte Carlo SQS methods, and select a few of the highest-ranked SQS structures of different supercell sizes. The supercell with the lowest energy/atom is then closest to the ground state

2 Likes

Thank you for your reply,sir! I have tried to use the Class OrderDisorderedStructureTransformation to model this material.