change of space group after optimization

[sayantanu]

Dear ATK users,
                         I tried to create Phosphorene-2D crystal using the steps provide by atk. I have made a supercell 2x2x1, while exporting the configuration to cif format, the corresponding file is having P -1 space group but in builder->bulk tools->crystal symmetry info  it is showing space group 53. Further added after optimizing the structure the space group has been changed showing space group 10, why is it so??
Please find the corresponding files  in the attachments. Please help me out......

[Anders Blom]

First of all, the CIF export in VNL is extremely crude. We need to improve on this - currently it can only export P1.

As for 53 vs 10, spacegroup 10 is monoclinic so it means the lattice angles have changed slightly in the optimization. To avoid this (i.e. constrain the Bravais lattice to simple orthorhombic which 53 is), change the lattice class from UnitCell to SimpleOrthorhombic in the Builder after making the supercell, before sending the structure to the Script Generator.

[lily840408]

I have the same problem. I used "fix Bravais lattice constrain" in the same way as Anders Blom said: "Before sending the structure to the script generator, change the lattice class from UnitCell to SimpleOrthorhombic in the Builder". But I used the output of "builder-> bulk tools-> crystal symmetry info" to observe, but it showed P1.
Please help me out......

First of all, the CIF export in VNL is extremely crude. We need to improve on this - currently it can only export P1.

As for 53 vs 10, spacegroup 10 is monoclinic so it means the lattice angles have changed slightly in the optimization. To avoid this (i.e. constrain the Bravais lattice to simple orthorhombic which 53 is), change the lattice class from UnitCell to SimpleOrthorhombic in the Builder after making the supercell, before sending the structure to the Script Generator.

[Anders Blom]

This is a slightly tricky one.

First, if you replace back the Al atom with Hf, then it is indeed recognized as space group 14. And, the structure as such in QuantumATK will correctly be  represented as monoclinic no matter what.

However, by changing one atom (Al@Hf  as a substitutional defect) you violate the symmetries of the atomic positions to such a degree that P1 is the only remaining space group symmetry. This is because there are so few symmetries in the first place in the monoclinic spacegroups.

So, you actually need to separate the "lattice symmetry" from the space group. In your system, the lattice is still monoclinic, but the atomic positions do not match any monoclinic space group. The same can actually be seen for a cubic case, although the many rotational and reflection operations mean that a single defect only lowers the symmetry but the space group can still be cubic. For instance, you can make a supercell of 8 Si atoms in fcc and replace one atom with Ge, which turns it into 215 instead of 227. However, if you perturb all atomic positions a little bit (there is a button in the Builder for that), then the space group is reduced to P1 also here.

[lily840408]

Thank you for your help.

The implication is to use "fix space group" and "fix Bravais lattice type", and the output will maintain "space group" or "Bravais lattice type". However, because the substitution defect violates the symmetry of the atomic position, the "crystal symmetry info" cannot find a match, resulting in the display of P1.

Is the "a little bit" you said "Nudged Elastic Band"? Thank you for your additional notes.

[Anders Blom]

Not NEB, just simply "shake" the atoms a bit, adding a small random perturbation. Sometimes it's a good idea to actually break the space group on purpose to avoid being stuck in a local minimum, but here I was just making a point about the lattice and space group symmetries being separate things.

The "fix space group" constraint should work perfectly fine also for this system, but as I think you realize, the system is already in P1 before you optimize it.

[lily840408]

I truly appreciate your help.
I can try to move the position of any atom in order to create adding a small random perturbation under the condition of non-fixing the space qroup or the Bravais lattice .