sumo-kgen

[isabelmoib]

Hi,

If I use sumo-kgen for orthorhombic cell only with substitute one atom by a dopant atom, sumo reads lattice type as monoclinic.

Still can we use the suggested path?

Best regards, Ibra

[ajjackson]

That's interesting, there are few things that could be happening. What spacegroup does it report, and what spacegroup were you expecting?

One possibility is that your lattice vectors are not quite orthorhombic (e.g. following relaxation). Another possibility is that spglib detected a higher-symmetry monoclinic system within your orthorhombic cell.

[isabelmoib]

I was expecting space group 62, same as the pristine structure (62), however I got space group 6.

The lattice parameters are same as the pristine structure. So, exactly spglib detected the monoclinic symmetry within the orthorhombic. So, is it OK to use the suggested path by Sumo in that case?

[ajjackson]

"Is it ok" is a bit subjective; are we looking for a technically valid calculation, or a useful one?

Is Sumo giving you the warning message about primitive cells? In order to get a technically-correct band structure for the detected spacegroup of the supercell you might need to extract a primitive cell that is smaller than the supercell; otherwise there will be extra bands.

That band-folding phenomenon from primitive to conventional cell is also a problem going from the conventional cell to a supercell. There are therefore a couple of reasons you may not really want to plot the band structure of a defect-containing supercell: firstly it will be very dense and difficult to interpret, secondly the high-symmetry positions will be related to the dimensions of your supercell, which is not really physically significant if your supercell is approximating a dilute defect.

For this reason it is useful to perform band-unfolding calculations that project the electronic structure of a supercell back onto the underlying lattice. Sumo does not provide this functionality, I suggest you look at BANDUP or easyunfold. (You might find kgen useful for suggesting a k-point path on the original lattice, though!)

[isabelmoib]

Yes, I used the suggested primitive cell for the suggested path However, in doped system, the suggested primitives cell is the input cell as it is, some times with standardization.

Also, for doped system, I compared the suggested path with other tools Seekpath, Vaspkit almost the same. For that reason, it is the spglib that defines that space group in case of doped systems.

[utf]

If I use sumo-kgen for orthorhombic cell only with substitute one atom by a dopant atom, sumo reads lattice type as monoclinic.

By substituting an atom you are reducing the symmetry of the cell. You cannot use the orthorhombic path as the structure is no longer orthorhombic, so the path will not cover the full irreducible Brillouin zone.

If you need the band structure on the orthorhombic path, you could try band unfolding to map the cell back onto the orthorhombic symmetry.

[ajjackson]

It is a bit unintuitive that you can get from an orthorhombic to a "monoclinic" space group without distorting the lattice vectors. I think the key point is that spacegroup 6 (P1m1) has only one symmetry operation: a mirror plane. I guess the lattice angles don't really matter if you break all the other symmetry? If the only symmetry is a mirror plane it has to be P1m1. A second mirror plane could get you into the orthorhombic Pmm2.

[utf]

Yes, α = β = γ = 90° does not mean the crystal is orthorhombic. But for a crystal to be orthorhombic the conventional cell must have α = β = γ = 90°.