Bravais lattice in conceptual dictionary

[pkruzikova]

Currently, the bravais_lattice assigned is either bcc, fcc or hcp based on the space group identified on the structure. Number of atoms in the cell is also identified to distinguish primitive/conventional unit cell, and used for other parameters (e.g. angles, volume) - this does not affect the Bravais lattice assigned.

The issue raised is that the space group alone is not sufficient to infer the Bravais lattice, for example if the unit cell has multiple species.

[Tara-Lakshmipathy]

In my opinion, identifying the correct space group is the main issue since the analysis algorithms tend to fail quite often. Even when there is just one species, having even slight temperature effects can mess up the algorithms.
But if the correct space group has been determined (irrespective of the number of species), then there is a direct mapping between the space group families and the bravais lattice types.

But maybe I'm misunderstanding what is meant here by "multiple species".

[krejiba]

Point 1

HCP is not a Bravais Lattice. For HCP, the lattice is primitive hexagonal (hP).

Classification 1

Crystal Family	Lattice System	Bravais Lattice(s)
Triclinic	Triclinic	Triclinic
Monoclinic	Monoclinic	Simple Monoclinic, Base-Centered Monoclinic
Orthorhombic	Orthorhombic	Simple Orthorhombic, Base-Centered Orthorhombic, Body-Centered Orthorhombic, Face-Centered Orthorhombic
Tetragonal	Tetragonal	Simple Tetragonal, Body-Centered Tetragonal
Hexagonal	Rhombohedral	Rhombohedral
Hexagonal	Hexagonal	Hexagonal
Cubic	Cubic	Simple Cubic, Body-Centered Cubic, Face-Centered Cubic

Classification 2

Crystal Family	Crystal System
Triclinic	Triclinic
Monoclinic	Monoclinic
Orthorhombic	Orthorhombic
Tetragonal	Tetragonal
Hexagonal	Trigonal
Hexagonal	Hexagonal
Cubic	Cubic

Why the distinction

Space Group	Crystal System	Lattice System	Bravais Lattice	Example
P3_1 (144)	Trigonal	Hexagonal	Hexagonal	ZnTe - high pressure
R3m (160)	Trigonal	Rhombohedral	Rhombohedral	3R-MoS2
R-3m(166)	Trigonal	Rhombohedral	Rhombohedral	α-Samarium
R-3c (167 )	Trigonal	Rhombohedral	Rhombohedral	Calcite - CaCO3
P6_3/mmc (194)	Hexagonal	Hexagonal	Hexagonal	Magnesium (HCP)
P6_3/mmc (194)	Hexagonal	Hexagonal	Hexagonal	α-graphite (not HCP)
P6_3/mmc (194)	Hexagonal	Hexagonal	Hexagonal	α-Lanthanum (not HCP)
P6_3/mmc (194)	Hexagonal	Hexagonal	Hexagonal	2H-MoS2 (not HCP)

Point 2

There is a difference between Bravais Lattice and close-packed/non close-packed.

1. Space Group 221:

• α-CsCl (cP Simple Cubic, non close-packed)
• Cu3Au (cP Simple Cubic, close-packed)

2. Space Group 225:

• γ-Fe (cF Face-Centered Cubic, close-packed)
• Ag (cF Face-Centered Cubic, close-packed)

3. Space Group 229:

• α-Fe (cI Body-Centered Cubic, non close-packed)
• Tl7Sb2 (cI Body-Centered Cubic, non close-packed)

If you want to keep the field (fcc/bcc/hcp), you should name it something else, atom arrangement for example.

[Tara-Lakshmipathy]

Agreed, hcp should be changed to hexagonal. Missed that one when creating the master data.

[Tara-Lakshmipathy]

But the larger problem still remains, space groups are very hard to detect when there are defects and/or temperature effects. So, I think the whole unit cell part of the concceptual dictionary (including space group, bravais lattice, lattice parameters etc.) should apply mainly to pyiron_workflow structures. In the case of pyiron_atomistics, I think these details should be filled only when the user provides the original structure before introducing defects and thermal effects. In other words, it shouldn't be filled automatically for the user.

[Tara-Lakshmipathy]

Just discussed this witth @pkruzikova. One possible approach for pyiron_atomistics would be to offer the user a boolean and an argument something like this:

def classic_structure(pr, structure, structure_name, is_original_structure=False, original_structure=None):
    '''
    Identifying unit cell parameters and crystal structure are tricky since the algorithms involved are not very reliable.
    is_original_structure: can the incoming structure be used to identify the unit cell parameters including space-group etc.
    original_structure: if the above is False, please provide a suitable structure for identifying unit cell parameters.
    NOTE: if is_original_structure is False and no original_structure is provided, unit cell parameters will be left empty
    '''
    structure_path = pr.name + '/'
    # structure_name = structure_name_prefix + '_input_structure'
    
    from pyiron_base.storage.hdfio import FileHDFio
    hdf = FileHDFio(structure_path + structure_name + '.h5')
    structure.to_hdf(hdf)

    # Cannot guarantee this will be before manipulations hence skipping get_unit_cell_parameters
    from ob.concept_dict import process_structure_crystal, get_unit_cell_parameters
    if is_original_structure == False and original_structure == None:
        struct_params = {}
    elif is_original_structure == True and original_structure == None:
        try:
            struct_params = get_unit_cell_parameters(structure)
        except:
            struct_params = {}
    elif is_original_structure == False and original_structure != None: 
        try:
            struct_params = get_unit_cell_parameters(original_structure)
        except:
            struct_params = {}
    else:
        try:
            struct_params = get_unit_cell_parameters(structure)
        except:
            struct_params = {}
    struct_cdict = process_structure_crystal(pr, structure, structure_name, structure_path, struct_params)
    
    return struct_cdict

So if the user confirms that the incoming structure is the original structure or if they are willing to provide an original structure, the unit cell parameters including space-group, bravais lattice etc. can be filled. If not, unit cell details will be left empty.