Coordination environments

I tried to get the environment type in crystal structure by using various approach available in pymatgen.

1: By computing local order parameters of a site as provided in pymatgen.analysis.local_env and use the one which corresponds to highest local order parameter to find the motif types:

structure_from_cif = Structure.from_file(‘Mn2V2O7_mp-19142_primitive.cif’)


y.get_local_order_parameters(structure_from_cif, 0)

Gives the value of local order parameter and corresponding type as follows:

{‘hexagonal planar’: 0.13616935377365502,

‘octahedral’: 0.5202731318416254,

‘pentagonal pyramidal’: 0.5052761742825485}

I choose the one with highest value of local order parameter as the motif type. So, in above ‘octahedral’ is the type. When I use this approach, like in MoS2 and many others, there are trigonal prism kind of structure, why the module did not capture such environment or how these environments are determined?
In addition pymatgen site_is_of_motif_type doesnot captures all types as mentioned in
the documentation.
In addition:
I used, pymatgen.analysis.chemenv.coordination_environments.coordination_geometry_finder import LocalGeometryFinder. There are many listed strategies. For some compounds there is discrepancy in type returned by using one vs the other strategy. As an example for Mn2V2O7 as mentioned above the default approach returns it as tetrahedra for V while the SimpleChemenvStrategy returns it as Trigonal pyramid. In the paper they referenced mentioned that the one with lowest value of continuous symmetry measure corresponds to the most perfect environment. By using these two strategy the CSM value for V is different and the type as well. First one measure CSM as 0.62 (minimum), while the second one gives it as 1.48. Which way we should use?
Thanks in advance.

The local order parameter value can range from 0 to 1, generally a value of 0.5 is not that great, i.e. in this case even though octahedral is the best match from its dictionary of known local environments, it is likely not actually octahedral.

By manual inspection, it does look like Mn is in a very-distorted octahedral-like environment, so this analysis seems accurate.

If you look at the V site, I would say the most accurate co-ordination is a distorted trigonal bipyramid. By definition, this also has character of tetrahedral (if you discount one bond) or trigonal pyramidal (if you discount two bonds), it really depends how many bonds you count. And, indeed, how many bonds are “actually there” is not something that can be determined from geometry alone and requires some chemical intuition.

I am less familiar with the chemenv strategies; these attempt to solve similar issues, though are older algorithms (this does not mean less correct, I’m simply less familiar with how well they’ve been tested).

Thank you so much for the explanation. Still, I am confused about the type obtained for MoS2 by computing local order parameters. The local order parameter value for ‘Mo’ in MoS2_mp_2815 is Octahedral (0.49), hexagonal planar (0.035) and pentagonal pyramidal (0.47). As, you explained the local order parameter value of 0.5 is even not good to make the decision about the type of environment present at any site, what other criteria we can use to assign the environment? In Material Project, site ‘Mo’ is explained as distorted edge sharing pentagonal pyramids . There are many compounds which contains site with Prism sharing environment, why such environments are not included? or it contains different name?

site_is_of_motif_typ from pymatgen.analysis.chem_env returns None for the environment type for site ‘Mo’ in MoS2 mentioned before. Is that because of local order parameter value less than 0.5?

I haven’t had a chance to look at the MoS2 case yet, but to answer the chemenv question: the chemenv module is completely independent from the order parameters / localenv analysis. Even though they try to answer similar questions, they were developed and implemented independently of each other.

Thank you so much. Yes, the chemenv module is independent of order parameters, it use continuous symmetry measure (CSM) and identify the one with lowest CSM as the environment type. Could you please take a look at MoS2, when you find some time and suggest me.
Thanks again for the prompt response.