Half-Heusler Structures in MP

I’ve been looking at some Half-Heusler structures (XYZ) in MP and noticed that some of the Y-sites may be occupied by the incorrect element. For example, in TaFeSb (mp-631267: TaFeSb (cubic, F-43m, 216)), Ta is on the Y-site, but it should be Fe. Also, in NbRhSn (mp-30796: NbSnRh (cubic, F-43m, 216)) (listed as NbSnRh in MP), Sn on the Y-site when it should be Rh. Having the incorrect element on the Y-site alters the electronic structure, so both of these are predicted to be metals in MP when they are actually insulators/semiconductors.

@mkhorton any thoughts on the best way for MP to correct this?

Hi @mbrod, welcome!

This is not actually an error – crystal structures on Materials Project are sourced from a variety of places, and quality varies. While the majority are calculated from experimentally-synthesized crystal structures, many are not, and some structures are definitely unrealistic!

The important thing is that Materials Project calculates the total energy, and can therefore estimate the formation energy, and the “energy above hull” (a measure of metastability, where a zero value roughly indicates the predicted thermodynamic ground state at 0 K and no applied pressure).

We include these unstable materials for completeness; we cannot say whether a material will have a large or a small energy above hull until after it’s been calculated, and they may still save work or effort for others who might want to try out that crystal structure.

In this case, both mp-631267 and mp-30796 have a very large energy above hull of about 0.7 eV/atom. Although, mp-30796 does have an ICSD entry associated with it, indicating this is an experimentally-seen structure.

To re-iterate, the structures you’re interested in (e.g. where Fe is on the Y-site, and Rh on the Y-site), would appear as different materials (with different mp-ids) on Materials Project if we had them and, if calculated, would presumably have a much lower energy above hull and different electronic properties.

Practically, I would advise two things:

  1. Depending on your application, filtering out materials with “large” energy above hull values is usually a good first step.

  2. If you want to have the “correct” Half-Heuslers included in Materials Project (and we’d want to see that too), then you can submit the structures to us via MPComplete and we will try to include them in a future database release. We definitely appreciate good crystal structures being submitted in this way since it allows us to improve the quality of our database over time, and ensures we have materials available that people are interested in.

Hope this helps,


1 Like

I see-- thank you for clarifying!


1 Like