Discrepancy between phases from API and materials project phase diagram

Hi, when I tried to retrieve the phases of a chemical system from mpr.thermo.search and looked for the same in Phase Diagram App, I see certain phases are missing in the retrieved data. For example, when I try for Fe-Mn-O via ;

for phase in phase_entries:
    if phase.composition.reduced_formula not in phase_list:

I am getting ['Mn2FeO4', 'Mn(FeO2)2', 'Mn5FeO12', 'Mn13Fe3O32', 'Mn7Fe3O20', 'Mn4FeO8', 'MnFeO2', 'Mn5Fe3O16', 'MnFeO3', 'Mn13Fe11O32', 'Mn3Fe5O12', 'Mn2FeO6', 'Mn3FeO8', 'Mn19Fe17O48', 'Mn5(FeO3)4', 'Mn9(FeO6)4', 'Mn9FeO10', 'MnFe4O5', 'Mn17Fe13O40', 'MnFe5O8', 'Mn23FeO32', 'MnFe2O3', 'Mn4FeO5', 'MnFeO4', 'Mn3Fe3O8', 'Mn2FeO3']
Here, I do not see any compound containing only Fe and Mn whereas in the actual phase diagram of Fe-Mn-O in the phase diagram app, there are phases like MnFe, MnFe3 with energy above hull less than 0.1 eV/atom. I kept the thermo_type same as GGA_GGA+U in both the cases. May I know what am I missing here?

Hi @Akhil_S, there’s an existing thread about the phase diagrams here: Build GGA/GGA+U/R2SCAN (Mixed) phase diagrams via new API - #19 by Ting_Wang Does that help? If not, @munrojm might be able to help out.

Hi @tschaume thanks for sharing. As per the above discussion, there was already a fix done to remove this conflict. However, I am still facing this issue. I will anyway check with an upgraded mp-api client though I would have expected the GGA_GGA+U entries should have been retrieved as close to what is being shown in the web app

You need to make sure you are querying for all sub-chemical systems as well. That query will only get you compounds with Mn, Fe, and O. This doc page might help you: Phase Diagrams (PDs) - Materials Project Documentation

– Jason

Thanks @munrojm, got it. I would like to clarify another doubt in the same context. I had noticed that when I use mpr.get_pourbaix_entries(chemsys) to retrieve entries used for constructing the Pourbaix Diagram, many of the entries in the Phase diagram (including the ones belonging to all sub-chemical systems) are being retrieved though many of them have an energy above hull quite high. May I know whether all the phases (stable and metastable) in the phase diagram are being used for constructing Pourbaix diagram or is there any filtering being applied?

I believe some may get filtered out depending on what ion reference data is available. You can see the guts of method here: https://github.com/materialsproject/api/blob/8a84e76763fe549c0ce1ebae5914da7d4e31c36f/mp_api/client/mprester.py#L951

– Jason