Inconsistent hull energies between legacy/new MP and within new MP

Hi everyone,

I am looking at stabilities of various ABF6 and ABF4 type structures and have noticed some inconsistencies between the new/legacy MP websites and within the new MP website. In the Sr-Cr-F system, the individual entry pages on the new website for SrCrF6 (mp-554332) and SrCrF4 (mp-558281) both state 0.0 eV/atom hull energies. However, on the legacy site these same entry pages state hull energies of 0.171 eV/atom and 0.510 eV/atom respectively. The formation energies and bandgaps are also different, so there is clearly some sort of difference in calculating the properties.

Additionally, there is inconsistency in hull energy between the specific entry pages and the phase diagram of the system on the new website. The phase diagram on the new website says that SrCrF6 SrCrF4 are both unstable with quite large hull energies, 1.98 eV/atom and 2.81 eV/atom, despite the individual entry pages saying 0.0 eV/atom. The formation energies are also different on the phase diagram. This is true for Mg-Cr-F and Ca-Cr-F as well, and likely others.

I wondered if this was due to unequal application of error correction with the GGA/GGA+U mixing, but this does not seem to be the case. I’ve calculated some of the formation energies by hand using the Sr, Ca, and F2 entries and they are consistent with the new website individual entries. I’m most confused by the difference between the phase diagram and structure pages on the new website. I hope I’m just missing something trivial, so please let me know if you have any insight or ideas, thanks!

Hi @donmct,

To start, please review:


There have certainly been a lot of bugs fixed in the newer database versions, but we also switched to a new correction scheme which caused additional changes.




Thanks Matt, should have done a more thorough check through those in the first place. Looks like selecting GGA/GGA+U/R2SCAN for phase diagram formation to give agreement with the specific entry pages. Thanks!