Using the new Materials Project and the legacy version, I analyzed the interface reactivity between a pair of phases. The results are actually different and I don’t quite understand the result produced by the new Materials Project.
For example, with Li3B11O18-C interface, the legacy MP predicts stability, while the new MP predicts reactivity via Li3B11O18 → Li3BO3 + 5 B2O3. As Li3B11O18 is on the convex hull, I suppose the this reaction is not energetically favorable.
Could someone help explain the result, and the discrepancy between two MP versions?
This looks like a temporary bug on the MP app. I do not see the predicted decomposition of Li3B11O18 when manually creating the interface reaction plot. If you want to avoid the bug in the meantime, you can build your own diagrams using the InterfacialReactivity class in pymatgen.
@munrojm, if you get a chance, can you investigate if this is a data-sourcing issue? My guess is the Interface Reactions App is being provided the wrong set of entries (maybe an older build or something).