I was comparing some of my calculated formation energies with those from the Materials Project webpage. In general I find a good agreement, an exception are compounds containing S, where I find large differences.
An example is FeS (mp-505531) with a formation energy Ef -0.847 eV/atom:
[data from the Materials project for FeS and the reference elements I downloaded two days ago]
#id fu Ef[ev/at] E[eV/at]
mp-505531 FeS -0.847 -6.809
mp-96 S 0.000 -4.128
mp-13 Fe 0.000 -8.461
If I calculate the formation energy directly from the energies above I get
which is close to the value I find in my calculations, but quite different from the value on the web page Ef=-0.847 eV/atom.
Is there something I am missing?
You might be missing the anion correction term, which in the case of FeS mp-505531 is 1.3269 eV or 0.33 eV/atom, which accounts for your discrepancy.
These correction terms are listed at the bottom of the materials page under “Calculation Summary”, more information of their meaning can be found there or in our wiki. In this specific case, it’s a fit to ensure that the correct formation energies are found in sulphide compounds, and is fit to Li2S, Na2S, K2S, Rb2S, CsS, MgS, CaS, BaS, Al2S3.
Hope this helps,