Computational Irregularities in the CoN System: Analysis and Insights

Materials Project
1

Issue with Convex Hull Calculations for the Co-N System in VASP

I am using VASP to construct the convex hull of the Co-N system but have encountered discrepancies when comparing my results to the Co-N phase diagram from the Materials Project (MP) (Co-N MP phase diagram).

I am using the PBE-GGA functional with the latest POTCAR potpaw.64. After noticing inconsistencies in my results, I focused specifically on CoN (CoN MP entry), which is reported as the ground-state structure in the MP database.

To investigate further, I used the VASP input files provided by MP to calculate the formation energy of CoN. However, I obtained a positive formation energy, which is problematic since CoN is an experimentally observed material and should be stable.

Additionally, I noticed a discrepancy in MP’s reported formation energies:

  • The CoN material entry lists its formation energy as 0.103 eV/atom—which is positive despite CoN being experimentally confirmed.
  • However, in the Co-N phase diagram, its formation energy is given as -0.074 eV/atom, indicating stability.

It seems there may be some issues with this material entry. If these discrepancies are due to the transition from the old Materials Project site to the new one, it would be helpful if MP could clarify or address these inconsistencies.

Additionally, I would like to understand what “atom” refers to in eV/atom. Does it correspond to the number of N atoms in the chemical formula, the number of N atoms in the unit cell I used, or is it simply the total number of atoms in the unit cell or formula unit?

Has anyone else encountered similar issues? Could this be due to differences in the computational setup between the phase diagram and individual material entries? Any insights or recommendations would be greatly appreciated!

2

Hi @Nadav_Moav, this is expected and due to the corrections that the Materials Project applies to thermodynamic data based on composition, level of theory used, etc. Sometimes DFT, fails to capture more complex electronic interactions, and MP applies empirical corrections to bring computed formation energies into alignment with experiment.

In this case, the net correction is -0.1805 eV/atom to the formation energy of Co-N. You can see the specific values of these corrections using the Materials Project API:

from mp_api.client import MPRester

with MPRester("your_api_key") as mpr:
    entries = mpr.get_entries("mp-448")

entries is a list of pymatgen ComputedStructureEntry objects with information about the corrections applied.

3

Hi Aaron,

Thank you for your response. However, this raises another question: how can I correct the results for my own compounds to accurately determine the true position of my Co-N compound in the convex hull calculation?

4

For each of your calculations, you’d need to make a pymatgen ComputedStructureEntry and apply the corrections scheme (MaterialsProject2020Compatibility):

from pymatgen.io.vasp import Vasprun
from pymatgen.entries.computed_entries import ComputedStructureEntry
from pymatgen.entries.compatibility import MaterialsProject2020Compatibility

vasprun = Vasprun("vasprun.xml")
cse = ComputedStructureEntry(
    structure = vasprun.structures[-1],
    energy = vasprun.final_energy,
    parameters = {'run_type': 'GGA', 'potcar_symbols': vasprun.potcar_symbols}
)
cse.energy_adjustments = MaterialsProject2020Compatibility().get_adjustments(cse)

Then the computed structure entry will contain an individual list of energy corrections, cse.energy_adjustments, an overall correction value in cse.corrections, and a final corrected energy in cse.energy or cse.energy_per_atom