DOS quality for finding the Fermi level

[Steven Hartman]

Hello,

To learn how to use PyCDT, I'm studying GaAs based on the work presented in the published paper.  Currently, I'm trying to solve for the Fermi energy and equilibrium defect concentrations.  The solve_for_fermi_energy() method requires the density of states from the bulk material. 

Do I need to obtain a high-quality DOS for this, such as that produced by an NSCF-uniform parameter set in Pymatgen?  For GaAs, the 2x2x2 Monkhorst grid used for the default PyCDT bulk calculation doesn't find the CBM or VBM, both located at the Gamma point, so the band gap is overestimated.  I suspect this would lead to bad results, but if you have tested this, I'd appreciate any insights.

Thanks

[Bharat Medasani]

My suggestion is to go for high quality DOS obtained with either the primitive or the unit cell. PyCDT calculations are focused on supercell calculations and the resulting energetics. PyCDT doesn't make an effort in generating accurate DOS since it can be obtained easily from MP website.