Dear Alex,

I have some concerns related to the use of ISMEAR. When I look at the example of Si, you used Gaussian smearing ISMEAR=0. Although VASP is recommending not to use this type of smearing for semiconductors, and rather recommends the tetrahedron method ISMEAR=-5.

Is there a specific reason why to use this type of smearing for the AMSET calculations? Do you expect large differences to appear?

Kind regards,

Dear Simon,

The silicon example calculations are rather old and just there as quick example. I wouldn’t read too much into the calculation settings.

That said, the primary benefits of tetrahedron smearing in VASP are: (i) to obtain better single-point energies in metals, (ii) to avoid partial occupancies in small band gap semiconductors, and (iii) to obtain nicely interpolated spectral properties such as the density of states or frequency-dependent dielectric constant.

ISMEAR of 0 or -5 won’t make much difference in this case (although I do use tetrahedron method for all semiconductors these days). The PBE band gap of silicon is large enough that there will be no partial occupancies caused by gaussian broadening.