k-points Fermi level, band structure and transmission spectrum

[Alexander Romero]

Dear QuantumATK users,

1) How is the Fermi level computed in QuantumATK?  How it depends on the k-points number? The variation of the Fermi level with respect to the k-points number indicates a change in the number of energy levels of the band structure? or it is possible the band structure converges with respect to the k-points number but the Fermi level still varies?

2) The Fermi level and the transmission spectrum (calculated from bulk) should converge, with respect to the k-points number, at the same k-points number?

[Anders Blom]

That depends a bit on what you mean by Fermi level here.

If you are talking about the number reported in the log file sometimes, it's a meaningless quantity that cannot be used for anything except (internally) to set the reference level for other energies, whose difference compared to this energy is meaningful of course.

If you mean how we determine the zero level for band structure plots, this is simply the level which makes the Fermi function integrate to the correct number of electrons. Note that the default electron temperature (1000 K) sometimes means this level is not exactly in the middle of the band gap for semiconductors. The high number means convergence is quicker, but not as accurate.