search

lucydot / effmass

Calculates various definitions of effective mass from the electronic bandstructure of a semiconductor.
https://effmass.readthedocs.io/en/latest/
MIT License
65 stars 37 forks source link

Effective masses undetectable in a command line? #66

Open FPrith opened 5 months ago

FPrith commented 5 months ago

In my use of effmass in a command line, the detection of effective masses in an electronic band structure sometimes fails.

My purpose in the use of effmass is the quick evalution of effective masses in various insulators. Below is an example of my calculation and analysis. ･ DFT code: VASP ･ chemical composition: BaO2 ･ space group: _I_4/mmm ･ cell: primitive unit cell (fully relaxed before the evaluation of a band structure) ･ k points to evaluate band structures : created by the SeeK-path program in vise, which has been developed by Y. Kumagai. ･ how to use effmass: command line

As an important supplement, regardless of the type of XC potential (GGA or HF/DFT hybrid), band structures in my calculations have been obtained through SCF calculations with explicit lists of k points.

This is a band structure of BaO2, which was drawn by vise. Its VBM and CBM are on the M points. The names of high-symmetry k points in this band structure are defined in SeeK-path. band_BaO2_PBEsol.pdf

Two cases of k-point distance (k-point density) in zero-weighed k points were tested. k-point distance is controlled by vise. Case 1) rougher k points As shown in a png file and a txt file, six hole effective masses along the M-to-gamma and M-to-S directions, and three electron effective masses along the same directions were detected. RougherKpoints.zip

Case 2) finer k points Unlike in Case 1, hole/electron effective masses along the M-to-S direction were NOT detected. In the case of even finer k points, the detection of effective masses still failed. To improve the reliability of effective mass, I would like to use the finest possible k-point distance. FinerKpoints.zip

What could be the source of this difference? Could you please provide any methods to improve the accuracy of detection?

lucydot commented 4 months ago

Hi @FPrith -

Apologies for late reply. I have not yet developed a habit of checking my github mentions frequently; something to work on.

Could it be that for the finer sampling there are two points which are equal in energy? Effmass is not particularly clever :) If there are two points equal in energy it will not detect it as a maximum/minimum (as technically it would be an inflexion point).