If I calculate the DFT (Density Functional Theory, with hybrid functional) band-structures of a semiconductor, says GaAs, for different values of the lattice constant (equivalent to applying different hydrostatic strains), then does the change in the valence band edge versus the strain reflect the that of a realistic GaAs. In other words, does it make sense to compare the valence band offset between the pure DFT calculations (even with hybrid functional) for the same material but different in lattice constant?
Anh-Luan Phan In my opinion, vacuum reference energy could be a reference for comparision.
Dai-Nam Le Yes, but we have only the band edges and not the vacuum level. I wonder whether we can use the pure valence band edge values of DFT calculation without extra reference level or not.
Anh-Luan Phan No, there is always a energy reference when doing any calculations. You cannot say that there is no reference, even for a single calculation. If there is no reference, you will get different value of energy even for the same method, same system, same parameters...
Can you state you problem on what physical effect you want to see? This might give us a more easy way to understand what you want.
In general, you calculations are compatible in the sense that the reference energy is defined in the pseudopotential. If you maintain those fixed between calculations you are implicitly assuming the same reference. Clearly, you might want to check that by some other method, for example the vacuum energy.
That said, DFT is giving you total energies. The position of the valence band edge is not a total energy and you are thus introducing a systematic error that is impossible to control. And yet, DFT seems to work quite well... With this caveat, why you are interested in the Valence band edge rather than in the band gap and its nature or on the Density of States?
Roberto D'Agosta Thank you. I'm interested in the position of the valence band edge because I wanted to fit an empirical model to DFT so that it can reflect properly the band offset alignment of the same material but in different strain profiles. [Of course band gap and density of states are also of interest]. As you said, to do that we should have the same reference level. It turned out that our DFT calculation software (namely QuantumATK) supplies such a reference level. In summary, I got the answer for my question: without knowing the reference level of each calculation, NO.
You should test that the difference is the valence band edge remains the same as there is no change of reference if you are using the same pseudopotentials.
Regards,
Roberto
- Badges
- Science topic