bulk InSe band gap is wrong in the 2017.a2 version

[njuxyh]

Hi all

i tried the newest version 2017.a2 to calculate the band gap of bulk Inse.

I used GGA-PBE. as we know, the local EXC-Function will underestimate the band gap, but my results is opposite, which is 1.35eV, While the experimental value is 0.5eV


i test the calculation with the older version such 13.8.1, the band gap is almost the same with the experimental value.

so i am wondering what cause the error.


please help me.

thanks very much

[Jess Wellendorff]

Running your script without the geometry optimization produces a band structure with bands crossing the Fermi level (no fundamental band gap). Are you sure this 3D bulk material is a semiconductor? Also: You are considering a bulk, so it would be wise to use at least 2 k-points along the C-direction in order to capture the periodicity.

[njuxyh]

thank you for your kindly reply.

yes, i confirmed that the bulk inse is semiconductor, please go to the material project web:
https://materialsproject.org/#search/materials/{"reduced_cell_formula"%3A"InSe"}

in fact, i used 2 kpoints in the z direction, unfortunately,the result does not change .

[Daniele Stradi]

Hi,
did you use PBE xc-functional also for the calculations with ATK 13.8.1?
Daniele

[njuxyh]

yes, i used "normconserving/SE.GGAPBE.zip in the 13.8.1

[Daniele Stradi]

The electronic structure of III-V depends a lot on the lattice constant. Could you please check the optimized lattice constants calculated with 13.8.1 and compare to those calculated using 2017.a2?
Thanks,
Daniele.

[njuxyh]

yes, the optimized lattice is not the same using two different version,  in older 13.8.1 version, i used QuasiNewton method. but  the newest 17.a2 version i used LBFGS.

So i am wondering why 17.a2  and 13.8.1 give so much different lattice constant.

[Jess Wellendorff]

How much different? Are the computational settings equivalent?