I am using Amset with HSE calculations. The system is a nodal line semimetal in PBE and a 0.15 eV band gap appears on applying HSE. The experimental data is already published on that system.

INPUT:

wavefunction.h5 from two step HSE calculations. (PBE + HSE on a 9 X 9 X15 Gamma centered KPOINTS.)

deformation.h5 is also generated on 9 X 9 X 15 Gamma-centered KPOINTS, using HSE on a pre-converged PBE calculation.

Elastic constants are calculated using PBE since it is computationally expensive.

Dielectric constants are generated using PBE.

I have calculated using;

ADP, IMP

ADP

CRT

We are getting thermal conductivity as high as 100 W/mK. The published experimental results reported it to be around 3.2 W/mK. I need your help to understand the reason behind this.

Also, Iām worried about your PBE dielectric constants. The dielectric constant is very sensitive to the band gap, and indeed a metal has an infinite dielectric constant. So I would suggest trying to calculate the dielectric constant using HSE (unfortunately this is very expensive), or if that is not possible, see if something like SCAN or mBJ will help open the band gap to obtain the dielectric constant. Lastly, for small band gap materials, you need very dense k-point meshes to converge the dielectric. You should carefully check convergence of the dielectric with respect to the k-point mesh.

Thank you for your instant reply!
I have not used POP.

I was using vasprun.xml from fake k-point HSE band calculations and the thermal conductivity was increasing with temperature. Now, I used vasprun.xml from two step (PBE+HSE) calculation on gamma centered KPOINTS (9 X 9 X 15) and the kappa_e is decreasing with temperature, which is matching with experimental findings. The kappa_e is two order larger and relaxation time is of the order of E-12 s. I will try implementing your suggestions.