Hi Alex
Seebeck coefficient is known to decrease with doping. For heavily doped systems, the Seebeck coefficient may increase and then decrease with high doping levels. However, at fixed 300K and small doping concentrations, I am observing a increase in Seebeck coefficient. This is unusual to me. Could you please help me with this?
This is very unusual. Do you have very large SOC splitting in this material?
Thanks for your response.
Band splitting isn’t there. However, the curvature and slope of the bands have changed. But I wonder if it would be significant for this to happen. If you have time, please have a look at the band structure (Fig. 2) here
https://journals.aps.org/prmaterials/pdf/10.1103/PhysRevMaterials.6.085404
With SOC your band gap is very small ~0.14 eV, which means that at low doping concentrations, the Fermi level will be in the middle of the gap and you’re therefore getting bipolar conduction (you have both free electrons and free holes at 300 K, as the Fermi-Dirac functions overlap both the conduction and valence bands). Because you’re using a semilocal functional, the band gap is likely underestimated. If you calculate the bandgap with HSE+SOC you can then use the scissor argument in AMSET to artificially widen the band gap of the PBE+SOC calculation. This will prevent bipolar conduction and you won’t see the strange behaviour in the Seebeck coefficient.
Thank you very much for the clarification.