Electronic Relaxation Time (tau ) for electrical Conductivity

[mohamed_daoud]

Hi,
I am trying to calculate electrical conductivity using Boltztrap, but the output is a ratio (electrical conductivity (sigma)/relaxation time (tau)). Is there a way to calculate the relaxation time independently and then use it to calculate the actual electrical conductivity? Your help will be much appreciated.

[alexey.tal]

Hi,

BoltzTraP uses the constant relaxation time approximation, so τ is not calculated and has to come from somewhere else, for example from fitting to an experimental conductivity.

You can avoid this approximation in VASP (version 6.5 and later) by calculating the transport coefficients directly from electron-phonon scattering with ELPH_MODE = TRANSPORT. Instead of assuming one τ, VASP computes a state- and
temperature-dependent lifetime for every electronic state. It then solves the Boltzmann transport equation and gives you the electrical conductivity, mobility, Seebeck coefficient and electronic thermal conductivity in absolute units,
with no fitted parameters.

The VASP Wiki explains the theory and the full workflow:
https://vasp.at/wiki/Transport_coeffici ... scattering

Keep in mind that this treats only electron-phonon scattering. Impurity, defect and grain-boundary scattering are not included.
For heavily doped or polycrystalline samples, the result is therefore an upper bound on the conductivity.

Best regards,
Alexey