Dear all,
Based on the AMSET documentations, the software calculate the effective POP frequency by weighing each optical branch with its mode averaged Born effective charge.
For polar semiconductors, the Lyddane–Sachs–Teller relation indicates that we should observe LO-TO splitting at the GAMMA center. However, in my DFPT calculations using VASP, I observed that the non-analytical term correction (NAC) seems not to be included, as indicated by the absence of LO-TO splitting in the OUTCAR results.
I have attached an example of the DFPT calculation OUTCAR of PbTe. Search for keywords “Eigenvectors and eigenvalues of the dynamical matrix”, you can see the three optical branches have the exact same frequency.
OUTCAR (618.6 KB)
Normally, if we don’t consider NAC, the LO frequency will be underestimated. Doesn’t this make the calculated pop-frequency inaccurate? If so, how do we solve it?
Best regards,
Zane