Hi @alex
I have been trying to understand the electronic transport properties on few materials which are polar. In these materials, LO-TO splitting in the phonon dispersion is significant and is not captured in POP frequency calculation using amset phonon-frequency utility. Usually, to observe the LO-TO splitting we have add a non-analytical term correction (NAC) to the dynamical matrix and re-calculate the phonon frequencies. I’m attaching the phonon dispersion with and without LO-TO splitting for your reference below.
Without LO-TO splitting
With LO-TO splitting
Using amset phonon-frequency utility, calculated phonon frequencies and weights are shown below.
Here we can see that the highest three optical phonon modes has the same frequency which should not be the case as can be observed from phonon dispersion with LO-TO splitting.
I tried to resolve this issue by performing phonon dispersion calculation using VASP (Look for LPHON_DISPERSION tag) which incorporates NAC (Look for LPHON_POLAR tag). This is implemented only in the VASP versions above 6.3.2 and I was able to get the NAC corrected phonon eigenvectors and eigenvalues at gamma point.
I have looked into the amset/tools/phonon_frequency.py code and noticed that its extracting the phonon eigenvectors and eigenvalues from vasprun.xml file of DFPT run. However, the calculation performed with NAC correction, phonon eigenvectors and eigenvalues are written to OUTCAR file but not in vasprun.xml file. The output format is provided below.
--------------------------------------------------------------------------------
Phonons
q-point No. q_x q_y q_z q-path length
1 0.000000 0.000000 0.000000 0.000000
frequencies
[THz] -0.000000 -0.000000 0.000003 6.038520 6.038520 6.103634 7.757204 7.757204 10.218882
[2Pi THz] -0.000002 -0.000002 0.000016 37.941139 37.941139 38.350264 48.739953 48.739953 64.207130
[cm-1] -0.000011 -0.000010 0.000087 201.423343 201.423343 203.595316 258.752490 258.752490 340.865217
[meV] -0.000001 -0.000001 0.000011 24.973300 24.973300 25.242591 32.081206 32.081206 42.261882
eigenvectors
0.347476 -0.000023i 0.131143 -0.000028i -0.261753 +0.000000i 0.000051 -0.000000i 0.015098 +0.000000i -0.065653 -0.000000i -0.529888 -0.000062i -0.498018 -0.000207i 0.510103 -0.000000i
0.058767 -0.000023i 0.366725 -0.000028i 0.261749 +0.000000i -0.013050 -0.000000i 0.007593 +0.000000i 0.065653 +0.000000i 0.166352 -0.000062i -0.707906 -0.000207i -0.510103 +0.000000i
0.286809 -0.000000i -0.234025 -0.000000i 0.263488 -0.000000i 0.013101 -0.000000i 0.007505 -0.000000i 0.065653 -0.000000i -0.696240 +0.000000i 0.209888 +0.000000i -0.510103 +0.000000i
0.448184 -0.000030i 0.169152 -0.000036i -0.337616 +0.000000i -0.002094 +0.000004i -0.619843 -0.000006i -0.409220 -0.000000i 0.168512 +0.000020i 0.158377 +0.000066i -0.226679 -0.000000i
0.075799 -0.000030i 0.473011 -0.000036i 0.337611 +0.000000i 0.535753 +0.000004i -0.311735 -0.000006i 0.409220 +0.000000i -0.052902 +0.000020i 0.225124 +0.000066i 0.226679 +0.000000i
0.369934 +0.000000i -0.301851 +0.000000i 0.339854 +0.000000i -0.537847 +0.000000i -0.308108 +0.000000i 0.409220 +0.000000i 0.221415 -0.000000i -0.066747 +0.000000i 0.226679 +0.000000i
0.513050 -0.000034i 0.193634 -0.000041i -0.386479 +0.000000i 0.001795 -0.000003i 0.531249 +0.000005i 0.401947 +0.000000i 0.211673 +0.000025i 0.198942 +0.000083i -0.147461 -0.000000i
0.086770 -0.000034i 0.541470 -0.000041i 0.386473 +0.000000i -0.459178 -0.000003i 0.267179 +0.000005i -0.401947 -0.000000i -0.066452 +0.000025i 0.282786 +0.000083i 0.147461 +0.000000i
0.423475 +0.000000i -0.345538 +0.000000i 0.389041 +0.000000i 0.460973 +0.000000i 0.264071 +0.000000i -0.401947 0.000000i 0.278126 +0.000000i -0.083843 +0.000000i 0.147461 +0.000000i
--------------------------------------------------------------------------------
We can observe that after NAC correction, phonon frequencies in OUTCAR matches with the phonon dispersion (with LO-TO splitting) provided above.
As LO-TO splitting is a signature of polar materials and POP scattering rates mainly depends on pop_frequency, I request you to address this issue by updating the amset/tools/phonon_frequency.py code to read the phonon eigenvectors and eigenvalues from OUTCAR file instead of vasprun.xml file. Also, please update the documentation to include necessary tags such as LPHON_DISPERSION and LPHON_POLAR to get the correct phonon eigenvalues and eigenvectors for polar materials.
Thank you.
Best regards,
Shiva.