Phonon calculations

by **Algerien1970** Fri 29 May - 18:52

From the userguide :

Calculations of phonons is based on a program PHONON by K.Parlinski, which runs under MS-Windows and must be ordered separately (see http://wolf.ifj.edu.pl/phonon/ )
You would define the structure of your compound in PHONON together with a supercell of sufficient size (e.g. 64 atoms). PHONON will then generate a list of necessary displacements of the individual atoms. The resulting file case.d45 must be transfered to UNIX. Here you would run WIEN2k-scf calculations for all displacements and collect the resulting forces, which will be transfered back to PHONON (case.dat and/or case.dsy). With these force information PHONON calculates phonon at arbitrary q-vectors together with several thermodynamic properties.

1 init_phonon_lapw

init_phonon_lapw uses case.d45 from PHONON and creates subdirectories case_XX and case_XX.struct files for all required displacements. It allows you to define globally RMT values for the different atoms and

- initializes every case individually (batch option of init_lapw is now supported) or
- initializes every second case (useful for pos. and neg. displacements, which have the same symmetry and thus only one initialization is necessary), or
- initializes only the first case and copies the files from the first case to all others. This is most convenient in low symmetry cases with P1 symmetry for all cases and thus just one init_lapw needs to be executed (while for higher symmetry a separate initialization is required (but computational effort is reduced).
Please use mainly nn to reduce equivalent atoms. sgroup might change the unitcell and than the collection of forces into the original supercell is not possible (or quite difficult).

A script run_phonon has been created. Modify it according to your needs (parallelization,....) and run all cases to selfconsistency.

Note that good force convergence is essential (at least 0.1 mRy/bohr) and if your structure has free parameters, either very good equillibrium positions must have been found before, or even better, use both, positive and negative displacements to average out any resulting error from non-equillibrium positions.

by **Algerien1970** Fri 29 May - 18:53

2 analyse_phonon_lapw

analyse_phonon_lapw uses the resulting scf files and generates the ``Hellmann-Feynman''-file required by PHONON. When you have positive and negative displacements an automatic averaging will be performed. The resulting case.dat and case.dsy filse should be transfered back to MS-Windows and imported into PHONON.

Phonon calculations