I assume that the mechanism that you refer to is the peak position shift in Raman peaks?
The Raman peaks will shifts position due to anharmonic effects with temperature.
The best option currently would be to perform MD for a supercell to include anharmonic interaction.
In general, ensemble averaging the Raman spectra should do the job but that will of cause be computational heavy for a supercell.
A better solution would be to calculate anharmonic shifts in the phonon frequencies directly. QuantumATK currently doesn't support that but it is in the pipeline so stay tuned.
