Soft mode dispersion and 'waterfall' phenomenon in relaxors revisited

Results of recent inelastic neutron scattering studies of lead-based relaxor ferroelectrics by Gvasaliya et al. [J. Phys.: Condens. Matter 17, 4343 (2005); J. Phys.: Condens. Matter 19, 016219 (2007)] have put in question the existence of the waterfall anomaly-an apparent vertical dispersion segment joining the TA and TO branches-observed earlier in low-energy [00] phonon dispersion curves of these materials. In the present article, we review the results of earlier experiments and model calculations together with the outcome of our recent measurements on PMN using the same instrumental set-up as Gvasaliya et al. to conclude that the waterfall feature is not an experimental artefact. We also give some hints on a possible explanation of the results of Gvasaliya et al., by exploring the fact that the reported dispersion of the underdamped transverse optic branch follows the longitudinal acoustic (LA) branch dispersion surprisingly closely.