Reply to "Comments on `Theoretical study of impurity-induced first-order Raman spectra for the alkali halides' "

Abstract

The preceding paper1 is concerned with our recent paper2 in which we presented the results of calculations, based on the deformation-dipole model (DDM), for the E, components of the firstorder Raman spectra of KC1, KBr, KI, and RbCl doped with TI+. In that paper we compared our results with those obtained by Harley, Page, and walker3 using the breathing-shell model (BSM) and attributed the discrepancies between the two to the fact that the BSM had been fitted to known dispersion while the DDM was not so adjusted. In his paper Page is concerned that by this statement we might have been implying that a large number of parameters (e. g., 11) had been used. This was not our intention, but the fact remains that the monopole charge Z was used as a fitting parameter, at least for KI and KBr. In this sense information from the dispersion curves has been employed and the Harley-Page-Walker (HPw) calculations are not based solely on macroscopic input data. Previously we held Z fixed at unity, the value indicated by the highly ionic character of the bonding in alkali halides and most consistent with the observed cohesive energies. Any model which one uses for the dynamics should be reasonably consistent with the interpretation of these static properties and it appears to us that models with Z = 0.9, which imply 10% covalent bonding do not satisfy this requirement and should be regarded as parametrized to fit the dynamical data. Recently, we have obtained new results for the E, spectra5 calculated using DDM phonons derived from a model which was identical with that used earlier2 except that instead of holding Z fixed at unity we used values of Z which were adjusted to optimize the fit to the measured dispersion curves. These are thus directly comparable with the HPW results since both models (DDM and BSM) are now parametrized in the same way. The only difference is that the elastic constants are not used as input data in the DDM calculations.