Calculation of molecular polarizabilities using a semiclassical Slater-type orbital-point dipole interaction (STOPDI) model

Abstract

The point dipole interaction model for molecular polarizability proposed by Applequist, Carl, and Fung [J. Am. Chem. Soc. 94, 2952 (1972)] is modified by replacing the point dipole interaction tensor with a descaled distributed charge interaction tensor. Our procedure is based on the descaled tensor algorithm proposed by Thole [Chem. Phys. 59, 341 (1981)] and uses a Slater-type orbital (STO) function to represent the charge distribution. The resulting STOPDI formalism calculates mean molecular polarizabilities and the components of the molecular polarizabilities with errors comparable to experimental uncertainty. Furthermore, these procedures require only one optimized parameter per atom, the average atomic polarizability. The formalism is invariant to coordinate transformations and avoids the discontinuities and/or false resonances that are characteristic of previous classical and semiclassical formalisms. The STOPDI algorithm requires less parameterization and computation time than the anisotropic atom point dipole interaction (AAPDI) model of Birge [J. Chem. Phys. 72, 5312 (1980)] and is more reliable for the calculation of polarizability derivatives and Raman cross sections. We demonstrate, however, that none of the above formalisms are reliable for calculating absolute Raman cross sections for normal modes involving significant bond stretching components. This is an inherent limitation of any formalism which does not explicitly account for electron density redistribution accompanying changes in the internuclear distances of covalently bonded atoms.