NONLINEAR OPTICAL-PROPERTIES OF PARA-NITROANILINE - AN ABINITIO TIME-DEPENDENT COUPLED PERTURBED HARTREE-FOCK STUDY

For p-nitroaniline the ab initio method with a double-zeta basis set which includes semidiffuse polarization functions has been used to calculate the dipole moment mu, frequency-dependent linear polarizability alpha, and nonlinear hyperpolarizabilities beta and gamma using the time-dependent coupled perturbed Hartree-Fock approach. The computation procedure used here yields information on the dispersion behavior of all the tensor components of polarizability and various hyperpolarizability terms. The largest dispersion effect is observed for the diagonal components of the polarizability and hyperpolarizability tensors along the long in-plane axis. The magnitudes of the various hyperpolarizability terms which describe the various second-order nonlinear processes show the following trend: beta( - 2-omega; omega, omega) > beta(0; omega, - omega) = beta(- omega; 0, omega) > beta(0; 0, 0), with beta(- 2-omega; omega, omega) exhibiting the largest frequency dispersion. The various second hyperpolarizability terms which describe the various third-order nonlinear optical processes show the following trend: gamma(- 3-omega; omega, omega, omega) > gamma(- 2-omega; 0, omega, omega) > gamma(- omega; omega, -?? omega, omega) > gamma(- omega; 0, 0, omega) almost-equal-to gamma(0; 0, omega, - omega) > gamma(0; 0, 0, 0). Again gamma(- 3-omega; omega, omega, omega) shows the largest dispersion effect. The results of existing semiempirical calculations on p-nitroaniline are compared with that of the present ab initio calculation, and the problem due to the arbitrary parametrization procedure adopted in the past for semiempirical calculation is discussed. The computed values of the first resonance energy, the dipole moment, and the polarizability are in good agreement with the respective values experimentally observed, within the spread of the existing experimental data. In contrast, the computed beta and gamma values are considerably smaller than the respective experimentally determined values. We attribute this discrepancy to two sources. First, in the theoretical calculation electron correlation has been neglected, and the basis set used, although large, may not still be adequate. Second, there is a considerable spread in the reported experimental values for a given nonlinear coefficient making any comparison between the theory and the experiment difficult. In contrast, the computed beta and gamma values are considerably smaller than the respective experimentally determined values. We attribute this discrepancy to two sources. First, in the theoretical calculation electron correlation has been neglected, and the basis set used, although large, may not still be adequate. Second, there is a considerable spread in the reported experimental values for a given nonlinear coefficient making any comparison between the theory and the experiment difficult.