EFFECTS OF THE CRYSTAL-FIELD ON TRANSITION MOMENTS IN 9-ETHYLGUANINE

Molecular orbital calculations, both semiempirical and ab initio, give transition moment directions for purine and pyrimidine bases which frequently differ significantly from experimentally determined polarization directions. A particularly serious discrepancy exists in the case of 9-ethylguanine, in which the two lowest energy pi-pi* transitions are predicted to have transition moment directions differing by 40-50-degrees from the results of polarized reflection measurements on single crystals. The large ground-state dipole moments of the purine and pyrimidine bases lead to strong electrostatic interactions in the crystal which have been neglected in the MO calculations. The INDO/S Hamiltonian has been modified to include the effect of static charges in the environment of the molecule. The resulting modifications in the Fock matrix require the calculation of the electrostatic potential and field at each atomic center. These have been evaluated for 9-methylguanine as a model for 9-ethylguanine by using the known crystal structure of the latter. Ground-state charges calculated in the Mulliken approximation were evaluated from INDO/S wave functions, and the potentials and fields included contributions from all molecules in four unit cells in each direction from the central molecule. Polarization effects were included by an iterative procedure in which the initial charges, calculated for an isolated molecule, were replaced by those calculated for the molecule in the crystal field. This process was repeated until the resulting transition moment directions differed by less than 0.1-degrees between successive cycles. Additional calculations were performed with charges which had been scaled down so that the dipole moment calculated for the isolated molecule agrees with that estimated from ab initio calculations. The results of these calculations show marked improvement in agreement with experiment. The residual discrepancy for the first pi-pi* transition is 6-18-degrees and that for the second is 24-29-degrees, where the smaller value is from unscaled calculations and the larger from scaled. Comparison with data for higher energy transitions also shows greatly improved agreement with experiment, especially for the scaled calculations. The results also indicate substantial mixing of n-pi* and pi-pi* excited states by the crystal field, which could be significant for the circular dichroism of nucleic acids. We conclude that local electrostatic fields in crystals can substantially modify the excited states and transition parameters of the bases of nucleic acids. Clearly these effects must be considered in theoretical treatments of the optical properties of DNA and RNA.