Relativistic effects in a b i n i t i o effective core potential studies of heavy metal compounds. Application to HgCl2, AuCl, and PtH

Publisher Website
Google Scholar
Abstract
A method is described for obtaining l‐dependent relativistic effective core potentials (ECPs) from Dirac–Fock self‐consistent field atomic wave functions. These potentials are designed for use in nonrelativistic (NR) valence electron Hartree–Fock calculations on atoms and molecules. The novel aspect of this approach involves the averaging of the separately generated nodeless pseudovalence atomic orbitals belonging to each of the spin–orbit j components for every l value greater than zero. For the mercury and gold atoms the ECPs obtained here are shown to be virtually identical in the valence region with the corresponding ECPs obtained by Hay et al. [J. Chem. Phys. 69, 984 (1978)] using different type relativistic (R) atomic wave functions. Ab initio valence electron calculations using relativistic potentials on Hg, Au, HgCl2, and AuCl show corresponding agreement with all electron calculations, experimental values, and previous relativistic core potential results. For the Pt atom relativistic state energy calculations give good agreement with experiment, whereas the NR approach gives grossly incorrect values for relative multiplet energies, including the wrong ground state. These errors are reflected in valence electron calculations on the PtH diatomic molecule where using the NR potential predicts incorrectly a 2Σ ground state. On the other hand, using the R potential the proper 2Δ ground state is obtained along with good agreement with experiment for the equilibrium bond distance and stretching frequency. Both the R and NR type wave functions in their respective ground states predict the major bonding orbital on the Pt atom to be the 5d.