Ab initio effective core potential calculations of the electronic structures of silane and chlorosilanes

Abstract

Pseudopotential methods, in which atomic core electrons are replaced by the appropriate potentials in such a manner that only valence electrons are explicitly treated in molecular calculations, offer the advantage of considerable time saving compared with all-electron(AE) calculation with little loss in accuracy. Ab initio effective core potentials(ECP) of Wadt and Hay are used for Si and Cl atoms. We performed AE and ECP calculations of the ionization and excitation energies for Cl atom in order to assess the reliability of the ECP generated by Wadt and Hay. We learned from those calculations that the differences between AE and ECP results are small being less than 0.15 eV. Furthermore, these differences are mainly due to the basis set truncation. The magnitude of errors due to the approximation which are used to generate the individual ECP is far less than that due to the basis set choice. ECP calculations with and without d functions upon Si and Cl atoms are carried out for silane and chlorosilanes at experimental geometries. The results of ECP calculations with and without d functions on third-period atoms agree well with those of corresponding AE calculations. The influence of d functions on thirdperiod atoms in ECP calculations is almost identical to that in AE calculations. Clearly, d function for the third-period atoms must be included to study the electronic structure of a molecule and to predict the correct trend in a molecular series such as $SiH_{4-n}Cl_n(n=1-4)$. We optimized the geometries of $SiH_4$, $SiCl_4$ and $SiH_3Cl$ with and without d functions on the third-period atoms. When we compare the optimized geometries with those from AE calculations, the optimized bond lengths are shorter than those in AE calculations by some hundredth of angstrom. The ECP``s generated by Wadt and Hay are slightly more attractive than the AE potentials but still useful for many purposes.