Relativistic small-core energy-consistent pseudopotential is a commonly used method to deal with relativistic effects. Unfortunately, spin-orbit coupling effects (SOC) are generally not considered in developing basis sets for this pseudopotential except for the dhf-nZVPP-2c basis sets. Even the dhf-nZVPP-2c basis sets are validated mainly in density functional calculations. In this work, the SOC-CCSD(T) calculations with the scalar-relativistic Hartree-Fock determinant as reference using the small-core energy-consistent pseudopotential are carried out to investigate effects of 5s5p electrons and performance of various basis sets on properties and SOC effects of closed-shell diatomic molecules containing 6 row elements. Our results indicate that it is essential to include 5s5p electrons in SOC-CCSD(T) calculations to provide a reliable description on SOC effects. The cc-pVnZ-PP basis sets and the dhf-nZVPP basis set inadequately capture SOC effects, while the dhf-nZVPP-2c and cc-pwCVnZ-PP basis sets provide reasonable results. The error of these basis sets on properties of heavy-element systems aligns with the number of basis functions in these basis sets and the order is as follows: dhf-TZVPP-2c, cc-pwCVTZ, dhf-QZVPP-2c, cc-pwCVQZ, and cc-pwCV5Z. To obtain accurate bond lengths and harmonic vibrational frequencies of these heavy-element systems, at least the dhf-QZVPP-2c basis set should be employed. On the other hand, error of even the cc-pwCVQZ basis set is still sizeable for dissociation energies.

Key words: basis set, spin-orbit coupling, coupled-cluster theory, relativistic effect, electron correlation