Gaussian basis sets for use in correlated molecular calculations. V. Core-valence basis sets for boron through neon

TLDR

The authors extended the cc‑pVXZ basis sets to core‑valence sets (cc‑pCVXZ) by adding core functions (1s1p, 2s2p1d, 3s3p2d1f, 4s4p3d2f1g) and optimizing their exponents to minimize the difference between all‑electron and valence‑only correlation energies from HF+1+2 calculations on the ground states of B–Ne. With the cc‑pCVXZ sets, core, core‑valence, and valence correlation energies, as well as all‑electron singles‑and‑doubles CI energies, converge exponentially toward complete‑basis‑set limits, and test applications to boron ionization potentials, carbon 3P–5S splitting, and CH₂ state separations confirm that core‑electron correlation effects are small yet essential for high‑accuracy predictions.

Abstract

The correlation-consistent polarized valence basis sets (cc-pVXZ) for the atoms boron through neon have been extended to treat core and core-valence correlation effects. Basis functions were added to the existing cc-pVXZ sets to form correlation-consistent polarized core-valence sets (cc-pCVXZ) in the usual pattern: Double zeta added (1s1p), triple zeta added (2s2p1d), quadruple zeta added (3s3p2d1f), and quintuple zeta added (4s4p3d2f1g). The exponents of the core functions were determined by minimizing the difference between all-electron and valence-only correlation energies obtained from HF+1+2 calculations on the ground states of the atoms. With the cc-pCVXZ sets, core, core-valence, and valence correlation energies all converge exponentially toward apparent complete basis set (CBS) limits, as do the corresponding all-electron singles and doubles CI energies. Several test applications of the new sets are presented: The first two ionization potentials of boron, the 3P–5S separation in carbon, and the X̃ 3B1–ã 1A1 state separation in CH2. As expected, correlation effects involving the core electrons of the first row atoms, B–Ne, are small but must be included if high accuracy is required.