In the past, basis sets for use in correlated molecular calculations have largely been taken from single configuration calculations. Recently, AlmlOf, Taylor, and co-workers have found that basis sets of natural orbitals derived from correlated atomic calculations (ANOs) provide an excellent description of molecular correlation effects. We report here a careful study of correlation effects in the oxygen atom, establishing that compact sets of primitive Gaussian functions effectively and efficiently describe correlation effects if the exponents of the functions are optimized in atomic correlated calculations, although the primitive (sp) functions for describing correlation effects can be taken from atomic Hartree-Fock calculations if the appropriate primitive set is used. Test calculations on oxygen-containing molecules indicate that these primitive basis sets describe molecular correlation effects as well as the ANO sets of Alml6f and Taylor. Guided by the calculations on oxygen, basis sets for use in correlated atomic and molecular calculations were developed for all of the first row atoms from boron through neon and for hydrogen. As in the oxygen atom calculations, it was found that the incremental energy lowerings due to the addition of correlating functions fall into distinct groups. This leads to the concept of correlation consistent basis sets, i.e., sets which include all functions in a given group as well as all functions in any higher groups. Correlation consistent sets are given for all of the atoms considered. The most accurate sets determined in this way, [5s4p3d 2flg] , consistently yield 99% of the correlation energy obtained with the corresponding ANO sets, even though the latter contains 50% more primitive functions and twice as many primitive polarization functions. It is estimated that this set yields 94%-97% of the total (HF + 1 + 2) correlation energy for the atoms neon through boron.

An accurate description of the electrical properties of atoms and molecules is critical for quantitative predictions of the nonlinear properties of molecules and of long-range atomic and molecular interactions between both neutral and charged species. We report a systematic study of the basis sets required to obtain accurate correlated values for the static dipole (α1), quadrupole (α2), and octopole (α3) polarizabilities and the hyperpolarizability (γ) of the rare gas atoms He, Ne, and Ar. Several methods of correlation treatment were examined, including various orders of Moller–Plesset perturbation theory (MP2, MP3, MP4), coupled-cluster theory with and without perturbative treatment of triple excitations [CCSD, CCSD(T)], and singles and doubles configuration interaction (CISD). All of the basis sets considered here were constructed by adding even-tempered sets of diffuse functions to the correlation consistent basis sets of Dunning and co-workers. With multiply-augmented sets we find that the electrical properties of the rare gas atoms converge smoothly to values that are in excellent agreement with the available experimental data and/or previously computed results. As a further test of the basis sets presented here, the dipole polarizabilities of the F− and Cl− anions and of the HCl and N2 molecules are also reported.

Valence correlation consistent and augmented correlation consistent basis sets have been determined for the third row, main group atoms gallium through krypton. The methodology, originally developed for the first row atoms, was first applied to the selenium atom, resulting in the expected natural groupings of correlation functions ͑although higher angular momentum functions tend to be relatively more important for the third row atoms as they were for the second row atoms͒. After testing the generality of the conclusions for the gallium atom, the procedure was used to generate correlation consistent basis sets for all of the atoms gallium through krypton. The correlation consistent basis sets for the third row main group atoms are as follows: cc-pVDZ: (14s11p6d)/͓5s4p2d͔; cc-pVTZ: (20s13p9d1 f)/͓6s5p3d1 f ͔; cc-pVQZ: (21s16p12d2 f 1g)/ ͓7s6p4d2 f 1g͔; cc-pV5Z: (26s17p13d3 f 2g1h)/͓8s7p5d3 f 2g1h͔. Augmented sets were obtained by adding diffuse functions to the above sets ͑one for each angular momentum present in the set͒, with the exponents of the additional functions optimized in calculations on the atomic anions. Test calculations on the atoms as well as selected molecules with the new basis sets show good convergence to an apparent complete basis set limit.

To gain new insight into the nature of hypervalency, we have characterized the bonding across the entire SF(n) sequence (n = 1-6) with high-level quantum chemical theory (multireference configuration interaction and coupled cluster calculations using correlation consistent basis sets). In contrast to most previous studies, this work examined both the stable equilibrium structures and the process of SF(n)-F bond formation. We conclude that two different types of bonding can occur in these species: normal polar covalent bonding and a new type that we call recoupled pair bonding. The two bonding processes can be seen in diatomic SF, where hypervalent behavior first occurs. In the covalently bonded (2)Pi ground state, the bond is formed by straightforward singlet coupling of electrons in the singly occupied S 3p and F 2p orbitals. But there is also a low-lying (4)Sigma(-) excited state where the S 3p(2) pair of electrons must first be decoupled so that one of the electrons can singlet couple with the electron in the F 2p orbital, hence the term recoupled pair bonding. Energy is required to decouple the electron pair, but the bond energy of SF((4)Sigma(-)) is still a substantial fraction (about 40%) of the bond energy of SF((2)Pi). Recoupled pair bonding is the basis for hypervalent behavior: for example, the three unpaired electrons of SF((4)Sigma(-)) are available for further bond formation, and their spatial orientations clearly anticipate the structure of SF(4). The new model of hypervalent bonding introduced in this work accounts for the observed trends in the structures of SF(n) molecules and the variations in the (SF(n)-F) bond energies. The model also predicts the existence of low-lying excited states in some SF(n) species and provides explanations for their energetic separations and orderings.

Ab initio POL–CI calculations, augmented by a dispersion term, are used to predict the potential energy surface for the reaction Cl+HCl. The saddle point is found to be nonlinear. The surface is represented by a rotated-Morse-oscillator spline fit for collinear geometries plus an analytic bend potential. Variational transition state theory calculations, based on a linear reference path, are carried out, and they yield much smaller rate constants than conventional transition state theory, confirming that earlier similar results for this heavy–light–heavy mass combination were consequences of the small skew angle and were not artifacts of the more approximate potential energy surfaces used in those studies. Transmission coefficients are calculated using approximations valid for large-reaction-path curvature and the potential along the reference path is scaled so that the calculated rate constant agrees with experiment. The resulting surface is used to compute the H/D kinetic isotope effect which is in qualitative agreement with experiment.