Conventional correlation methods suffer from the slow convergence of electron correlation energies with respect to the size of orbital expansions. This problem is due to the fact that orbital products alone cannot describe the behavior of the exact wave function at short inter-electronic distances. Explicitly correlated methods overcome this basis set problem by including the inter-electronic distances (rij) explicitly in wave function expansions. Here, the origin of the basis set problem of conventional wave function methods is reviewed, and a short history of explicitly correlated methods is presented. The F12 methods are the focus herein, as they are the most practical explicitly correlated methods to date. Moreover, some of the key developments in modern F12 technology, which have significantly improved the efficiency and accuracy of these methods, are also reviewed.

In this work, the extension of the perturbative coupled-cluster F12 method, CCSD(T)F12, developed in our group for the treatment of high-spin open-shell molecules (J. Zhang and E. F. Valeev, J. Chem. Theory Comput., 2012, 8, 3175.), is also documented. Its performance is assessed for accurate prediction of chemical reactivity. The reference data include reaction barrier heights, electronic reaction energies, atomization energies, and enthalpies of formation from the following sources: (1) the DBH24/08 database of 22 reaction barriers (Truhlar et al., J. Chem. Theory Comput., 2007, 3, 569.), (2) the HJO12 set of isogyric reaction energies (Helgaker et al., Modern Electronic Structure Theory, Wiley, Chichester, first ed., 2000.), and (3) the HEAT set of atomization energies and heats of formation (Stanton et al., J. Chem. Phys., 2004, 121, 11599.). Two types of analyses were performed, which target the two distinct uses of explicitly correlated CCSD(T) models: as a replacement for the basis-set-extrapolated CCSD(T) in highly accurate composite methods like HEAT and as a distinct model chemistry for standalone applications. Hence, (1) the basis set error of each component of the CCSD(T)F12 contribution to the chemical energy difference in question and (2) the total error of the CCSD(T)F12 model chemistry relative to the benchmark values are analyzed in detail. Two basis set families were utilized in the calculations: the standard aug-cc-p(C)VXZ (X = D, T, Q) basis sets for the conventional correlation methods and the cc-p(C)VXZ-F12 (X = D, T, Q) basis sets of Peterson and co-workers that are specifically designed for explicitly correlated methods. The conclusion is that the performance of the two families for CCSD correlation contributions (which are the only components affected by the explicitly correlated terms in our formulation) are nearly identical with triple- and quadruple-ζ quality basis sets, with some differences at the double-ζ level. Chemical accuracy (~4.18 kJ/mol) for reaction barrier heights, electronic reaction energies, atomization energies, and enthalpies of formation is attained, on average, with the aug-cc-pVDZ, aug-cc-pVTZ, cc- pCVTZ-F12/aug-cc-pCVTZ, and cc-pCVDZ-F12 basis sets, respectively, at the CCSD(T)F12 level of theory. The corresponding mean unsigned errors are 1.72 kJ/ mol, 1.5 kJ/mol, ~ 2 kJ/mol, and 2.17 kJ/mol, and the corresponding maximum unsigned errors are 4.44 kJ/mol, 3.6 kJ/mol, ~ 5 kJ/mol, and 5.75 kJ/mol.

In addition to accurate energy calculations, our studies were extended to the computation of molecular properties with the MP2-F12 method, and its performance was assessed for prediction of the electric dipole and quadrupole moments of the BH, CO, H2O, and HF molecules (J. Zhang and E. F. Valeev, in preparation for submission). First, various MP2- F12 contributions to the electric dipole and quadrupole moments were analyzed. It was found that the unrelaxed one-electron density contribution is much larger than the orbital response contribution in the CABS singles correction, while both contributions are important in the MP2 correlation contribution. In contrast, the majority of the F12 correction originates from orbital response effects. In the calculations, the two basis set families, the aug-cc-pVXZ (X = D, T, Q) and cc-pVXZ-F12 (X = D, T, Q) basis sets, were also employed. The two basis set series show noticeably different performances at the double-ζ level, though the difference is smaller at triple- and quadruple-ζ levels. In general, the F12 calculations with the aug-cc- pVXZ series give better results than those with the cc-pVXZ-F12 family. In addition, the contribution of the coupling from the MP2 and F12 corrections was investigated. Although the computational cost of the F12 calculations can be significantly reduced by neglecting the coupling terms, this does increase the errors in most cases. With the MP2-F12C/aug-cc-pVDZ calculations, dipole moments close to the basis set limits can be obtained; the errors are around 0.001 a.u. For quadrupole moments, the MP2-F12C/aug-cc-pVTZ calculations can accurately approximate the MP2 basis set limits (within 0.001 a.u.).