The accuracy of geometries and harmonic vibrational frequencies is evaluated for two equation-of-motion ionization potential coupled-cluster methods including CC3 and CCSDT-3 triples corrections. The first two states and first state of the , CN, and BO diatomic radicals are studied. The calculations show a tendency for the CC3 variant to overestimate the bond lengths and to underestimate the vibrational frequencies, while the CCSDT-3 variant seems to be more reliable. It is also demonstrated that the accuracy of such methods is comparable to sophisticated traditional multireference approaches and the full configuration interaction method.
REFERENCES
Specifically, the EOMIP method gives a balanced treatment of all states that are best described (in zeroth order) as those resulting from removal of a single electron from the Slater determinant that describes the nearly closed-shell state. If the state of interest mixes strongly with other, qualitatively different, states such as “” states, EOMIP is not a good choice.
The same method was implemented earlier with a nonproper scaling by Hirata et al. (Ref. 45).
The EOM designation used throughout this paper has its origin in the derivation of excitation energies through the equation of motion of the excitation operator. However, it should be noted that such a derivation applies only to complete methods such as EOM-CCSD, EOM-CCSDT, etc. A more general approach is provided by linear response theory, which applies equally well to approximate methods like CC2, CC3, CCSDT-, etc. Nevertheless, we will use the EOM designation to emphasize that the methods studied here are the logical (and consistent) extensions of EOMIP-CCSD that include iterative treatment of triple excitation effects.
