New benzene dimers: a benchmark theoretical investigation

Abstract

A wide variety of methods have been used to study the retro-dimerization reactions of one cis-fused (1) and four new trans-fused benzene dimers (2–5). These methods, which included both single-reference and multi-reference approaches, were benchmarked against multi-reference Moller-Plesset perturbation theory of the second order (MRMP2) based on CASSCF(12,12) wavefunctions. Each of the single-reference approaches was found to be overestimating the activation barrier for retro-dimerization process. Interestingly, N-electron valence perturbation theory (NEVPT2) energies also diverged for these reactions from MRMP2 values. Canonical MP2 barriers were in better agreement with MRMP2 results than the NEVPT2 numbers, with mean unsigned deviations (MUD) of 5.4 kcal/mol and 6.6 kcal/mol, respectively. The behavior of each of the methods is discussed and compared with the application of the same methods to the standard pericyclic reaction set proposed by Houk et al. (J Phys Chem A 107:11445, 2003). The comparison between NEVPT2 and MRMP2 reaction barriers, when applied to the cycloadditions in the Houk set and the synchronous symmetry-allowed retro-dimerization of the benzene dimer 1, showed that these two methods give almost identical results for these reaction types (MUD of 0.5 kcal/mol).