EXTENDED AB-INITIO AND THEORETICAL THERMODYNAMICS STUDIES OF THE BERGMAN REACTION AND THE ENERGY SPLITTING OF THE SINGLET O-BENZYNES, M-BENZYNES, AND P-BENZYNES

The autoaromatization of (Z)-hex-3-ene-1,5-diyne to the singlet biradi cal p-benzyne has been reinvestigated by state of the art ab initio me thods. Previous CCSD(T)/6-31G(d,p) and CASPT2[0]/ANO[C(5s4p2dlf)/H(3s2 p)] calculations estimated the reaction heat at 298 K to be 8-10 and 3 .9 +/- 3.2 kcal/mol, respectively. Recent NO- and oxygen-dependent tra pping experiments and collision-induced dissociation threshold energy experiments estimate the heat of reaction to be 8.5 +/- 1.0 kcal/mol a t 470 K (corrected to 9.5 +/- 1.0 kcal/mol at 298 K) and 8.4 +/- 3.0 k cal/mol at 298 K, respectively. New theoretical estimates at 298 K pre dict the values at the basis set Limit for the CCSD(T) and CASPT2[gl] methods to be 12.7 +/- 2.0 and 5.4 +/- 2.0 kcal/mol, respectively. The experimentally predicted electronic contribution to the heat of activ ation is 28.6 kcal/mol. This can be compared with 25.5 and 29.8 kcal/m ol from the CASPT2[gl] and the CCSD(T) methods, respectively. The new study has a much larger one-particle basis set for the CCSD(T) method as compared to earlier studies. For the CASPT2 investigation the bette r suited CASPT2[gl] approximation is utilized. The original CASPT2 met hod, CASPT2[0], systematically favors open-shell systems relative to c losed-shell systems. This was previously corrected empirically. The cu rrent study shows that the energy difference between CCSD(T) and CASPT 2[gl] at the basis set limit is estimated to be 7 +/- 2 kcal/mol. The study also demonstrates that the estimated heat of reaction is very se nsitive to the quality of the basis set. In particular CCSD(T)/6-31G(d ,p) approach underestimates the basis set limit of the enthalpy by app roximately 5 kcal/mol. Furthermore, the relative energies of the p-, m -, and o-benzynes are computed at the CASPT2[gl] and CCSD(T) levels of theory. These results help to explain the discrepancy between the two methods in the case of the Bergman reaction. The deficiency of the CA SPT2 method is mainly attributed to the approximate way in which the d ynamic correlation is included by perturbation theory. A similar sized error is attributed to the CCSD(T) method due to the approximate way in which near degeneracy effects are included. This combined CCSD(T) a nd CASPT2[gl] study indicates that the most recent experimental value of the p-benzyne-o-benzyne energy splitting is overestimated.