Detection and Identification of Reaction Intermediates in the Photorearrangement of Pyridazine N‑Oxide: Discrepancies between Experiment and Theory

Photolysis of pyridazine N-oxide (PNO) results in the detection of a complex series of transient phenomena. On the ultrafast (fs) timescale, we could detect the decay of the first singlet excited state of PNO and the formation of a short-lived transient species, which, based on its time-resolved resonance Raman (TR³) spectrum, we assign to oxaziridine 1,2-diaza-7-oxa-bicyclo[4.1.0]­hepta-2,4-diene. On a longer (hundreds of ns) timescale, this species rearranges to a ring-opened diazo compound, which we have also detected by time-resolved infrared and TR³ spectroscopy. In addition, we identify 1-oxa-3,4-diazepine as a long-lived species formed in the photochemistry of PNO. This species is formed via its oxirane isomer, which in turn is likely formed directly from the S₁ state of PNO via a conical intersection. The barrier determined experimentally for the decay of 1,2-diaza-7-oxa-bicyclo[4.1.0]­hepta-2,4-diene (E_a = (7.1 ± 0.5) kcal mol^–1) is far larger than any barrier calculated by any method that includes dynamic electron correlation but very close to the barriers calculated at the RHF or CASSCF levels of theory. Many methods (B3LYP, MP2, and MP4) fail to give a minimum structure for 1,2-diaza-7-oxa-bicyclo[4.1.0]­hepta-2,4-diene, while M06, M06-2X, MP3, CCSD, or CCSD­(T) yield activation energies for its electrocyclic ring opening that are far too small. In addition, we note that several important geometric parameters, both of 1,2-diaza-7-oxa-bicyclo[4.1.0]­hepta-2,4-diene and of the transition state of its ring opening reaction, clearly have reached no convergence, even at the fully optimized CCSD­(T)/cc-pVTZ level of theory. We therefore suggest that the transient species described in this contribution might be excellent test molecules for further development of highly correlated and density functional theory methods.